[SCSI] fcoe, libfcoe: Add support for FIP. FCoE discovery and keep-alive.

FIP is the new standard way to discover Fibre-Channel Forwarders (FCFs)
by sending solicitations and listening for advertisements from FCFs.

It also provides for keep-alives and period advertisements so that both
parties know they have connectivity.  If the FCF loses connectivity to
the storage fabric, it can send a Link Reset to inform the E_node.

This version is also compatible with pre-FIP implementations, so no
configured selection between FIP mode and non-FIP mode is required.

We wait a couple seconds after sending the initial solicitation
and then send an old-style FLOGI.  If we receive any FIP frames,
we use FIP only mode.  If the old FLOGI receives a response,
we disable FIP mode.  After every reset or link up, this
determination is repeated.

Signed-off-by: Joe Eykholt <jeykholt@cisco.com>
Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Signed-off-by: Robert Love <robert.w.love@intel.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>

4 files changed, 1499 insertions, 137 deletions

diff --git a/drivers/scsi/fcoe/fcoe.c b/drivers/scsi/fcoe/fcoe.c
index d5f009ad038..94e1e318977 100644
--- a/drivers/scsi/fcoe/fcoe.c
+++ b/drivers/scsi/fcoe/fcoe.c
@@ -37,6 +37,7 @@
 #include <net/rtnetlink.h>
 
 #include <scsi/fc/fc_encaps.h>
+#include <scsi/fc/fc_fip.h>
 
 #include <scsi/libfc.h>
 #include <scsi/fc_frame.h>
@@ -71,7 +72,6 @@ static int fcoe_hostlist_add(const struct fc_lport *);
 static int fcoe_hostlist_remove(const struct fc_lport *);
 
 static int fcoe_check_wait_queue(struct fc_lport *);
-static void fcoe_recv_flogi(struct fcoe_softc *, struct fc_frame *, u8 *);
 static int fcoe_device_notification(struct notifier_block *, ulong, void *);
 static void fcoe_dev_setup(void);
 static void fcoe_dev_cleanup(void);
@@ -185,7 +185,7 @@ static int fcoe_netdev_config(struct fc_lport *lp, struct net_device *netdev)
 
 	/* Setup lport private data to point to fcoe softc */
 	fc = lport_priv(lp);
-	fc->lp = lp;
+	fc->ctlr.lp = lp;
 	fc->real_dev = netdev;
 	fc->phys_dev = netdev;
 
@@ -210,9 +210,6 @@ static int fcoe_netdev_config(struct fc_lport *lp, struct net_device *netdev)
 	if (fc_set_mfs(lp, mfs))
 		return -EINVAL;
 
-	if (!fcoe_link_ok(lp))
-		lp->link_up = 1;
-
 	/* offload features support */
 	if (fc->real_dev->features & NETIF_F_SG)
 		lp->sg_supp = 1;
@@ -242,7 +239,7 @@ static int fcoe_netdev_config(struct fc_lport *lp, struct net_device *netdev)
 	fc->fcoe_pending_queue_active = 0;
 
 	/* setup Source Mac Address */
-	memcpy(fc->ctl_src_addr, fc->real_dev->dev_addr,
+	memcpy(fc->ctlr.ctl_src_addr, fc->real_dev->dev_addr,
 	       fc->real_dev->addr_len);
 
 	wwnn = fcoe_wwn_from_mac(fc->real_dev->dev_addr, 1, 0);
@@ -358,6 +355,8 @@ static int fcoe_if_destroy(struct net_device *netdev)
 
 	/* Don't listen for Ethernet packets anymore */
 	dev_remove_pack(&fc->fcoe_packet_type);
+	dev_remove_pack(&fc->fip_packet_type);
+	fcoe_ctlr_destroy(&fc->ctlr);
 
 	/* Cleanup the fc_lport */
 	fc_lport_destroy(lp);
@@ -375,8 +374,10 @@ static int fcoe_if_destroy(struct net_device *netdev)
 	rtnl_lock();
 	memcpy(flogi_maddr, (u8[6]) FC_FCOE_FLOGI_MAC, ETH_ALEN);
 	dev_unicast_delete(fc->real_dev, flogi_maddr, ETH_ALEN);
-	if (compare_ether_addr(fc->data_src_addr, (u8[6]) { 0 }))
-		dev_unicast_delete(fc->real_dev, fc->data_src_addr, ETH_ALEN);
+	if (!is_zero_ether_addr(fc->ctlr.data_src_addr))
+		dev_unicast_delete(fc->real_dev,
+				   fc->ctlr.data_src_addr, ETH_ALEN);
+	dev_mc_delete(fc->real_dev, FIP_ALL_ENODE_MACS, ETH_ALEN, 0);
 	rtnl_unlock();
 
 	/* Free the per-CPU revieve threads */
@@ -438,6 +439,58 @@ static struct libfc_function_template fcoe_libfc_fcn_templ = {
 };
 
