Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / scsi / fcoe / fcoe.c
diff --git a/kernel/drivers/scsi/fcoe/fcoe.c b/kernel/drivers/scsi/fcoe/fcoe.c
new file mode 100644 (file)
index 0000000..d6b6dde
--- /dev/null
@@ -0,0 +1,2965 @@
+/*
+ * Copyright(c) 2007 - 2009 Intel Corporation. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License along with
+ * this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Maintained at www.Open-FCoE.org
+ */
+
+#include <linux/module.h>
+#include <linux/spinlock.h>
+#include <linux/netdevice.h>
+#include <linux/etherdevice.h>
+#include <linux/ethtool.h>
+#include <linux/if_ether.h>
+#include <linux/if_vlan.h>
+#include <linux/crc32.h>
+#include <linux/slab.h>
+#include <linux/cpu.h>
+#include <linux/fs.h>
+#include <linux/sysfs.h>
+#include <linux/ctype.h>
+#include <linux/workqueue.h>
+#include <net/dcbnl.h>
+#include <net/dcbevent.h>
+#include <scsi/scsi_tcq.h>
+#include <scsi/scsicam.h>
+#include <scsi/scsi_transport.h>
+#include <scsi/scsi_transport_fc.h>
+#include <net/rtnetlink.h>
+
+#include <scsi/fc/fc_encaps.h>
+#include <scsi/fc/fc_fip.h>
+#include <scsi/fc/fc_fcoe.h>
+
+#include <scsi/libfc.h>
+#include <scsi/fc_frame.h>
+#include <scsi/libfcoe.h>
+
+#include "fcoe.h"
+
+MODULE_AUTHOR("Open-FCoE.org");
+MODULE_DESCRIPTION("FCoE");
+MODULE_LICENSE("GPL v2");
+
+/* Performance tuning parameters for fcoe */
+static unsigned int fcoe_ddp_min = 4096;
+module_param_named(ddp_min, fcoe_ddp_min, uint, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(ddp_min, "Minimum I/O size in bytes for "     \
+                "Direct Data Placement (DDP).");
+
+unsigned int fcoe_debug_logging;
+module_param_named(debug_logging, fcoe_debug_logging, int, S_IRUGO|S_IWUSR);
+MODULE_PARM_DESC(debug_logging, "a bit mask of logging levels");
+
+static DEFINE_MUTEX(fcoe_config_mutex);
+
+static struct workqueue_struct *fcoe_wq;
+
+/* fcoe_percpu_clean completion.  Waiter protected by fcoe_create_mutex */
+static DECLARE_COMPLETION(fcoe_flush_completion);
+
+/* fcoe host list */
+/* must only by accessed under the RTNL mutex */
+static LIST_HEAD(fcoe_hostlist);
+static DEFINE_PER_CPU(struct fcoe_percpu_s, fcoe_percpu);
+
+/* Function Prototypes */
+static int fcoe_reset(struct Scsi_Host *);
+static int fcoe_xmit(struct fc_lport *, struct fc_frame *);
+static int fcoe_rcv(struct sk_buff *, struct net_device *,
+                   struct packet_type *, struct net_device *);
+static int fcoe_percpu_receive_thread(void *);
+static void fcoe_percpu_clean(struct fc_lport *);
+static int fcoe_link_ok(struct fc_lport *);
+
+static struct fc_lport *fcoe_hostlist_lookup(const struct net_device *);
+static int fcoe_hostlist_add(const struct fc_lport *);
+static void fcoe_hostlist_del(const struct fc_lport *);
+
+static int fcoe_device_notification(struct notifier_block *, ulong, void *);
+static void fcoe_dev_setup(void);
+static void fcoe_dev_cleanup(void);
+static struct fcoe_interface
+*fcoe_hostlist_lookup_port(const struct net_device *);
+
+static int fcoe_fip_recv(struct sk_buff *, struct net_device *,
+                        struct packet_type *, struct net_device *);
+
+static void fcoe_fip_send(struct fcoe_ctlr *, struct sk_buff *);
+static void fcoe_update_src_mac(struct fc_lport *, u8 *);
+static u8 *fcoe_get_src_mac(struct fc_lport *);
+static void fcoe_destroy_work(struct work_struct *);
+
+static int fcoe_ddp_setup(struct fc_lport *, u16, struct scatterlist *,
+                         unsigned int);
+static int fcoe_ddp_done(struct fc_lport *, u16);
+static int fcoe_ddp_target(struct fc_lport *, u16, struct scatterlist *,
+                          unsigned int);
+static int fcoe_cpu_callback(struct notifier_block *, unsigned long, void *);
+static int fcoe_dcb_app_notification(struct notifier_block *notifier,
+                                    ulong event, void *ptr);
+
+static bool fcoe_match(struct net_device *netdev);
+static int fcoe_create(struct net_device *netdev, enum fip_state fip_mode);
+static int fcoe_destroy(struct net_device *netdev);
+static int fcoe_enable(struct net_device *netdev);
+static int fcoe_disable(struct net_device *netdev);
+
+/* fcoe_syfs control interface handlers */
+static int fcoe_ctlr_alloc(struct net_device *netdev);
+static int fcoe_ctlr_enabled(struct fcoe_ctlr_device *cdev);
+
+
+static struct fc_seq *fcoe_elsct_send(struct fc_lport *,
+                                     u32 did, struct fc_frame *,
+                                     unsigned int op,
+                                     void (*resp)(struct fc_seq *,
+                                                  struct fc_frame *,
+                                                  void *),
+                                     void *, u32 timeout);
+static void fcoe_recv_frame(struct sk_buff *skb);
+
+/* notification function for packets from net device */
+static struct notifier_block fcoe_notifier = {
+       .notifier_call = fcoe_device_notification,
+};
+
+/* notification function for CPU hotplug events */
+static struct notifier_block fcoe_cpu_notifier = {
+       .notifier_call = fcoe_cpu_callback,
+};
+
+/* notification function for DCB events */
+static struct notifier_block dcb_notifier = {
+       .notifier_call = fcoe_dcb_app_notification,
+};
+
+static struct scsi_transport_template *fcoe_nport_scsi_transport;
+static struct scsi_transport_template *fcoe_vport_scsi_transport;
+
+static int fcoe_vport_destroy(struct fc_vport *);
+static int fcoe_vport_create(struct fc_vport *, bool disabled);
+static int fcoe_vport_disable(struct fc_vport *, bool disable);
+static void fcoe_set_vport_symbolic_name(struct fc_vport *);
+static void fcoe_set_port_id(struct fc_lport *, u32, struct fc_frame *);
+static void fcoe_fcf_get_vlan_id(struct fcoe_fcf_device *);
+
+static struct fcoe_sysfs_function_template fcoe_sysfs_templ = {
+       .set_fcoe_ctlr_mode = fcoe_ctlr_set_fip_mode,
+       .set_fcoe_ctlr_enabled = fcoe_ctlr_enabled,
+       .get_fcoe_ctlr_link_fail = fcoe_ctlr_get_lesb,
+       .get_fcoe_ctlr_vlink_fail = fcoe_ctlr_get_lesb,
+       .get_fcoe_ctlr_miss_fka = fcoe_ctlr_get_lesb,
+       .get_fcoe_ctlr_symb_err = fcoe_ctlr_get_lesb,
+       .get_fcoe_ctlr_err_block = fcoe_ctlr_get_lesb,
+       .get_fcoe_ctlr_fcs_error = fcoe_ctlr_get_lesb,
+
+       .get_fcoe_fcf_selected = fcoe_fcf_get_selected,
+       .get_fcoe_fcf_vlan_id = fcoe_fcf_get_vlan_id,
+};
+
+static struct libfc_function_template fcoe_libfc_fcn_templ = {
+       .frame_send = fcoe_xmit,
+       .ddp_setup = fcoe_ddp_setup,
+       .ddp_done = fcoe_ddp_done,
+       .ddp_target = fcoe_ddp_target,
+       .elsct_send = fcoe_elsct_send,
+       .get_lesb = fcoe_get_lesb,
+       .lport_set_port_id = fcoe_set_port_id,
+};
+
+static struct fc_function_template fcoe_nport_fc_functions = {
+       .show_host_node_name = 1,
+       .show_host_port_name = 1,
+       .show_host_supported_classes = 1,
+       .show_host_supported_fc4s = 1,
+       .show_host_active_fc4s = 1,
+       .show_host_maxframe_size = 1,
+       .show_host_serial_number = 1,
+       .show_host_manufacturer = 1,
+       .show_host_model = 1,
+       .show_host_model_description = 1,
+       .show_host_hardware_version = 1,
+       .show_host_driver_version = 1,
+       .show_host_firmware_version = 1,
+       .show_host_optionrom_version = 1,
+
+       .show_host_port_id = 1,
+       .show_host_supported_speeds = 1,
+       .get_host_speed = fc_get_host_speed,
+       .show_host_speed = 1,
+       .show_host_port_type = 1,
+       .get_host_port_state = fc_get_host_port_state,
+       .show_host_port_state = 1,
+       .show_host_symbolic_name = 1,
+
+       .dd_fcrport_size = sizeof(struct fc_rport_libfc_priv),
+       .show_rport_maxframe_size = 1,
+       .show_rport_supported_classes = 1,
+
+       .show_host_fabric_name = 1,
+       .show_starget_node_name = 1,
+       .show_starget_port_name = 1,
+       .show_starget_port_id = 1,
+       .set_rport_dev_loss_tmo = fc_set_rport_loss_tmo,
+       .show_rport_dev_loss_tmo = 1,
+       .get_fc_host_stats = fc_get_host_stats,
+       .issue_fc_host_lip = fcoe_reset,
+
+       .terminate_rport_io = fc_rport_terminate_io,
+
+       .vport_create = fcoe_vport_create,
+       .vport_delete = fcoe_vport_destroy,
+       .vport_disable = fcoe_vport_disable,
+       .set_vport_symbolic_name = fcoe_set_vport_symbolic_name,
+
+       .bsg_request = fc_lport_bsg_request,
+};
+
+static struct fc_function_template fcoe_vport_fc_functions = {
+       .show_host_node_name = 1,
+       .show_host_port_name = 1,
+       .show_host_supported_classes = 1,
+       .show_host_supported_fc4s = 1,
+       .show_host_active_fc4s = 1,
+       .show_host_maxframe_size = 1,
+       .show_host_serial_number = 1,
+       .show_host_manufacturer = 1,
+       .show_host_model = 1,
+       .show_host_model_description = 1,
+       .show_host_hardware_version = 1,
+       .show_host_driver_version = 1,
+       .show_host_firmware_version = 1,
+       .show_host_optionrom_version = 1,
+
+       .show_host_port_id = 1,
+       .show_host_supported_speeds = 1,
+       .get_host_speed = fc_get_host_speed,
+       .show_host_speed = 1,
+       .show_host_port_type = 1,
+       .get_host_port_state = fc_get_host_port_state,
+       .show_host_port_state = 1,
+       .show_host_symbolic_name = 1,
+
+       .dd_fcrport_size = sizeof(struct fc_rport_libfc_priv),
+       .show_rport_maxframe_size = 1,
+       .show_rport_supported_classes = 1,
+
+       .show_host_fabric_name = 1,
+       .show_starget_node_name = 1,
+       .show_starget_port_name = 1,
+       .show_starget_port_id = 1,
+       .set_rport_dev_loss_tmo = fc_set_rport_loss_tmo,
+       .show_rport_dev_loss_tmo = 1,
+       .get_fc_host_stats = fc_get_host_stats,
+       .issue_fc_host_lip = fcoe_reset,
+
+       .terminate_rport_io = fc_rport_terminate_io,
+
+       .bsg_request = fc_lport_bsg_request,
+};
+
+static struct scsi_host_template fcoe_shost_template = {
+       .module = THIS_MODULE,
+       .name = "FCoE Driver",
+       .proc_name = FCOE_NAME,
+       .queuecommand = fc_queuecommand,
+       .eh_abort_handler = fc_eh_abort,
+       .eh_device_reset_handler = fc_eh_device_reset,
+       .eh_host_reset_handler = fc_eh_host_reset,
+       .slave_alloc = fc_slave_alloc,
+       .change_queue_depth = scsi_change_queue_depth,
+       .this_id = -1,
+       .cmd_per_lun = 3,
+       .can_queue = FCOE_MAX_OUTSTANDING_COMMANDS,
+       .use_clustering = ENABLE_CLUSTERING,
+       .sg_tablesize = SG_ALL,
+       .max_sectors = 0xffff,
+       .use_blk_tags = 1,
+       .track_queue_depth = 1,
+};
+
+/**
+ * fcoe_interface_setup() - Setup a FCoE interface
+ * @fcoe:   The new FCoE interface
+ * @netdev: The net device that the fcoe interface is on
+ *
+ * Returns : 0 for success
+ * Locking: must be called with the RTNL mutex held
+ */
+static int fcoe_interface_setup(struct fcoe_interface *fcoe,
+                               struct net_device *netdev)
+{
+       struct fcoe_ctlr *fip = fcoe_to_ctlr(fcoe);
+       struct netdev_hw_addr *ha;
+       struct net_device *real_dev;
+       u8 flogi_maddr[ETH_ALEN];
+       const struct net_device_ops *ops;
+
+       fcoe->netdev = netdev;
+
+       /* Let LLD initialize for FCoE */
+       ops = netdev->netdev_ops;
+       if (ops->ndo_fcoe_enable) {
+               if (ops->ndo_fcoe_enable(netdev))
+                       FCOE_NETDEV_DBG(netdev, "Failed to enable FCoE"
+                                       " specific feature for LLD.\n");
+       }
+
+       /* Do not support for bonding device */
+       if (netdev->priv_flags & IFF_BONDING && netdev->flags & IFF_MASTER) {
+               FCOE_NETDEV_DBG(netdev, "Bonded interfaces not supported\n");
+               return -EOPNOTSUPP;
+       }
+
+       /* look for SAN MAC address, if multiple SAN MACs exist, only
+        * use the first one for SPMA */
+       real_dev = (netdev->priv_flags & IFF_802_1Q_VLAN) ?
