tipc: fix stale link problem during synchronization
[cascardo/linux.git] / net / tipc / link.c
index eaa9fe5..75db07c 100644 (file)
@@ -48,9 +48,8 @@
 /*
  * Error message prefixes
  */
-static const char *link_co_err = "Link changeover error, ";
+static const char *link_co_err = "Link tunneling error, ";
 static const char *link_rst_msg = "Resetting link ";
-static const char *link_unk_evt = "Unknown link event ";
 
 static const struct nla_policy tipc_nl_link_policy[TIPC_NLA_LINK_MAX + 1] = {
        [TIPC_NLA_LINK_UNSPEC]          = { .type = NLA_UNSPEC },
@@ -76,257 +75,414 @@ static const struct nla_policy tipc_nl_prop_policy[TIPC_NLA_PROP_MAX + 1] = {
        [TIPC_NLA_PROP_WIN]             = { .type = NLA_U32 }
 };
 
+/*
+ * Interval between NACKs when packets arrive out of order
+ */
+#define TIPC_NACK_INTV (TIPC_MIN_LINK_WIN * 2)
 /*
  * Out-of-range value for link session numbers
  */
-#define INVALID_SESSION 0x10000
+#define WILDCARD_SESSION 0x10000
 
-/*
- * Link state events:
+/* Link FSM states:
  */
-#define  STARTING_EVT    856384768     /* link processing trigger */
-#define  TRAFFIC_MSG_EVT 560815u       /* rx'd ??? */
-#define  SILENCE_EVT     560817u       /* timer dicovered silence from peer */
+enum {
+       LINK_ESTABLISHED     = 0xe,
+       LINK_ESTABLISHING    = 0xe  << 4,
+       LINK_RESET           = 0x1  << 8,
+       LINK_RESETTING       = 0x2  << 12,
+       LINK_PEER_RESET      = 0xd  << 16,
+       LINK_FAILINGOVER     = 0xf  << 20,
+       LINK_SYNCHING        = 0xc  << 24
+};
 
-/*
- * State value stored in 'failover_pkts'
+/* Link FSM state checking routines
  */
-#define FIRST_FAILOVER 0xffffu
-
-static void link_handle_out_of_seq_msg(struct tipc_link *link,
-                                      struct sk_buff *skb);
-static void tipc_link_proto_rcv(struct tipc_link *link,
-                               struct sk_buff *skb);
-static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tol);
-static void link_state_event(struct tipc_link *l_ptr, u32 event);
+static int link_is_up(struct tipc_link *l)
+{
+       return l->state & (LINK_ESTABLISHED | LINK_SYNCHING);
+}
+
+static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
+                              struct sk_buff_head *xmitq);
+static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
+                                     u16 rcvgap, int tolerance, int priority,
+                                     struct sk_buff_head *xmitq);
 static void link_reset_statistics(struct tipc_link *l_ptr);
 static void link_print(struct tipc_link *l_ptr, const char *str);
-static void tipc_link_sync_xmit(struct tipc_link *l);
 static void tipc_link_sync_rcv(struct tipc_node *n, struct sk_buff *buf);
-static void tipc_link_input(struct tipc_link *l, struct sk_buff *skb);
-static bool tipc_data_input(struct tipc_link *l, struct sk_buff *skb);
-static bool tipc_link_failover_rcv(struct tipc_link *l, struct sk_buff **skb);
-static void link_set_timer(struct tipc_link *link, unsigned long time);
+
 /*
- *  Simple link routines
+ *  Simple non-static link routines (i.e. referenced outside this file)
  */
-static unsigned int align(unsigned int i)
+bool tipc_link_is_up(struct tipc_link *l)
 {
-       return (i + 3) & ~3u;
+       return link_is_up(l);
 }
 
-static void tipc_link_release(struct kref *kref)
+bool tipc_link_is_reset(struct tipc_link *l)
 {
-       kfree(container_of(kref, struct tipc_link, ref));
+       return l->state & (LINK_RESET | LINK_FAILINGOVER | LINK_ESTABLISHING);
 }
 
-static void tipc_link_get(struct tipc_link *l_ptr)
+bool tipc_link_is_synching(struct tipc_link *l)
 {
-       kref_get(&l_ptr->ref);
+       return l->state == LINK_SYNCHING;
 }
 
-static void tipc_link_put(struct tipc_link *l_ptr)
+bool tipc_link_is_failingover(struct tipc_link *l)
 {
-       kref_put(&l_ptr->ref, tipc_link_release);
+       return l->state == LINK_FAILINGOVER;
 }
 
-static struct tipc_link *tipc_parallel_link(struct tipc_link *l)
+bool tipc_link_is_blocked(struct tipc_link *l)
 {
-       if (l->owner->active_links[0] != l)
-               return l->owner->active_links[0];
-       return l->owner->active_links[1];
+       return l->state & (LINK_RESETTING | LINK_PEER_RESET | LINK_FAILINGOVER);
 }
 
-/*
- *  Simple non-static link routines (i.e. referenced outside this file)
- */
-int tipc_link_is_up(struct tipc_link *l_ptr)
+int tipc_link_is_active(struct tipc_link *l)
 {
-       if (!l_ptr)
-               return 0;
-       return link_working_working(l_ptr) || link_working_unknown(l_ptr);
+       struct tipc_node *n = l->owner;
+
+       return (node_active_link(n, 0) == l) || (node_active_link(n, 1) == l);
 }
 
-int tipc_link_is_active(struct tipc_link *l_ptr)
+static u32 link_own_addr(struct tipc_link *l)
 {
-       return  (l_ptr->owner->active_links[0] == l_ptr) ||
-               (l_ptr->owner->active_links[1] == l_ptr);
+       return msg_prevnode(l->pmsg);
 }
 
 /**
- * link_timeout - handle expiration of link timer
- * @l_ptr: pointer to link
+ * tipc_link_create - create a new link
+ * @n: pointer to associated node
+ * @b: pointer to associated bearer
+ * @ownnode: identity of own node
+ * @peer: identity of peer node
+ * @maddr: media address to be used
+ * @inputq: queue to put messages ready for delivery
+ * @namedq: queue to put binding table update messages ready for delivery
+ * @link: return value, pointer to put the created link
+ *
+ * Returns true if link was created, otherwise false
  */
-static void link_timeout(unsigned long data)
+bool tipc_link_create(struct tipc_node *n, struct tipc_bearer *b, u32 session,
+                     u32 ownnode, u32 peer, struct tipc_media_addr *maddr,
+                     struct sk_buff_head *inputq, struct sk_buff_head *namedq,
+                     struct tipc_link **link)
 {
-       struct tipc_link *l_ptr = (struct tipc_link *)data;
-       struct sk_buff *skb;
+       struct tipc_link *l;
+       struct tipc_msg *hdr;
+       char *if_name;
+
+       l = kzalloc(sizeof(*l), GFP_ATOMIC);
+       if (!l)
+               return false;
+       *link = l;
+
+       /* Note: peer i/f name is completed by reset/activate message */
+       if_name = strchr(b->name, ':') + 1;
+       sprintf(l->name, "%u.%u.%u:%s-%u.%u.%u:unknown",
+               tipc_zone(ownnode), tipc_cluster(ownnode), tipc_node(ownnode),
+               if_name, tipc_zone(peer), tipc_cluster(peer), tipc_node(peer));
+
+       l->addr = peer;
+       l->media_addr = maddr;
+       l->owner = n;
+       l->peer_session = WILDCARD_SESSION;
+       l->bearer_id = b->identity;
+       l->tolerance = b->tolerance;
+       l->net_plane = b->net_plane;
+       l->advertised_mtu = b->mtu;
+       l->mtu = b->mtu;
+       l->priority = b->priority;
+       tipc_link_set_queue_limits(l, b->window);
+       l->inputq = inputq;
+       l->namedq = namedq;
+       l->state = LINK_RESETTING;
+       l->pmsg = (struct tipc_msg *)&l->proto_msg;
+       hdr = l->pmsg;
+       tipc_msg_init(ownnode, hdr, LINK_PROTOCOL, RESET_MSG, INT_H_SIZE, peer);
+       msg_set_size(hdr, sizeof(l->proto_msg));
+       msg_set_session(hdr, session);
+       msg_set_bearer_id(hdr, l->bearer_id);
+       strcpy((char *)msg_data(hdr), if_name);
+       __skb_queue_head_init(&l->transmq);
+       __skb_queue_head_init(&l->backlogq);
+       __skb_queue_head_init(&l->deferdq);
+       skb_queue_head_init(&l->wakeupq);
+       skb_queue_head_init(l->inputq);
+       return true;
+}
 
-       tipc_node_lock(l_ptr->owner);
+/* tipc_link_build_bcast_sync_msg() - synchronize broadcast link endpoints.
+ *
+ * Give a newly added peer node the sequence number where it should
+ * start receiving and acking broadcast packets.
+ */
+void tipc_link_build_bcast_sync_msg(struct tipc_link *l,
+                                   struct sk_buff_head *xmitq)
+{
+       struct sk_buff *skb;
+       struct sk_buff_head list;
+       u16 last_sent;
 
-       /* update counters used in statistical profiling of send traffic */
-       l_ptr->stats.accu_queue_sz += skb_queue_len(&l_ptr->transmq);
-       l_ptr->stats.queue_sz_counts++;
+       skb = tipc_msg_create(BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE,
+                             0, l->addr, link_own_addr(l), 0, 0, 0);
+       if (!skb)
+               return;
+       last_sent = tipc_bclink_get_last_sent(l->owner->net);
+       msg_set_last_bcast(buf_msg(skb), last_sent);
+       __skb_queue_head_init(&list);
+       __skb_queue_tail(&list, skb);
+       tipc_link_xmit(l, &list, xmitq);
+}
 
-       skb = skb_peek(&l_ptr->transmq);
-       if (skb) {
-               struct tipc_msg *msg = buf_msg(skb);
-               u32 length = msg_size(msg);
+/**
+ * tipc_link_fsm_evt - link finite state machine
+ * @l: pointer to link
+ * @evt: state machine event to be processed
+ */
+int tipc_link_fsm_evt(struct tipc_link *l, int evt)
+{
+       int rc = 0;
 
