1 files changed, 258 insertions, 301 deletions

diff --git a/net/tipc/link.c b/net/tipc/link.c
index 14f09b3cb87c..514466efc25c 100644
--- a/net/tipc/link.c
+++ b/net/tipc/link.c
@@ -1,7 +1,7 @@
 /*
  * net/tipc/link.c: TIPC link code
  *
- * Copyright (c) 1996-2007, 2012-2014, Ericsson AB
+ * Copyright (c) 1996-2007, 2012-2015, Ericsson AB
  * Copyright (c) 2004-2007, 2010-2013, Wind River Systems
  * All rights reserved.
  *
@@ -35,6 +35,7 @@
  */
 
 #include "core.h"
+#include "subscr.h"
 #include "link.h"
 #include "bcast.h"
 #include "socket.h"
@@ -138,6 +139,13 @@ static void tipc_link_put(struct tipc_link *l_ptr)
 	kref_put(&l_ptr->ref, tipc_link_release);
 }
 
+static struct tipc_link *tipc_parallel_link(struct tipc_link *l)
+{
+	if (l->owner->active_links[0] != l)
+		return l->owner->active_links[0];
+	return l->owner->active_links[1];
+}
+
 static void link_init_max_pkt(struct tipc_link *l_ptr)
 {
 	struct tipc_node *node = l_ptr->owner;
@@ -194,10 +202,10 @@ static void link_timeout(unsigned long data)
 	tipc_node_lock(l_ptr->owner);
 
 	/* update counters used in statistical profiling of send traffic */
-	l_ptr->stats.accu_queue_sz += skb_queue_len(&l_ptr->outqueue);
+	l_ptr->stats.accu_queue_sz += skb_queue_len(&l_ptr->transmq);
 	l_ptr->stats.queue_sz_counts++;
 
-	skb = skb_peek(&l_ptr->outqueue);
+	skb = skb_peek(&l_ptr->transmq);
 	if (skb) {
 		struct tipc_msg *msg = buf_msg(skb);
 		u32 length = msg_size(msg);
@@ -229,7 +237,7 @@ static void link_timeout(unsigned long data)
 	/* do all other link processing performed on a periodic basis */
 	link_state_event(l_ptr, TIMEOUT_EVT);
 
-	if (l_ptr->next_out)
+	if (skb_queue_len(&l_ptr->backlogq))
 		tipc_link_push_packets(l_ptr);
 
 	tipc_node_unlock(l_ptr->owner);
@@ -305,16 +313,14 @@ struct tipc_link *tipc_link_create(struct tipc_node *n_ptr,
 	msg_set_session(msg, (tn->random & 0xffff));
 	msg_set_bearer_id(msg, b_ptr->identity);
 	strcpy((char *)msg_data(msg), if_name);
-
-	l_ptr->priority = b_ptr->priority;
-	tipc_link_set_queue_limits(l_ptr, b_ptr->window);
-
 	l_ptr->net_plane = b_ptr->net_plane;
 	link_init_max_pkt(l_ptr);
-
+	l_ptr->priority = b_ptr->priority;
+	tipc_link_set_queue_limits(l_ptr, b_ptr->window);
 	l_ptr->next_out_no = 1;
-	__skb_queue_head_init(&l_ptr->outqueue);
-	__skb_queue_head_init(&l_ptr->deferred_queue);
+	__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);
@@ -344,6 +350,7 @@ 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;
+	bool del_link;
 
 	rcu_read_lock();
 	list_for_each_entry_rcu(node, &tn->node_list, list) {
@@ -353,12 +360,13 @@ void tipc_link_delete_list(struct net *net, unsigned int bearer_id,
 			tipc_node_unlock(node);
 			continue;
 		}
+		del_link = !tipc_link_is_up(link) && !link->exp_msg_count;
 		tipc_link_reset(link);
 		if (del_timer(&link->timer))
 			tipc_link_put(link);
 		link->flags |= LINK_STOPPED;
 		/* Delete link now, or when failover is finished: */
-		if (shutting_down || !tipc_node_is_up(node))
+		if (shutting_down || !tipc_node_is_up(node) || del_link)
 			tipc_link_delete(link);
 		tipc_node_unlock(node);
 	}
@@ -366,28 +374,43 @@ void tipc_link_delete_list(struct net *net, unsigned int bearer_id,
 }
 