+               vlan_dev_real_dev(netdev) : netdev;
+       fcoe->realdev = real_dev;
+       rcu_read_lock();
+       for_each_dev_addr(real_dev, ha) {
+               if ((ha->type == NETDEV_HW_ADDR_T_SAN) &&
+                   (is_valid_ether_addr(ha->addr))) {
+                       memcpy(fip->ctl_src_addr, ha->addr, ETH_ALEN);
+                       fip->spma = 1;
+                       break;
+               }
+       }
+       rcu_read_unlock();
+
+       /* setup Source Mac Address */
+       if (!fip->spma)
+               memcpy(fip->ctl_src_addr, netdev->dev_addr, netdev->addr_len);
+
+       /*
+        * Add FCoE MAC address as second unicast MAC address
+        * or enter promiscuous mode if not capable of listening
+        * for multiple unicast MACs.
+        */
+       memcpy(flogi_maddr, (u8[6]) FC_FCOE_FLOGI_MAC, ETH_ALEN);
+       dev_uc_add(netdev, flogi_maddr);
+       if (fip->spma)
+               dev_uc_add(netdev, fip->ctl_src_addr);
+       if (fip->mode == FIP_MODE_VN2VN) {
+               dev_mc_add(netdev, FIP_ALL_VN2VN_MACS);
+               dev_mc_add(netdev, FIP_ALL_P2P_MACS);
+       } else
+               dev_mc_add(netdev, FIP_ALL_ENODE_MACS);
+
+       /*
+        * setup the receive function from ethernet driver
+        * on the ethertype for the given device
+        */
+       fcoe->fcoe_packet_type.func = fcoe_rcv;
+       fcoe->fcoe_packet_type.type = __constant_htons(ETH_P_FCOE);
+       fcoe->fcoe_packet_type.dev = netdev;
+       dev_add_pack(&fcoe->fcoe_packet_type);
+
+       fcoe->fip_packet_type.func = fcoe_fip_recv;
+       fcoe->fip_packet_type.type = htons(ETH_P_FIP);
+       fcoe->fip_packet_type.dev = netdev;
+       dev_add_pack(&fcoe->fip_packet_type);
+
+       return 0;
+}
+
+/**
+ * fcoe_interface_create() - Create a FCoE interface on a net device
+ * @netdev: The net device to create the FCoE interface on
+ * @fip_mode: The mode to use for FIP
+ *
+ * Returns: pointer to a struct fcoe_interface or NULL on error
+ */
+static struct fcoe_interface *fcoe_interface_create(struct net_device *netdev,
+                                                   enum fip_state fip_mode)
+{
+       struct fcoe_ctlr_device *ctlr_dev;
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       int size;
+       int err;
+
+       if (!try_module_get(THIS_MODULE)) {
+               FCOE_NETDEV_DBG(netdev,
+                               "Could not get a reference to the module\n");
+               fcoe = ERR_PTR(-EBUSY);
+               goto out;
+       }
+
+       size = sizeof(struct fcoe_ctlr) + sizeof(struct fcoe_interface);
+       ctlr_dev = fcoe_ctlr_device_add(&netdev->dev, &fcoe_sysfs_templ,
+                                       size);
+       if (!ctlr_dev) {
+               FCOE_DBG("Failed to add fcoe_ctlr_device\n");
+               fcoe = ERR_PTR(-ENOMEM);
+               goto out_putmod;
+       }
+
+       ctlr = fcoe_ctlr_device_priv(ctlr_dev);
+       ctlr->cdev = ctlr_dev;
+       fcoe = fcoe_ctlr_priv(ctlr);
+
+       dev_hold(netdev);
+
+       /*
+        * Initialize FIP.
+        */
+       fcoe_ctlr_init(ctlr, fip_mode);
+       ctlr->send = fcoe_fip_send;
+       ctlr->update_mac = fcoe_update_src_mac;
+       ctlr->get_src_addr = fcoe_get_src_mac;
+
+       err = fcoe_interface_setup(fcoe, netdev);
+       if (err) {
+               fcoe_ctlr_destroy(ctlr);
+               fcoe_ctlr_device_delete(ctlr_dev);
+               dev_put(netdev);
+               fcoe = ERR_PTR(err);
+               goto out_putmod;
+       }
+
+       goto out;
+
+out_putmod:
+       module_put(THIS_MODULE);
+out:
+       return fcoe;
+}
+
+/**
+ * fcoe_interface_remove() - remove FCoE interface from netdev
+ * @fcoe: The FCoE interface to be cleaned up
+ *
+ * Caller must be holding the RTNL mutex
+ */
+static void fcoe_interface_remove(struct fcoe_interface *fcoe)
+{
+       struct net_device *netdev = fcoe->netdev;
+       struct fcoe_ctlr *fip = fcoe_to_ctlr(fcoe);
+       u8 flogi_maddr[ETH_ALEN];
+       const struct net_device_ops *ops;
+
+       /*
+        * Don't listen for Ethernet packets anymore.
+        * synchronize_net() ensures that the packet handlers are not running
+        * on another CPU. dev_remove_pack() would do that, this calls the
+        * unsyncronized version __dev_remove_pack() to avoid multiple delays.
+        */
+       __dev_remove_pack(&fcoe->fcoe_packet_type);
+       __dev_remove_pack(&fcoe->fip_packet_type);
+       synchronize_net();
+
+       /* Delete secondary MAC addresses */
+       memcpy(flogi_maddr, (u8[6]) FC_FCOE_FLOGI_MAC, ETH_ALEN);
+       dev_uc_del(netdev, flogi_maddr);
+       if (fip->spma)
+               dev_uc_del(netdev, fip->ctl_src_addr);
+       if (fip->mode == FIP_MODE_VN2VN) {
+               dev_mc_del(netdev, FIP_ALL_VN2VN_MACS);
+               dev_mc_del(netdev, FIP_ALL_P2P_MACS);
+       } else
+               dev_mc_del(netdev, FIP_ALL_ENODE_MACS);
+
+       /* Tell the LLD we are done w/ FCoE */
+       ops = netdev->netdev_ops;
+       if (ops->ndo_fcoe_disable) {
+               if (ops->ndo_fcoe_disable(netdev))
+                       FCOE_NETDEV_DBG(netdev, "Failed to disable FCoE"
+                                       " specific feature for LLD.\n");
+       }
+       fcoe->removed = 1;
+}
+
+
+/**
+ * fcoe_interface_cleanup() - Clean up a FCoE interface
+ * @fcoe: The FCoE interface to be cleaned up
+ */
+static void fcoe_interface_cleanup(struct fcoe_interface *fcoe)
+{
+       struct net_device *netdev = fcoe->netdev;
+       struct fcoe_ctlr *fip = fcoe_to_ctlr(fcoe);
+
+       rtnl_lock();
+       if (!fcoe->removed)
+               fcoe_interface_remove(fcoe);
+       rtnl_unlock();
+
+       /* Release the self-reference taken during fcoe_interface_create() */
+       /* tear-down the FCoE controller */
+       fcoe_ctlr_destroy(fip);
+       scsi_host_put(fip->lp->host);
+       dev_put(netdev);
+       module_put(THIS_MODULE);
+}
+
+/**
+ * fcoe_fip_recv() - Handler for received FIP frames
+ * @skb:      The receive skb
+ * @netdev:   The associated net device
+ * @ptype:    The packet_type structure which was used to register this handler
+ * @orig_dev: The original net_device the the skb was received on.
+ *           (in case dev is a bond)
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_fip_recv(struct sk_buff *skb, struct net_device *netdev,
+                        struct packet_type *ptype,
+                        struct net_device *orig_dev)
+{
+       struct fcoe_interface *fcoe;
+       struct fcoe_ctlr *ctlr;
+
+       fcoe = container_of(ptype, struct fcoe_interface, fip_packet_type);
+       ctlr = fcoe_to_ctlr(fcoe);
+       fcoe_ctlr_recv(ctlr, skb);
+       return 0;
+}
+
+/**
+ * fcoe_port_send() - Send an Ethernet-encapsulated FIP/FCoE frame
+ * @port: The FCoE port
+ * @skb: The FIP/FCoE packet to be sent
+ */
+static void fcoe_port_send(struct fcoe_port *port, struct sk_buff *skb)
+{
+       if (port->fcoe_pending_queue.qlen)
+               fcoe_check_wait_queue(port->lport, skb);
+       else if (fcoe_start_io(skb))
+               fcoe_check_wait_queue(port->lport, skb);
+}
+
+/**
+ * fcoe_fip_send() - Send an Ethernet-encapsulated FIP frame
+ * @fip: The FCoE controller
+ * @skb: The FIP packet to be sent
+ */
+static void fcoe_fip_send(struct fcoe_ctlr *fip, struct sk_buff *skb)
+{
+       skb->dev = fcoe_from_ctlr(fip)->netdev;
+       fcoe_port_send(lport_priv(fip->lp), skb);
+}
+
+/**
+ * fcoe_update_src_mac() - Update the Ethernet MAC filters
+ * @lport: The local port to update the source MAC on
+ * @addr:  Unicast MAC address to add
+ *
+ * Remove any previously-set unicast MAC filter.
+ * Add secondary FCoE MAC address filter for our OUI.
+ */
+static void fcoe_update_src_mac(struct fc_lport *lport, u8 *addr)
+{
+       struct fcoe_port *port = lport_priv(lport);
+       struct fcoe_interface *fcoe = port->priv;
+
+       if (!is_zero_ether_addr(port->data_src_addr))
+               dev_uc_del(fcoe->netdev, port->data_src_addr);
+       if (!is_zero_ether_addr(addr))
+               dev_uc_add(fcoe->netdev, addr);
+       memcpy(port->data_src_addr, addr, ETH_ALEN);
+}
+
+/**
+ * fcoe_get_src_mac() - return the Ethernet source address for an lport
+ * @lport: libfc lport
+ */
+static u8 *fcoe_get_src_mac(struct fc_lport *lport)
+{
+       struct fcoe_port *port = lport_priv(lport);
+
+       return port->data_src_addr;
+}
+
+/**
+ * fcoe_lport_config() - Set up a local port
+ * @lport: The local port to be setup
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_lport_config(struct fc_lport *lport)
+{
+       lport->link_up = 0;
+       lport->qfull = 0;
+       lport->max_retry_count = 3;
+       lport->max_rport_retry_count = 3;
+       lport->e_d_tov = 2 * 1000;      /* FC-FS default */
+       lport->r_a_tov = 2 * 2 * 1000;
+       lport->service_params = (FCP_SPPF_INIT_FCN | FCP_SPPF_RD_XRDY_DIS |
+                                FCP_SPPF_RETRY | FCP_SPPF_CONF_COMPL);
+       lport->does_npiv = 1;
+
+       fc_lport_init_stats(lport);
+
+       /* lport fc_lport related configuration */
+       fc_lport_config(lport);
+
+       /* offload related configuration */
+       lport->crc_offload = 0;
+       lport->seq_offload = 0;
+       lport->lro_enabled = 0;
+       lport->lro_xid = 0;
+       lport->lso_max = 0;
+
+       return 0;
+}
+
+/**
+ * fcoe_netdev_features_change - Updates the lport's offload flags based
+ * on the LLD netdev's FCoE feature flags
+ */
+static void fcoe_netdev_features_change(struct fc_lport *lport,
+                                       struct net_device *netdev)
+{
+       mutex_lock(&lport->lp_mutex);
+
+       if (netdev->features & NETIF_F_SG)
+               lport->sg_supp = 1;
+       else
+               lport->sg_supp = 0;
+
+       if (netdev->features & NETIF_F_FCOE_CRC) {
+               lport->crc_offload = 1;
+               FCOE_NETDEV_DBG(netdev, "Supports FCCRC offload\n");
+       } else {
+               lport->crc_offload = 0;
+       }
+
+       if (netdev->features & NETIF_F_FSO) {
+               lport->seq_offload = 1;
+               lport->lso_max = netdev->gso_max_size;
+               FCOE_NETDEV_DBG(netdev, "Supports LSO for max len 0x%x\n",
+                               lport->lso_max);
+       } else {
+               lport->seq_offload = 0;
+               lport->lso_max = 0;
+       }
+
+       if (netdev->fcoe_ddp_xid) {
+               lport->lro_enabled = 1;
+               lport->lro_xid = netdev->fcoe_ddp_xid;
+               FCOE_NETDEV_DBG(netdev, "Supports LRO for max xid 0x%x\n",
+                               lport->lro_xid);
+       } else {
+               lport->lro_enabled = 0;
+               lport->lro_xid = 0;
+       }
+
+       mutex_unlock(&lport->lp_mutex);
+}
+
+/**
+ * fcoe_netdev_config() - Set up net devive for SW FCoE
+ * @lport:  The local port that is associated with the net device
+ * @netdev: The associated net device
+ *
+ * Must be called after fcoe_lport_config() as it will use local port mutex
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_netdev_config(struct fc_lport *lport, struct net_device *netdev)
+{
+       u32 mfs;
+       u64 wwnn, wwpn;
+       struct fcoe_interface *fcoe;
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_port *port;
+
+       /* Setup lport private data to point to fcoe softc */
+       port = lport_priv(lport);
+       fcoe = port->priv;
+       ctlr = fcoe_to_ctlr(fcoe);
+
+       /*
+        * Determine max frame size based on underlying device and optional
+        * user-configured limit.  If the MFS is too low, fcoe_link_ok()
+        * will return 0, so do this first.