-               if ((msg_user(msg) == MSG_FRAGMENTER) &&
-                   (msg_type(msg) == FIRST_FRAGMENT)) {
-                       length = msg_size(msg_get_wrapped(msg));
+       switch (l->state) {
+       case LINK_RESETTING:
+               switch (evt) {
+               case LINK_PEER_RESET_EVT:
+                       l->state = LINK_PEER_RESET;
+                       break;
+               case LINK_RESET_EVT:
+                       l->state = LINK_RESET;
+                       break;
+               case LINK_FAILURE_EVT:
+               case LINK_FAILOVER_BEGIN_EVT:
+               case LINK_ESTABLISH_EVT:
+               case LINK_FAILOVER_END_EVT:
+               case LINK_SYNCH_BEGIN_EVT:
+               case LINK_SYNCH_END_EVT:
+               default:
+                       goto illegal_evt;
+               }
+               break;
+       case LINK_RESET:
+               switch (evt) {
+               case LINK_PEER_RESET_EVT:
+                       l->state = LINK_ESTABLISHING;
+                       break;
+               case LINK_FAILOVER_BEGIN_EVT:
+                       l->state = LINK_FAILINGOVER;
+               case LINK_FAILURE_EVT:
+               case LINK_RESET_EVT:
+               case LINK_ESTABLISH_EVT:
+               case LINK_FAILOVER_END_EVT:
+                       break;
+               case LINK_SYNCH_BEGIN_EVT:
+               case LINK_SYNCH_END_EVT:
+               default:
+                       goto illegal_evt;
+               }
+               break;
+       case LINK_PEER_RESET:
+               switch (evt) {
+               case LINK_RESET_EVT:
+                       l->state = LINK_ESTABLISHING;
+                       break;
+               case LINK_PEER_RESET_EVT:
+               case LINK_ESTABLISH_EVT:
+               case LINK_FAILURE_EVT:
+                       break;
+               case LINK_SYNCH_BEGIN_EVT:
+               case LINK_SYNCH_END_EVT:
+               case LINK_FAILOVER_BEGIN_EVT:
+               case LINK_FAILOVER_END_EVT:
+               default:
+                       goto illegal_evt;
+               }
+               break;
+       case LINK_FAILINGOVER:
+               switch (evt) {
+               case LINK_FAILOVER_END_EVT:
+                       l->state = LINK_RESET;
+                       break;
+               case LINK_PEER_RESET_EVT:
+               case LINK_RESET_EVT:
+               case LINK_ESTABLISH_EVT:
+               case LINK_FAILURE_EVT:
+                       break;
+               case LINK_FAILOVER_BEGIN_EVT:
+               case LINK_SYNCH_BEGIN_EVT:
+               case LINK_SYNCH_END_EVT:
+               default:
+                       goto illegal_evt;
+               }
+               break;
+       case LINK_ESTABLISHING:
+               switch (evt) {
+               case LINK_ESTABLISH_EVT:
+                       l->state = LINK_ESTABLISHED;
+                       rc |= TIPC_LINK_UP_EVT;
+                       break;
+               case LINK_FAILOVER_BEGIN_EVT:
+                       l->state = LINK_FAILINGOVER;
+                       break;
+               case LINK_PEER_RESET_EVT:
+               case LINK_RESET_EVT:
+               case LINK_FAILURE_EVT:
+               case LINK_SYNCH_BEGIN_EVT:
+               case LINK_FAILOVER_END_EVT:
+                       break;
+               case LINK_SYNCH_END_EVT:
+               default:
+                       goto illegal_evt;
+               }
+               break;
+       case LINK_ESTABLISHED:
+               switch (evt) {
+               case LINK_PEER_RESET_EVT:
+                       l->state = LINK_PEER_RESET;
+                       rc |= TIPC_LINK_DOWN_EVT;
+                       break;
+               case LINK_FAILURE_EVT:
+                       l->state = LINK_RESETTING;
+                       rc |= TIPC_LINK_DOWN_EVT;
+                       break;
+               case LINK_RESET_EVT:
+                       l->state = LINK_RESET;
+                       break;
+               case LINK_ESTABLISH_EVT:
+               case LINK_SYNCH_END_EVT:
+                       break;
+               case LINK_SYNCH_BEGIN_EVT:
+                       l->state = LINK_SYNCHING;
+                       break;
+               case LINK_FAILOVER_BEGIN_EVT:
+               case LINK_FAILOVER_END_EVT:
+               default:
+                       goto illegal_evt;
                }
-               if (length) {
-                       l_ptr->stats.msg_lengths_total += length;
-                       l_ptr->stats.msg_length_counts++;
-                       if (length <= 64)
-                               l_ptr->stats.msg_length_profile[0]++;
-                       else if (length <= 256)
-                               l_ptr->stats.msg_length_profile[1]++;
-                       else if (length <= 1024)
-                               l_ptr->stats.msg_length_profile[2]++;
-                       else if (length <= 4096)
-                               l_ptr->stats.msg_length_profile[3]++;
-                       else if (length <= 16384)
-                               l_ptr->stats.msg_length_profile[4]++;
-                       else if (length <= 32768)
-                               l_ptr->stats.msg_length_profile[5]++;
-                       else
-                               l_ptr->stats.msg_length_profile[6]++;
+               break;
+       case LINK_SYNCHING:
+               switch (evt) {
+               case LINK_PEER_RESET_EVT:
+                       l->state = LINK_PEER_RESET;
+                       rc |= TIPC_LINK_DOWN_EVT;
+                       break;
+               case LINK_FAILURE_EVT:
+                       l->state = LINK_RESETTING;
+                       rc |= TIPC_LINK_DOWN_EVT;
+                       break;
+               case LINK_RESET_EVT:
+                       l->state = LINK_RESET;
+                       break;
+               case LINK_ESTABLISH_EVT:
+               case LINK_SYNCH_BEGIN_EVT:
+                       break;
+               case LINK_SYNCH_END_EVT:
+                       l->state = LINK_ESTABLISHED;
+                       break;
+               case LINK_FAILOVER_BEGIN_EVT:
+               case LINK_FAILOVER_END_EVT:
+               default:
+                       goto illegal_evt;
                }
+               break;
+       default:
+               pr_err("Unknown FSM state %x in %s\n", l->state, l->name);
        }
-
-       /* do all other link processing performed on a periodic basis */
-       if (l_ptr->silent_intv_cnt || tipc_bclink_acks_missing(l_ptr->owner))
-               link_state_event(l_ptr, SILENCE_EVT);
-       l_ptr->silent_intv_cnt++;
-       if (skb_queue_len(&l_ptr->backlogq))
-               tipc_link_push_packets(l_ptr);
-       link_set_timer(l_ptr, l_ptr->keepalive_intv);
-       tipc_node_unlock(l_ptr->owner);
-       tipc_link_put(l_ptr);
-}
-
-static void link_set_timer(struct tipc_link *link, unsigned long time)
-{
-       if (!mod_timer(&link->timer, jiffies + time))
-               tipc_link_get(link);
+       return rc;
+illegal_evt:
+       pr_err("Illegal FSM event %x in state %x on link %s\n",
+              evt, l->state, l->name);
+       return rc;
 }
 
-/**
- * tipc_link_create - create a new link
- * @n_ptr: pointer to associated node
- * @b_ptr: pointer to associated bearer
- * @media_addr: media address to use when sending messages over link
- *
- * Returns pointer to link.
+/* link_profile_stats - update statistical profiling of traffic
  */
-struct tipc_link *tipc_link_create(struct tipc_node *n_ptr,
-                                  struct tipc_bearer *b_ptr,
-                                  const struct tipc_media_addr *media_addr)
+static void link_profile_stats(struct tipc_link *l)
 {
-       struct tipc_net *tn = net_generic(n_ptr->net, tipc_net_id);
-       struct tipc_link *l_ptr;
+       struct sk_buff *skb;
        struct tipc_msg *msg;
-       char *if_name;
-       char addr_string[16];
-       u32 peer = n_ptr->addr;
+       int length;
 
-       if (n_ptr->link_cnt >= MAX_BEARERS) {
-               tipc_addr_string_fill(addr_string, n_ptr->addr);
-               pr_err("Cannot establish %uth link to %s. Max %u allowed.\n",
-                      n_ptr->link_cnt, addr_string, MAX_BEARERS);
-               return NULL;
-       }
+       /* Update counters used in statistical profiling of send traffic */
+       l->stats.accu_queue_sz += skb_queue_len(&l->transmq);
+       l->stats.queue_sz_counts++;
 
-       if (n_ptr->links[b_ptr->identity]) {
-               tipc_addr_string_fill(addr_string, n_ptr->addr);
-               pr_err("Attempt to establish second link on <%s> to %s\n",
-                      b_ptr->name, addr_string);
-               return NULL;
-       }
+       skb = skb_peek(&l->transmq);
+       if (!skb)
+               return;
+       msg = buf_msg(skb);
+       length = msg_size(msg);
 
-       l_ptr = kzalloc(sizeof(*l_ptr), GFP_ATOMIC);
-       if (!l_ptr) {
-               pr_warn("Link creation failed, no memory\n");
-               return NULL;
+       if (msg_user(msg) == MSG_FRAGMENTER) {
+               if (msg_type(msg) != FIRST_FRAGMENT)
+                       return;
+               length = msg_size(msg_get_wrapped(msg));
        }
-       kref_init(&l_ptr->ref);
-       l_ptr->addr = peer;
-       if_name = strchr(b_ptr->name, ':') + 1;
-       sprintf(l_ptr->name, "%u.%u.%u:%s-%u.%u.%u:unknown",
-               tipc_zone(tn->own_addr), tipc_cluster(tn->own_addr),
-               tipc_node(tn->own_addr),
-               if_name,
-               tipc_zone(peer), tipc_cluster(peer), tipc_node(peer));
-               /* note: peer i/f name is updated by reset/activate message */
-       memcpy(&l_ptr->media_addr, media_addr, sizeof(*media_addr));
-       l_ptr->owner = n_ptr;
-       l_ptr->peer_session = INVALID_SESSION;
-       l_ptr->bearer_id = b_ptr->identity;
-       link_set_supervision_props(l_ptr, b_ptr->tolerance);
-       l_ptr->state = RESET_UNKNOWN;
-
-       l_ptr->pmsg = (struct tipc_msg *)&l_ptr->proto_msg;
-       msg = l_ptr->pmsg;
-       tipc_msg_init(tn->own_addr, msg, LINK_PROTOCOL, RESET_MSG, INT_H_SIZE,
-                     l_ptr->addr);
-       msg_set_size(msg, sizeof(l_ptr->proto_msg));
-       msg_set_session(msg, (tn->random & 0xffff));
-       msg_set_bearer_id(msg, b_ptr->identity);
-       strcpy((char *)msg_data(msg), if_name);
-       l_ptr->net_plane = b_ptr->net_plane;
-       l_ptr->advertised_mtu = b_ptr->mtu;
-       l_ptr->mtu = l_ptr->advertised_mtu;
-       l_ptr->priority = b_ptr->priority;
-       tipc_link_set_queue_limits(l_ptr, b_ptr->window);
-       l_ptr->snd_nxt = 1;
-       __skb_queue_head_init(&l_ptr->transmq);
-       __skb_queue_head_init(&l_ptr->backlogq);
-       __skb_queue_head_init(&l_ptr->deferdq);
-       skb_queue_head_init(&l_ptr->wakeupq);
-       skb_queue_head_init(&l_ptr->inputq);
-       skb_queue_head_init(&l_ptr->namedq);
-       link_reset_statistics(l_ptr);
-       tipc_node_attach_link(n_ptr, l_ptr);
-       setup_timer(&l_ptr->timer, link_timeout, (unsigned long)l_ptr);
-       link_state_event(l_ptr, STARTING_EVT);
-
-       return l_ptr;
+       l->stats.msg_lengths_total += length;
+       l->stats.msg_length_counts++;
+       if (length <= 64)
+               l->stats.msg_length_profile[0]++;
+       else if (length <= 256)
+               l->stats.msg_length_profile[1]++;
+       else if (length <= 1024)
+               l->stats.msg_length_profile[2]++;
+       else if (length <= 4096)
+               l->stats.msg_length_profile[3]++;
+       else if (length <= 16384)
+               l->stats.msg_length_profile[4]++;
+       else if (length <= 32768)
+               l->stats.msg_length_profile[5]++;
+       else
+               l->stats.msg_length_profile[6]++;
 }
 
-/**
- * tipc_link_delete - Delete a link
- * @l: link to be deleted
+/* tipc_link_timeout - perform periodic task as instructed from node timeout
  */
-void tipc_link_delete(struct tipc_link *l)
+int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq)
 {
-       tipc_link_reset(l);
-       if (del_timer(&l->timer))
-               tipc_link_put(l);
-       l->flags |= LINK_STOPPED;
-       /* Delete link now, or when timer is finished: */
-       tipc_link_reset_fragments(l);
-       tipc_node_detach_link(l->owner, l);
-       tipc_link_put(l);
-}
+       int rc = 0;
+       int mtyp = STATE_MSG;
+       bool xmit = false;
+       bool prb = false;
+
+       link_profile_stats(l);
+
+       switch (l->state) {
+       case LINK_ESTABLISHED:
+       case LINK_SYNCHING:
+               if (!l->silent_intv_cnt) {
+                       if (tipc_bclink_acks_missing(l->owner))
+                               xmit = true;
+               } else if (l->silent_intv_cnt <= l->abort_limit) {
+                       xmit = true;
+                       prb = true;
+               } else {
+                       rc |= tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
+               }
+               l->silent_intv_cnt++;
+               break;
+       case LINK_RESET:
+               xmit = true;
+               mtyp = RESET_MSG;
+               break;
+       case LINK_ESTABLISHING:
+               xmit = true;
+               mtyp = ACTIVATE_MSG;
+               break;
+       case LINK_PEER_RESET:
+       case LINK_RESETTING:
+       case LINK_FAILINGOVER:
+               break;
+       default:
+               break;
+       }
 
-void tipc_link_delete_list(struct net *net, unsigned int bearer_id)
-{
-       struct tipc_net *tn = net_generic(net, tipc_net_id);
-       struct tipc_link *link;
-       struct tipc_node *node;
+       if (xmit)
+               tipc_link_build_proto_msg(l, mtyp, prb, 0, 0, 0, xmitq);
 