 /**
+ * fcoe_fip_recv - handle a received FIP frame.
+ * @skb: the receive skb
+ * @dev: associated &net_device
+ * @ptype: the &packet_type structure which was used to register this handler.
+ * @orig_dev: original receive &net_device, in case @dev is a bond.
+ *
+ * Returns: 0 for success
+ */
+static int fcoe_fip_recv(struct sk_buff *skb, struct net_device *dev,
+			 struct packet_type *ptype,
+			 struct net_device *orig_dev)
+{
+	struct fcoe_softc *fc;
+
+	fc = container_of(ptype, struct fcoe_softc, fip_packet_type);
+	fcoe_ctlr_recv(&fc->ctlr, skb);
+	return 0;
+}
+
+/**
+ * fcoe_fip_send() - send an Ethernet-encapsulated FIP frame.
+ * @fip: FCoE controller.
+ * @skb: FIP Packet.
+ */
+static void fcoe_fip_send(struct fcoe_ctlr *fip, struct sk_buff *skb)
+{
+	skb->dev = fcoe_from_ctlr(fip)->real_dev;
+	dev_queue_xmit(skb);
+}
+
+/**
+ * fcoe_update_src_mac() - Update Ethernet MAC filters.
+ * @fip: FCoE controller.
+ * @old: Unicast MAC address to delete if the MAC is non-zero.
+ * @new: 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 fcoe_ctlr *fip, u8 *old, u8 *new)
+{
+	struct fcoe_softc *fc;
+
+	fc = fcoe_from_ctlr(fip);
+	rtnl_lock();
+	if (!is_zero_ether_addr(old))
+		dev_unicast_delete(fc->real_dev, old, ETH_ALEN);
+	dev_unicast_add(fc->real_dev, new, ETH_ALEN);
+	rtnl_unlock();
+}
+
+/**
  * fcoe_if_create() - this function creates the fcoe interface
  * @netdev: pointer the associated netdevice
  *
@@ -485,6 +538,18 @@ static int fcoe_if_create(struct net_device *netdev)
 		goto out_host_put;
 	}
 
+	/*
+	 * Initialize FIP.
+	 */
+	fcoe_ctlr_init(&fc->ctlr);
+	fc->ctlr.send = fcoe_fip_send;
+	fc->ctlr.update_mac = fcoe_update_src_mac;
+
+	fc->fip_packet_type.func = fcoe_fip_recv;
+	fc->fip_packet_type.type = htons(ETH_P_FIP);
+	fc->fip_packet_type.dev = fc->real_dev;
+	dev_add_pack(&fc->fip_packet_type);
+
 	/* configure lport scsi host properties */
 	rc = fcoe_shost_config(lp, shost, &netdev->dev);
 	if (rc) {
@@ -513,6 +578,9 @@ static int fcoe_if_create(struct net_device *netdev)
 
 	fc_fabric_login(lp);
 
+	if (!fcoe_link_ok(lp))
+		fcoe_ctlr_link_up(&fc->ctlr);
+
 	dev_hold(netdev);
 
 	return rc;
@@ -727,11 +795,13 @@ int fcoe_rcv(struct sk_buff *skb, struct net_device *dev,
 	unsigned int cpu = 0;
 
 	fc = container_of(ptype, struct fcoe_softc, fcoe_packet_type);
-	lp = fc->lp;
+	lp = fc->ctlr.lp;
 	if (unlikely(lp == NULL)) {
 		FC_DBG("cannot find hba structure");
 		goto err2;
 	}
+	if (!lp->link_up)
+		goto err2;
 
 	if (unlikely(debug_fcoe)) {
 		FC_DBG("skb_info: len:%d data_len:%d head:%p data:%p tail:%p "
@@ -929,7 +999,6 @@ int fcoe_xmit(struct fc_lport *lp, struct fc_frame *fp)
 	unsigned int hlen;		/* header length implies the version */
 	unsigned int tlen;		/* trailer length */
 	unsigned int elen;		/* eth header, may include vlan */
-	int flogi_in_progress = 0;
 	struct fcoe_softc *fc;
 	u8 sof, eof;
 	struct fcoe_hdr *hp;
@@ -937,31 +1006,19 @@ int fcoe_xmit(struct fc_lport *lp, struct fc_frame *fp)
 	WARN_ON((fr_len(fp) % sizeof(u32)) != 0);
 
 	fc = lport_priv(lp);
-	/*
-	 * if it is a flogi then we need to learn gw-addr
-	 * and my own fcid
-	 */
 	fh = fc_frame_header_get(fp);
-	if (unlikely(fh->fh_r_ctl == FC_RCTL_ELS_REQ)) {
-		if (fc_frame_payload_op(fp) == ELS_FLOGI) {
-			fc->flogi_oxid = ntohs(fh->fh_ox_id);
-			fc->address_mode = FCOE_FCOUI_ADDR_MODE;
-			fc->flogi_progress = 1;
-			flogi_in_progress = 1;
-		} else if (fc->flogi_progress && ntoh24(fh->fh_s_id) != 0) {
-			/*
-			 * Here we must've gotten an SID by accepting an FLOGI
-			 * from a point-to-point connection.  Switch to using
-			 * the source mac based on the SID.  The destination
-			 * MAC in this case would have been set by receving the
-			 * FLOGI.
-			 */
-			fc_fcoe_set_mac(fc->data_src_addr, fh->fh_s_id);
-			fc->flogi_progress = 0;
-		}
+	skb = fp_skb(fp);
+	wlen = skb->len / FCOE_WORD_TO_BYTE;
+
+	if (!lp->link_up) {
+		kfree(skb);
+		return 0;
 	}
 
-	skb = fp_skb(fp);
+	if (unlikely(fh->fh_r_ctl == FC_RCTL_ELS_REQ) &&
+	    fcoe_ctlr_els_send(&fc->ctlr, skb))
+		return 0;
+
 	sof = fr_sof(fp);
 	eof = fr_eof(fp);
 
@@ -1016,16 +1073,16 @@ int fcoe_xmit(struct fc_lport *lp, struct fc_frame *fp)
 	/* fill up mac and fcoe headers */
 	eh = eth_hdr(skb);
 	eh->h_proto = htons(ETH_P_FCOE);
-	if (fc->address_mode == FCOE_FCOUI_ADDR_MODE)
+	if (fc->ctlr.map_dest)
 		fc_fcoe_set_mac(eh->h_dest, fh->fh_d_id);
 	else
 		/* insert GW address */
-		memcpy(eh->h_dest, fc->dest_addr, ETH_ALEN);
+		memcpy(eh->h_dest, fc->ctlr.dest_addr, ETH_ALEN);
 