+        */
+       mfs = netdev->mtu;
+       if (netdev->features & NETIF_F_FCOE_MTU) {
+               mfs = FCOE_MTU;
+               FCOE_NETDEV_DBG(netdev, "Supports FCOE_MTU of %d bytes\n", mfs);
+       }
+       mfs -= (sizeof(struct fcoe_hdr) + sizeof(struct fcoe_crc_eof));
+       if (fc_set_mfs(lport, mfs))
+               return -EINVAL;
+
+       /* offload features support */
+       fcoe_netdev_features_change(lport, netdev);
+
+       skb_queue_head_init(&port->fcoe_pending_queue);
+       port->fcoe_pending_queue_active = 0;
+       setup_timer(&port->timer, fcoe_queue_timer, (unsigned long)lport);
+
+       fcoe_link_speed_update(lport);
+
+       if (!lport->vport) {
+               if (fcoe_get_wwn(netdev, &wwnn, NETDEV_FCOE_WWNN))
+                       wwnn = fcoe_wwn_from_mac(ctlr->ctl_src_addr, 1, 0);
+               fc_set_wwnn(lport, wwnn);
+               if (fcoe_get_wwn(netdev, &wwpn, NETDEV_FCOE_WWPN))
+                       wwpn = fcoe_wwn_from_mac(ctlr->ctl_src_addr,
+                                                2, 0);
+               fc_set_wwpn(lport, wwpn);
+       }
+
+       return 0;
+}
+
+/**
+ * fcoe_shost_config() - Set up the SCSI host associated with a local port
+ * @lport: The local port
+ * @dev:   The device associated with the SCSI host
+ *
+ * Must be called after fcoe_lport_config() and fcoe_netdev_config()
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_shost_config(struct fc_lport *lport, struct device *dev)
+{
+       int rc = 0;
+
+       /* lport scsi host config */
+       lport->host->max_lun = FCOE_MAX_LUN;
+       lport->host->max_id = FCOE_MAX_FCP_TARGET;
+       lport->host->max_channel = 0;
+       lport->host->max_cmd_len = FCOE_MAX_CMD_LEN;
+
+       if (lport->vport)
+               lport->host->transportt = fcoe_vport_scsi_transport;
+       else
+               lport->host->transportt = fcoe_nport_scsi_transport;
+
+       /* add the new host to the SCSI-ml */
+       rc = scsi_add_host(lport->host, dev);
+       if (rc) {
+               FCOE_NETDEV_DBG(fcoe_netdev(lport), "fcoe_shost_config: "
+                               "error on scsi_add_host\n");
+               return rc;
+       }
+
+       if (!lport->vport)
+               fc_host_max_npiv_vports(lport->host) = USHRT_MAX;
+
+       snprintf(fc_host_symbolic_name(lport->host), FC_SYMBOLIC_NAME_SIZE,
+                "%s v%s over %s", FCOE_NAME, FCOE_VERSION,
+                fcoe_netdev(lport)->name);
+
+       return 0;
+}
+
+
+/**
+ * fcoe_fdmi_info() - Get FDMI related info from net devive for SW FCoE
+ * @lport:  The local port that is associated with the net device
+ * @netdev: The associated net device
+ *
+ * Must be called after fcoe_shost_config() as it will use local port mutex
+ *
+ */
+static void fcoe_fdmi_info(struct fc_lport *lport, struct net_device *netdev)
+{
+       struct fcoe_interface *fcoe;
+       struct fcoe_port *port;
+       struct net_device *realdev;
+       int rc;
+
+       port = lport_priv(lport);
+       fcoe = port->priv;
+       realdev = fcoe->realdev;
+
+       if (!realdev)
+               return;
+
+       /* No FDMI state m/c for NPIV ports */
+       if (lport->vport)
+               return;
+
+       if (realdev->netdev_ops->ndo_fcoe_get_hbainfo) {
+               struct netdev_fcoe_hbainfo *fdmi;
+               fdmi = kzalloc(sizeof(*fdmi), GFP_KERNEL);
+               if (!fdmi)
+                       return;
+
+               rc = realdev->netdev_ops->ndo_fcoe_get_hbainfo(realdev,
+                                                              fdmi);
+               if (rc) {
+                       printk(KERN_INFO "fcoe: Failed to retrieve FDMI "
+                                       "information from netdev.\n");
+                       return;
+               }
+
+               snprintf(fc_host_serial_number(lport->host),
+                        FC_SERIAL_NUMBER_SIZE,
+                        "%s",
+                        fdmi->serial_number);
+               snprintf(fc_host_manufacturer(lport->host),
+                        FC_SERIAL_NUMBER_SIZE,
+                        "%s",
+                        fdmi->manufacturer);
+               snprintf(fc_host_model(lport->host),
+                        FC_SYMBOLIC_NAME_SIZE,
+                        "%s",
+                        fdmi->model);
+               snprintf(fc_host_model_description(lport->host),
+                        FC_SYMBOLIC_NAME_SIZE,
+                        "%s",
+                        fdmi->model_description);
+               snprintf(fc_host_hardware_version(lport->host),
+                        FC_VERSION_STRING_SIZE,
+                        "%s",
+                        fdmi->hardware_version);
+               snprintf(fc_host_driver_version(lport->host),
+                        FC_VERSION_STRING_SIZE,
+                        "%s",
+                        fdmi->driver_version);
+               snprintf(fc_host_optionrom_version(lport->host),
+                        FC_VERSION_STRING_SIZE,
+                        "%s",
+                        fdmi->optionrom_version);
+               snprintf(fc_host_firmware_version(lport->host),
+                        FC_VERSION_STRING_SIZE,
+                        "%s",
+                        fdmi->firmware_version);
+
+               /* Enable FDMI lport states */
+               lport->fdmi_enabled = 1;
+               kfree(fdmi);
+       } else {
+               lport->fdmi_enabled = 0;
+               printk(KERN_INFO "fcoe: No FDMI support.\n");
+       }
+}
+
+/**
+ * fcoe_oem_match() - The match routine for the offloaded exchange manager
+ * @fp: The I/O frame
+ *
+ * This routine will be associated with an exchange manager (EM). When
+ * the libfc exchange handling code is looking for an EM to use it will
+ * call this routine and pass it the frame that it wishes to send. This
+ * routine will return True if the associated EM is to be used and False
+ * if the echange code should continue looking for an EM.
+ *
+ * The offload EM that this routine is associated with will handle any
+ * packets that are for SCSI read requests.
+ *
+ * This has been enhanced to work when FCoE stack is operating in target
+ * mode.
+ *
+ * Returns: True for read types I/O, otherwise returns false.
+ */
+static bool fcoe_oem_match(struct fc_frame *fp)
+{
+       struct fc_frame_header *fh = fc_frame_header_get(fp);
+       struct fcp_cmnd *fcp;
+
+       if (fc_fcp_is_read(fr_fsp(fp)) &&
+           (fr_fsp(fp)->data_len > fcoe_ddp_min))
+               return true;
+       else if ((fr_fsp(fp) == NULL) &&
+                (fh->fh_r_ctl == FC_RCTL_DD_UNSOL_CMD) &&
+                (ntohs(fh->fh_rx_id) == FC_XID_UNKNOWN)) {
+               fcp = fc_frame_payload_get(fp, sizeof(*fcp));
+               if ((fcp->fc_flags & FCP_CFL_WRDATA) &&
+                   (ntohl(fcp->fc_dl) > fcoe_ddp_min))
+                       return true;
+       }
+       return false;
+}
+
+/**
+ * fcoe_em_config() - Allocate and configure an exchange manager
+ * @lport: The local port that the new EM will be associated with
+ *
+ * Returns: 0 on success
+ */
+static inline int fcoe_em_config(struct fc_lport *lport)
+{
+       struct fcoe_port *port = lport_priv(lport);
+       struct fcoe_interface *fcoe = port->priv;
+       struct fcoe_interface *oldfcoe = NULL;
+       struct net_device *old_real_dev, *cur_real_dev;
+       u16 min_xid = FCOE_MIN_XID;
+       u16 max_xid = FCOE_MAX_XID;
+
+       /*
+        * Check if need to allocate an em instance for
+        * offload exchange ids to be shared across all VN_PORTs/lport.
+        */
+       if (!lport->lro_enabled || !lport->lro_xid ||
+           (lport->lro_xid >= max_xid)) {
+               lport->lro_xid = 0;
+               goto skip_oem;
+       }
+
+       /*
+        * Reuse existing offload em instance in case
+        * it is already allocated on real eth device
+        */
+       if (fcoe->netdev->priv_flags & IFF_802_1Q_VLAN)
+               cur_real_dev = vlan_dev_real_dev(fcoe->netdev);
+       else
+               cur_real_dev = fcoe->netdev;
+
+       list_for_each_entry(oldfcoe, &fcoe_hostlist, list) {
+               if (oldfcoe->netdev->priv_flags & IFF_802_1Q_VLAN)
+                       old_real_dev = vlan_dev_real_dev(oldfcoe->netdev);
+               else
+                       old_real_dev = oldfcoe->netdev;
+
+               if (cur_real_dev == old_real_dev) {
+                       fcoe->oem = oldfcoe->oem;
+                       break;
+               }
+       }
+
+       if (fcoe->oem) {
+               if (!fc_exch_mgr_add(lport, fcoe->oem, fcoe_oem_match)) {
+                       printk(KERN_ERR "fcoe_em_config: failed to add "
+                              "offload em:%p on interface:%s\n",
+                              fcoe->oem, fcoe->netdev->name);
+                       return -ENOMEM;
+               }
+       } else {
+               fcoe->oem = fc_exch_mgr_alloc(lport, FC_CLASS_3,
+                                             FCOE_MIN_XID, lport->lro_xid,
+                                             fcoe_oem_match);
+               if (!fcoe->oem) {
+                       printk(KERN_ERR "fcoe_em_config: failed to allocate "
+                              "em for offload exches on interface:%s\n",
+                              fcoe->netdev->name);
+                       return -ENOMEM;
+               }
+       }
+
+       /*
+        * Exclude offload EM xid range from next EM xid range.
+        */
+       min_xid += lport->lro_xid + 1;
+
+skip_oem:
+       if (!fc_exch_mgr_alloc(lport, FC_CLASS_3, min_xid, max_xid, NULL)) {
+               printk(KERN_ERR "fcoe_em_config: failed to "
+                      "allocate em on interface %s\n", fcoe->netdev->name);
+               return -ENOMEM;
+       }
+
+       return 0;
+}
+
+/**
+ * fcoe_if_destroy() - Tear down a SW FCoE instance
+ * @lport: The local port to be destroyed
+ *
+ */
+static void fcoe_if_destroy(struct fc_lport *lport)
+{
+       struct fcoe_port *port = lport_priv(lport);
+       struct fcoe_interface *fcoe = port->priv;
+       struct net_device *netdev = fcoe->netdev;
+
+       FCOE_NETDEV_DBG(netdev, "Destroying interface\n");
+
+       /* Logout of the fabric */
+       fc_fabric_logoff(lport);
+
+       /* Cleanup the fc_lport */
+       fc_lport_destroy(lport);
+
+       /* Stop the transmit retry timer */
+       del_timer_sync(&port->timer);
+
+       /* Free existing transmit skbs */
+       fcoe_clean_pending_queue(lport);
+
+       rtnl_lock();
+       if (!is_zero_ether_addr(port->data_src_addr))
+               dev_uc_del(netdev, port->data_src_addr);
+       if (lport->vport)
+               synchronize_net();
+       else
+               fcoe_interface_remove(fcoe);
+       rtnl_unlock();
+
+       /* Free queued packets for the per-CPU receive threads */
+       fcoe_percpu_clean(lport);
+
+       /* Detach from the scsi-ml */
+       fc_remove_host(lport->host);
+       scsi_remove_host(lport->host);
+
+       /* Destroy lport scsi_priv */
+       fc_fcp_destroy(lport);
+
+       /* There are no more rports or I/O, free the EM */
+       fc_exch_mgr_free(lport);
+
+       /* Free memory used by statistical counters */
+       fc_lport_free_stats(lport);
+
+       /*
+        * Release the Scsi_Host for vport but hold on to
+        * master lport until it fcoe interface fully cleaned-up.
+        */
+       if (lport->vport)
+               scsi_host_put(lport->host);
+}
+
+/**
+ * fcoe_ddp_setup() - Call a LLD's ddp_setup through the net device
+ * @lport: The local port to setup DDP for
+ * @xid:   The exchange ID for this DDP transfer
+ * @sgl:   The scatterlist describing this transfer
+ * @sgc:   The number of sg items
+ *
+ * Returns: 0 if the DDP context was not configured
+ */
+static int fcoe_ddp_setup(struct fc_lport *lport, u16 xid,
+                         struct scatterlist *sgl, unsigned int sgc)
+{
+       struct net_device *netdev = fcoe_netdev(lport);
+
+       if (netdev->netdev_ops->ndo_fcoe_ddp_setup)
+               return netdev->netdev_ops->ndo_fcoe_ddp_setup(netdev,
+                                                             xid, sgl,
+                                                             sgc);
+
+       return 0;
+}
+
+/**
+ * fcoe_ddp_target() - Call a LLD's ddp_target through the net device
+ * @lport: The local port to setup DDP for
+ * @xid:   The exchange ID for this DDP transfer
+ * @sgl:   The scatterlist describing this transfer
+ * @sgc:   The number of sg items
+ *
+ * Returns: 0 if the DDP context was not configured
+ */
+static int fcoe_ddp_target(struct fc_lport *lport, u16 xid,
+                          struct scatterlist *sgl, unsigned int sgc)
+{
+       struct net_device *netdev = fcoe_netdev(lport);
+
+       if (netdev->netdev_ops->ndo_fcoe_ddp_target)
+               return netdev->netdev_ops->ndo_fcoe_ddp_target(netdev, xid,
+                                                              sgl, sgc);
+
+       return 0;
+}
+
+
+/**
+ * fcoe_ddp_done() - Call a LLD's ddp_done through the net device
+ * @lport: The local port to complete DDP on
+ * @xid:   The exchange ID for this DDP transfer
+ *
+ * Returns: the length of data that have been completed by DDP
+ */
+static int fcoe_ddp_done(struct fc_lport *lport, u16 xid)
+{
+       struct net_device *netdev = fcoe_netdev(lport);
+
+       if (netdev->netdev_ops->ndo_fcoe_ddp_done)
+               return netdev->netdev_ops->ndo_fcoe_ddp_done(netdev, xid);
+       return 0;
+}
+
+/**
+ * fcoe_if_create() - Create a FCoE instance on an interface
+ * @fcoe:   The FCoE interface to create a local port on
+ * @parent: The device pointer to be the parent in sysfs for the SCSI host
+ * @npiv:   Indicates if the port is a vport or not
+ *
+ * Creates a fc_lport instance and a Scsi_Host instance and configure them.