-       rcu_read_lock();
-       list_for_each_entry_rcu(node, &tn->node_list, list) {
-               tipc_node_lock(node);
-               link = node->links[bearer_id];
-               if (link)
-                       tipc_link_delete(link);
-               tipc_node_unlock(node);
-       }
-       rcu_read_unlock();
+       return rc;
 }
 
 /**
@@ -334,7 +490,7 @@ void tipc_link_delete_list(struct net *net, unsigned int bearer_id)
  * @link: congested link
  * @list: message that was attempted sent
  * Create pseudo msg to send back to user when congestion abates
- * Only consumes message if there is an error
+ * Does not consume buffer list
  */
 static int link_schedule_user(struct tipc_link *link, struct sk_buff_head *list)
 {
@@ -347,8 +503,7 @@ static int link_schedule_user(struct tipc_link *link, struct sk_buff_head *list)
        /* This really cannot happen...  */
        if (unlikely(imp > TIPC_CRITICAL_IMPORTANCE)) {
                pr_warn("%s<%s>, send queue full", link_rst_msg, link->name);
-               tipc_link_reset(link);
-               goto err;
+               return -ENOBUFS;
        }
        /* Non-blocking sender: */
        if (TIPC_SKB_CB(skb_peek(list))->wakeup_pending)
@@ -358,15 +513,12 @@ static int link_schedule_user(struct tipc_link *link, struct sk_buff_head *list)
        skb = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
                              addr, addr, oport, 0, 0);
        if (!skb)
-               goto err;
+               return -ENOBUFS;
        TIPC_SKB_CB(skb)->chain_sz = skb_queue_len(list);
        TIPC_SKB_CB(skb)->chain_imp = imp;
        skb_queue_tail(&link->wakeupq, skb);
        link->stats.link_congs++;
        return -ELINKCONG;
-err:
-       __skb_queue_purge(list);
-       return -ENOBUFS;
 }
 
 /**
@@ -388,9 +540,7 @@ void link_prepare_wakeup(struct tipc_link *l)
                if ((pnd[imp] + l->backlog[imp].len) >= lim)
                        break;
                skb_unlink(skb, &l->wakeupq);
-               skb_queue_tail(&l->inputq, skb);
-               l->owner->inputq = &l->inputq;
-               l->owner->action_flags |= TIPC_MSG_EVT;
+               skb_queue_tail(l->inputq, skb);
        }
 }
 
@@ -426,208 +576,36 @@ void tipc_link_purge_queues(struct tipc_link *l_ptr)
        tipc_link_reset_fragments(l_ptr);
 }
 
-void tipc_link_reset(struct tipc_link *l_ptr)
+void tipc_link_reset(struct tipc_link *l)
 {
-       u32 prev_state = l_ptr->state;
-       int was_active_link = tipc_link_is_active(l_ptr);
-       struct tipc_node *owner = l_ptr->owner;
-       struct tipc_link *pl = tipc_parallel_link(l_ptr);
-
-       msg_set_session(l_ptr->pmsg, ((msg_session(l_ptr->pmsg) + 1) & 0xffff));
+       tipc_link_fsm_evt(l, LINK_RESET_EVT);
 
        /* Link is down, accept any session */
-       l_ptr->peer_session = INVALID_SESSION;
-
-       /* Prepare for renewed mtu size negotiation */
-       l_ptr->mtu = l_ptr->advertised_mtu;
-
-       l_ptr->state = RESET_UNKNOWN;
-
-       if ((prev_state == RESET_UNKNOWN) || (prev_state == RESET_RESET))
-               return;
-
-       tipc_node_link_down(l_ptr->owner, l_ptr);
-       tipc_bearer_remove_dest(owner->net, l_ptr->bearer_id, l_ptr->addr);
-
-       if (was_active_link && tipc_node_is_up(l_ptr->owner) && (pl != l_ptr)) {
-               l_ptr->flags |= LINK_FAILINGOVER;
-               l_ptr->failover_checkpt = l_ptr->rcv_nxt;
-               pl->failover_pkts = FIRST_FAILOVER;
-               pl->failover_checkpt = l_ptr->rcv_nxt;
-               pl->failover_skb = l_ptr->reasm_buf;
-       } else {
-               kfree_skb(l_ptr->reasm_buf);
-       }
-       /* Clean up all queues, except inputq: */
-       __skb_queue_purge(&l_ptr->transmq);
-       __skb_queue_purge(&l_ptr->deferdq);
-       if (!owner->inputq)
-               owner->inputq = &l_ptr->inputq;
-       skb_queue_splice_init(&l_ptr->wakeupq, owner->inputq);
-       if (!skb_queue_empty(owner->inputq))
-               owner->action_flags |= TIPC_MSG_EVT;
-       tipc_link_purge_backlog(l_ptr);
-       l_ptr->reasm_buf = NULL;
-       l_ptr->rcv_unacked = 0;
-       l_ptr->snd_nxt = 1;
-       l_ptr->silent_intv_cnt = 0;
-       l_ptr->stale_count = 0;
-       link_reset_statistics(l_ptr);
-}
-
-static void link_activate(struct tipc_link *link)
-{
-       struct tipc_node *node = link->owner;
-
-       link->rcv_nxt = 1;
-       link->stats.recv_info = 1;
-       link->silent_intv_cnt = 0;
-       tipc_node_link_up(node, link);
-       tipc_bearer_add_dest(node->net, link->bearer_id, link->addr);
-}
-
-/**
- * link_state_event - link finite state machine
- * @l_ptr: pointer to link
- * @event: state machine event to process
- */
-static void link_state_event(struct tipc_link *l_ptr, unsigned int event)
-{
-       struct tipc_link *other;
-       unsigned long timer_intv = l_ptr->keepalive_intv;
+       l->peer_session = WILDCARD_SESSION;
 
-       if (l_ptr->flags & LINK_STOPPED)
-               return;
-
-       if (!(l_ptr->flags & LINK_STARTED) && (event != STARTING_EVT))
-               return;         /* Not yet. */
-
-       if (l_ptr->flags & LINK_FAILINGOVER)
-               return;
+       /* If peer is up, it only accepts an incremented session number */
+       msg_set_session(l->pmsg, msg_session(l->pmsg) + 1);
 
-       switch (l_ptr->state) {
-       case WORKING_WORKING:
-               switch (event) {
-               case TRAFFIC_MSG_EVT:
-               case ACTIVATE_MSG:
-                       l_ptr->silent_intv_cnt = 0;
-                       break;
-               case SILENCE_EVT:
-                       if (!l_ptr->silent_intv_cnt) {
-                               if (tipc_bclink_acks_missing(l_ptr->owner))
-                                       tipc_link_proto_xmit(l_ptr, STATE_MSG,
-                                                            0, 0, 0, 0);
-                               break;
-                       }
-                       l_ptr->state = WORKING_UNKNOWN;
-                       tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0);
-                       break;
-               case RESET_MSG:
-                       pr_debug("%s<%s>, requested by peer\n",
-                                link_rst_msg, l_ptr->name);
-                       tipc_link_reset(l_ptr);
-                       l_ptr->state = RESET_RESET;
-                       tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
-                                            0, 0, 0, 0);
-                       break;
-               default:
-                       pr_debug("%s%u in WW state\n", link_unk_evt, event);
-               }
-               break;
-       case WORKING_UNKNOWN:
-               switch (event) {
-               case TRAFFIC_MSG_EVT:
-               case ACTIVATE_MSG:
-                       l_ptr->state = WORKING_WORKING;
-                       l_ptr->silent_intv_cnt = 0;
-                       break;
-               case RESET_MSG:
-                       pr_debug("%s<%s>, requested by peer while probing\n",
-                                link_rst_msg, l_ptr->name);
-                       tipc_link_reset(l_ptr);
-                       l_ptr->state = RESET_RESET;
-                       tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
-                                            0, 0, 0, 0);
-                       break;
-               case SILENCE_EVT:
-                       if (!l_ptr->silent_intv_cnt) {
-                               l_ptr->state = WORKING_WORKING;
-                               if (tipc_bclink_acks_missing(l_ptr->owner))
-                                       tipc_link_proto_xmit(l_ptr, STATE_MSG,
-                                                            0, 0, 0, 0);
-                       } else if (l_ptr->silent_intv_cnt <
-                                  l_ptr->abort_limit) {
-                               tipc_link_proto_xmit(l_ptr, STATE_MSG,
-                                                    1, 0, 0, 0);
-                       } else {        /* Link has failed */
-                               pr_debug("%s<%s>, peer not responding\n",
-                                        link_rst_msg, l_ptr->name);
-                               tipc_link_reset(l_ptr);
-                               l_ptr->state = RESET_UNKNOWN;
-                               tipc_link_proto_xmit(l_ptr, RESET_MSG,
-                                                    0, 0, 0, 0);
-                       }
-                       break;
-               default:
-                       pr_err("%s%u in WU state\n", link_unk_evt, event);
-               }
-               break;
-       case RESET_UNKNOWN:
-               switch (event) {
-               case TRAFFIC_MSG_EVT:
-                       break;
-               case ACTIVATE_MSG:
-                       other = l_ptr->owner->active_links[0];
-                       if (other && link_working_unknown(other))
-                               break;
-                       l_ptr->state = WORKING_WORKING;
-                       link_activate(l_ptr);
-                       tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0);
-                       if (l_ptr->owner->working_links == 1)
-                               tipc_link_sync_xmit(l_ptr);
-                       break;
-               case RESET_MSG:
-                       l_ptr->state = RESET_RESET;
-                       tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
-                                            1, 0, 0, 0);
-                       break;
-               case STARTING_EVT:
-                       l_ptr->flags |= LINK_STARTED;
-                       link_set_timer(l_ptr, timer_intv);
-                       break;
-               case SILENCE_EVT:
-                       tipc_link_proto_xmit(l_ptr, RESET_MSG, 0, 0, 0, 0);
-                       break;
-               default:
-                       pr_err("%s%u in RU state\n", link_unk_evt, event);
-               }
-               break;
-       case RESET_RESET:
-               switch (event) {
-               case TRAFFIC_MSG_EVT:
-               case ACTIVATE_MSG:
-                       other = l_ptr->owner->active_links[0];
-                       if (other && link_working_unknown(other))
-                               break;
-                       l_ptr->state = WORKING_WORKING;
-                       link_activate(l_ptr);
-                       tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0);
-                       if (l_ptr->owner->working_links == 1)
-                               tipc_link_sync_xmit(l_ptr);
-                       break;
-               case RESET_MSG:
-                       break;
-               case SILENCE_EVT:
-                       tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
-                                            0, 0, 0, 0);
-                       break;
-               default:
-                       pr_err("%s%u in RR state\n", link_unk_evt, event);
-               }
-               break;
-       default:
-               pr_err("Unknown link state %u/%u\n", l_ptr->state, event);
-       }
+       /* Prepare for renewed mtu size negotiation */
+       l->mtu = l->advertised_mtu;
+
+       /* Clean up all queues: */
+       __skb_queue_purge(&l->transmq);
+       __skb_queue_purge(&l->deferdq);
+       skb_queue_splice_init(&l->wakeupq, l->inputq);
+
+       tipc_link_purge_backlog(l);
+       kfree_skb(l->reasm_buf);
+       kfree_skb(l->failover_reasm_skb);
+       l->reasm_buf = NULL;
+       l->failover_reasm_skb = NULL;
+       l->rcv_unacked = 0;
+       l->snd_nxt = 1;
+       l->rcv_nxt = 1;
+       l->silent_intv_cnt = 0;
+       l->stats.recv_info = 0;
+       l->stale_count = 0;
+       link_reset_statistics(l);
 }
 