-	if (unlikely(flogi_in_progress))
-		memcpy(eh->h_source, fc->ctl_src_addr, ETH_ALEN);
+	if (unlikely(fc->ctlr.flogi_oxid != FC_XID_UNKNOWN))
+		memcpy(eh->h_source, fc->ctlr.ctl_src_addr, ETH_ALEN);
 	else
-		memcpy(eh->h_source, fc->data_src_addr, ETH_ALEN);
+		memcpy(eh->h_source, fc->ctlr.data_src_addr, ETH_ALEN);
 
 	hp = (struct fcoe_hdr *)(eh + 1);
 	memset(hp, 0, sizeof(*hp));
@@ -1125,11 +1182,9 @@ int fcoe_percpu_receive_thread(void *arg)
 		 * Save source MAC address before discarding header.
 		 */
 		fc = lport_priv(lp);
-		if (unlikely(fc->flogi_progress))
-			mac = eth_hdr(skb)->h_source;
-
 		if (skb_is_nonlinear(skb))
 			skb_linearize(skb);	/* not ideal */
+		mac = eth_hdr(skb)->h_source;
 
 		/*
 		 * Frame length checks and setting up the header pointers
@@ -1204,72 +1259,17 @@ int fcoe_percpu_receive_thread(void *arg)
 			}
 			fr_flags(fp) &= ~FCPHF_CRC_UNCHECKED;
 		}
-		/* non flogi and non data exchanges are handled here */
-		if (unlikely(fc->flogi_progress))
-			fcoe_recv_flogi(fc, fp, mac);
+		if (unlikely(fc->ctlr.flogi_oxid != FC_XID_UNKNOWN) &&
+		    fcoe_ctlr_recv_flogi(&fc->ctlr, fp, mac)) {
+			fc_frame_free(fp);
+			continue;
+		}
 		fc_exch_recv(lp, lp->emp, fp);
 	}
 	return 0;
 }
 
 /**
- * fcoe_recv_flogi() - flogi receive function
- * @fc: associated fcoe_softc
- * @fp: the recieved frame
- * @sa: the source address of this flogi
- *
- * This is responsible to parse the flogi response and sets the corresponding
- * mac address for the initiator, eitehr OUI based or GW based.
- *
- * Returns: none
- */
-static void fcoe_recv_flogi(struct fcoe_softc *fc, struct fc_frame *fp, u8 *sa)
-{
-	struct fc_frame_header *fh;
-	u8 op;
-
-	fh = fc_frame_header_get(fp);
-	if (fh->fh_type != FC_TYPE_ELS)
-		return;
-	op = fc_frame_payload_op(fp);
-	if (op == ELS_LS_ACC && fh->fh_r_ctl == FC_RCTL_ELS_REP &&
-	    fc->flogi_oxid == ntohs(fh->fh_ox_id)) {
-		/*
-		 * FLOGI accepted.
-		 * If the src mac addr is FC_OUI-based, then we mark the
-		 * address_mode flag to use FC_OUI-based Ethernet DA.
-		 * Otherwise we use the FCoE gateway addr
-		 */
-		if (!compare_ether_addr(sa, (u8[6]) FC_FCOE_FLOGI_MAC)) {
-			fc->address_mode = FCOE_FCOUI_ADDR_MODE;
-		} else {
-			memcpy(fc->dest_addr, sa, ETH_ALEN);
-			fc->address_mode = FCOE_GW_ADDR_MODE;
-		}
-
-		/*
-		 * Remove any previously-set unicast MAC filter.
-		 * Add secondary FCoE MAC address filter for our OUI.
-		 */
-		rtnl_lock();
-		if (compare_ether_addr(fc->data_src_addr, (u8[6]) { 0 }))
-			dev_unicast_delete(fc->real_dev, fc->data_src_addr,
-					   ETH_ALEN);
-		fc_fcoe_set_mac(fc->data_src_addr, fh->fh_d_id);
-		dev_unicast_add(fc->real_dev, fc->data_src_addr, ETH_ALEN);
-		rtnl_unlock();
-
-		fc->flogi_progress = 0;
-	} else if (op == ELS_FLOGI && fh->fh_r_ctl == FC_RCTL_ELS_REQ && sa) {
-		/*
-		 * Save source MAC for point-to-point responses.
-		 */
-		memcpy(fc->dest_addr, sa, ETH_ALEN);
-		fc->address_mode = FCOE_GW_ADDR_MODE;
-	}
-}
-
-/**
  * fcoe_watchdog() - fcoe timer callback
  * @vp:
  *
@@ -1285,8 +1285,8 @@ void fcoe_watchdog(ulong vp)
 
 	read_lock(&fcoe_hostlist_lock);
 	list_for_each_entry(fc, &fcoe_hostlist, list) {
-		if (fc->lp)
-			fcoe_check_wait_queue(fc->lp);
+		if (fc->ctlr.lp)
+			fcoe_check_wait_queue(fc->ctlr.lp);
 	}
 	read_unlock(&fcoe_hostlist_lock);
 
@@ -1387,14 +1387,14 @@ static int fcoe_device_notification(struct notifier_block *notifier,
 	struct net_device *real_dev = ptr;
 	struct fcoe_softc *fc;
 	struct fcoe_dev_stats *stats;
-	u32 new_link_up;
+	u32 link_possible = 1;
 	u32 mfs;
 	int rc = NOTIFY_OK;
 
 	read_lock(&fcoe_hostlist_lock);
 	list_for_each_entry(fc, &fcoe_hostlist, list) {
 		if (fc->real_dev == real_dev) {
-			lp = fc->lp;
+			lp = fc->ctlr.lp;
 			break;
 		}
 	}
@@ -1404,15 +1404,13 @@ static int fcoe_device_notification(struct notifier_block *notifier,
 		goto out;
 	}
 