+ *
+ * Returns: The allocated fc_lport or an error pointer
+ */
+static struct fc_lport *fcoe_if_create(struct fcoe_interface *fcoe,
+                                      struct device *parent, int npiv)
+{
+       struct fcoe_ctlr *ctlr = fcoe_to_ctlr(fcoe);
+       struct net_device *netdev = fcoe->netdev;
+       struct fc_lport *lport, *n_port;
+       struct fcoe_port *port;
+       struct Scsi_Host *shost;
+       int rc;
+       /*
+        * parent is only a vport if npiv is 1,
+        * but we'll only use vport in that case so go ahead and set it
+        */
+       struct fc_vport *vport = dev_to_vport(parent);
+
+       FCOE_NETDEV_DBG(netdev, "Create Interface\n");
+
+       if (!npiv)
+               lport = libfc_host_alloc(&fcoe_shost_template, sizeof(*port));
+       else
+               lport = libfc_vport_create(vport, sizeof(*port));
+
+       if (!lport) {
+               FCOE_NETDEV_DBG(netdev, "Could not allocate host structure\n");
+               rc = -ENOMEM;
+               goto out;
+       }
+       port = lport_priv(lport);
+       port->lport = lport;
+       port->priv = fcoe;
+       port->get_netdev = fcoe_netdev;
+       port->max_queue_depth = FCOE_MAX_QUEUE_DEPTH;
+       port->min_queue_depth = FCOE_MIN_QUEUE_DEPTH;
+       INIT_WORK(&port->destroy_work, fcoe_destroy_work);
+
+       /*
+        * Need to add the lport to the hostlist
+        * so we catch NETDEV_CHANGE events.
+        */
+       fcoe_hostlist_add(lport);
+
+       /* configure a fc_lport including the exchange manager */
+       rc = fcoe_lport_config(lport);
+       if (rc) {
+               FCOE_NETDEV_DBG(netdev, "Could not configure lport for the "
+                               "interface\n");
+               goto out_host_put;
+       }
+
+       if (npiv) {
+               FCOE_NETDEV_DBG(netdev, "Setting vport names, "
+                               "%16.16llx %16.16llx\n",
+                               vport->node_name, vport->port_name);
+               fc_set_wwnn(lport, vport->node_name);
+               fc_set_wwpn(lport, vport->port_name);
+       }
+
+       /* configure lport network properties */
+       rc = fcoe_netdev_config(lport, netdev);
+       if (rc) {
+               FCOE_NETDEV_DBG(netdev, "Could not configure netdev for the "
+                               "interface\n");
+               goto out_lp_destroy;
+       }
+
+       /* configure lport scsi host properties */
+       rc = fcoe_shost_config(lport, parent);
+       if (rc) {
+               FCOE_NETDEV_DBG(netdev, "Could not configure shost for the "
+                               "interface\n");
+               goto out_lp_destroy;
+       }
+
+       /* Initialize the library */
+       rc = fcoe_libfc_config(lport, ctlr, &fcoe_libfc_fcn_templ, 1);
+       if (rc) {
+               FCOE_NETDEV_DBG(netdev, "Could not configure libfc for the "
+                               "interface\n");
+               goto out_lp_destroy;
+       }
+
+       /* Initialized FDMI information */
+       fcoe_fdmi_info(lport, netdev);
+
+       /*
+        * fcoe_em_alloc() and fcoe_hostlist_add() both
+        * need to be atomic with respect to other changes to the
+        * hostlist since fcoe_em_alloc() looks for an existing EM
+        * instance on host list updated by fcoe_hostlist_add().
+        *
+        * This is currently handled through the fcoe_config_mutex
+        * begin held.
+        */
+       if (!npiv)
+               /* lport exch manager allocation */
+               rc = fcoe_em_config(lport);
+       else {
+               shost = vport_to_shost(vport);
+               n_port = shost_priv(shost);
+               rc = fc_exch_mgr_list_clone(n_port, lport);
+       }
+
+       if (rc) {
+               FCOE_NETDEV_DBG(netdev, "Could not configure the EM\n");
+               goto out_lp_destroy;
+       }
+
+       return lport;
+
+out_lp_destroy:
+       fc_exch_mgr_free(lport);
+out_host_put:
+       fcoe_hostlist_del(lport);
+       scsi_host_put(lport->host);
+out:
+       return ERR_PTR(rc);
+}
+
+/**
+ * fcoe_if_init() - Initialization routine for fcoe.ko
+ *
+ * Attaches the SW FCoE transport to the FC transport
+ *
+ * Returns: 0 on success
+ */
+static int __init fcoe_if_init(void)
+{
+       /* attach to scsi transport */
+       fcoe_nport_scsi_transport =
+               fc_attach_transport(&fcoe_nport_fc_functions);
+       fcoe_vport_scsi_transport =
+               fc_attach_transport(&fcoe_vport_fc_functions);
+
+       if (!fcoe_nport_scsi_transport) {
+               printk(KERN_ERR "fcoe: Failed to attach to the FC transport\n");
+               return -ENODEV;
+       }
+
+       return 0;
+}
+
+/**
+ * fcoe_if_exit() - Tear down fcoe.ko
+ *
+ * Detaches the SW FCoE transport from the FC transport
+ *
+ * Returns: 0 on success
+ */
+static int __exit fcoe_if_exit(void)
+{
+       fc_release_transport(fcoe_nport_scsi_transport);
+       fc_release_transport(fcoe_vport_scsi_transport);
+       fcoe_nport_scsi_transport = NULL;
+       fcoe_vport_scsi_transport = NULL;
+       return 0;
+}
+
+/**
+ * fcoe_percpu_thread_create() - Create a receive thread for an online CPU
+ * @cpu: The CPU index of the CPU to create a receive thread for
+ */
+static void fcoe_percpu_thread_create(unsigned int cpu)
+{
+       struct fcoe_percpu_s *p;
+       struct task_struct *thread;
+
+       p = &per_cpu(fcoe_percpu, cpu);
+
+       thread = kthread_create_on_node(fcoe_percpu_receive_thread,
+                                       (void *)p, cpu_to_node(cpu),
+                                       "fcoethread/%d", cpu);
+
+       if (likely(!IS_ERR(thread))) {
+               kthread_bind(thread, cpu);
+               wake_up_process(thread);
+
+               spin_lock_bh(&p->fcoe_rx_list.lock);
+               p->thread = thread;
+               spin_unlock_bh(&p->fcoe_rx_list.lock);
+       }
+}
+
+/**
+ * fcoe_percpu_thread_destroy() - Remove the receive thread of a CPU
+ * @cpu: The CPU index of the CPU whose receive thread is to be destroyed
+ *
+ * Destroys a per-CPU Rx thread. Any pending skbs are moved to the
+ * current CPU's Rx thread. If the thread being destroyed is bound to
+ * the CPU processing this context the skbs will be freed.
+ */
+static void fcoe_percpu_thread_destroy(unsigned int cpu)
+{
+       struct fcoe_percpu_s *p;
+       struct task_struct *thread;
+       struct page *crc_eof;
+       struct sk_buff *skb;
+#ifdef CONFIG_SMP
+       struct fcoe_percpu_s *p0;
+       unsigned targ_cpu = get_cpu_light();
+#endif /* CONFIG_SMP */
+
+       FCOE_DBG("Destroying receive thread for CPU %d\n", cpu);
+
+       /* Prevent any new skbs from being queued for this CPU. */
+       p = &per_cpu(fcoe_percpu, cpu);
+       spin_lock_bh(&p->fcoe_rx_list.lock);
+       thread = p->thread;
+       p->thread = NULL;
+       crc_eof = p->crc_eof_page;
+       p->crc_eof_page = NULL;
+       p->crc_eof_offset = 0;
+       spin_unlock_bh(&p->fcoe_rx_list.lock);
+
+#ifdef CONFIG_SMP
+       /*
+        * Don't bother moving the skb's if this context is running
+        * on the same CPU that is having its thread destroyed. This
+        * can easily happen when the module is removed.
+        */
+       if (cpu != targ_cpu) {
+               p0 = &per_cpu(fcoe_percpu, targ_cpu);
+               spin_lock_bh(&p0->fcoe_rx_list.lock);
+               if (p0->thread) {
+                       FCOE_DBG("Moving frames from CPU %d to CPU %d\n",
+                                cpu, targ_cpu);
+
+                       while ((skb = __skb_dequeue(&p->fcoe_rx_list)) != NULL)
+                               __skb_queue_tail(&p0->fcoe_rx_list, skb);
+                       spin_unlock_bh(&p0->fcoe_rx_list.lock);
+               } else {
+                       /*
+                        * The targeted CPU is not initialized and cannot accept
+                        * new  skbs. Unlock the targeted CPU and drop the skbs
+                        * on the CPU that is going offline.
+                        */
+                       while ((skb = __skb_dequeue(&p->fcoe_rx_list)) != NULL)
+                               kfree_skb(skb);
+                       spin_unlock_bh(&p0->fcoe_rx_list.lock);
+               }
+       } else {
+               /*
+                * This scenario occurs when the module is being removed
+                * and all threads are being destroyed. skbs will continue
+                * to be shifted from the CPU thread that is being removed
+                * to the CPU thread associated with the CPU that is processing
+                * the module removal. Once there is only one CPU Rx thread it
+                * will reach this case and we will drop all skbs and later
+                * stop the thread.
+                */
+               spin_lock_bh(&p->fcoe_rx_list.lock);
+               while ((skb = __skb_dequeue(&p->fcoe_rx_list)) != NULL)
+                       kfree_skb(skb);
+               spin_unlock_bh(&p->fcoe_rx_list.lock);
+       }
+       put_cpu_light();
+#else
+       /*
+        * This a non-SMP scenario where the singular Rx thread is
+        * being removed. Free all skbs and stop the thread.
+        */
+       spin_lock_bh(&p->fcoe_rx_list.lock);
+       while ((skb = __skb_dequeue(&p->fcoe_rx_list)) != NULL)
+               kfree_skb(skb);
+       spin_unlock_bh(&p->fcoe_rx_list.lock);
+#endif
+
+       if (thread)
+               kthread_stop(thread);
+
+       if (crc_eof)
+               put_page(crc_eof);
+}
+
+/**
+ * fcoe_cpu_callback() - Handler for CPU hotplug events
+ * @nfb:    The callback data block
+ * @action: The event triggering the callback
+ * @hcpu:   The index of the CPU that the event is for
+ *
+ * This creates or destroys per-CPU data for fcoe
+ *
+ * Returns NOTIFY_OK always.
+ */
+static int fcoe_cpu_callback(struct notifier_block *nfb,
+                            unsigned long action, void *hcpu)
+{
+       unsigned cpu = (unsigned long)hcpu;
+
+       switch (action) {
+       case CPU_ONLINE:
+       case CPU_ONLINE_FROZEN:
+               FCOE_DBG("CPU %x online: Create Rx thread\n", cpu);
+               fcoe_percpu_thread_create(cpu);
+               break;
+       case CPU_DEAD:
+       case CPU_DEAD_FROZEN:
+               FCOE_DBG("CPU %x offline: Remove Rx thread\n", cpu);
+               fcoe_percpu_thread_destroy(cpu);
+               break;
+       default:
+               break;
+       }
+       return NOTIFY_OK;
+}
+
+/**
+ * fcoe_select_cpu() - Selects CPU to handle post-processing of incoming
+ *                     command.
+ *
+ * This routine selects next CPU based on cpumask to distribute
+ * incoming requests in round robin.
+ *
+ * Returns: int CPU number
+ */
+static inline unsigned int fcoe_select_cpu(void)
+{
+       static unsigned int selected_cpu;
+
+       selected_cpu = cpumask_next(selected_cpu, cpu_online_mask);
+       if (selected_cpu >= nr_cpu_ids)
+               selected_cpu = cpumask_first(cpu_online_mask);
+
+       return selected_cpu;
+}
+
+/**
+ * fcoe_rcv() - Receive packets from a net device
+ * @skb:    The received packet
+ * @netdev: The net device that the packet was received on
+ * @ptype:  The packet type context
+ * @olddev: The last device net device
+ *
+ * This routine is called by NET_RX_SOFTIRQ. It receives a packet, builds a
+ * FC frame and passes the frame to libfc.