 /**
@@ -635,8 +613,7 @@ static void link_state_event(struct tipc_link *l_ptr, unsigned int event)
  * @link: link to use
  * @list: chain of buffers containing message
  *
- * Consumes the buffer chain, except when returning -ELINKCONG,
- * since the caller then may want to make more send attempts.
+ * Consumes the buffer chain, except when returning an error code,
  * Returns 0 if success, or errno: -ELINKCONG, -EMSGSIZE or -ENOBUFS
  * Messages at TIPC_SYSTEM_IMPORTANCE are always accepted
  */
@@ -650,7 +627,7 @@ int __tipc_link_xmit(struct net *net, struct tipc_link *link,
        u16 ack = mod(link->rcv_nxt - 1);
        u16 seqno = link->snd_nxt;
        u16 bc_last_in = link->owner->bclink.last_in;
-       struct tipc_media_addr *addr = &link->media_addr;
+       struct tipc_media_addr *addr = link->media_addr;
        struct sk_buff_head *transmq = &link->transmq;
        struct sk_buff_head *backlogq = &link->backlogq;
        struct sk_buff *skb, *bskb;
@@ -660,10 +637,9 @@ int __tipc_link_xmit(struct net *net, struct tipc_link *link,
                if (unlikely(link->backlog[i].len >= link->backlog[i].limit))
                        return link_schedule_user(link, list);
        }
-       if (unlikely(msg_size(msg) > mtu)) {
-               __skb_queue_purge(list);
+       if (unlikely(msg_size(msg) > mtu))
                return -EMSGSIZE;
-       }
+
        /* Prepare each packet for sending, and add to relevant queue: */
        while (skb_queue_len(list)) {
                skb = skb_peek(list);
@@ -700,101 +676,76 @@ int __tipc_link_xmit(struct net *net, struct tipc_link *link,
        return 0;
 }
 
-static void skb2list(struct sk_buff *skb, struct sk_buff_head *list)
-{
-       skb_queue_head_init(list);
-       __skb_queue_tail(list, skb);
-}
-
-static int __tipc_link_xmit_skb(struct tipc_link *link, struct sk_buff *skb)
-{
-       struct sk_buff_head head;
-
-       skb2list(skb, &head);
-       return __tipc_link_xmit(link->owner->net, link, &head);
-}
-
-/* tipc_link_xmit_skb(): send single buffer to destination
- * Buffers sent via this functon are generally TIPC_SYSTEM_IMPORTANCE
- * messages, which will not be rejected
- * The only exception is datagram messages rerouted after secondary
- * lookup, which are rare and safe to dispose of anyway.
- * TODO: Return real return value, and let callers use
- * tipc_wait_for_sendpkt() where applicable
- */
-int tipc_link_xmit_skb(struct net *net, struct sk_buff *skb, u32 dnode,
-                      u32 selector)
-{
-       struct sk_buff_head head;
-       int rc;
-
-       skb2list(skb, &head);
-       rc = tipc_link_xmit(net, &head, dnode, selector);
-       if (rc == -ELINKCONG)
-               kfree_skb(skb);
-       return 0;
-}
-
 /**
- * tipc_link_xmit() is the general link level function for message sending
- * @net: the applicable net namespace
+ * tipc_link_xmit(): enqueue buffer list according to queue situation
+ * @link: link to use
  * @list: chain of buffers containing message
- * @dsz: amount of user data to be sent
- * @dnode: address of destination node
- * @selector: a number used for deterministic link selection
- * Consumes the buffer chain, except when returning -ELINKCONG
- * Returns 0 if success, otherwise errno: -ELINKCONG,-EHOSTUNREACH,-EMSGSIZE
+ * @xmitq: returned list of packets to be sent by caller
+ *
+ * Consumes the buffer chain, except when returning -ELINKCONG,
+ * since the caller then may want to make more send attempts.
+ * Returns 0 if success, or errno: -ELINKCONG, -EMSGSIZE or -ENOBUFS
+ * Messages at TIPC_SYSTEM_IMPORTANCE are always accepted
  */
-int tipc_link_xmit(struct net *net, struct sk_buff_head *list, u32 dnode,
-                  u32 selector)
+int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list,
+                  struct sk_buff_head *xmitq)
 {
-       struct tipc_link *link = NULL;
-       struct tipc_node *node;
-       int rc = -EHOSTUNREACH;
+       struct tipc_msg *hdr = buf_msg(skb_peek(list));
+       unsigned int maxwin = l->window;
+       unsigned int i, imp = msg_importance(hdr);
+       unsigned int mtu = l->mtu;
+       u16 ack = l->rcv_nxt - 1;
+       u16 seqno = l->snd_nxt;
+       u16 bc_last_in = l->owner->bclink.last_in;
+       struct sk_buff_head *transmq = &l->transmq;
+       struct sk_buff_head *backlogq = &l->backlogq;
+       struct sk_buff *skb, *_skb, *bskb;
 
-       node = tipc_node_find(net, dnode);
-       if (node) {
-               tipc_node_lock(node);
-               link = node->active_links[selector & 1];
-               if (link)
-                       rc = __tipc_link_xmit(net, link, list);
-               tipc_node_unlock(node);
-               tipc_node_put(node);
-       }
-       if (link)
-               return rc;
-
-       if (likely(in_own_node(net, dnode))) {
-               tipc_sk_rcv(net, list);
-               return 0;
+       /* Match msg importance against this and all higher backlog limits: */
+       for (i = imp; i <= TIPC_SYSTEM_IMPORTANCE; i++) {
+               if (unlikely(l->backlog[i].len >= l->backlog[i].limit))
+                       return link_schedule_user(l, list);
        }
+       if (unlikely(msg_size(hdr) > mtu))
+               return -EMSGSIZE;
 
-       __skb_queue_purge(list);
-       return rc;
-}
-
-/*
- * tipc_link_sync_xmit - synchronize broadcast link endpoints.
- *
- * Give a newly added peer node the sequence number where it should
- * start receiving and acking broadcast packets.
- *
- * Called with node locked
- */
-static void tipc_link_sync_xmit(struct tipc_link *link)
-{
-       struct sk_buff *skb;
-       struct tipc_msg *msg;
-
-       skb = tipc_buf_acquire(INT_H_SIZE);
-       if (!skb)
-               return;
+       /* Prepare each packet for sending, and add to relevant queue: */
+       while (skb_queue_len(list)) {
+               skb = skb_peek(list);
+               hdr = buf_msg(skb);
+               msg_set_seqno(hdr, seqno);
+               msg_set_ack(hdr, ack);
+               msg_set_bcast_ack(hdr, bc_last_in);
 
-       msg = buf_msg(skb);
-       tipc_msg_init(link_own_addr(link), msg, BCAST_PROTOCOL, STATE_MSG,
-                     INT_H_SIZE, link->addr);
-       msg_set_last_bcast(msg, link->owner->bclink.acked);
-       __tipc_link_xmit_skb(link, skb);
+               if (likely(skb_queue_len(transmq) < maxwin)) {
+                       _skb = skb_clone(skb, GFP_ATOMIC);
+                       if (!_skb)
+                               return -ENOBUFS;
+                       __skb_dequeue(list);
+                       __skb_queue_tail(transmq, skb);
+                       __skb_queue_tail(xmitq, _skb);
+                       l->rcv_unacked = 0;
+                       seqno++;
+                       continue;
+               }
+               if (tipc_msg_bundle(skb_peek_tail(backlogq), hdr, mtu)) {
+                       kfree_skb(__skb_dequeue(list));
+                       l->stats.sent_bundled++;
+                       continue;
+               }
+               if (tipc_msg_make_bundle(&bskb, hdr, mtu, l->addr)) {
+                       kfree_skb(__skb_dequeue(list));
+                       __skb_queue_tail(backlogq, bskb);
+                       l->backlog[msg_importance(buf_msg(bskb))].len++;
+                       l->stats.sent_bundled++;
+                       l->stats.sent_bundles++;
+                       continue;
+               }
+               l->backlog[imp].len += skb_queue_len(list);
+               skb_queue_splice_tail_init(list, backlogq);
+       }
+       l->snd_nxt = seqno;
+       return 0;
 }
 
 /*
@@ -842,29 +793,37 @@ void tipc_link_push_packets(struct tipc_link *link)
                link->rcv_unacked = 0;
                __skb_queue_tail(&link->transmq, skb);
                tipc_bearer_send(link->owner->net, link->bearer_id,
-                                skb, &link->media_addr);
+                                skb, link->media_addr);
        }
        link->snd_nxt = seqno;
 }
 
-void tipc_link_reset_all(struct tipc_node *node)
+void tipc_link_advance_backlog(struct tipc_link *l, struct sk_buff_head *xmitq)
 {
-       char addr_string[16];
-       u32 i;
+       struct sk_buff *skb, *_skb;
+       struct tipc_msg *hdr;
+       u16 seqno = l->snd_nxt;
+       u16 ack = l->rcv_nxt - 1;
 
-       tipc_node_lock(node);
-
-       pr_warn("Resetting all links to %s\n",
-               tipc_addr_string_fill(addr_string, node->addr));
-
-       for (i = 0; i < MAX_BEARERS; i++) {
-               if (node->links[i]) {
-                       link_print(node->links[i], "Resetting link\n");
-                       tipc_link_reset(node->links[i]);
-               }
+       while (skb_queue_len(&l->transmq) < l->window) {
+               skb = skb_peek(&l->backlogq);
+               if (!skb)
+                       break;
+               _skb = skb_clone(skb, GFP_ATOMIC);
+               if (!_skb)
+                       break;
+               __skb_dequeue(&l->backlogq);
+               hdr = buf_msg(skb);
+               l->backlog[msg_importance(hdr)].len--;
+               __skb_queue_tail(&l->transmq, skb);
+               __skb_queue_tail(xmitq, _skb);
+               msg_set_ack(hdr, ack);
+               msg_set_seqno(hdr, seqno);
+               msg_set_bcast_ack(hdr, l->owner->bclink.last_in);
+               l->rcv_unacked = 0;
+               seqno++;
        }
-
-       tipc_node_unlock(node);
+       l->snd_nxt = seqno;
 }
 
 static void link_retransmit_failure(struct tipc_link *l_ptr,
@@ -877,9 +836,12 @@ static void link_retransmit_failure(struct tipc_link *l_ptr,
 
        if (l_ptr->addr) {
                /* Handle failure on standard link */
-               link_print(l_ptr, "Resetting link\n");
-               tipc_link_reset(l_ptr);
-
+               link_print(l_ptr, "Resetting link ");
+               pr_info("Failed msg: usr %u, typ %u, len %u, err %u\n",
+                       msg_user(msg), msg_type(msg), msg_size(msg),
+                       msg_errcode(msg));
+               pr_info("sqno %u, prev: %x, src: %x\n",
+                       msg_seqno(msg), msg_prevnode(msg), msg_orignode(msg));
        } else {
                /* Handle failure on broadcast link */
                struct tipc_node *n_ptr;
@@ -934,191 +896,45 @@ void tipc_link_retransmit(struct tipc_link *l_ptr, struct sk_buff *skb,
                msg_set_ack(msg, mod(l_ptr->rcv_nxt - 1));
                msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
                tipc_bearer_send(l_ptr->owner->net, l_ptr->bearer_id, skb,
-                                &l_ptr->media_addr);
+                                l_ptr->media_addr);
                retransmits--;
                l_ptr->stats.retransmitted++;
        }
 }
 
-/* link_synch(): check if all packets arrived before the synch
- *               point have been consumed
- * Returns true if the parallel links are synched, otherwise false
- */
-static bool link_synch(struct tipc_link *l)
-{
-       unsigned int post_synch;
-       struct tipc_link *pl;
-
-       pl  = tipc_parallel_link(l);
-       if (pl == l)
-               goto synched;
-
-       /* Was last pre-synch packet added to input queue ? */
-       if (less_eq(pl->rcv_nxt, l->synch_point))
-               return false;
-
-       /* Is it still in the input queue ? */
-       post_synch = mod(pl->rcv_nxt - l->synch_point) - 1;
-       if (skb_queue_len(&pl->inputq) > post_synch)
-               return false;
-synched:
-       l->flags &= ~LINK_SYNCHING;
-       return true;
-}
-
-static void link_retrieve_defq(struct tipc_link *link,
-                              struct sk_buff_head *list)
+static int tipc_link_retransm(struct tipc_link *l, int retransm,
+                             struct sk_buff_head *xmitq)
 {
-       u16 seq_no;
-
-       if (skb_queue_empty(&link->deferdq))
-               return;
-
-       seq_no = buf_seqno(skb_peek(&link->deferdq));
-       if (seq_no == link->rcv_nxt)
-               skb_queue_splice_tail_init(&link->deferdq, list);
-}
-
-/**
- * tipc_rcv - process TIPC packets/messages arriving from off-node
- * @net: the applicable net namespace
- * @skb: TIPC packet
- * @b_ptr: pointer to bearer message arrived on
- *
- * Invoked with no locks held.  Bearer pointer must point to a valid bearer
- * structure (i.e. cannot be NULL), but bearer can be inactive.
- */
-void tipc_rcv(struct net *net, struct sk_buff *skb, struct tipc_bearer *b_ptr)
-{
-       struct tipc_net *tn = net_generic(net, tipc_net_id);
-       struct sk_buff_head head;
-       struct tipc_node *n_ptr;
-       struct tipc_link *l_ptr;
-       struct sk_buff *skb1, *tmp;
-       struct tipc_msg *msg;
-       u16 seq_no;
-       u16 ackd;
-       u32 released;
+       struct sk_buff *_skb, *skb = skb_peek(&l->transmq);
+       struct tipc_msg *hdr;
 