-	new_link_up = lp->link_up;
 	switch (event) {
 	case NETDEV_DOWN:
 	case NETDEV_GOING_DOWN:
-		new_link_up = 0;
+		link_possible = 0;
 		break;
 	case NETDEV_UP:
 	case NETDEV_CHANGE:
-		new_link_up = !fcoe_link_ok(lp);
 		break;
 	case NETDEV_CHANGEMTU:
 		mfs = fc->real_dev->mtu -
@@ -1420,22 +1418,18 @@ static int fcoe_device_notification(struct notifier_block *notifier,
 			 sizeof(struct fcoe_crc_eof));
 		if (mfs >= FC_MIN_MAX_FRAME)
 			fc_set_mfs(lp, mfs);
-		new_link_up = !fcoe_link_ok(lp);
 		break;
 	case NETDEV_REGISTER:
 		break;
 	default:
-		FC_DBG("unknown event %ld call", event);
+		FC_DBG("Unknown event %ld from netdev netlink\n", event);
 	}
-	if (lp->link_up != new_link_up) {
-		if (new_link_up)
-			fc_linkup(lp);
-		else {
-			stats = fc_lport_get_stats(lp);
-			stats->LinkFailureCount++;
-			fc_linkdown(lp);
-			fcoe_clean_pending_queue(lp);
-		}
+	if (link_possible && !fcoe_link_ok(lp))
+		fcoe_ctlr_link_up(&fc->ctlr);
+	else if (fcoe_ctlr_link_down(&fc->ctlr)) {
+		stats = fc_lport_get_stats(lp);
+		stats->LinkFailureCount++;
+		fcoe_clean_pending_queue(lp);
 	}
 out:
 	return rc;
@@ -1761,7 +1755,7 @@ struct fc_lport *fcoe_hostlist_lookup(const struct net_device *netdev)
 
 	fc = fcoe_hostlist_lookup_softc(netdev);
 
-	return (fc) ? fc->lp : NULL;
+	return (fc) ? fc->ctlr.lp : NULL;
 }
 EXPORT_SYMBOL_GPL(fcoe_hostlist_lookup);
 
diff --git a/drivers/scsi/fcoe/fcoe.h b/drivers/scsi/fcoe/fcoe.h
index 4a9611f5a58..917aae88689 100644
--- a/drivers/scsi/fcoe/fcoe.h
+++ b/drivers/scsi/fcoe/fcoe.h
@@ -26,10 +26,6 @@
 #define FCOE_MAX_QUEUE_DEPTH	256
 #define FCOE_LOW_QUEUE_DEPTH	32
 
-/* destination address mode */
-#define FCOE_GW_ADDR_MODE	0x00
-#define FCOE_FCOUI_ADDR_MODE	0x01
-
 #define FCOE_WORD_TO_BYTE	4
 
 #define FCOE_VERSION	"0.1"
@@ -59,24 +55,17 @@ struct fcoe_percpu_s {
  */
 struct fcoe_softc {
 	struct list_head list;
-	struct fc_lport *lp;
 	struct net_device *real_dev;
 	struct net_device *phys_dev;		/* device with ethtool_ops */
 	struct packet_type  fcoe_packet_type;
+	struct packet_type  fip_packet_type;
 	struct sk_buff_head fcoe_pending_queue;
 	u8	fcoe_pending_queue_active;
-
-	u8 dest_addr[ETH_ALEN];
-	u8 ctl_src_addr[ETH_ALEN];
-	u8 data_src_addr[ETH_ALEN];
-	/*
-	 * fcoe protocol address learning related stuff
-	 */
-	u16 flogi_oxid;
-	u8 flogi_progress;
-	u8 address_mode;
+	struct fcoe_ctlr ctlr;
 };
 
+#define fcoe_from_ctlr(fc) container_of(fc, struct fcoe_softc, ctlr)
+
 static inline struct net_device *fcoe_netdev(
 	const struct fc_lport *lp)
 {
diff --git a/drivers/scsi/fcoe/libfcoe.c b/drivers/scsi/fcoe/libfcoe.c
index dff97fcf883..f410f4abb54 100644
--- a/drivers/scsi/fcoe/libfcoe.c
+++ b/drivers/scsi/fcoe/libfcoe.c
@@ -1,5 +1,6 @@
 /*
- * Copyright(c) 2009 Intel Corporation. All rights reserved.
+ * Copyright (c) 2008-2009 Cisco Systems, Inc.  All rights reserved.
+ * Copyright (c) 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,
@@ -17,15 +18,1261 @@
  * Maintained at www.Open-FCoE.org
  */
 
+#include <linux/types.h>
 #include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/spinlock.h>
+#include <linux/timer.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/netdevice.h>
+#include <linux/errno.h>
+#include <linux/bitops.h>
+#include <net/rtnetlink.h>
+
+#include <scsi/fc/fc_els.h>
+#include <scsi/fc/fc_fs.h>
+#include <scsi/fc/fc_fip.h>
+#include <scsi/fc/fc_encaps.h>
+#include <scsi/fc/fc_fcoe.h>
 
 #include <scsi/libfc.h>
+#include <scsi/libfcoe.h>
 
 MODULE_AUTHOR("Open-FCoE.org");
-MODULE_DESCRIPTION("FCoE");
+MODULE_DESCRIPTION("FIP discovery protocol support for FCoE HBAs");
 MODULE_LICENSE("GPL v2");
 