 /**
- * link_schedule_user - schedule user for wakeup after congestion
+ * link_schedule_user - schedule a message sender for wakeup after congestion
  * @link: congested link
- * @oport: sending port
- * @chain_sz: size of buffer chain that was attempted sent
- * @imp: importance of message attempted sent
+ * @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
  */
-static bool link_schedule_user(struct tipc_link *link, u32 oport,
-			       uint chain_sz, uint imp)
+static int link_schedule_user(struct tipc_link *link, struct sk_buff_head *list)
 {
-	struct sk_buff *buf;
+	struct tipc_msg *msg = buf_msg(skb_peek(list));
+	int imp = msg_importance(msg);
+	u32 oport = msg_origport(msg);
+	u32 addr = link_own_addr(link);
+	struct sk_buff *skb;
 
-	buf = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
-			      link_own_addr(link), link_own_addr(link),
-			      oport, 0, 0);
-	if (!buf)
-		return false;
-	TIPC_SKB_CB(buf)->chain_sz = chain_sz;
-	TIPC_SKB_CB(buf)->chain_imp = imp;
-	skb_queue_tail(&link->wakeupq, buf);
+	/* 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;
+	}
+	/* Non-blocking sender: */
+	if (TIPC_SKB_CB(skb_peek(list))->wakeup_pending)
+		return -ELINKCONG;
+
+	/* Create and schedule wakeup pseudo message */
+	skb = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
+			      addr, addr, oport, 0, 0);
+	if (!skb)
+		goto err;
+	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 true;
+	return -ELINKCONG;
+err:
+	__skb_queue_purge(list);
+	return -ENOBUFS;
 }
 
 /**
@@ -396,19 +419,22 @@ static bool link_schedule_user(struct tipc_link *link, u32 oport,
  * Move a number of waiting users, as permitted by available space in
  * the send queue, from link wait queue to node wait queue for wakeup
  */
-void link_prepare_wakeup(struct tipc_link *link)
+void link_prepare_wakeup(struct tipc_link *l)
 {
-	uint pend_qsz = skb_queue_len(&link->outqueue);
+	int pnd[TIPC_SYSTEM_IMPORTANCE + 1] = {0,};
+	int imp, lim;
 	struct sk_buff *skb, *tmp;
 
-	skb_queue_walk_safe(&link->wakeupq, skb, tmp) {
-		if (pend_qsz >= link->queue_limit[TIPC_SKB_CB(skb)->chain_imp])
+	skb_queue_walk_safe(&l->wakeupq, skb, tmp) {
+		imp = TIPC_SKB_CB(skb)->chain_imp;
+		lim = l->window + l->backlog[imp].limit;
+		pnd[imp] += TIPC_SKB_CB(skb)->chain_sz;
+		if ((pnd[imp] + l->backlog[imp].len) >= lim)
 			break;
-		pend_qsz += TIPC_SKB_CB(skb)->chain_sz;
-		skb_unlink(skb, &link->wakeupq);
-		skb_queue_tail(&link->inputq, skb);
-		link->owner->inputq = &link->inputq;
-		link->owner->action_flags |= TIPC_MSG_EVT;
+		skb_unlink(skb, &l->wakeupq);
+		skb_queue_tail(&l->inputq, skb);
+		l->owner->inputq = &l->inputq;
+		l->owner->action_flags |= TIPC_MSG_EVT;
 	}
 }
 
@@ -422,14 +448,25 @@ void tipc_link_reset_fragments(struct tipc_link *l_ptr)
 	l_ptr->reasm_buf = NULL;
 }
 
+static void tipc_link_purge_backlog(struct tipc_link *l)
+{
+	__skb_queue_purge(&l->backlogq);
+	l->backlog[TIPC_LOW_IMPORTANCE].len = 0;
+	l->backlog[TIPC_MEDIUM_IMPORTANCE].len = 0;
+	l->backlog[TIPC_HIGH_IMPORTANCE].len = 0;
+	l->backlog[TIPC_CRITICAL_IMPORTANCE].len = 0;
+	l->backlog[TIPC_SYSTEM_IMPORTANCE].len = 0;
+}
+
 /**
  * tipc_link_purge_queues - purge all pkt queues associated with link
  * @l_ptr: pointer to link
  */
 void tipc_link_purge_queues(struct tipc_link *l_ptr)
 {
-	__skb_queue_purge(&l_ptr->deferred_queue);
-	__skb_queue_purge(&l_ptr->outqueue);
+	__skb_queue_purge(&l_ptr->deferdq);
+	__skb_queue_purge(&l_ptr->transmq);
+	tipc_link_purge_backlog(l_ptr);
 	tipc_link_reset_fragments(l_ptr);
 }
 