-       skb2list(skb, &head);
-
-       while ((skb = __skb_dequeue(&head))) {
-               /* Ensure message is well-formed */
-               if (unlikely(!tipc_msg_validate(skb)))
-                       goto discard;
-
-               /* Handle arrival of a non-unicast link message */
-               msg = buf_msg(skb);
-               if (unlikely(msg_non_seq(msg))) {
-                       if (msg_user(msg) ==  LINK_CONFIG)
-                               tipc_disc_rcv(net, skb, b_ptr);
-                       else
-                               tipc_bclink_rcv(net, skb);
-                       continue;
-               }
-
-               /* Discard unicast link messages destined for another node */
-               if (unlikely(!msg_short(msg) &&
-                            (msg_destnode(msg) != tn->own_addr)))
-                       goto discard;
-
-               /* Locate neighboring node that sent message */
-               n_ptr = tipc_node_find(net, msg_prevnode(msg));
-               if (unlikely(!n_ptr))
-                       goto discard;
-
-               tipc_node_lock(n_ptr);
-               /* Locate unicast link endpoint that should handle message */
-               l_ptr = n_ptr->links[b_ptr->identity];
-               if (unlikely(!l_ptr))
-                       goto unlock;
-
-               /* Verify that communication with node is currently allowed */
-               if ((n_ptr->action_flags & TIPC_WAIT_PEER_LINKS_DOWN) &&
-                   msg_user(msg) == LINK_PROTOCOL &&
-                   (msg_type(msg) == RESET_MSG ||
-                   msg_type(msg) == ACTIVATE_MSG) &&
-                   !msg_redundant_link(msg))
-                       n_ptr->action_flags &= ~TIPC_WAIT_PEER_LINKS_DOWN;
-
-               if (tipc_node_blocked(n_ptr))
-                       goto unlock;
-
-               /* Validate message sequence number info */
-               seq_no = msg_seqno(msg);
-               ackd = msg_ack(msg);
-
-               /* Release acked messages */
-               if (unlikely(n_ptr->bclink.acked != msg_bcast_ack(msg)))
-                       tipc_bclink_acknowledge(n_ptr, msg_bcast_ack(msg));
-
-               released = 0;
-               skb_queue_walk_safe(&l_ptr->transmq, skb1, tmp) {
-                       if (more(buf_seqno(skb1), ackd))
-                               break;
-                        __skb_unlink(skb1, &l_ptr->transmq);
-                        kfree_skb(skb1);
-                        released = 1;
-               }
-
-               /* Try sending any messages link endpoint has pending */
-               if (unlikely(skb_queue_len(&l_ptr->backlogq)))
-                       tipc_link_push_packets(l_ptr);
-
-               if (released && !skb_queue_empty(&l_ptr->wakeupq))
-                       link_prepare_wakeup(l_ptr);
-
-               /* Process the incoming packet */
-               if (unlikely(!link_working_working(l_ptr))) {
-                       if (msg_user(msg) == LINK_PROTOCOL) {
-                               tipc_link_proto_rcv(l_ptr, skb);
-                               link_retrieve_defq(l_ptr, &head);
-                               skb = NULL;
-                               goto unlock;
-                       }
-
-                       /* Traffic message. Conditionally activate link */
-                       link_state_event(l_ptr, TRAFFIC_MSG_EVT);
-
-                       if (link_working_working(l_ptr)) {
-                               /* Re-insert buffer in front of queue */
-                               __skb_queue_head(&head, skb);
-                               skb = NULL;
-                               goto unlock;
-                       }
-                       goto unlock;
-               }
-
-               /* Link is now in state WORKING_WORKING */
-               if (unlikely(seq_no != l_ptr->rcv_nxt)) {
-                       link_handle_out_of_seq_msg(l_ptr, skb);
-                       link_retrieve_defq(l_ptr, &head);
-                       skb = NULL;
-                       goto unlock;
-               }
-               l_ptr->silent_intv_cnt = 0;
+       if (!skb)
+               return 0;
 
-               /* Synchronize with parallel link if applicable */
-               if (unlikely((l_ptr->flags & LINK_SYNCHING) && !msg_dup(msg))) {
-                       if (!link_synch(l_ptr))
-                               goto unlock;
-               }
-               l_ptr->rcv_nxt++;
-               if (unlikely(!skb_queue_empty(&l_ptr->deferdq)))
-                       link_retrieve_defq(l_ptr, &head);
-               if (unlikely(++l_ptr->rcv_unacked >= TIPC_MIN_LINK_WIN)) {
-                       l_ptr->stats.sent_acks++;
-                       tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0, 0, 0);
-               }
-               tipc_link_input(l_ptr, skb);
-               skb = NULL;
-unlock:
-               tipc_node_unlock(n_ptr);
-               tipc_node_put(n_ptr);
-discard:
-               if (unlikely(skb))
-                       kfree_skb(skb);
+       /* Detect repeated retransmit failures on same packet */
+       if (likely(l->last_retransm != buf_seqno(skb))) {
+               l->last_retransm = buf_seqno(skb);
+               l->stale_count = 1;
+       } else if (++l->stale_count > 100) {
+               link_retransmit_failure(l, skb);
+               return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
+       }
+       skb_queue_walk(&l->transmq, skb) {
+               if (!retransm)
+                       return 0;
+               hdr = buf_msg(skb);
+               _skb = __pskb_copy(skb, MIN_H_SIZE, GFP_ATOMIC);
+               if (!_skb)
+                       return 0;
+               hdr = buf_msg(_skb);
+               msg_set_ack(hdr, l->rcv_nxt - 1);
+               msg_set_bcast_ack(hdr, l->owner->bclink.last_in);
+               _skb->priority = TC_PRIO_CONTROL;
+               __skb_queue_tail(xmitq, _skb);
+               retransm--;
+               l->stats.retransmitted++;
        }
+       return 0;
 }
 
 /* tipc_data_input - deliver data and name distr msgs to upper layer
@@ -1126,29 +942,22 @@ discard:
  * Consumes buffer if message is of right type
  * Node lock must be held
  */
-static bool tipc_data_input(struct tipc_link *link, struct sk_buff *skb)
+static bool tipc_data_input(struct tipc_link *link, struct sk_buff *skb,
+                           struct sk_buff_head *inputq)
 {
        struct tipc_node *node = link->owner;
-       struct tipc_msg *msg = buf_msg(skb);
-       u32 dport = msg_destport(msg);
 
-       switch (msg_user(msg)) {
+       switch (msg_user(buf_msg(skb))) {
        case TIPC_LOW_IMPORTANCE:
        case TIPC_MEDIUM_IMPORTANCE:
        case TIPC_HIGH_IMPORTANCE:
        case TIPC_CRITICAL_IMPORTANCE:
        case CONN_MANAGER:
-               if (tipc_skb_queue_tail(&link->inputq, skb, dport)) {
-                       node->inputq = &link->inputq;
-                       node->action_flags |= TIPC_MSG_EVT;
-               }
+               __skb_queue_tail(inputq, skb);
                return true;
        case NAME_DISTRIBUTOR:
                node->bclink.recv_permitted = true;
-               node->namedq = &link->namedq;
-               skb_queue_tail(&link->namedq, skb);
-               if (skb_queue_len(&link->namedq) == 1)
-                       node->action_flags |= TIPC_NAMED_MSG_EVT;
+               skb_queue_tail(link->namedq, skb);
                return true;
        case MSG_BUNDLER:
        case TUNNEL_PROTOCOL:
@@ -1165,54 +974,160 @@ static bool tipc_data_input(struct tipc_link *link, struct sk_buff *skb)
 /* tipc_link_input - process packet that has passed link protocol check
  *
  * Consumes buffer
- * Node lock must be held
  */
-static void tipc_link_input(struct tipc_link *link, struct sk_buff *skb)
+static int tipc_link_input(struct tipc_link *l, struct sk_buff *skb,
+                          struct sk_buff_head *inputq)
 {
-       struct tipc_node *node = link->owner;
-       struct tipc_msg *msg = buf_msg(skb);
+       struct tipc_node *node = l->owner;
+       struct tipc_msg *hdr = buf_msg(skb);
+       struct sk_buff **reasm_skb = &l->reasm_buf;
        struct sk_buff *iskb;
+       int usr = msg_user(hdr);
+       int rc = 0;
        int pos = 0;
+       int ipos = 0;
 
-       if (likely(tipc_data_input(link, skb)))
-               return;
+       if (unlikely(usr == TUNNEL_PROTOCOL)) {
+               if (msg_type(hdr) == SYNCH_MSG) {
+                       __skb_queue_purge(&l->deferdq);
+                       goto drop;
+               }
+               if (!tipc_msg_extract(skb, &iskb, &ipos))
+                       return rc;
+               kfree_skb(skb);
+               skb = iskb;
+               hdr = buf_msg(skb);
+               if (less(msg_seqno(hdr), l->drop_point))
+                       goto drop;
+               if (tipc_data_input(l, skb, inputq))
+                       return rc;
+               usr = msg_user(hdr);
+               reasm_skb = &l->failover_reasm_skb;
+       }
 
-       switch (msg_user(msg)) {
-       case TUNNEL_PROTOCOL:
-               if (msg_dup(msg)) {
-                       link->flags |= LINK_SYNCHING;
-                       link->synch_point = msg_seqno(msg_get_wrapped(msg));
-                       kfree_skb(skb);
-                       break;
+       if (usr == MSG_BUNDLER) {
+               l->stats.recv_bundles++;
+               l->stats.recv_bundled += msg_msgcnt(hdr);
+               while (tipc_msg_extract(skb, &iskb, &pos))
+                       tipc_data_input(l, iskb, inputq);
+               return 0;
+       } else if (usr == MSG_FRAGMENTER) {
+               l->stats.recv_fragments++;
+               if (tipc_buf_append(reasm_skb, &skb)) {
+                       l->stats.recv_fragmented++;
+                       tipc_data_input(l, skb, inputq);
+               } else if (!*reasm_skb) {
+                       return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
                }
-               if (!tipc_link_failover_rcv(link, &skb))
-                       break;
-               if (msg_user(buf_msg(skb)) != MSG_BUNDLER) {
-                       tipc_data_input(link, skb);
+               return 0;
+       } else if (usr == BCAST_PROTOCOL) {
+               tipc_link_sync_rcv(node, skb);
+               return 0;
+       }
+drop:
+       kfree_skb(skb);
+       return 0;
+}
+
+static bool tipc_link_release_pkts(struct tipc_link *l, u16 acked)
+{
+       bool released = false;
+       struct sk_buff *skb, *tmp;
+
+       skb_queue_walk_safe(&l->transmq, skb, tmp) {
+               if (more(buf_seqno(skb), acked))
                        break;
+               __skb_unlink(skb, &l->transmq);
+               kfree_skb(skb);
+               released = true;
+       }
+       return released;
+}
+
+/* tipc_link_rcv - process TIPC packets/messages arriving from off-node
+ * @link: the link that should handle the message
+ * @skb: TIPC packet
+ * @xmitq: queue to place packets to be sent after this call
+ */
+int tipc_link_rcv(struct tipc_link *l, struct sk_buff *skb,
+                 struct sk_buff_head *xmitq)
+{
+       struct sk_buff_head *arrvq = &l->deferdq;
+       struct sk_buff_head tmpq;
+       struct tipc_msg *hdr;
+       u16 seqno, rcv_nxt;
+       int rc = 0;
+
+       __skb_queue_head_init(&tmpq);
+
+       if (unlikely(!__tipc_skb_queue_sorted(arrvq, skb))) {
+               if (!(skb_queue_len(arrvq) % TIPC_NACK_INTV))
+                       tipc_link_build_proto_msg(l, STATE_MSG, 0,
+                                                 0, 0, 0, xmitq);
+               return rc;
+       }
+
+       while ((skb = skb_peek(arrvq))) {
+               hdr = buf_msg(skb);
+
+               /* Verify and update link state */
+               if (unlikely(msg_user(hdr) == LINK_PROTOCOL)) {
+                       __skb_dequeue(arrvq);
+                       rc = tipc_link_proto_rcv(l, skb, xmitq);
+                       continue;
                }
-       case MSG_BUNDLER:
-               link->stats.recv_bundles++;
-               link->stats.recv_bundled += msg_msgcnt(msg);
 