+#define	FCOE_CTLR_MIN_FKA	500		/* min keep alive (mS) */
+#define	FCOE_CTLR_DEF_FKA	FIP_DEF_FKA	/* default keep alive (mS) */
+
+static void fcoe_ctlr_timeout(unsigned long);
+static void fcoe_ctlr_link_work(struct work_struct *);
+static void fcoe_ctlr_recv_work(struct work_struct *);
+
+static u8 fcoe_all_fcfs[ETH_ALEN] = FIP_ALL_FCF_MACS;
+
+static u32 fcoe_ctlr_debug;	/* 1 for basic, 2 for noisy debug */
+
+#define FIP_DBG_LVL(level, fmt, args...) 				\
+		do {							\
+			if (fcoe_ctlr_debug >= (level))			\
+				FC_DBG(fmt, ##args);			\
+		} while (0)
+
+#define FIP_DBG(fmt, args...)	FIP_DBG_LVL(1, fmt, ##args)
+
+/*
+ * Return non-zero if FCF fcoe_size has been validated.
+ */
+static inline int fcoe_ctlr_mtu_valid(const struct fcoe_fcf *fcf)
+{
+	return (fcf->flags & FIP_FL_SOL) != 0;
+}
+
+/*
+ * Return non-zero if the FCF is usable.
+ */
+static inline int fcoe_ctlr_fcf_usable(struct fcoe_fcf *fcf)
+{
+	u16 flags = FIP_FL_SOL | FIP_FL_AVAIL;
+
+	return (fcf->flags & flags) == flags;
+}
+
+/**
+ * fcoe_ctlr_init() - Initialize the FCoE Controller instance.
+ * @fip:	FCoE controller.
+ */
+void fcoe_ctlr_init(struct fcoe_ctlr *fip)
+{
+	fip->state = FIP_ST_LINK_WAIT;
+	INIT_LIST_HEAD(&fip->fcfs);
+	spin_lock_init(&fip->lock);
+	fip->flogi_oxid = FC_XID_UNKNOWN;
+	setup_timer(&fip->timer, fcoe_ctlr_timeout, (unsigned long)fip);
+	INIT_WORK(&fip->link_work, fcoe_ctlr_link_work);
+	INIT_WORK(&fip->recv_work, fcoe_ctlr_recv_work);
+	skb_queue_head_init(&fip->fip_recv_list);
+}
+EXPORT_SYMBOL(fcoe_ctlr_init);
+
+/**
+ * fcoe_ctlr_reset_fcfs() - Reset and free all FCFs for a controller.
+ * @fip:	FCoE controller.
+ *
+ * Called with &fcoe_ctlr lock held.
+ */
+static void fcoe_ctlr_reset_fcfs(struct fcoe_ctlr *fip)
+{
+	struct fcoe_fcf *fcf;
+	struct fcoe_fcf *next;
+
+	fip->sel_fcf = NULL;
+	list_for_each_entry_safe(fcf, next, &fip->fcfs, list) {
+		list_del(&fcf->list);
+		kfree(fcf);
+	}
+	fip->fcf_count = 0;
+	fip->sel_time = 0;
+}
+
+/**
+ * fcoe_ctrl_destroy() - Disable and tear-down the FCoE controller.
+ * @fip:	FCoE controller.
+ *
+ * This is called by FCoE drivers before freeing the &fcoe_ctlr.
+ *
+ * The receive handler will have been deleted before this to guarantee
+ * that no more recv_work will be scheduled.
+ *
+ * The timer routine will simply return once we set FIP_ST_DISABLED.
+ * This guarantees that no further timeouts or work will be scheduled.
+ */
+void fcoe_ctlr_destroy(struct fcoe_ctlr *fip)
+{
+	flush_work(&fip->recv_work);
+	spin_lock_bh(&fip->lock);
+	fip->state = FIP_ST_DISABLED;
+	fcoe_ctlr_reset_fcfs(fip);
+	spin_unlock_bh(&fip->lock);
+	del_timer_sync(&fip->timer);
+	flush_work(&fip->link_work);
+}
+EXPORT_SYMBOL(fcoe_ctlr_destroy);
+
+/**
+ * fcoe_ctlr_fcoe_size() - Return the maximum FCoE size required for VN_Port.
+ * @fip:	FCoE controller.
+ *
+ * Returns the maximum packet size including the FCoE header and trailer,
+ * but not including any Ethernet or VLAN headers.
+ */
+static inline u32 fcoe_ctlr_fcoe_size(struct fcoe_ctlr *fip)
+{
+	/*
+	 * Determine the max FCoE frame size allowed, including
+	 * FCoE header and trailer.
+	 * Note:  lp->mfs is currently the payload size, not the frame size.
+	 */
+	return fip->lp->mfs + sizeof(struct fc_frame_header) +
+		sizeof(struct fcoe_hdr) + sizeof(struct fcoe_crc_eof);
+}
+
+/**
+ * fcoe_ctlr_solicit() - Send a solicitation.
+ * @fip:	FCoE controller.
+ * @fcf:	Destination FCF.  If NULL, a multicast solicitation is sent.
+ */
+static void fcoe_ctlr_solicit(struct fcoe_ctlr *fip, struct fcoe_fcf *fcf)
+{
+	struct sk_buff *skb;
+	struct fip_sol {
+		struct ethhdr eth;
+		struct fip_header fip;
+		struct {
+			struct fip_mac_desc mac;
+			struct fip_wwn_desc wwnn;
+			struct fip_size_desc size;
+		} __attribute__((packed)) desc;
+	}  __attribute__((packed)) *sol;
+	u32 fcoe_size;
+
+	skb = dev_alloc_skb(sizeof(*sol));
+	if (!skb)
+		return;
+
+	sol = (struct fip_sol *)skb->data;
+
+	memset(sol, 0, sizeof(*sol));
+	memcpy(sol->eth.h_dest, fcf ? fcf->fcf_mac : fcoe_all_fcfs, ETH_ALEN);
+	memcpy(sol->eth.h_source, fip->ctl_src_addr, ETH_ALEN);
+	sol->eth.h_proto = htons(ETH_P_FIP);