@@ -462,15 +499,15 @@ void tipc_link_reset(struct tipc_link *l_ptr)
 	}
 
 	/* Clean up all queues, except inputq: */
-	__skb_queue_purge(&l_ptr->outqueue);
-	__skb_queue_purge(&l_ptr->deferred_queue);
+	__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;
-	l_ptr->next_out = NULL;
-	l_ptr->unacked_window = 0;
+	tipc_link_purge_backlog(l_ptr);
+	l_ptr->rcv_unacked = 0;
 	l_ptr->checkpoint = 1;
 	l_ptr->next_out_no = 1;
 	l_ptr->fsm_msg_cnt = 0;
@@ -693,101 +730,65 @@ static void link_state_event(struct tipc_link *l_ptr, unsigned int event)
 	}
 }
 
-/* tipc_link_cong: determine return value and how to treat the
- * sent buffer during link congestion.
- * - For plain, errorless user data messages we keep the buffer and
- *   return -ELINKONG.
- * - For all other messages we discard the buffer and return -EHOSTUNREACH
- * - For TIPC internal messages we also reset the link
- */
-static int tipc_link_cong(struct tipc_link *link, struct sk_buff_head *list)
-{
-	struct sk_buff *skb = skb_peek(list);
-	struct tipc_msg *msg = buf_msg(skb);
-	uint imp = tipc_msg_tot_importance(msg);
-	u32 oport = msg_tot_origport(msg);
-
-	if (unlikely(imp > TIPC_CRITICAL_IMPORTANCE)) {
-		pr_warn("%s<%s>, send queue full", link_rst_msg, link->name);
-		tipc_link_reset(link);
-		goto drop;
-	}
-	if (unlikely(msg_errcode(msg)))
-		goto drop;
-	if (unlikely(msg_reroute_cnt(msg)))
-		goto drop;
-	if (TIPC_SKB_CB(skb)->wakeup_pending)
-		return -ELINKCONG;
-	if (link_schedule_user(link, oport, skb_queue_len(list), imp))
-		return -ELINKCONG;
-drop:
-	__skb_queue_purge(list);
-	return -EHOSTUNREACH;
-}
-
 /**
  * __tipc_link_xmit(): same as tipc_link_xmit, but destlink is known & locked
  * @link: link to use
  * @list: chain of buffers containing message
  *
- * Consumes the buffer chain, except when returning -ELINKCONG
- * Returns 0 if success, otherwise errno: -ELINKCONG, -EMSGSIZE (plain socket
- * user data messages) or -EHOSTUNREACH (all other messages/senders)
- * Only the socket functions tipc_send_stream() and tipc_send_packet() need
- * to act on the return value, since they may need to do more send attempts.
+ * 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 tipc_link *link,
 		     struct sk_buff_head *list)
 {
 	struct tipc_msg *msg = buf_msg(skb_peek(list));
-	uint psz = msg_size(msg);
-	uint sndlim = link->queue_limit[0];
-	uint imp = tipc_msg_tot_importance(msg);
+	unsigned int maxwin = link->window;
+	unsigned int imp = msg_importance(msg);
 	uint mtu = link->max_pkt;
 	uint ack = mod(link->next_in_no - 1);
 	uint seqno = link->next_out_no;
 	uint bc_last_in = link->owner->bclink.last_in;
 	struct tipc_media_addr *addr = &link->media_addr;
-	struct sk_buff_head *outqueue = &link->outqueue;
+	struct sk_buff_head *transmq = &link->transmq;
+	struct sk_buff_head *backlogq = &link->backlogq;
 	struct sk_buff *skb, *tmp;
 