-               while (tipc_msg_extract(skb, &iskb, &pos))
-                       tipc_data_input(link, iskb);
-               break;
-       case MSG_FRAGMENTER:
-               link->stats.recv_fragments++;
-               if (tipc_buf_append(&link->reasm_buf, &skb)) {
-                       link->stats.recv_fragmented++;
-                       tipc_data_input(link, skb);
-               } else if (!link->reasm_buf) {
-                       tipc_link_reset(link);
+               if (unlikely(!link_is_up(l))) {
+                       rc = tipc_link_fsm_evt(l, LINK_ESTABLISH_EVT);
+                       if (!link_is_up(l)) {
+                               kfree_skb(__skb_dequeue(arrvq));
+                               goto exit;
+                       }
                }
-               break;
-       case BCAST_PROTOCOL:
-               tipc_link_sync_rcv(node, skb);
-               break;
-       default:
-               break;
-       };
+
+               l->silent_intv_cnt = 0;
+
+               /* Forward queues and wake up waiting users */
+               if (likely(tipc_link_release_pkts(l, msg_ack(hdr)))) {
+                       tipc_link_advance_backlog(l, xmitq);
+                       if (unlikely(!skb_queue_empty(&l->wakeupq)))
+                               link_prepare_wakeup(l);
+               }
+
+               /* Defer reception if there is a gap in the sequence */
+               seqno = msg_seqno(hdr);
+               rcv_nxt = l->rcv_nxt;
+               if (unlikely(less(rcv_nxt, seqno))) {
+                       l->stats.deferred_recv++;
+                       goto exit;
+               }
+
+               __skb_dequeue(arrvq);
+
+               /* Drop if packet already received */
+               if (unlikely(more(rcv_nxt, seqno))) {
+                       l->stats.duplicates++;
+                       kfree_skb(skb);
+                       goto exit;
+               }
+
+               /* Packet can be delivered */
+               l->rcv_nxt++;
+               l->stats.recv_info++;
+               if (unlikely(!tipc_data_input(l, skb, &tmpq)))
+                       rc = tipc_link_input(l, skb, &tmpq);
+
+               /* Ack at regular intervals */
+               if (unlikely(++l->rcv_unacked >= TIPC_MIN_LINK_WIN)) {
+                       l->rcv_unacked = 0;
+                       l->stats.sent_acks++;
+                       tipc_link_build_proto_msg(l, STATE_MSG,
+                                                 0, 0, 0, 0, xmitq);
+               }
+       }
+exit:
+       tipc_skb_queue_splice_tail(&tmpq, l->inputq);
+       return rc;
 }
 
 /**
@@ -1255,458 +1170,250 @@ u32 tipc_link_defer_pkt(struct sk_buff_head *list, struct sk_buff *skb)
 }
 
 /*
- * link_handle_out_of_seq_msg - handle arrival of out-of-sequence packet
+ * Send protocol message to the other endpoint.
  */
-static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
-                                      struct sk_buff *buf)
+void tipc_link_proto_xmit(struct tipc_link *l, u32 msg_typ, int probe_msg,
+                         u32 gap, u32 tolerance, u32 priority)
 {
-       u32 seq_no = buf_seqno(buf);
+       struct sk_buff *skb = NULL;
+       struct sk_buff_head xmitq;
 
-       if (likely(msg_user(buf_msg(buf)) == LINK_PROTOCOL)) {
-               tipc_link_proto_rcv(l_ptr, buf);
+       __skb_queue_head_init(&xmitq);
+       tipc_link_build_proto_msg(l, msg_typ, probe_msg, gap,
+                                 tolerance, priority, &xmitq);
+       skb = __skb_dequeue(&xmitq);
+       if (!skb)
                return;
-       }
-
-       /* Record OOS packet arrival */
-       l_ptr->silent_intv_cnt = 0;
+       tipc_bearer_send(l->owner->net, l->bearer_id, skb, l->media_addr);
+       l->rcv_unacked = 0;
+       kfree_skb(skb);
+}
 
-       /*
-        * Discard packet if a duplicate; otherwise add it to deferred queue
-        * and notify peer of gap as per protocol specification
-        */
-       if (less(seq_no, l_ptr->rcv_nxt)) {
-               l_ptr->stats.duplicates++;
-               kfree_skb(buf);
+/* tipc_link_build_proto_msg: prepare link protocol message for transmission
+ */
+static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
+                                     u16 rcvgap, int tolerance, int priority,
+                                     struct sk_buff_head *xmitq)
+{
+       struct sk_buff *skb = NULL;
+       struct tipc_msg *hdr = l->pmsg;
+       u16 snd_nxt = l->snd_nxt;
+       u16 rcv_nxt = l->rcv_nxt;
+       u16 rcv_last = rcv_nxt - 1;
+       int node_up = l->owner->bclink.recv_permitted;
+
+       /* Don't send protocol message during reset or link failover */
+       if (tipc_link_is_blocked(l))
                return;
-       }
 
-       if (tipc_link_defer_pkt(&l_ptr->deferdq, buf)) {
-               l_ptr->stats.deferred_recv++;
-               if ((skb_queue_len(&l_ptr->deferdq) % TIPC_MIN_LINK_WIN) == 1)
-                       tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0, 0, 0);
+       msg_set_type(hdr, mtyp);
+       msg_set_net_plane(hdr, l->net_plane);
+       msg_set_bcast_ack(hdr, l->owner->bclink.last_in);
+       msg_set_last_bcast(hdr, tipc_bclink_get_last_sent(l->owner->net));
+       msg_set_link_tolerance(hdr, tolerance);
+       msg_set_linkprio(hdr, priority);
+       msg_set_redundant_link(hdr, node_up);
+       msg_set_seq_gap(hdr, 0);
+
+       /* Compatibility: created msg must not be in sequence with pkt flow */
+       msg_set_seqno(hdr, snd_nxt + U16_MAX / 2);
+
+       if (mtyp == STATE_MSG) {
+               if (!tipc_link_is_up(l))
+                       return;
+               msg_set_next_sent(hdr, snd_nxt);
+
+               /* Override rcvgap if there are packets in deferred queue */
+               if (!skb_queue_empty(&l->deferdq))
+                       rcvgap = buf_seqno(skb_peek(&l->deferdq)) - rcv_nxt;
+               if (rcvgap) {
+                       msg_set_seq_gap(hdr, rcvgap);
+                       l->stats.sent_nacks++;
+               }
+               msg_set_ack(hdr, rcv_last);
+               msg_set_probe(hdr, probe);
+               if (probe)
+                       l->stats.sent_probes++;
+               l->stats.sent_states++;
        } else {
-               l_ptr->stats.duplicates++;
+               /* RESET_MSG or ACTIVATE_MSG */
+               msg_set_max_pkt(hdr, l->advertised_mtu);
+               msg_set_ack(hdr, l->rcv_nxt - 1);
+               msg_set_next_sent(hdr, 1);
        }
+       skb = tipc_buf_acquire(msg_size(hdr));
+       if (!skb)
+               return;
+       skb_copy_to_linear_data(skb, hdr, msg_size(hdr));
+       skb->priority = TC_PRIO_CONTROL;
+       __skb_queue_tail(xmitq, skb);
 }
 
-/*
- * Send protocol message to the other endpoint.
+/* tipc_link_tnl_prepare(): prepare and return a list of tunnel packets
+ * with contents of the link's tranmsit and backlog queues.
  */
-void tipc_link_proto_xmit(struct tipc_link *l_ptr, u32 msg_typ, int probe_msg,
-                         u32 gap, u32 tolerance, u32 priority)
+void tipc_link_tnl_prepare(struct tipc_link *l, struct tipc_link *tnl,
+                          int mtyp, struct sk_buff_head *xmitq)
 {
-       struct sk_buff *buf = NULL;
-       struct tipc_msg *msg = l_ptr->pmsg;
-       u32 msg_size = sizeof(l_ptr->proto_msg);
-       int r_flag;
-       u16 last_rcv;
-
-       /* Don't send protocol message during link failover */
-       if (l_ptr->flags & LINK_FAILINGOVER)
-               return;
+       struct sk_buff *skb, *tnlskb;
+       struct tipc_msg *hdr, tnlhdr;
+       struct sk_buff_head *queue = &l->transmq;
+       struct sk_buff_head tmpxq, tnlq;
+       u16 pktlen, pktcnt, seqno = l->snd_nxt;
 
-       /* Abort non-RESET send if communication with node is prohibited */
-       if ((tipc_node_blocked(l_ptr->owner)) && (msg_typ != RESET_MSG))
+       if (!tnl)
                return;
 
-       /* Create protocol message with "out-of-sequence" sequence number */
-       msg_set_type(msg, msg_typ);
-       msg_set_net_plane(msg, l_ptr->net_plane);
-       msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
-       msg_set_last_bcast(msg, tipc_bclink_get_last_sent(l_ptr->owner->net));
-
-       if (msg_typ == STATE_MSG) {
-               u16 next_sent = l_ptr->snd_nxt;
+       skb_queue_head_init(&tnlq);
+       skb_queue_head_init(&tmpxq);
 
-               if (!tipc_link_is_up(l_ptr))
+       /* At least one packet required for safe algorithm => add dummy */
+       skb = tipc_msg_create(TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
+                             BASIC_H_SIZE, 0, l->addr, link_own_addr(l),
+                             0, 0, TIPC_ERR_NO_PORT);
+       if (!skb) {
+               pr_warn("%sunable to create tunnel packet\n", link_co_err);
+               return;
+       }
+       skb_queue_tail(&tnlq, skb);
+       tipc_link_xmit(l, &tnlq, &tmpxq);
+       __skb_queue_purge(&tmpxq);
+
+       /* Initialize reusable tunnel packet header */
+       tipc_msg_init(link_own_addr(l), &tnlhdr, TUNNEL_PROTOCOL,
+                     mtyp, INT_H_SIZE, l->addr);
+       pktcnt = skb_queue_len(&l->transmq) + skb_queue_len(&l->backlogq);
+       msg_set_msgcnt(&tnlhdr, pktcnt);
+       msg_set_bearer_id(&tnlhdr, l->peer_bearer_id);
+tnl:
+       /* Wrap each packet into a tunnel packet */
+       skb_queue_walk(queue, skb) {
+               hdr = buf_msg(skb);
+               if (queue == &l->backlogq)
+                       msg_set_seqno(hdr, seqno++);
+               pktlen = msg_size(hdr);
+               msg_set_size(&tnlhdr, pktlen + INT_H_SIZE);
+               tnlskb = tipc_buf_acquire(pktlen + INT_H_SIZE);
+               if (!tnlskb) {
+                       pr_warn("%sunable to send packet\n", link_co_err);
                        return;
-               msg_set_next_sent(msg, next_sent);
-               if (!skb_queue_empty(&l_ptr->deferdq)) {
-                       last_rcv = buf_seqno(skb_peek(&l_ptr->deferdq));
-                       gap = mod(last_rcv - l_ptr->rcv_nxt);
                }
-               msg_set_seq_gap(msg, gap);
-               if (gap)
-                       l_ptr->stats.sent_nacks++;
-               msg_set_link_tolerance(msg, tolerance);
-               msg_set_linkprio(msg, priority);
-               msg_set_max_pkt(msg, l_ptr->mtu);
-               msg_set_ack(msg, mod(l_ptr->rcv_nxt - 1));
-               msg_set_probe(msg, probe_msg != 0);
-               if (probe_msg)
-                       l_ptr->stats.sent_probes++;
-               l_ptr->stats.sent_states++;
-       } else {                /* RESET_MSG or ACTIVATE_MSG */
-               msg_set_ack(msg, mod(l_ptr->failover_checkpt - 1));
-               msg_set_seq_gap(msg, 0);
-               msg_set_next_sent(msg, 1);
-               msg_set_probe(msg, 0);
-               msg_set_link_tolerance(msg, l_ptr->tolerance);
-               msg_set_linkprio(msg, l_ptr->priority);
-               msg_set_max_pkt(msg, l_ptr->advertised_mtu);
+               skb_copy_to_linear_data(tnlskb, &tnlhdr, INT_H_SIZE);
+               skb_copy_to_linear_data_offset(tnlskb, INT_H_SIZE, hdr, pktlen);
+               __skb_queue_tail(&tnlq, tnlskb);
+       }
+       if (queue != &l->backlogq) {
+               queue = &l->backlogq;
+               goto tnl;
        }
 