+
+	sol->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
+	sol->fip.fip_op = htons(FIP_OP_DISC);
+	sol->fip.fip_subcode = FIP_SC_SOL;
+	sol->fip.fip_dl_len = htons(sizeof(sol->desc) / FIP_BPW);
+	sol->fip.fip_flags = htons(FIP_FL_FPMA);
+
+	sol->desc.mac.fd_desc.fip_dtype = FIP_DT_MAC;
+	sol->desc.mac.fd_desc.fip_dlen = sizeof(sol->desc.mac) / FIP_BPW;
+	memcpy(sol->desc.mac.fd_mac, fip->ctl_src_addr, ETH_ALEN);
+
+	sol->desc.wwnn.fd_desc.fip_dtype = FIP_DT_NAME;
+	sol->desc.wwnn.fd_desc.fip_dlen = sizeof(sol->desc.wwnn) / FIP_BPW;
+	put_unaligned_be64(fip->lp->wwnn, &sol->desc.wwnn.fd_wwn);
+
+	fcoe_size = fcoe_ctlr_fcoe_size(fip);
+	sol->desc.size.fd_desc.fip_dtype = FIP_DT_FCOE_SIZE;
+	sol->desc.size.fd_desc.fip_dlen = sizeof(sol->desc.size) / FIP_BPW;
+	sol->desc.size.fd_size = htons(fcoe_size);
+
+	skb_put(skb, sizeof(*sol));
+	skb->protocol = htons(ETH_P_802_3);
+	skb_reset_mac_header(skb);
+	skb_reset_network_header(skb);
+	fip->send(fip, skb);
+
+	if (!fcf)
+		fip->sol_time = jiffies;
+}
+
+/**
+ * fcoe_ctlr_link_up() - Start FCoE controller.
+ * @fip:	FCoE controller.
+ *
+ * Called from the LLD when the network link is ready.
+ */
+void fcoe_ctlr_link_up(struct fcoe_ctlr *fip)
+{
+	spin_lock_bh(&fip->lock);
+	if (fip->state == FIP_ST_NON_FIP || fip->state == FIP_ST_AUTO) {
+		fip->last_link = 1;
+		fip->link = 1;
+		spin_unlock_bh(&fip->lock);
+		fc_linkup(fip->lp);
+	} else if (fip->state == FIP_ST_LINK_WAIT) {
+		fip->state = FIP_ST_AUTO;
+		fip->last_link = 1;
+		fip->link = 1;
+		spin_unlock_bh(&fip->lock);
+		FIP_DBG("%s", "setting AUTO mode.\n");
+		fc_linkup(fip->lp);
+		fcoe_ctlr_solicit(fip, NULL);
+	} else
+		spin_unlock_bh(&fip->lock);
+}
+EXPORT_SYMBOL(fcoe_ctlr_link_up);
+
+/**
+ * fcoe_ctlr_reset() - Reset FIP.
+ * @fip:	FCoE controller.
+ * @new_state:	FIP state to be entered.
+ *
+ * Returns non-zero if the link was up and now isn't.
+ */
+static int fcoe_ctlr_reset(struct fcoe_ctlr *fip, enum fip_state new_state)
+{
+	struct fc_lport *lp = fip->lp;
+	int link_dropped;
+
+	spin_lock_bh(&fip->lock);
+	fcoe_ctlr_reset_fcfs(fip);
+	del_timer(&fip->timer);
+	fip->state = new_state;
+	fip->ctlr_ka_time = 0;
+	fip->port_ka_time = 0;
+	fip->sol_time = 0;
+	fip->flogi_oxid = FC_XID_UNKNOWN;
+	fip->map_dest = 0;
+	fip->last_link = 0;
+	link_dropped = fip->link;
+	fip->link = 0;
+	spin_unlock_bh(&fip->lock);
+
+	if (link_dropped)
+		fc_linkdown(lp);
+
+	if (new_state == FIP_ST_ENABLED) {
+		fcoe_ctlr_solicit(fip, NULL);
+		fc_linkup(lp);
+		link_dropped = 0;
+	}
+	return link_dropped;
+}
+
+/**
+ * fcoe_ctlr_link_down() - Stop FCoE controller.
+ * @fip:	FCoE controller.
+ *
+ * Returns non-zero if the link was up and now isn't.
+ *
+ * Called from the LLD when the network link is not ready.
+ * There may be multiple calls while the link is down.
+ */
+int fcoe_ctlr_link_down(struct fcoe_ctlr *fip)
+{
+	return fcoe_ctlr_reset(fip, FIP_ST_LINK_WAIT);
+}
+EXPORT_SYMBOL(fcoe_ctlr_link_down);
+
+/**
+ * fcoe_ctlr_send_keep_alive() - Send a keep-alive to the selected FCF.
+ * @fip:	FCoE controller.
+ * @ports:	0 for controller keep-alive, 1 for port keep-alive.
+ * @sa:		source MAC address.
+ *
+ * A controller keep-alive is sent every fka_period (typically 8 seconds).
+ * The source MAC is the native MAC address.
+ *
+ * A port keep-alive is sent every 90 seconds while logged in.
+ * The source MAC is the assigned mapped source address.
+ * The destination is the FCF's F-port.
+ */
+static void fcoe_ctlr_send_keep_alive(struct fcoe_ctlr *fip, int ports, u8 *sa)
+{
+	struct sk_buff *skb;
+	struct fip_kal {
+		struct ethhdr eth;
+		struct fip_header fip;
+		struct fip_mac_desc mac;
+	} __attribute__((packed)) *kal;
+	struct fip_vn_desc *vn;
+	u32 len;
+	struct fc_lport *lp;
+	struct fcoe_fcf *fcf;
+
+	fcf = fip->sel_fcf;
+	lp = fip->lp;
+	if (!fcf || !fc_host_port_id(lp->host))
+		return;
+
+	len = fcoe_ctlr_fcoe_size(fip) + sizeof(struct ethhdr);
+	BUG_ON(len < sizeof(*kal) + sizeof(*vn));
+	skb = dev_alloc_skb(len);
+	if (!skb)
+		return;
+
+	kal = (struct fip_kal *)skb->data;
+	memset(kal, 0, len);