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_rcv(struct sk_buff *skb, struct net_device *netdev,
+            struct packet_type *ptype, struct net_device *olddev)
+{
+       struct fc_lport *lport;
+       struct fcoe_rcv_info *fr;
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       struct fc_frame_header *fh;
+       struct fcoe_percpu_s *fps;
+       struct ethhdr *eh;
+       unsigned int cpu;
+
+       fcoe = container_of(ptype, struct fcoe_interface, fcoe_packet_type);
+       ctlr = fcoe_to_ctlr(fcoe);
+       lport = ctlr->lp;
+       if (unlikely(!lport)) {
+               FCOE_NETDEV_DBG(netdev, "Cannot find hba structure\n");
+               goto err2;
+       }
+       if (!lport->link_up)
+               goto err2;
+
+       FCOE_NETDEV_DBG(netdev,
+                       "skb_info: len:%d data_len:%d head:%p data:%p tail:%p end:%p sum:%d dev:%s\n",
+                       skb->len, skb->data_len, skb->head, skb->data,
+                       skb_tail_pointer(skb), skb_end_pointer(skb),
+                       skb->csum, skb->dev ? skb->dev->name : "<NULL>");
+
+
+       skb = skb_share_check(skb, GFP_ATOMIC);
+
+       if (skb == NULL)
+               return NET_RX_DROP;
+
+       eh = eth_hdr(skb);
+
+       if (is_fip_mode(ctlr) &&
+           !ether_addr_equal(eh->h_source, ctlr->dest_addr)) {
+               FCOE_NETDEV_DBG(netdev, "wrong source mac address:%pM\n",
+                               eh->h_source);
+               goto err;
+       }
+
+       /*
+        * Check for minimum frame length, and make sure required FCoE
+        * and FC headers are pulled into the linear data area.
+        */
+       if (unlikely((skb->len < FCOE_MIN_FRAME) ||
+                    !pskb_may_pull(skb, FCOE_HEADER_LEN)))
+               goto err;
+
+       skb_set_transport_header(skb, sizeof(struct fcoe_hdr));
+       fh = (struct fc_frame_header *) skb_transport_header(skb);
+
+       if (ntoh24(&eh->h_dest[3]) != ntoh24(fh->fh_d_id)) {
+               FCOE_NETDEV_DBG(netdev, "FC frame d_id mismatch with MAC:%pM\n",
+                               eh->h_dest);
+               goto err;
+       }
+
+       fr = fcoe_dev_from_skb(skb);
+       fr->fr_dev = lport;
+
+       /*
+        * In case the incoming frame's exchange is originated from
+        * the initiator, then received frame's exchange id is ANDed
+        * with fc_cpu_mask bits to get the same cpu on which exchange
+        * was originated, otherwise select cpu using rx exchange id
+        * or fcoe_select_cpu().
+        */
+       if (ntoh24(fh->fh_f_ctl) & FC_FC_EX_CTX)
+               cpu = ntohs(fh->fh_ox_id) & fc_cpu_mask;
+       else {
+               if (ntohs(fh->fh_rx_id) == FC_XID_UNKNOWN)
+                       cpu = fcoe_select_cpu();
+               else
+                       cpu = ntohs(fh->fh_rx_id) & fc_cpu_mask;
+       }
+
+       if (cpu >= nr_cpu_ids)
+               goto err;
+
+       fps = &per_cpu(fcoe_percpu, cpu);
+       spin_lock(&fps->fcoe_rx_list.lock);
+       if (unlikely(!fps->thread)) {
+               /*
+                * The targeted CPU is not ready, let's target
+                * the first CPU now. For non-SMP systems this
+                * will check the same CPU twice.
+                */
+               FCOE_NETDEV_DBG(netdev, "CPU is online, but no receive thread "
+                               "ready for incoming skb- using first online "
+                               "CPU.\n");
+
+               spin_unlock(&fps->fcoe_rx_list.lock);
+               cpu = cpumask_first(cpu_online_mask);
+               fps = &per_cpu(fcoe_percpu, cpu);
+               spin_lock(&fps->fcoe_rx_list.lock);
+               if (!fps->thread) {
+                       spin_unlock(&fps->fcoe_rx_list.lock);
+                       goto err;
+               }
+       }
+
+       /*
+        * We now have a valid CPU that we're targeting for
+        * this skb. We also have this receive thread locked,
+        * so we're free to queue skbs into it's queue.
+        */
+
+       /*
+        * Note: We used to have a set of conditions under which we would
+        * call fcoe_recv_frame directly, rather than queuing to the rx list
+        * as it could save a few cycles, but doing so is prohibited, as
+        * fcoe_recv_frame has several paths that may sleep, which is forbidden
+        * in softirq context.
+        */
+       __skb_queue_tail(&fps->fcoe_rx_list, skb);
+       if (fps->thread->state == TASK_INTERRUPTIBLE)
+               wake_up_process(fps->thread);
+       spin_unlock(&fps->fcoe_rx_list.lock);
+
+       return NET_RX_SUCCESS;
+err:
+       per_cpu_ptr(lport->stats, get_cpu())->ErrorFrames++;
+       put_cpu();
+err2:
+       kfree_skb(skb);
+       return NET_RX_DROP;
+}
+
+/**
+ * fcoe_alloc_paged_crc_eof() - Allocate a page to be used for the trailer CRC
+ * @skb:  The packet to be transmitted
+ * @tlen: The total length of the trailer
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_alloc_paged_crc_eof(struct sk_buff *skb, int tlen)
+{
+       struct fcoe_percpu_s *fps;
+       int rc, cpu = get_cpu_light();
+
+       fps = &per_cpu(fcoe_percpu, cpu);
+       rc = fcoe_get_paged_crc_eof(skb, tlen, fps);
+       put_cpu_light();
+
+       return rc;
+}
+
+/**
+ * fcoe_xmit() - Transmit a FCoE frame
+ * @lport: The local port that the frame is to be transmitted for
+ * @fp:           The frame to be transmitted
+ *
+ * Return: 0 for success
+ */
+static int fcoe_xmit(struct fc_lport *lport, struct fc_frame *fp)
+{
+       int wlen;
+       u32 crc;
+       struct ethhdr *eh;
+       struct fcoe_crc_eof *cp;
+       struct sk_buff *skb;
+       struct fc_stats *stats;
+       struct fc_frame_header *fh;
+       unsigned int hlen;              /* header length implies the version */
+       unsigned int tlen;              /* trailer length */
+       unsigned int elen;              /* eth header, may include vlan */
+       struct fcoe_port *port = lport_priv(lport);
+       struct fcoe_interface *fcoe = port->priv;
+       struct fcoe_ctlr *ctlr = fcoe_to_ctlr(fcoe);
+       u8 sof, eof;
+       struct fcoe_hdr *hp;
+
+       WARN_ON((fr_len(fp) % sizeof(u32)) != 0);
+
+       fh = fc_frame_header_get(fp);
+       skb = fp_skb(fp);
+       wlen = skb->len / FCOE_WORD_TO_BYTE;
+
+       if (!lport->link_up) {
+               kfree_skb(skb);
+               return 0;
+       }
+
+       if (unlikely(fh->fh_type == FC_TYPE_ELS) &&
+           fcoe_ctlr_els_send(ctlr, lport, skb))
+               return 0;
+
+       sof = fr_sof(fp);
+       eof = fr_eof(fp);
+
+       elen = sizeof(struct ethhdr);
+       hlen = sizeof(struct fcoe_hdr);
+       tlen = sizeof(struct fcoe_crc_eof);
+       wlen = (skb->len - tlen + sizeof(crc)) / FCOE_WORD_TO_BYTE;
+
+       /* crc offload */
+       if (likely(lport->crc_offload)) {
+               skb->ip_summed = CHECKSUM_UNNECESSARY;
+               skb->csum_start = skb_headroom(skb);
+               skb->csum_offset = skb->len;
+               crc = 0;
+       } else {
+               skb->ip_summed = CHECKSUM_NONE;
+               crc = fcoe_fc_crc(fp);
+       }
+
+       /* copy port crc and eof to the skb buff */
+       if (skb_is_nonlinear(skb)) {
+               skb_frag_t *frag;
+               if (fcoe_alloc_paged_crc_eof(skb, tlen)) {
+                       kfree_skb(skb);
+                       return -ENOMEM;
+               }
+               frag = &skb_shinfo(skb)->frags[skb_shinfo(skb)->nr_frags - 1];
+               cp = kmap_atomic(skb_frag_page(frag))
+                       + frag->page_offset;
+       } else {
+               cp = (struct fcoe_crc_eof *)skb_put(skb, tlen);
+       }
+
+       memset(cp, 0, sizeof(*cp));
+       cp->fcoe_eof = eof;
+       cp->fcoe_crc32 = cpu_to_le32(~crc);
+
+       if (skb_is_nonlinear(skb)) {
+               kunmap_atomic(cp);
+               cp = NULL;
+       }
+
+       /* adjust skb network/transport offsets to match mac/fcoe/port */
+       skb_push(skb, elen + hlen);
+       skb_reset_mac_header(skb);
+       skb_reset_network_header(skb);
+       skb->mac_len = elen;
+       skb->protocol = htons(ETH_P_FCOE);
+       skb->priority = fcoe->priority;
+
+       if (fcoe->netdev->priv_flags & IFF_802_1Q_VLAN &&
+           fcoe->realdev->features & NETIF_F_HW_VLAN_CTAG_TX) {
+               /* must set skb->dev before calling vlan_put_tag */
+               skb->dev = fcoe->realdev;
+               __vlan_hwaccel_put_tag(skb, htons(ETH_P_8021Q),
+                                      vlan_dev_vlan_id(fcoe->netdev));
+       } else
+               skb->dev = fcoe->netdev;
+
+       /* fill up mac and fcoe headers */
+       eh = eth_hdr(skb);
+       eh->h_proto = htons(ETH_P_FCOE);
+       memcpy(eh->h_dest, ctlr->dest_addr, ETH_ALEN);
+       if (ctlr->map_dest)
+               memcpy(eh->h_dest + 3, fh->fh_d_id, 3);
+
+       if (unlikely(ctlr->flogi_oxid != FC_XID_UNKNOWN))
+               memcpy(eh->h_source, ctlr->ctl_src_addr, ETH_ALEN);
+       else
+               memcpy(eh->h_source, port->data_src_addr, ETH_ALEN);
+
+       hp = (struct fcoe_hdr *)(eh + 1);
+       memset(hp, 0, sizeof(*hp));
+       if (FC_FCOE_VER)
+               FC_FCOE_ENCAPS_VER(hp, FC_FCOE_VER);
+       hp->fcoe_sof = sof;
+
+       /* fcoe lso, mss is in max_payload which is non-zero for FCP data */
+       if (lport->seq_offload && fr_max_payload(fp)) {
+               skb_shinfo(skb)->gso_type = SKB_GSO_FCOE;
+               skb_shinfo(skb)->gso_size = fr_max_payload(fp);
+       } else {
+               skb_shinfo(skb)->gso_type = 0;
+               skb_shinfo(skb)->gso_size = 0;
+       }
+       /* update tx stats: regardless if LLD fails */
+       stats = per_cpu_ptr(lport->stats, get_cpu());
+       stats->TxFrames++;
+       stats->TxWords += wlen;
+       put_cpu();
+
+       /* send down to lld */
+       fr_dev(fp) = lport;
+       fcoe_port_send(port, skb);
+       return 0;
+}
+
+/**
+ * fcoe_percpu_flush_done() - Indicate per-CPU queue flush completion
+ * @skb: The completed skb (argument required by destructor)
+ */
+static void fcoe_percpu_flush_done(struct sk_buff *skb)
+{
+       complete(&fcoe_flush_completion);
+}
+
+/**
+ * fcoe_filter_frames() - filter out bad fcoe frames, i.e. bad CRC
+ * @lport: The local port the frame was received on
+ * @fp:           The received frame
+ *
+ * Return: 0 on passing filtering checks
+ */
+static inline int fcoe_filter_frames(struct fc_lport *lport,
+                                    struct fc_frame *fp)
+{
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       struct fc_frame_header *fh;
+       struct sk_buff *skb = (struct sk_buff *)fp;
+       struct fc_stats *stats;
+
+       /*
+        * We only check CRC if no offload is available and if it is
+        * it's solicited data, in which case, the FCP layer would
+        * check it during the copy.
+        */
+       if (lport->crc_offload && skb->ip_summed == CHECKSUM_UNNECESSARY)
+               fr_flags(fp) &= ~FCPHF_CRC_UNCHECKED;
+       else
+               fr_flags(fp) |= FCPHF_CRC_UNCHECKED;
+
+       fh = (struct fc_frame_header *) skb_transport_header(skb);
+       fh = fc_frame_header_get(fp);
+       if (fh->fh_r_ctl == FC_RCTL_DD_SOL_DATA && fh->fh_type == FC_TYPE_FCP)
+               return 0;
+
+       fcoe = ((struct fcoe_port *)lport_priv(lport))->priv;
+       ctlr = fcoe_to_ctlr(fcoe);
+       if (is_fip_mode(ctlr) && fc_frame_payload_op(fp) == ELS_LOGO &&
+           ntoh24(fh->fh_s_id) == FC_FID_FLOGI) {
+               FCOE_DBG("fcoe: dropping FCoE lport LOGO in fip mode\n");
+               return -EINVAL;
+       }
+
+       if (!(fr_flags(fp) & FCPHF_CRC_UNCHECKED) ||
+           le32_to_cpu(fr_crc(fp)) == ~crc32(~0, skb->data, skb->len)) {
+               fr_flags(fp) &= ~FCPHF_CRC_UNCHECKED;
+               return 0;
+       }
+
+       stats = per_cpu_ptr(lport->stats, get_cpu_light());
+       stats->InvalidCRCCount++;
+       if (stats->InvalidCRCCount < 5)
+               printk(KERN_WARNING "fcoe: dropping frame with CRC error\n");
+       put_cpu_light();
+       return -EINVAL;
+}
+
+/**
+ * fcoe_recv_frame() - process a single received frame
+ * @skb: frame to process
+ */
+static void fcoe_recv_frame(struct sk_buff *skb)
+{
+       u32 fr_len;
+       struct fc_lport *lport;
+       struct fcoe_rcv_info *fr;
+       struct fc_stats *stats;
+       struct fcoe_crc_eof crc_eof;
+       struct fc_frame *fp;
+       struct fcoe_port *port;
+       struct fcoe_hdr *hp;
+
+       fr = fcoe_dev_from_skb(skb);
+       lport = fr->fr_dev;
+       if (unlikely(!lport)) {
+               if (skb->destructor != fcoe_percpu_flush_done)
+                       FCOE_NETDEV_DBG(skb->dev, "NULL lport in skb\n");
+               kfree_skb(skb);
+               return;
+       }
+
+       FCOE_NETDEV_DBG(skb->dev,
+                       "skb_info: len:%d data_len:%d head:%p data:%p tail:%p end:%p sum:%d dev:%s\n",
+                       skb->len, skb->data_len,
+                       skb->head, skb->data, skb_tail_pointer(skb),
+                       skb_end_pointer(skb), skb->csum,
+                       skb->dev ? skb->dev->name : "<NULL>");
+
+       port = lport_priv(lport);
+       skb_linearize(skb); /* check for skb_is_nonlinear is within skb_linearize */
+
+       /*
+        * Frame length checks and setting up the header pointers
+        * was done in fcoe_rcv already.