-       r_flag = (l_ptr->owner->working_links > tipc_link_is_up(l_ptr));
-       msg_set_redundant_link(msg, r_flag);
-       msg_set_linkprio(msg, l_ptr->priority);
-       msg_set_size(msg, msg_size);
-
-       msg_set_seqno(msg, mod(l_ptr->snd_nxt + (0xffff / 2)));
-
-       buf = tipc_buf_acquire(msg_size);
-       if (!buf)
-               return;
+       tipc_link_xmit(tnl, &tnlq, xmitq);
 
-       skb_copy_to_linear_data(buf, msg, sizeof(l_ptr->proto_msg));
-       buf->priority = TC_PRIO_CONTROL;
-       tipc_bearer_send(l_ptr->owner->net, l_ptr->bearer_id, buf,
-                        &l_ptr->media_addr);
-       l_ptr->rcv_unacked = 0;
-       kfree_skb(buf);
+       if (mtyp == FAILOVER_MSG) {
+               tnl->drop_point = l->rcv_nxt;
+               tnl->failover_reasm_skb = l->reasm_buf;
+               l->reasm_buf = NULL;
+       }
 }
 
-/*
- * Receive protocol message :
+/* tipc_link_proto_rcv(): receive link level protocol message :
  * Note that network plane id propagates through the network, and may
- * change at any time. The node with lowest address rules
+ * change at any time. The node with lowest numerical id determines
+ * network plane
  */
-static void tipc_link_proto_rcv(struct tipc_link *l_ptr,
-                               struct sk_buff *buf)
+static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
+                              struct sk_buff_head *xmitq)
 {
-       u32 rec_gap = 0;
-       u32 msg_tol;
-       struct tipc_msg *msg = buf_msg(buf);
+       struct tipc_msg *hdr = buf_msg(skb);
+       u16 rcvgap = 0;
+       u16 nacked_gap = msg_seq_gap(hdr);
+       u16 peers_snd_nxt =  msg_next_sent(hdr);
+       u16 peers_tol = msg_link_tolerance(hdr);
+       u16 peers_prio = msg_linkprio(hdr);
+       u16 rcv_nxt = l->rcv_nxt;
+       char *if_name;
+       int rc = 0;
 
-       if (l_ptr->flags & LINK_FAILINGOVER)
+       if (tipc_link_is_blocked(l))
                goto exit;
 
-       if (l_ptr->net_plane != msg_net_plane(msg))
-               if (link_own_addr(l_ptr) > msg_prevnode(msg))
-                       l_ptr->net_plane = msg_net_plane(msg);
-
-       switch (msg_type(msg)) {
+       if (link_own_addr(l) > msg_prevnode(hdr))
+               l->net_plane = msg_net_plane(hdr);
 
+       switch (msg_type(hdr)) {
        case RESET_MSG:
-               if (!link_working_unknown(l_ptr) &&
-                   (l_ptr->peer_session != INVALID_SESSION)) {
-                       if (less_eq(msg_session(msg), l_ptr->peer_session))
-                               break; /* duplicate or old reset: ignore */
-               }
-
-               if (!msg_redundant_link(msg) && (link_working_working(l_ptr) ||
-                               link_working_unknown(l_ptr))) {
-                       /*
-                        * peer has lost contact -- don't allow peer's links
-                        * to reactivate before we recognize loss & clean up
-                        */
-                       l_ptr->owner->action_flags |= TIPC_WAIT_OWN_LINKS_DOWN;
-               }
-
-               link_state_event(l_ptr, RESET_MSG);
 
+               /* Ignore duplicate RESET with old session number */
+               if ((less_eq(msg_session(hdr), l->peer_session)) &&
+                   (l->peer_session != WILDCARD_SESSION))
+                       break;
                /* fall thru' */
-       case ACTIVATE_MSG:
-               /* Update link settings according other endpoint's values */
-               strcpy((strrchr(l_ptr->name, ':') + 1), (char *)msg_data(msg));
-
-               msg_tol = msg_link_tolerance(msg);
-               if (msg_tol > l_ptr->tolerance)
-                       link_set_supervision_props(l_ptr, msg_tol);
-
-               if (msg_linkprio(msg) > l_ptr->priority)
-                       l_ptr->priority = msg_linkprio(msg);
-
-               if (l_ptr->mtu > msg_max_pkt(msg))
-                       l_ptr->mtu = msg_max_pkt(msg);
 
-               /* Synchronize broadcast link info, if not done previously */
-               if (!tipc_node_is_up(l_ptr->owner)) {
-                       l_ptr->owner->bclink.last_sent =
-                               l_ptr->owner->bclink.last_in =
-                               msg_last_bcast(msg);
-                       l_ptr->owner->bclink.oos_state = 0;
-               }
-
-               l_ptr->peer_session = msg_session(msg);
-               l_ptr->peer_bearer_id = msg_bearer_id(msg);
-
-               if (msg_type(msg) == ACTIVATE_MSG)
-                       link_state_event(l_ptr, ACTIVATE_MSG);
-               break;
-       case STATE_MSG:
+       case ACTIVATE_MSG:
 
-               msg_tol = msg_link_tolerance(msg);
-               if (msg_tol)
-                       link_set_supervision_props(l_ptr, msg_tol);
-
-               if (msg_linkprio(msg) &&
-                   (msg_linkprio(msg) != l_ptr->priority)) {
-                       pr_debug("%s<%s>, priority change %u->%u\n",
-                                link_rst_msg, l_ptr->name,
-                                l_ptr->priority, msg_linkprio(msg));
-                       l_ptr->priority = msg_linkprio(msg);
-                       tipc_link_reset(l_ptr); /* Enforce change to take effect */
+               /* Complete own link name with peer's interface name */
+               if_name =  strrchr(l->name, ':') + 1;
+               if (sizeof(l->name) - (if_name - l->name) <= TIPC_MAX_IF_NAME)
                        break;
-               }
-
-               /* Record reception; force mismatch at next timeout: */
-               l_ptr->silent_intv_cnt = 0;
-
-               link_state_event(l_ptr, TRAFFIC_MSG_EVT);
-               l_ptr->stats.recv_states++;
-               if (link_reset_unknown(l_ptr))
+               if (msg_data_sz(hdr) < TIPC_MAX_IF_NAME)
                        break;
+               strncpy(if_name, msg_data(hdr), TIPC_MAX_IF_NAME);
 
-               if (less_eq(l_ptr->rcv_nxt, msg_next_sent(msg)))
-                       rec_gap = mod(msg_next_sent(msg) - l_ptr->rcv_nxt);
-
-               if (msg_probe(msg))
-                       l_ptr->stats.recv_probes++;
+               /* Update own tolerance if peer indicates a non-zero value */
+               if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL))
+                       l->tolerance = peers_tol;
 
-               /* Protocol message before retransmits, reduce loss risk */
-               if (l_ptr->owner->bclink.recv_permitted)
-                       tipc_bclink_update_link_state(l_ptr->owner,
-                                                     msg_last_bcast(msg));
+               /* Update own priority if peer's priority is higher */
+               if (in_range(peers_prio, l->priority + 1, TIPC_MAX_LINK_PRI))
+                       l->priority = peers_prio;
 
-               if (rec_gap || (msg_probe(msg))) {
-                       tipc_link_proto_xmit(l_ptr, STATE_MSG, 0,
-                                            rec_gap, 0, 0);
-               }
-               if (msg_seq_gap(msg)) {
-                       l_ptr->stats.recv_nacks++;
-                       tipc_link_retransmit(l_ptr, skb_peek(&l_ptr->transmq),
-                                            msg_seq_gap(msg));
+               if (msg_type(hdr) == RESET_MSG) {
+                       rc |= tipc_link_fsm_evt(l, LINK_PEER_RESET_EVT);
+               } else if (!link_is_up(l)) {
+                       tipc_link_fsm_evt(l, LINK_PEER_RESET_EVT);
+                       rc |= tipc_link_fsm_evt(l, LINK_ESTABLISH_EVT);
                }
+               l->peer_session = msg_session(hdr);
+               l->peer_bearer_id = msg_bearer_id(hdr);
+               if (l->mtu > msg_max_pkt(hdr))
+                       l->mtu = msg_max_pkt(hdr);
                break;
-       }
-exit:
-       kfree_skb(buf);
-}
-
-
-/* tipc_link_tunnel_xmit(): Tunnel one packet via a link belonging to
- * a different bearer. Owner node is locked.
- */
-static void tipc_link_tunnel_xmit(struct tipc_link *l_ptr,
-                                 struct tipc_msg *tunnel_hdr,
-                                 struct tipc_msg *msg,
-                                 u32 selector)
-{
-       struct tipc_link *tunnel;
-       struct sk_buff *skb;
-       u32 length = msg_size(msg);
-
-       tunnel = l_ptr->owner->active_links[selector & 1];
-       if (!tipc_link_is_up(tunnel)) {
-               pr_warn("%stunnel link no longer available\n", link_co_err);
-               return;
-       }
-       msg_set_size(tunnel_hdr, length + INT_H_SIZE);
-       skb = tipc_buf_acquire(length + INT_H_SIZE);
-       if (!skb) {
-               pr_warn("%sunable to send tunnel msg\n", link_co_err);
-               return;
-       }
-       skb_copy_to_linear_data(skb, tunnel_hdr, INT_H_SIZE);
-       skb_copy_to_linear_data_offset(skb, INT_H_SIZE, msg, length);
-       __tipc_link_xmit_skb(tunnel, skb);
-}
-
 
-/* tipc_link_failover_send_queue(): A link has gone down, but a second
- * link is still active. We can do failover. Tunnel the failing link's
- * whole send queue via the remaining link. This way, we don't lose
- * any packets, and sequence order is preserved for subsequent traffic
- * sent over the remaining link. Owner node is locked.
- */
-void tipc_link_failover_send_queue(struct tipc_link *l_ptr)
-{
-       int msgcount;
-       struct tipc_link *tunnel = l_ptr->owner->active_links[0];
-       struct tipc_msg tunnel_hdr;
-       struct sk_buff *skb;
-       int split_bundles;
-
-       if (!tunnel)
-               return;
-
-       tipc_msg_init(link_own_addr(l_ptr), &tunnel_hdr, TUNNEL_PROTOCOL,
-                     FAILOVER_MSG, INT_H_SIZE, l_ptr->addr);
-
-       skb_queue_walk(&l_ptr->backlogq, skb) {
-               msg_set_seqno(buf_msg(skb), l_ptr->snd_nxt);
-               l_ptr->snd_nxt = mod(l_ptr->snd_nxt + 1);
-       }
-       skb_queue_splice_tail_init(&l_ptr->backlogq, &l_ptr->transmq);
-       tipc_link_purge_backlog(l_ptr);
-       msgcount = skb_queue_len(&l_ptr->transmq);
-       msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
-       msg_set_msgcnt(&tunnel_hdr, msgcount);
-
-       if (skb_queue_empty(&l_ptr->transmq)) {
-               skb = tipc_buf_acquire(INT_H_SIZE);
-               if (skb) {
-                       skb_copy_to_linear_data(skb, &tunnel_hdr, INT_H_SIZE);
-                       msg_set_size(&tunnel_hdr, INT_H_SIZE);
-                       __tipc_link_xmit_skb(tunnel, skb);
-               } else {
-                       pr_warn("%sunable to send changeover msg\n",
-                               link_co_err);
-               }
-               return;
-       }
-
-       split_bundles = (l_ptr->owner->active_links[0] !=
-                        l_ptr->owner->active_links[1]);
-
-       skb_queue_walk(&l_ptr->transmq, skb) {
-               struct tipc_msg *msg = buf_msg(skb);
+       case STATE_MSG:
 