-	/* Match queue limits against msg importance: */
-	if (unlikely(skb_queue_len(outqueue) >= link->queue_limit[imp]))
-		return tipc_link_cong(link, list);
+	/* Match backlog limit against msg importance: */
+	if (unlikely(link->backlog[imp].len >= link->backlog[imp].limit))
+		return link_schedule_user(link, list);
 
-	/* Has valid packet limit been used ? */
-	if (unlikely(psz > mtu)) {
+	if (unlikely(msg_size(msg) > mtu)) {
 		__skb_queue_purge(list);
 		return -EMSGSIZE;
 	}
-
-	/* Prepare each packet for sending, and add to outqueue: */
+	/* Prepare each packet for sending, and add to relevant queue: */
 	skb_queue_walk_safe(list, skb, tmp) {
 		__skb_unlink(skb, list);
 		msg = buf_msg(skb);
-		msg_set_word(msg, 2, ((ack << 16) | mod(seqno)));
+		msg_set_seqno(msg, seqno);
+		msg_set_ack(msg, ack);
 		msg_set_bcast_ack(msg, bc_last_in);
 
-		if (skb_queue_len(outqueue) < sndlim) {
-			__skb_queue_tail(outqueue, skb);
-			tipc_bearer_send(net, link->bearer_id,
-					 skb, addr);
-			link->next_out = NULL;
-			link->unacked_window = 0;
-		} else if (tipc_msg_bundle(outqueue, skb, mtu)) {
+		if (likely(skb_queue_len(transmq) < maxwin)) {
+			__skb_queue_tail(transmq, skb);
+			tipc_bearer_send(net, link->bearer_id, skb, addr);
+			link->rcv_unacked = 0;
+			seqno++;
+			continue;
+		}
+		if (tipc_msg_bundle(skb_peek_tail(backlogq), skb, mtu)) {
 			link->stats.sent_bundled++;
 			continue;
-		} else if (tipc_msg_make_bundle(outqueue, skb, mtu,
-						link->addr)) {
+		}
+		if (tipc_msg_make_bundle(&skb, mtu, link->addr)) {
 			link->stats.sent_bundled++;
 			link->stats.sent_bundles++;
-			if (!link->next_out)
-				link->next_out = skb_peek_tail(outqueue);
-		} else {
-			__skb_queue_tail(outqueue, skb);
-			if (!link->next_out)
-				link->next_out = skb;
+			imp = msg_importance(buf_msg(skb));
 		}
+		__skb_queue_tail(backlogq, skb);
+		link->backlog[imp].len++;
 		seqno++;
 	}
 	link->next_out_no = seqno;
@@ -808,13 +809,25 @@ static int __tipc_link_xmit_skb(struct tipc_link *link, struct sk_buff *skb)
 	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);
-	return tipc_link_xmit(net, &head, dnode, selector);
+	rc = tipc_link_xmit(net, &head, dnode, selector);
+	if (rc == -ELINKCONG)
+		kfree_skb(skb);
+	return 0;
 }
 
 /**
@@ -841,12 +854,15 @@ int tipc_link_xmit(struct net *net, struct sk_buff_head *list, u32 dnode,
 		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)))
-		return tipc_sk_rcv(net, list);
+	if (likely(in_own_node(net, dnode))) {
+		tipc_sk_rcv(net, list);
+		return 0;
+	}
 
 	__skb_queue_purge(list);
 	return rc;
@@ -893,14 +909,6 @@ static void tipc_link_sync_rcv(struct tipc_node *n, struct sk_buff *buf)
 	kfree_skb(buf);
 }
 
-struct sk_buff *tipc_skb_queue_next(const struct sk_buff_head *list,
-				    const struct sk_buff *skb)
-{
-	if (skb_queue_is_last(list, skb))
-		return NULL;
-	return skb->next;
-}
-
 /*
  * tipc_link_push_packets - push unsent packets to bearer
  *
@@ -909,30 +917,24 @@ struct sk_buff *tipc_skb_queue_next(const struct sk_buff_head *list,
  *
  * Called with node locked
  */
-void tipc_link_push_packets(struct tipc_link *l_ptr)
+void tipc_link_push_packets(struct tipc_link *link)
 {
-	struct sk_buff_head *outqueue = &l_ptr->outqueue;
-	struct sk_buff *skb = l_ptr->next_out;
+	struct sk_buff *skb;
 	struct tipc_msg *msg;
-	u32 next, first;
+	unsigned int ack = mod(link->next_in_no - 1);
 