+        */
+       hp = (struct fcoe_hdr *) skb_network_header(skb);
+
+       stats = per_cpu_ptr(lport->stats, get_cpu());
+       if (unlikely(FC_FCOE_DECAPS_VER(hp) != FC_FCOE_VER)) {
+               if (stats->ErrorFrames < 5)
+                       printk(KERN_WARNING "fcoe: FCoE version "
+                              "mismatch: The frame has "
+                              "version %x, but the "
+                              "initiator supports version "
+                              "%x\n", FC_FCOE_DECAPS_VER(hp),
+                              FC_FCOE_VER);
+               goto drop;
+       }
+
+       skb_pull(skb, sizeof(struct fcoe_hdr));
+       fr_len = skb->len - sizeof(struct fcoe_crc_eof);
+
+       stats->RxFrames++;
+       stats->RxWords += fr_len / FCOE_WORD_TO_BYTE;
+
+       fp = (struct fc_frame *)skb;
+       fc_frame_init(fp);
+       fr_dev(fp) = lport;
+       fr_sof(fp) = hp->fcoe_sof;
+
+       /* Copy out the CRC and EOF trailer for access */
+       if (skb_copy_bits(skb, fr_len, &crc_eof, sizeof(crc_eof)))
+               goto drop;
+       fr_eof(fp) = crc_eof.fcoe_eof;
+       fr_crc(fp) = crc_eof.fcoe_crc32;
+       if (pskb_trim(skb, fr_len))
+               goto drop;
+
+       if (!fcoe_filter_frames(lport, fp)) {
+               put_cpu_light();
+               fc_exch_recv(lport, fp);
+               return;
+       }
+drop:
+       stats->ErrorFrames++;
+       put_cpu_light();
+       kfree_skb(skb);
+}
+
+/**
+ * fcoe_percpu_receive_thread() - The per-CPU packet receive thread
+ * @arg: The per-CPU context
+ *
+ * Return: 0 for success
+ */
+static int fcoe_percpu_receive_thread(void *arg)
+{
+       struct fcoe_percpu_s *p = arg;
+       struct sk_buff *skb;
+       struct sk_buff_head tmp;
+
+       skb_queue_head_init(&tmp);
+
+       set_user_nice(current, MIN_NICE);
+
+retry:
+       while (!kthread_should_stop()) {
+
+               spin_lock_bh(&p->fcoe_rx_list.lock);
+               skb_queue_splice_init(&p->fcoe_rx_list, &tmp);
+
+               if (!skb_queue_len(&tmp)) {
+                       set_current_state(TASK_INTERRUPTIBLE);
+                       spin_unlock_bh(&p->fcoe_rx_list.lock);
+                       schedule();
+                       goto retry;
+               }
+
+               spin_unlock_bh(&p->fcoe_rx_list.lock);
+
+               while ((skb = __skb_dequeue(&tmp)) != NULL)
+                       fcoe_recv_frame(skb);
+
+       }
+       return 0;
+}
+
+/**
+ * fcoe_dev_setup() - Setup the link change notification interface
+ */
+static void fcoe_dev_setup(void)
+{
+       register_dcbevent_notifier(&dcb_notifier);
+       register_netdevice_notifier(&fcoe_notifier);
+}
+
+/**
+ * fcoe_dev_cleanup() - Cleanup the link change notification interface
+ */
+static void fcoe_dev_cleanup(void)
+{
+       unregister_dcbevent_notifier(&dcb_notifier);
+       unregister_netdevice_notifier(&fcoe_notifier);
+}
+
+static struct fcoe_interface *
+fcoe_hostlist_lookup_realdev_port(struct net_device *netdev)
+{
+       struct fcoe_interface *fcoe;
+       struct net_device *real_dev;
+
+       list_for_each_entry(fcoe, &fcoe_hostlist, list) {
+               if (fcoe->netdev->priv_flags & IFF_802_1Q_VLAN)
+                       real_dev = vlan_dev_real_dev(fcoe->netdev);
+               else
+                       real_dev = fcoe->netdev;
+
+               if (netdev == real_dev)
+                       return fcoe;
+       }
+       return NULL;
+}
+
+static int fcoe_dcb_app_notification(struct notifier_block *notifier,
+                                    ulong event, void *ptr)
+{
+       struct dcb_app_type *entry = ptr;
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       struct net_device *netdev;
+       int prio;
+
+       if (entry->app.selector != DCB_APP_IDTYPE_ETHTYPE)
+               return NOTIFY_OK;
+
+       netdev = dev_get_by_index(&init_net, entry->ifindex);
+       if (!netdev)
+               return NOTIFY_OK;
+
+       fcoe = fcoe_hostlist_lookup_realdev_port(netdev);
+       dev_put(netdev);
+       if (!fcoe)
+               return NOTIFY_OK;
+
+       ctlr = fcoe_to_ctlr(fcoe);
+
+       if (entry->dcbx & DCB_CAP_DCBX_VER_CEE)
+               prio = ffs(entry->app.priority) - 1;
+       else
+               prio = entry->app.priority;
+
+       if (prio < 0)
+               return NOTIFY_OK;
+
+       if (entry->app.protocol == ETH_P_FIP ||
+           entry->app.protocol == ETH_P_FCOE)
+               ctlr->priority = prio;
+
+       if (entry->app.protocol == ETH_P_FCOE)
+               fcoe->priority = prio;
+
+       return NOTIFY_OK;
+}
+
+/**
+ * fcoe_device_notification() - Handler for net device events
+ * @notifier: The context of the notification
+ * @event:    The type of event
+ * @ptr:      The net device that the event was on
+ *
+ * This function is called by the Ethernet driver in case of link change event.
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_device_notification(struct notifier_block *notifier,
+                                   ulong event, void *ptr)
+{
+       struct fcoe_ctlr_device *cdev;
+       struct fc_lport *lport = NULL;
+       struct net_device *netdev = netdev_notifier_info_to_dev(ptr);
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       struct fcoe_port *port;
+       struct fc_stats *stats;
+       u32 link_possible = 1;
+       u32 mfs;
+       int rc = NOTIFY_OK;
+
+       list_for_each_entry(fcoe, &fcoe_hostlist, list) {
+               if (fcoe->netdev == netdev) {
+                       ctlr = fcoe_to_ctlr(fcoe);
+                       lport = ctlr->lp;
+                       break;
+               }
+       }
+       if (!lport) {
+               rc = NOTIFY_DONE;
+               goto out;
+       }
+
+       switch (event) {
+       case NETDEV_DOWN:
+       case NETDEV_GOING_DOWN:
+               link_possible = 0;
+               break;
+       case NETDEV_UP:
+       case NETDEV_CHANGE:
+               break;
+       case NETDEV_CHANGEMTU:
+               if (netdev->features & NETIF_F_FCOE_MTU)
+                       break;
+               mfs = netdev->mtu - (sizeof(struct fcoe_hdr) +
+                                    sizeof(struct fcoe_crc_eof));
+               if (mfs >= FC_MIN_MAX_FRAME)
+                       fc_set_mfs(lport, mfs);
+               break;
+       case NETDEV_REGISTER:
+               break;
+       case NETDEV_UNREGISTER:
+               list_del(&fcoe->list);
+               port = lport_priv(ctlr->lp);
+               queue_work(fcoe_wq, &port->destroy_work);
+               goto out;
+               break;
+       case NETDEV_FEAT_CHANGE:
+               fcoe_netdev_features_change(lport, netdev);
+               break;
+       default:
+               FCOE_NETDEV_DBG(netdev, "Unknown event %ld "
+                               "from netdev netlink\n", event);
+       }
+
+       fcoe_link_speed_update(lport);
+
+       cdev = fcoe_ctlr_to_ctlr_dev(ctlr);
+
+       if (link_possible && !fcoe_link_ok(lport)) {
+               switch (cdev->enabled) {
+               case FCOE_CTLR_DISABLED:
+                       pr_info("Link up while interface is disabled.\n");
+                       break;
+               case FCOE_CTLR_ENABLED:
+               case FCOE_CTLR_UNUSED:
+                       fcoe_ctlr_link_up(ctlr);
+               };
+       } else if (fcoe_ctlr_link_down(ctlr)) {
+               switch (cdev->enabled) {
+               case FCOE_CTLR_DISABLED:
+                       pr_info("Link down while interface is disabled.\n");
+                       break;
+               case FCOE_CTLR_ENABLED:
+               case FCOE_CTLR_UNUSED:
+                       stats = per_cpu_ptr(lport->stats, get_cpu());
+                       stats->LinkFailureCount++;
+                       put_cpu();
+                       fcoe_clean_pending_queue(lport);
+               };
+       }
+out:
+       return rc;
+}
+
+/**
+ * fcoe_disable() - Disables a FCoE interface
+ * @netdev  : The net_device object the Ethernet interface to create on
+ *
+ * Called from fcoe transport.
+ *
+ * Returns: 0 for success
+ *
+ * Deprecated: use fcoe_ctlr_enabled()
+ */
+static int fcoe_disable(struct net_device *netdev)
+{
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       int rc = 0;
+
+       mutex_lock(&fcoe_config_mutex);
+
+       rtnl_lock();
+       fcoe = fcoe_hostlist_lookup_port(netdev);
+       rtnl_unlock();
+
+       if (fcoe) {
+               ctlr = fcoe_to_ctlr(fcoe);
+               fcoe_ctlr_link_down(ctlr);
+               fcoe_clean_pending_queue(ctlr->lp);
+       } else
+               rc = -ENODEV;
+
+       mutex_unlock(&fcoe_config_mutex);
+       return rc;
+}
+
+/**
+ * fcoe_enable() - Enables a FCoE interface
+ * @netdev  : The net_device object the Ethernet interface to create on
+ *
+ * Called from fcoe transport.
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_enable(struct net_device *netdev)
+{
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       int rc = 0;
+
+       mutex_lock(&fcoe_config_mutex);
+       rtnl_lock();
+       fcoe = fcoe_hostlist_lookup_port(netdev);
+       rtnl_unlock();
+
+       if (!fcoe) {
+               rc = -ENODEV;
+               goto out;
+       }
+
+       ctlr = fcoe_to_ctlr(fcoe);
+
+       if (!fcoe_link_ok(ctlr->lp))
+               fcoe_ctlr_link_up(ctlr);
+
+out:
+       mutex_unlock(&fcoe_config_mutex);
+       return rc;
+}
+
+/**
+ * fcoe_ctlr_enabled() - Enable or disable an FCoE Controller
+ * @cdev: The FCoE Controller that is being enabled or disabled
+ *
+ * fcoe_sysfs will ensure that the state of 'enabled' has
+ * changed, so no checking is necessary here. This routine simply
+ * calls fcoe_enable or fcoe_disable, both of which are deprecated.
+ * When those routines are removed the functionality can be merged
+ * here.