-               if ((msg_user(msg) == MSG_BUNDLER) && split_bundles) {
-                       struct tipc_msg *m = msg_get_wrapped(msg);
-                       unchar *pos = (unchar *)m;
+               /* Update own tolerance if peer indicates a non-zero value */
+               if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL))
+                       l->tolerance = peers_tol;
 
-                       msgcount = msg_msgcnt(msg);
-                       while (msgcount--) {
-                               msg_set_seqno(m, msg_seqno(msg));
-                               tipc_link_tunnel_xmit(l_ptr, &tunnel_hdr, m,
-                                                     msg_link_selector(m));
-                               pos += align(msg_size(m));
-                               m = (struct tipc_msg *)pos;
-                       }
-               } else {
-                       tipc_link_tunnel_xmit(l_ptr, &tunnel_hdr, msg,
-                                             msg_link_selector(msg));
-               }
-       }
-}
-
-/* tipc_link_dup_queue_xmit(): A second link has become active. Tunnel a
- * duplicate of the first link's send queue via the new link. This way, we
- * are guaranteed that currently queued packets from a socket are delivered
- * before future traffic from the same socket, even if this is using the
- * new link. The last arriving copy of each duplicate packet is dropped at
- * the receiving end by the regular protocol check, so packet cardinality
- * and sequence order is preserved per sender/receiver socket pair.
- * Owner node is locked.
- */
-void tipc_link_dup_queue_xmit(struct tipc_link *link,
-                             struct tipc_link *tnl)
-{
-       struct sk_buff *skb;
-       struct tipc_msg tnl_hdr;
-       struct sk_buff_head *queue = &link->transmq;
-       int mcnt;
-       u16 seqno;
-
-       tipc_msg_init(link_own_addr(link), &tnl_hdr, TUNNEL_PROTOCOL,
-                     SYNCH_MSG, INT_H_SIZE, link->addr);
-       mcnt = skb_queue_len(&link->transmq) + skb_queue_len(&link->backlogq);
-       msg_set_msgcnt(&tnl_hdr, mcnt);
-       msg_set_bearer_id(&tnl_hdr, link->peer_bearer_id);
-
-tunnel_queue:
-       skb_queue_walk(queue, skb) {
-               struct sk_buff *outskb;
-               struct tipc_msg *msg = buf_msg(skb);
-               u32 len = msg_size(msg);
+               l->silent_intv_cnt = 0;
+               l->stats.recv_states++;
+               if (msg_probe(hdr))
+                       l->stats.recv_probes++;
+               rc = tipc_link_fsm_evt(l, LINK_ESTABLISH_EVT);
+               if (!link_is_up(l))
+                       break;
 
-               msg_set_ack(msg, mod(link->rcv_nxt - 1));
-               msg_set_bcast_ack(msg, link->owner->bclink.last_in);
-               msg_set_size(&tnl_hdr, len + INT_H_SIZE);
-               outskb = tipc_buf_acquire(len + INT_H_SIZE);
-               if (outskb == NULL) {
-                       pr_warn("%sunable to send duplicate msg\n",
-                               link_co_err);
-                       return;
+               /* Send NACK if peer has sent pkts we haven't received yet */
+               if (more(peers_snd_nxt, rcv_nxt) && !tipc_link_is_synching(l))
+                       rcvgap = peers_snd_nxt - l->rcv_nxt;
+               if (rcvgap || (msg_probe(hdr)))
+                       tipc_link_build_proto_msg(l, STATE_MSG, 0, rcvgap,
+                                                 0, 0, xmitq);
+               tipc_link_release_pkts(l, msg_ack(hdr));
+
+               /* If NACK, retransmit will now start at right position */
+               if (nacked_gap) {
+                       rc = tipc_link_retransm(l, nacked_gap, xmitq);
+                       l->stats.recv_nacks++;
                }
-               skb_copy_to_linear_data(outskb, &tnl_hdr, INT_H_SIZE);
-               skb_copy_to_linear_data_offset(outskb, INT_H_SIZE,
-                                              skb->data, len);
-               __tipc_link_xmit_skb(tnl, outskb);
-               if (!tipc_link_is_up(link))
-                       return;
-       }
-       if (queue == &link->backlogq)
-               return;
-       seqno = link->snd_nxt;
-       skb_queue_walk(&link->backlogq, skb) {
-               msg_set_seqno(buf_msg(skb), seqno);
-               seqno = mod(seqno + 1);
-       }
-       queue = &link->backlogq;
-       goto tunnel_queue;
-}
 
-/*  tipc_link_failover_rcv(): Receive a tunnelled FAILOVER_MSG packet
- *  Owner node is locked.
- */
-static bool tipc_link_failover_rcv(struct tipc_link *link,
-                                  struct sk_buff **skb)
-{
-       struct tipc_msg *msg = buf_msg(*skb);
-       struct sk_buff *iskb = NULL;
-       struct tipc_link *pl = NULL;
-       int bearer_id = msg_bearer_id(msg);
-       int pos = 0;
-
-       if (msg_type(msg) != FAILOVER_MSG) {
-               pr_warn("%sunknown tunnel pkt received\n", link_co_err);
-               goto exit;
-       }
-       if (bearer_id >= MAX_BEARERS)
-               goto exit;
-
-       if (bearer_id == link->bearer_id)
-               goto exit;
-
-       pl = link->owner->links[bearer_id];
-       if (pl && tipc_link_is_up(pl))
-               tipc_link_reset(pl);
-
-       if (link->failover_pkts == FIRST_FAILOVER)
-               link->failover_pkts = msg_msgcnt(msg);
-
-       /* Should we expect an inner packet? */
-       if (!link->failover_pkts)
-               goto exit;
-
-       if (!tipc_msg_extract(*skb, &iskb, &pos)) {
-               pr_warn("%sno inner failover pkt\n", link_co_err);
-               *skb = NULL;
-               goto exit;
-       }
-       link->failover_pkts--;
-       *skb = NULL;
-
-       /* Was this packet already delivered? */
-       if (less(buf_seqno(iskb), link->failover_checkpt)) {
-               kfree_skb(iskb);
-               iskb = NULL;
-               goto exit;
-       }
-       if (msg_user(buf_msg(iskb)) == MSG_FRAGMENTER) {
-               link->stats.recv_fragments++;
-               tipc_buf_append(&link->failover_skb, &iskb);
+               tipc_link_advance_backlog(l, xmitq);
+               if (unlikely(!skb_queue_empty(&l->wakeupq)))
+                       link_prepare_wakeup(l);
        }
 exit:
-       if (!link->failover_pkts && pl)
-               pl->flags &= ~LINK_FAILINGOVER;
-       kfree_skb(*skb);
-       *skb = iskb;
-       return *skb;
-}
-
-static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tol)
-{
-       unsigned long intv = ((tol / 4) > 500) ? 500 : tol / 4;
-
-       if ((tol < TIPC_MIN_LINK_TOL) || (tol > TIPC_MAX_LINK_TOL))
-               return;
-
-       l_ptr->tolerance = tol;
-       l_ptr->keepalive_intv = msecs_to_jiffies(intv);
-       l_ptr->abort_limit = tol / (jiffies_to_msecs(l_ptr->keepalive_intv));
+       kfree_skb(skb);
+       return rc;
 }
 
 void tipc_link_set_queue_limits(struct tipc_link *l, u32 win)
@@ -1743,7 +1450,7 @@ static struct tipc_node *tipc_link_find_owner(struct net *net,
        list_for_each_entry_rcu(n_ptr, &tn->node_list, list) {
                tipc_node_lock(n_ptr);
                for (i = 0; i < MAX_BEARERS; i++) {
-                       l_ptr = n_ptr->links[i];
+                       l_ptr = n_ptr->links[i].link;
                        if (l_ptr && !strcmp(l_ptr->name, link_name)) {
                                *bearer_id = i;
                                found_node = n_ptr;
@@ -1770,27 +1477,16 @@ static void link_reset_statistics(struct tipc_link *l_ptr)
        l_ptr->stats.recv_info = l_ptr->rcv_nxt;
 }
 
-static void link_print(struct tipc_link *l_ptr, const char *str)
+static void link_print(struct tipc_link *l, const char *str)
 {
-       struct tipc_net *tn = net_generic(l_ptr->owner->net, tipc_net_id);
-       struct tipc_bearer *b_ptr;
-
-       rcu_read_lock();
-       b_ptr = rcu_dereference_rtnl(tn->bearer_list[l_ptr->bearer_id]);
-       if (b_ptr)
-               pr_info("%s Link %x<%s>:", str, l_ptr->addr, b_ptr->name);
-       rcu_read_unlock();
-
-       if (link_working_unknown(l_ptr))
-               pr_cont(":WU\n");
-       else if (link_reset_reset(l_ptr))
-               pr_cont(":RR\n");
-       else if (link_reset_unknown(l_ptr))
-               pr_cont(":RU\n");
-       else if (link_working_working(l_ptr))
-               pr_cont(":WW\n");
-       else
-               pr_cont("\n");
+       struct sk_buff *hskb = skb_peek(&l->transmq);
+       u16 head = hskb ? msg_seqno(buf_msg(hskb)) : l->snd_nxt;
+       u16 tail = l->snd_nxt - 1;
+
+       pr_info("%s Link <%s> state %x\n", str, l->name, l->state);
+       pr_info("XMTQ: %u [%u-%u], BKLGQ: %u, SNDNX: %u, RCVNX: %u\n",
+               skb_queue_len(&l->transmq), head, tail,
+               skb_queue_len(&l->backlogq), l->snd_nxt, l->rcv_nxt);
 }
 
 /* Parse and validate nested (link) properties valid for media, bearer and link
@@ -1865,7 +1561,7 @@ int tipc_nl_link_set(struct sk_buff *skb, struct genl_info *info)
 
        tipc_node_lock(node);
 
-       link = node->links[bearer_id];
+       link = node->links[bearer_id].link;
        if (!link) {
                res = -EINVAL;
                goto out;
@@ -1885,7 +1581,7 @@ int tipc_nl_link_set(struct sk_buff *skb, struct genl_info *info)
                        u32 tol;
 
                        tol = nla_get_u32(props[TIPC_NLA_PROP_TOL]);
-                       link_set_supervision_props(link, tol);
+                       link->tolerance = tol;
                        tipc_link_proto_xmit(link, STATE_MSG, 0, 0, tol, 0);
                }
                if (props[TIPC_NLA_PROP_PRIO]) {
@@ -2055,10 +1751,11 @@ static int __tipc_nl_add_node_links(struct net *net, struct tipc_nl_msg *msg,
        for (i = *prev_link; i < MAX_BEARERS; i++) {
                *prev_link = i;
 
-               if (!node->links[i])
+               if (!node->links[i].link)
                        continue;
 
-               err = __tipc_nl_add_link(net, msg, node->links[i], NLM_F_MULTI);
+               err = __tipc_nl_add_link(net, msg,
+                                        node->links[i].link, NLM_F_MULTI);
                if (err)
                        return err;
        }
@@ -2172,7 +1869,7 @@ int tipc_nl_link_get(struct sk_buff *skb, struct genl_info *info)
                        return -EINVAL;
 
                tipc_node_lock(node);
-               link = node->links[bearer_id];
+               link = node->links[bearer_id].link;
                if (!link) {
                        tipc_node_unlock(node);
                        nlmsg_free(msg.skb);
@@ -2227,7 +1924,7 @@ int tipc_nl_link_reset_stats(struct sk_buff *skb, struct genl_info *info)
 
        tipc_node_lock(node);
 
-       link = node->links[bearer_id];
+       link = node->links[bearer_id].link;
        if (!link) {
                tipc_node_unlock(node);
                return -EINVAL;