+	memcpy(kal->eth.h_dest, fcf->fcf_mac, ETH_ALEN);
+	memcpy(kal->eth.h_source, sa, ETH_ALEN);
+	kal->eth.h_proto = htons(ETH_P_FIP);
+
+	kal->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
+	kal->fip.fip_op = htons(FIP_OP_CTRL);
+	kal->fip.fip_subcode = FIP_SC_KEEP_ALIVE;
+	kal->fip.fip_dl_len = htons((sizeof(kal->mac) +
+				    ports * sizeof(*vn)) / FIP_BPW);
+	kal->fip.fip_flags = htons(FIP_FL_FPMA);
+
+	kal->mac.fd_desc.fip_dtype = FIP_DT_MAC;
+	kal->mac.fd_desc.fip_dlen = sizeof(kal->mac) / FIP_BPW;
+	memcpy(kal->mac.fd_mac, fip->ctl_src_addr, ETH_ALEN);
+
+	if (ports) {
+		vn = (struct fip_vn_desc *)(kal + 1);
+		vn->fd_desc.fip_dtype = FIP_DT_VN_ID;
+		vn->fd_desc.fip_dlen = sizeof(*vn) / FIP_BPW;
+		memcpy(vn->fd_mac, fip->data_src_addr, ETH_ALEN);
+		hton24(vn->fd_fc_id, fc_host_port_id(lp->host));
+		put_unaligned_be64(lp->wwpn, &vn->fd_wwpn);
+	}
+
+	skb_put(skb, len);
+	skb->protocol = htons(ETH_P_802_3);
+	skb_reset_mac_header(skb);
+	skb_reset_network_header(skb);
+	fip->send(fip, skb);
+}
+
+/**
+ * fcoe_ctlr_encaps() - Encapsulate an ELS frame for FIP, without sending it.
+ * @fip:	FCoE controller.
+ * @dtype:	FIP descriptor type for the frame.
+ * @skb:	FCoE ELS frame including FC header but no FCoE headers.
+ *
+ * Returns non-zero error code on failure.
+ *
+ * The caller must check that the length is a multiple of 4.
+ *
+ * The @skb must have enough headroom (28 bytes) and tailroom (8 bytes).
+ * Headroom includes the FIP encapsulation description, FIP header, and
+ * Ethernet header.  The tailroom is for the FIP MAC descriptor.
+ */
+static int fcoe_ctlr_encaps(struct fcoe_ctlr *fip,
+			    u8 dtype, struct sk_buff *skb)
+{
+	struct fip_encaps_head {
+		struct ethhdr eth;
+		struct fip_header fip;
+		struct fip_encaps encaps;
+	} __attribute__((packed)) *cap;
+	struct fip_mac_desc *mac;
+	struct fcoe_fcf *fcf;
+	size_t dlen;
+
+	fcf = fip->sel_fcf;
+	if (!fcf)
+		return -ENODEV;
+	dlen = sizeof(struct fip_encaps) + skb->len;	/* len before push */
+	cap = (struct fip_encaps_head *)skb_push(skb, sizeof(*cap));
+
+	memset(cap, 0, sizeof(*cap));
+	memcpy(cap->eth.h_dest, fcf->fcf_mac, ETH_ALEN);
+	memcpy(cap->eth.h_source, fip->ctl_src_addr, ETH_ALEN);
+	cap->eth.h_proto = htons(ETH_P_FIP);
+
+	cap->fip.fip_ver = FIP_VER_ENCAPS(FIP_VER);
+	cap->fip.fip_op = htons(FIP_OP_LS);
+	cap->fip.fip_subcode = FIP_SC_REQ;
+	cap->fip.fip_dl_len = htons((dlen + sizeof(*mac)) / FIP_BPW);
+	cap->fip.fip_flags = htons(FIP_FL_FPMA);
+
+	cap->encaps.fd_desc.fip_dtype = dtype;
+	cap->encaps.fd_desc.fip_dlen = dlen / FIP_BPW;
+
+	mac = (struct fip_mac_desc *)skb_put(skb, sizeof(*mac));
+	memset(mac, 0, sizeof(mac));
+	mac->fd_desc.fip_dtype = FIP_DT_MAC;
+	mac->fd_desc.fip_dlen = sizeof(*mac) / FIP_BPW;
+	if (dtype != ELS_FLOGI)
+		memcpy(mac->fd_mac, fip->data_src_addr, ETH_ALEN);
+
+	skb->protocol = htons(ETH_P_802_3);
+	skb_reset_mac_header(skb);
+	skb_reset_network_header(skb);
+	return 0;
+}
+
+/**
+ * fcoe_ctlr_els_send() - Send an ELS frame encapsulated by FIP if appropriate.
+ * @fip:	FCoE controller.
+ * @skb:	FCoE ELS frame including FC header but no FCoE headers.
+ *
+ * Returns a non-zero error code if the frame should not be sent.
+ * Returns zero if the caller should send the frame with FCoE encapsulation.
+ *
+ * The caller must check that the length is a multiple of 4.
+ * The SKB must have enough headroom (28 bytes) and tailroom (8 bytes).
+ */
+int fcoe_ctlr_els_send(struct fcoe_ctlr *fip, struct sk_buff *skb)
+{
+	struct fc_frame_header *fh;
+	u16 old_xid;
+	u8 op;
+
+	if (fip->state == FIP_ST_NON_FIP)
+		return 0;
+
+	fh = (struct fc_frame_header *)skb->data;
+	op = *(u8 *)(fh + 1);
+
+	switch (op) {
+	case ELS_FLOGI:
+		old_xid = fip->flogi_oxid;
+		fip->flogi_oxid = ntohs(fh->fh_ox_id);
+		if (fip->state == FIP_ST_AUTO) {
+			if (old_xid == FC_XID_UNKNOWN)
+				fip->flogi_count = 0;
+			fip->flogi_count++;
+			if (fip->flogi_count < 3)
+				goto drop;
+			fip->map_dest = 1;
+			return 0;
+		}
+		op = FIP_DT_FLOGI;
+		break;
+	case ELS_FDISC:
+		if (ntoh24(fh->fh_s_id))
+			return 0;
+		op = FIP_DT_FDISC;
+		break;
+	case ELS_LOGO:
+		if (fip->state != FIP_ST_ENABLED)
+			return 0;
+		if (ntoh24(fh->fh_d_id) != FC_FID_FLOGI)
+			return 0;
+		op = FIP_DT_LOGO;
+		break;
+	case ELS_LS_ACC:
+		if (fip->flogi_oxid == FC_XID_UNKNOWN)
+			return 0;
+		if (!ntoh24(fh->fh_s_id))
+			return 0;
+		if (fip->state == FIP_ST_AUTO)
+			return 0;
+		/*
+		 * Here we must've gotten an SID by accepting an FLOGI
+		 * from a point-to-point connection.  Switch to using
+		 * the source mac based on the SID.  The destination
+		 * MAC in this case would have been set by receving the
+		 * FLOGI.
+		 */
+		fip->flogi_oxid = FC_XID_UNKNOWN;
+		fc_fcoe_set_mac(fip->data_src_addr, fh->fh_s_id);
+		return 0;
+	default:
+		if (fip->state != FIP_ST_ENABLED)
+			goto drop;
+		return 0;
+	}
+	if (fcoe_ctlr_encaps(fip, op, skb))
+		goto drop;
+	fip->send(fip, skb);
+	return -EINPROGRESS;
+drop:
+	kfree_skb(skb);
+	return -EINVAL;
+}
+EXPORT_SYMBOL(fcoe_ctlr_els_send);
+
+/*
+ * fcoe_ctlr_age_fcfs() - Reset and free all old FCFs for a controller.
+ * @fip:	FCoE controller.
+ *
+ * Called with lock held.
+ *
+ * An FCF is considered old if we have missed three advertisements.
+ * That is, there have been no valid advertisement from it for three
+ * times its keep-alive period including fuzz.
+ *
+ * In addition, determine the time when an FCF selection can occur.
+ */
+static void fcoe_ctlr_age_fcfs(struct fcoe_ctlr *fip)
+{
+	struct fcoe_fcf *fcf;
+	struct fcoe_fcf *next;
+	unsigned long sel_time = 0;
+
+	list_for_each_entry_safe(fcf, next, &fip->fcfs, list) {
+		if (time_after(jiffies, fcf->time + fcf->fka_period * 3 +
+			       msecs_to_jiffies(FIP_FCF_FUZZ * 3))) {
+			if (fip->sel_fcf == fcf)
+				fip->sel_fcf = NULL;
+			list_del(&fcf->list);
+			WARN_ON(!fip->fcf_count);
+			fip->fcf_count--;
+			kfree(fcf);
+		} else if (fcoe_ctlr_mtu_valid(fcf) &&
+			   (!sel_time || time_before(sel_time, fcf->time))) {
+			sel_time = fcf->time;
+		}
+	}
+	if (sel_time) {
+		sel_time += msecs_to_jiffies(FCOE_CTLR_START_DELAY);
+		fip->sel_time = sel_time;
+		if (time_before(sel_time, fip->timer.expires))
+			mod_timer(&fip->timer, sel_time);
+	} else {
+		fip->sel_time = 0;
+	}
+}
+
+/**
+ * fcoe_ctlr_parse_adv() - Decode a FIP advertisement into a new FCF entry.
+ * @skb:	received FIP advertisement frame
+ * @fcf:	resulting FCF entry.
+ *
+ * Returns zero on a valid parsed advertisement,
+ * otherwise returns non zero value.
+ */
+static int fcoe_ctlr_parse_adv(struct sk_buff *skb, struct fcoe_fcf *fcf)
+{
+	struct fip_header *fiph;
+	struct fip_desc *desc = NULL;
+	struct fip_wwn_desc *wwn;
+	struct fip_fab_desc *fab;
+	struct fip_fka_desc *fka;
+	unsigned long t;
+	size_t rlen;
+	size_t dlen;
+
+	memset(fcf, 0, sizeof(*fcf));
+	fcf->fka_period = msecs_to_jiffies(FCOE_CTLR_DEF_FKA);
+
+	fiph = (struct fip_header *)skb->data;
+	fcf->flags = ntohs(fiph->fip_flags);
+
+	rlen = ntohs(fiph->fip_dl_len) * 4;
+	if (rlen + sizeof(*fiph) > skb->len)
+		return -EINVAL;
+
+	desc = (struct fip_desc *)(fiph + 1);
+	while (rlen > 0) {
+		dlen = desc->fip_dlen * FIP_BPW;
+		if (dlen < sizeof(*desc) || dlen > rlen)
+			return -EINVAL;
+		switch (desc->fip_dtype) {
+		case FIP_DT_PRI:
+			if (dlen != sizeof(struct fip_pri_desc))
+				goto len_err;
+			fcf->pri = ((struct fip_pri_desc *)desc)->fd_pri;
+			break;
+		case FIP_DT_MAC:
+			if (dlen != sizeof(struct fip_mac_desc))
+				goto len_err;
+			memcpy(fcf->fcf_mac,
+			       ((struct fip_mac_desc *)desc)->fd_mac,
+			       ETH_ALEN);
+			if (!is_valid_ether_addr(fcf->fcf_mac)) {
+				FIP_DBG("invalid MAC addr in FIP adv\n");
+				return -EINVAL;
+			}
+			break;
+		case FIP_DT_NAME:
+			if (dlen != sizeof(struct fip_wwn_desc))
+				goto len_err;
+			wwn = (struct fip_wwn_desc *)desc;
+			fcf->switch_name = get_unaligned_be64(&wwn->fd_wwn);
+			break;
+		case FIP_DT_FAB:
+			if (dlen != sizeof(struct fip_fab_desc))
+				goto len_err;
+			fab = (struct fip_fab_desc *)desc;
+			fcf->fabric_name = get_unaligned_be64(&fab->fd_wwn);
+			fcf->vfid = ntohs(fab->fd_vfid);
+			fcf->fc_map = ntoh24(fab->fd_map);
+			break;
+		case FIP_DT_FKA:
+			if (dlen != sizeof(struct fip_fka_desc))
+				goto l