+ */
+static int fcoe_ctlr_enabled(struct fcoe_ctlr_device *cdev)
+{
+       struct fcoe_ctlr *ctlr = fcoe_ctlr_device_priv(cdev);
+       struct fc_lport *lport = ctlr->lp;
+       struct net_device *netdev = fcoe_netdev(lport);
+
+       switch (cdev->enabled) {
+       case FCOE_CTLR_ENABLED:
+               return fcoe_enable(netdev);
+       case FCOE_CTLR_DISABLED:
+               return fcoe_disable(netdev);
+       case FCOE_CTLR_UNUSED:
+       default:
+               return -ENOTSUPP;
+       };
+}
+
+/**
+ * fcoe_destroy() - Destroy a FCoE interface
+ * @netdev  : The net_device object the Ethernet interface to create on
+ *
+ * Called from fcoe transport
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_destroy(struct net_device *netdev)
+{
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       struct fc_lport *lport;
+       struct fcoe_port *port;
+       int rc = 0;
+
+       mutex_lock(&fcoe_config_mutex);
+       rtnl_lock();
+       fcoe = fcoe_hostlist_lookup_port(netdev);
+       if (!fcoe) {
+               rc = -ENODEV;
+               goto out_nodev;
+       }
+       ctlr = fcoe_to_ctlr(fcoe);
+       lport = ctlr->lp;
+       port = lport_priv(lport);
+       list_del(&fcoe->list);
+       queue_work(fcoe_wq, &port->destroy_work);
+out_nodev:
+       rtnl_unlock();
+       mutex_unlock(&fcoe_config_mutex);
+       return rc;
+}
+
+/**
+ * fcoe_destroy_work() - Destroy a FCoE port in a deferred work context
+ * @work: Handle to the FCoE port to be destroyed
+ */
+static void fcoe_destroy_work(struct work_struct *work)
+{
+       struct fcoe_ctlr_device *cdev;
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_port *port;
+       struct fcoe_interface *fcoe;
+       struct Scsi_Host *shost;
+       struct fc_host_attrs *fc_host;
+       unsigned long flags;
+       struct fc_vport *vport;
+       struct fc_vport *next_vport;
+
+       port = container_of(work, struct fcoe_port, destroy_work);
+       shost = port->lport->host;
+       fc_host = shost_to_fc_host(shost);
+
+       /* Loop through all the vports and mark them for deletion */
+       spin_lock_irqsave(shost->host_lock, flags);
+       list_for_each_entry_safe(vport, next_vport, &fc_host->vports, peers) {
+               if (vport->flags & (FC_VPORT_DEL | FC_VPORT_CREATING)) {
+                       continue;
+               } else {
+                       vport->flags |= FC_VPORT_DELETING;
+                       queue_work(fc_host_work_q(shost),
+                                  &vport->vport_delete_work);
+               }
+       }
+       spin_unlock_irqrestore(shost->host_lock, flags);
+
+       flush_workqueue(fc_host_work_q(shost));
+
+       mutex_lock(&fcoe_config_mutex);
+
+       fcoe = port->priv;
+       ctlr = fcoe_to_ctlr(fcoe);
+       cdev = fcoe_ctlr_to_ctlr_dev(ctlr);
+
+       fcoe_if_destroy(port->lport);
+       fcoe_interface_cleanup(fcoe);
+
+       mutex_unlock(&fcoe_config_mutex);
+
+       fcoe_ctlr_device_delete(cdev);
+}
+
+/**
+ * fcoe_match() - Check if the FCoE is supported on the given netdevice
+ * @netdev  : The net_device object the Ethernet interface to create on
+ *
+ * Called from fcoe transport.
+ *
+ * Returns: always returns true as this is the default FCoE transport,
+ * i.e., support all netdevs.
+ */
+static bool fcoe_match(struct net_device *netdev)
+{
+       return true;
+}
+
+/**
+ * fcoe_dcb_create() - Initialize DCB attributes and hooks
+ * @netdev: The net_device object of the L2 link that should be queried
+ * @port: The fcoe_port to bind FCoE APP priority with
+ * @
+ */
+static void fcoe_dcb_create(struct fcoe_interface *fcoe)
+{
+#ifdef CONFIG_DCB
+       int dcbx;
+       u8 fup, up;
+       struct net_device *netdev = fcoe->realdev;
+       struct fcoe_ctlr *ctlr = fcoe_to_ctlr(fcoe);
+       struct dcb_app app = {
+                               .priority = 0,
+                               .protocol = ETH_P_FCOE
+                            };
+
+       /* setup DCB priority attributes. */
+       if (netdev && netdev->dcbnl_ops && netdev->dcbnl_ops->getdcbx) {
+               dcbx = netdev->dcbnl_ops->getdcbx(netdev);
+
+               if (dcbx & DCB_CAP_DCBX_VER_IEEE) {
+                       app.selector = IEEE_8021QAZ_APP_SEL_ETHERTYPE;
+                       up = dcb_ieee_getapp_mask(netdev, &app);
+                       app.protocol = ETH_P_FIP;
+                       fup = dcb_ieee_getapp_mask(netdev, &app);
+               } else {
+                       app.selector = DCB_APP_IDTYPE_ETHTYPE;
+                       up = dcb_getapp(netdev, &app);
+                       app.protocol = ETH_P_FIP;
+                       fup = dcb_getapp(netdev, &app);
+               }
+
+               fcoe->priority = ffs(up) ? ffs(up) - 1 : 0;
+               ctlr->priority = ffs(fup) ? ffs(fup) - 1 : fcoe->priority;
+       }
+#endif
+}
+
+enum fcoe_create_link_state {
+       FCOE_CREATE_LINK_DOWN,
+       FCOE_CREATE_LINK_UP,
+};
+
+/**
+ * _fcoe_create() - (internal) Create a fcoe interface
+ * @netdev  :   The net_device object the Ethernet interface to create on
+ * @fip_mode:   The FIP mode for this creation
+ * @link_state: The ctlr link state on creation
+ *
+ * Called from either the libfcoe 'create' module parameter
+ * via fcoe_create or from fcoe_syfs's ctlr_create file.
+ *
+ * libfcoe's 'create' module parameter is deprecated so some
+ * consolidation of code can be done when that interface is
+ * removed.
+ */
+static int _fcoe_create(struct net_device *netdev, enum fip_state fip_mode,
+                       enum fcoe_create_link_state link_state)
+{
+       int rc = 0;
+       struct fcoe_ctlr_device *ctlr_dev;
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+       struct fc_lport *lport;
+
+       mutex_lock(&fcoe_config_mutex);
+       rtnl_lock();
+
+       /* look for existing lport */
+       if (fcoe_hostlist_lookup(netdev)) {
+               rc = -EEXIST;
+               goto out_nodev;
+       }
+
+       fcoe = fcoe_interface_create(netdev, fip_mode);
+       if (IS_ERR(fcoe)) {
+               rc = PTR_ERR(fcoe);
+               goto out_nodev;
+       }
+
+       ctlr = fcoe_to_ctlr(fcoe);
+       ctlr_dev = fcoe_ctlr_to_ctlr_dev(ctlr);
+       lport = fcoe_if_create(fcoe, &ctlr_dev->dev, 0);
+       if (IS_ERR(lport)) {
+               printk(KERN_ERR "fcoe: Failed to create interface (%s)\n",
+                      netdev->name);
+               rc = -EIO;
+               rtnl_unlock();
+               fcoe_interface_cleanup(fcoe);
+               mutex_unlock(&fcoe_config_mutex);
+               fcoe_ctlr_device_delete(ctlr_dev);
+               goto out;
+       }
+
+       /* Make this the "master" N_Port */
+       ctlr->lp = lport;
+
+       /* setup DCB priority attributes. */
+       fcoe_dcb_create(fcoe);
+
+       /* start FIP Discovery and FLOGI */
+       lport->boot_time = jiffies;
+       fc_fabric_login(lport);
+
+       /*
+        * If the fcoe_ctlr_device is to be set to DISABLED
+        * it must be done after the lport is added to the
+        * hostlist, but before the rtnl_lock is released.
+        * This is because the rtnl_lock protects the
+        * hostlist that fcoe_device_notification uses. If
+        * the FCoE Controller is intended to be created
+        * DISABLED then 'enabled' needs to be considered
+        * handling link events. 'enabled' must be set
+        * before the lport can be found in the hostlist
+        * when a link up event is received.
+        */
+       if (link_state == FCOE_CREATE_LINK_UP)
+               ctlr_dev->enabled = FCOE_CTLR_ENABLED;
+       else
+               ctlr_dev->enabled = FCOE_CTLR_DISABLED;
+
+       if (link_state == FCOE_CREATE_LINK_UP &&
+           !fcoe_link_ok(lport)) {
+               rtnl_unlock();
+               fcoe_ctlr_link_up(ctlr);
+               mutex_unlock(&fcoe_config_mutex);
+               return rc;
+       }
+
+out_nodev:
+       rtnl_unlock();
+       mutex_unlock(&fcoe_config_mutex);
+out:
+       return rc;
+}
+
+/**
+ * fcoe_create() - Create a fcoe interface
+ * @netdev  : The net_device object the Ethernet interface to create on
+ * @fip_mode: The FIP mode for this creation
+ *
+ * Called from fcoe transport
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_create(struct net_device *netdev, enum fip_state fip_mode)
+{
+       return _fcoe_create(netdev, fip_mode, FCOE_CREATE_LINK_UP);
+}
+
+/**
+ * fcoe_ctlr_alloc() - Allocate a fcoe interface from fcoe_sysfs
+ * @netdev: The net_device to be used by the allocated FCoE Controller
+ *
+ * This routine is called from fcoe_sysfs. It will start the fcoe_ctlr
+ * in a link_down state. The allows the user an opportunity to configure
+ * the FCoE Controller from sysfs before enabling the FCoE Controller.
+ *
+ * Creating in with this routine starts the FCoE Controller in Fabric
+ * mode. The user can change to VN2VN or another mode before enabling.
+ */
+static int fcoe_ctlr_alloc(struct net_device *netdev)
+{
+       return _fcoe_create(netdev, FIP_MODE_FABRIC,
+                           FCOE_CREATE_LINK_DOWN);
+}
+
+/**
+ * fcoe_link_ok() - Check if the link is OK for a local port
+ * @lport: The local port to check link on
+ *
+ * Returns: 0 if link is UP and OK, -1 if not
+ *
+ */
+static int fcoe_link_ok(struct fc_lport *lport)
+{
+       struct net_device *netdev = fcoe_netdev(lport);
+
+       if (netif_oper_up(netdev))
+               return 0;
+       return -1;
+}
+
+/**
+ * fcoe_percpu_clean() - Clear all pending skbs for an local port
+ * @lport: The local port whose skbs are to be cleared
+ *
+ * Must be called with fcoe_create_mutex held to single-thread completion.
+ *
+ * This flushes the pending skbs by adding a new skb to each queue and
+ * waiting until they are all freed.  This assures us that not only are
+ * there no packets that will be handled by the lport, but also that any
+ * threads already handling packet have returned.
+ */
+static void fcoe_percpu_clean(struct fc_lport *lport)
+{
+       struct fcoe_percpu_s *pp;
+       struct sk_buff *skb;
+       unsigned int cpu;
+
+       for_each_possible_cpu(cpu) {
+               pp = &per_cpu(fcoe_percpu, cpu);
+
+               if (!pp->thread || !cpu_online(cpu))
+                       continue;
+
+               skb = dev_alloc_skb(0);
+               if (!skb)
+                       continue;
+
+               skb->destructor = fcoe_percpu_flush_done;
+
+               spin_lock_bh(&pp->fcoe_rx_list.lock);
+               __skb_queue_tail(&pp->fcoe_rx_list, skb);
+               if (pp->fcoe_rx_list.qlen == 1)
+                       wake_up_process(pp->thread);
+               spin_unlock_bh(&pp->fcoe_rx_list.lock);
+
+               wait_for_completion(&fcoe_flush_completion);
+       }
+}
+
+/**
+ * fcoe_reset() - Reset a local port
+ * @shost: The SCSI host associated with the local port to be reset
+ *
+ * Returns: Always 0 (return value required by FC transport template)
+ */
+static int fcoe_reset(struct Scsi_Host *shost)
+{
+       struct fc_lport *lport = shost_priv(shost);
+       struct fcoe_port *port = lport_priv(lport);
+       struct fcoe_interface *fcoe = port->priv;
+       struct fcoe_ctlr *ctlr = fcoe_to_ctlr(fcoe);
+       struct fcoe_ctlr_device *cdev = fcoe_ctlr_to_ctlr_dev(ctlr);
+
+       fcoe_ctlr_link_down(ctlr);
+       fcoe_clean_pending_queue(ctlr->lp);
+
+       if (cdev->enabled != FCOE_CTLR_DISABLED &&
+           !fcoe_link_ok(ctlr->lp))
+               fcoe_ctlr_link_up(ctlr);
+       return 0;
+}
+
+/**
+ * fcoe_hostlist_lookup_port() - Find the FCoE interface associated with a net device
+ * @netdev: The net device used as a key
+ *
+ * Locking: Must be called with the RNL mutex held.
+ *
+ * Returns: NULL or the FCoE interface
+ */
+static struct fcoe_interface *
+fcoe_hostlist_lookup_port(const struct net_device *netdev)
+{
+       struct fcoe_interface *fcoe;
+
+       list_for_each_entry(fcoe, &fcoe_hostlist, list) {
+               if (fcoe->netdev == netdev)
+                       return fcoe;
+       }
+       return NULL;
+}
+
+/**
+ * fcoe_hostlist_lookup() - Find the local port associated with a
+ *                         given net device
+ * @netdev: The netdevice used as a key
+ *
+ * Locking: Must be called with the RTNL mutex held
+ *
+ * Returns: NULL or the local port
+ */
+static struct fc_lport *fcoe_hostlist_lookup(const struct net_device *netdev)
+{
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_interface *fcoe;
+
+       fcoe = fcoe_hostlist_lookup_port(netdev);
+       ctlr = fcoe_to_ctlr(fcoe);
+       return (fcoe) ? ctlr->lp : NULL;
+}
+
+/**
+ * fcoe_hostlist_add() - Add the FCoE interface identified by a local
+ *                      port to the hostlist
+ * @lport: The local port that identifies the FCoE interface to be added
+ *
+ * Locking: must be called with the RTNL mutex held
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_hostlist_add(const struct fc_lport *lport)
+{
+       struct fcoe_interface *fcoe;
+       struct fcoe_port *port;
+
+       fcoe = fcoe_hostlist_lookup_port(fcoe_netdev(lport));
+       if (!fcoe) {
+               port = lport_priv(lport);
+               fcoe = port->priv;
+               list_add_tail(&fcoe->list, &fcoe_hostlist);
+       }
+       return 0;
+}
+
+/**
+ * fcoe_hostlist_del() - Remove the FCoE interface identified by a local
+ *                      port to the hostlist
+ * @lport: The local port that identifies the FCoE interface to be added
+ *
+ * Locking: must be called with the RTNL mutex held
+ *
+ */
+static void fcoe_hostlist_del(const struct fc_lport *lport)
+{
+       struct fcoe_interface *fcoe;
+       struct fcoe_port *port;
+
+       port = lport_priv(lport);
+       fcoe = port->priv;
+       list_del(&fcoe->list);
+       return;
+}
+
+static struct fcoe_transport fcoe_sw_transport = {
+       .name = {FCOE_TRANSPORT_DEFAULT},
+       .attached = false,
+       .list = LIST_HEAD_INIT(fcoe_sw_transport.list),
+       .match = fcoe_match,
+       .alloc = fcoe_ctlr_alloc,
+       .create = fcoe_create,
+       .destroy = fcoe_destroy,
+       .enable = fcoe_enable,
+       .disable = fcoe_disable,
+};
+
+/**
+ * fcoe_init() - Initialize fcoe.ko
+ *
+ * Returns: 0 on success, or a negative value on failure
+ */
+static int __init fcoe_init(void)
+{
+       struct fcoe_percpu_s *p;
+       unsigned int cpu;
+       int rc = 0;
+
+       fcoe_wq = alloc_workqueue("fcoe", 0, 0);
+       if (!fcoe_wq)
+               return -ENOMEM;
+
+       /* register as a fcoe transport */
+       rc = fcoe_transport_attach(&fcoe_sw_transport);
+       if (rc) {
+               printk(KERN_ERR "failed to register an fcoe transport, check "
+                       "if libfcoe is loaded\n");
+               return rc;
+       }
+
+       mutex_lock(&fcoe_config_mutex);
+
+       for_each_possible_cpu(cpu) {
+               p = &per_cpu(fcoe_percpu, cpu);
+               skb_queue_head_init(&p->fcoe_rx_list);
+       }
+
+       cpu_notifier_register_begin();
+
+       for_each_online_cpu(cpu)
+               fcoe_percpu_thread_create(cpu);
+
+       /* Initialize per CPU interrupt thread */
+       rc = __register_hotcpu_notifier(&fcoe_cpu_notifier);
+       if (rc)
+               goto out_free;
+
+       cpu_notifier_register_done();
+
+       /* Setup link change notification */
+       fcoe_dev_setup();
+
+       rc = fcoe_if_init();
+       if (rc)
+               goto out_free;
+
+       mutex_unlock(&fcoe_config_mutex);
+       return 0;
+
+out_free:
+       for_each_online_cpu(cpu) {
+               fcoe_percpu_thread_destroy(cpu);
+       }
+
+       cpu_notifier_register_done();
+
+       mutex_unlock(&fcoe_config_mutex);
+       destroy_workqueue(fcoe_wq);
+       return rc;
+}
+module_init(fcoe_init);
+
+/**
+ * fcoe_exit() - Clean up fcoe.ko
+ *
+ * Returns: 0 on success or a  negative value on failure
+ */
+static void __exit fcoe_exit(void)
+{
+       struct fcoe_interface *fcoe, *tmp;
+       struct fcoe_ctlr *ctlr;
+       struct fcoe_port *port;
+       unsigned int cpu;
+
+       mutex_lock(&fcoe_config_mutex);
+
+       fcoe_dev_cleanup();
+
+       /* releases the associated fcoe hosts */
+       rtnl_lock();
+       list_for_each_entry_safe(fcoe, tmp, &fcoe_hostlist, list) {
+               ctlr = fcoe_to_ctlr(fcoe);
+               port = lport_priv(ctlr->lp);
+               fcoe_hostlist_del(port->lport);
+               queue_work(fcoe_wq, &port->destroy_work);
+       }
+       rtnl_unlock();
+
+       cpu_notifier_register_begin();
+
+       for_each_online_cpu(cpu)
+               fcoe_percpu_thread_destroy(cpu);
+
+       __unregister_hotcpu_notifier(&fcoe_cpu_notifier);
+
+       cpu_notifier_register_done();
+
+       mutex_unlock(&fcoe_config_mutex);
+
+       /*
+        * destroy_work's may be chained but destroy_workqueue()
+        * can take care of them. Just kill the fcoe_wq.
+        */
+       destroy_workqueue(fcoe_wq);
+
+       /*
+        * Detaching from the scsi transport must happen after all
+        * destroys are done on the fcoe_wq. destroy_workqueue will
+        * enusre the fcoe_wq is flushed.
+        */
+       fcoe_if_exit();
+
+       /* detach from fcoe transport */
+       fcoe_transport_detach(&fcoe_sw_transport);
+}
+module_exit(fcoe_exit);
+
+/**
+ * fcoe_flogi_resp() - FCoE specific FLOGI and FDISC response handler
+ * @seq: active sequence in the FLOGI or FDISC exchange
+ * @fp: response frame, or error encoded in a pointer (timeout)
+ * @arg: pointer the the fcoe_ctlr structure
+ *
+ * This handles MAC address management for FCoE, then passes control on to
+ * the libfc FLOGI response handler.
+ */
+static void fcoe_flogi_resp(struct fc_seq *seq, struct fc_frame *fp, void *arg)
+{
+       struct fcoe_ctlr *fip = arg;
+       struct fc_exch *exch = fc_seq_exch(seq);
+       struct fc_lport *lport = exch->lp;
+       u8 *mac;
+
+       if (IS_ERR(fp))
+               goto done;
+
+       mac = fr_cb(fp)->granted_mac;
+       /* pre-FIP */
+       if (is_zero_ether_addr(mac))
+               fcoe_ctlr_recv_flogi(fip, lport, fp);
+       if (!is_zero_ether_addr(mac))
+               fcoe_update_src_mac(lport, mac);
+done:
+       fc_lport_flogi_resp(seq, fp, lport);
+}
+
+/**
+ * fcoe_logo_resp() - FCoE specific LOGO response handler
+ * @seq: active sequence in the LOGO exchange
+ * @fp: response frame, or error encoded in a pointer (timeout)
+ * @arg: pointer the the fcoe_ctlr structure
+ *
+ * This handles MAC address management for FCoE, then passes control on to
+ * the libfc LOGO response handler.
+ */
+static void fcoe_logo_resp(struct fc_seq *seq, struct fc_frame *fp, void *arg)
+{
+       struct fc_lport *lport = arg;
+       static u8 zero_mac[ETH_ALEN] = { 0 };
+
+       if (!IS_ERR(fp))
+               fcoe_update_src_mac(lport, zero_mac);
+       fc_lport_logo_resp(seq, fp, lport);
+}
+
+/**
+ * fcoe_elsct_send - FCoE specific ELS handler
+ *
+ * This does special case handling of FIP encapsualted ELS exchanges for FCoE,
+ * using FCoE specific response handlers and passing the FIP controller as
+ * the argument (the lport is still available from the exchange).
+ *
+ * Most of the work here is just handed off to the libfc routine.
+ */
+static struct fc_seq *fcoe_elsct_send(struct fc_lport *lport, u32 did,
+                                     struct fc_frame *fp, unsigned int op,
+                                     void (*resp)(struct fc_seq *,
+                                                  struct fc_frame *,
+                                                  void *),
+                                     void *arg, u32 timeout)
+{
+       struct fcoe_port *port = lport_priv(lport);
+       struct fcoe_interface *fcoe = port->priv;
+       struct fcoe_ctlr *fip = fcoe_to_ctlr(fcoe);
+       struct fc_frame_header *fh = fc_frame_header_get(fp);
+
+       switch (op) {
+       case ELS_FLOGI:
+       case ELS_FDISC:
+               if (lport->point_to_multipoint)
+                       break;
+               return fc_elsct_send(lport, did, fp, op, fcoe_flogi_resp,
+                                    fip, timeout);
+       case ELS_LOGO:
+               /* only hook onto fabric logouts, not port logouts */
+               if (ntoh24(fh->fh_d_id) != FC_FID_FLOGI)
+                       break;
+               return fc_elsct_send(lport, did, fp, op, fcoe_logo_resp,
+                                    lport, timeout);
+       }
+       return fc_elsct_send(lport, did, fp, op, resp, arg, timeout);
+}
+
+/**
+ * fcoe_vport_create() - create an fc_host/scsi_host for a vport
+ * @vport: fc_vport object to create a new fc_host for
+ * @disabled: start the new fc_host in a disabled state by default?
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_vport_create(struct fc_vport *vport, bool disabled)
+{
+       struct Scsi_Host *shost = vport_to_shost(vport);
+       struct fc_lport *n_port = shost_priv(shost);
+       struct fcoe_port *port = lport_priv(n_port);
+       struct fcoe_interface *fcoe = port->priv;
+       struct net_device *netdev = fcoe->netdev;
+       struct fc_lport *vn_port;
+       int rc;
+       char buf[32];
+
+       rc = fcoe_validate_vport_create(vport);
+       if (rc) {
+               fcoe_wwn_to_str(vport->port_name, buf, sizeof(buf));
+               printk(KERN_ERR "fcoe: Failed to create vport, "
+                       "WWPN (0x%s) already exists\n",
+                       buf);
+               return rc;
+       }
+
+       mutex_lock(&fcoe_config_mutex);
+       rtnl_lock();
+       vn_port = fcoe_if_create(fcoe, &vport->dev, 1);
+       rtnl_unlock();
+       mutex_unlock(&fcoe_config_mutex);
+
+       if (IS_ERR(vn_port)) {
+               printk(KERN_ERR "fcoe: fcoe_vport_create(%s) failed\n",
+                      netdev->name);
+               return -EIO;
+       }
+
+       if (disabled) {
+               fc_vport_set_state(vport, FC_VPORT_DISABLED);
+       } else {
+               vn_port->boot_time = jiffies;
+               fc_fabric_login(vn_port);
+               fc_vport_setlink(vn_port);
+       }
+       return 0;
+}
+
+/**
+ * fcoe_vport_destroy() - destroy the fc_host/scsi_host for a vport
+ * @vport: fc_vport object that is being destroyed
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_vport_destroy(struct fc_vport *vport)
+{
+       struct Scsi_Host *shost = vport_to_shost(vport);
+       struct fc_lport *n_port = shost_priv(shost);
+       struct fc_lport *vn_port = vport->dd_data;
+
+       mutex_lock(&n_port->lp_mutex);
+       list_del(&vn_port->list);
+       mutex_unlock(&n_port->lp_mutex);
+
+       mutex_lock(&fcoe_config_mutex);
+       fcoe_if_destroy(vn_port);
+       mutex_unlock(&fcoe_config_mutex);
+
+       return 0;
+}
+
+/**
+ * fcoe_vport_disable() - change vport state
+ * @vport: vport to bring online/offline
+ * @disable: should the vport be disabled?
+ */
+static int fcoe_vport_disable(struct fc_vport *vport, bool disable)
+{
+       struct fc_lport *lport = vport->dd_data;
+
+       if (disable) {
+               fc_vport_set_state(vport, FC_VPORT_DISABLED);
+               fc_fabric_logoff(lport);
+       } else {
+               lport->boot_time = jiffies;
+               fc_fabric_login(lport);
+               fc_vport_setlink(lport);
+       }
+
+       return 0;
+}
+
+/**
+ * fcoe_vport_set_symbolic_name() - append vport string to symbolic name
+ * @vport: fc_vport with a new symbolic name string
+ *
+ * After generating a new symbolic name string, a new RSPN_ID request is
+ * sent to the name server.  There is no response handler, so if it fails
+ * for some reason it will not be retried.
+ */
+static void fcoe_set_vport_symbolic_name(struct fc_vport *vport)
+{
+       struct fc_lport *lport = vport->dd_data;
+       struct fc_frame *fp;
+       size_t len;
+
+       snprintf(fc_host_symbolic_name(lport->host), FC_SYMBOLIC_NAME_SIZE,
+                "%s v%s over %s : %s", FCOE_NAME, FCOE_VERSION,
+                fcoe_netdev(lport)->name, vport->symbolic_name);
+
+       if (lport->state != LPORT_ST_READY)
+               return;
+
+       len = strnlen(fc_host_symbolic_name(lport->host), 255);
+       fp = fc_frame_alloc(lport,
+                           sizeof(struct fc_ct_hdr) +
+                           sizeof(struct fc_ns_rspn) + len);
+       if (!fp)
+               return;
+       lport->tt.elsct_send(lport, FC_FID_DIR_SERV, fp, FC_NS_RSPN_ID,
+                            NULL, NULL, 3 * lport->r_a_tov);
+}
+
+static void fcoe_fcf_get_vlan_id(struct fcoe_fcf_device *fcf_dev)
+{
+       struct fcoe_ctlr_device *ctlr_dev =
+               fcoe_fcf_dev_to_ctlr_dev(fcf_dev);
+       struct fcoe_ctlr *ctlr = fcoe_ctlr_device_priv(ctlr_dev);
+       struct fcoe_interface *fcoe = fcoe_ctlr_priv(ctlr);
+
+       fcf_dev->vlan_id = vlan_dev_vlan_id(fcoe->netdev);
+}
+
+/**
+ * fcoe_set_port_id() - Callback from libfc when Port_ID is set.
+ * @lport: the local port
+ * @port_id: the port ID
+ * @fp: the received frame, if any, that caused the port_id to be set.
+ *
+ * This routine handles the case where we received a FLOGI and are
+ * entering point-to-point mode.  We need to call fcoe_ctlr_recv_flogi()
+ * so it can set the non-mapped mode and gateway address.
+ *
+ * The FLOGI LS_ACC is handled by fcoe_flogi_resp().
+ */
+static void fcoe_set_port_id(struct fc_lport *lport,
+                            u32 port_id, struct fc_frame *fp)
+{
+       struct fcoe_port *port = lport_priv(lport);
+       struct fcoe_interface *fcoe = port->priv;
+       struct fcoe_ctlr *ctlr = fcoe_to_ctlr(fcoe);
+
+       if (fp && fc_frame_payload_op(fp) == ELS_FLOGI)
+               fcoe_ctlr_recv_flogi(ctlr, lport, fp);
+}