-	skb_queue_walk_from(outqueue, skb) {
-		msg = buf_msg(skb);
-		next = msg_seqno(msg);
-		first = buf_seqno(skb_peek(outqueue));
-
-		if (mod(next - first) < l_ptr->queue_limit[0]) {
-			msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
-			msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
-			if (msg_user(msg) == MSG_BUNDLER)
-				TIPC_SKB_CB(skb)->bundling = false;
-			tipc_bearer_send(l_ptr->owner->net,
-					 l_ptr->bearer_id, skb,
-					 &l_ptr->media_addr);
-			l_ptr->next_out = tipc_skb_queue_next(outqueue, skb);
-		} else {
+	while (skb_queue_len(&link->transmq) < link->window) {
+		skb = __skb_dequeue(&link->backlogq);
+		if (!skb)
 			break;
-		}
+		msg = buf_msg(skb);
+		link->backlog[msg_importance(msg)].len--;
+		msg_set_ack(msg, ack);
+		msg_set_bcast_ack(msg, link->owner->bclink.last_in);
+		link->rcv_unacked = 0;
+		__skb_queue_tail(&link->transmq, skb);
+		tipc_bearer_send(link->owner->net, link->bearer_id,
+				 skb, &link->media_addr);
 	}
 }
 
@@ -979,7 +981,6 @@ static void link_retransmit_failure(struct tipc_link *l_ptr,
 			(unsigned long) TIPC_SKB_CB(buf)->handle);
 
 		n_ptr = tipc_bclink_retransmit_to(net);
-		tipc_node_lock(n_ptr);
 
 		tipc_addr_string_fill(addr_string, n_ptr->addr);
 		pr_info("Broadcast link info for %s\n", addr_string);
@@ -991,9 +992,7 @@ static void link_retransmit_failure(struct tipc_link *l_ptr,
 			n_ptr->bclink.oos_state,
 			n_ptr->bclink.last_sent);
 
-		tipc_node_unlock(n_ptr);
-
-		tipc_bclink_set_flags(net, TIPC_BCLINK_RESET);
+		n_ptr->action_flags |= TIPC_BCAST_RESET;
 		l_ptr->stale_count = 0;
 	}
 }
@@ -1019,8 +1018,8 @@ void tipc_link_retransmit(struct tipc_link *l_ptr, struct sk_buff *skb,
 		l_ptr->stale_count = 1;
 	}
 
-	skb_queue_walk_from(&l_ptr->outqueue, skb) {
-		if (!retransmits || skb == l_ptr->next_out)
+	skb_queue_walk_from(&l_ptr->transmq, skb) {
+		if (!retransmits)
 			break;
 		msg = buf_msg(skb);
 		msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
@@ -1032,72 +1031,43 @@ void tipc_link_retransmit(struct tipc_link *l_ptr, struct sk_buff *skb,
 	}
 }
 
-static void link_retrieve_defq(struct tipc_link *link,
-			       struct sk_buff_head *list)
-{
-	u32 seq_no;
-
-	if (skb_queue_empty(&link->deferred_queue))
-		return;
-
-	seq_no = buf_seqno(skb_peek(&link->deferred_queue));
-	if (seq_no == mod(link->next_in_no))
-		skb_queue_splice_tail_init(&link->deferred_queue, list);
-}
-
-/**
- * link_recv_buf_validate - validate basic format of received message
- *
- * This routine ensures a TIPC message has an acceptable header, and at least
- * as much data as the header indicates it should.  The routine also ensures
- * that the entire message header is stored in the main fragment of the message
- * buffer, to simplify future access to message header fields.
- *
- * Note: Having extra info present in the message header or data areas is OK.
- * TIPC will ignore the excess, under the assumption that it is optional info
- * introduced by a later release of the protocol.
+/* 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 int link_recv_buf_validate(struct sk_buff *buf)
+static bool link_synch(struct tipc_link *l)
 {
-	static u32 min_data_hdr_size[8] = {
-		SHORT_H_SIZE, MCAST_H_SIZE, NAMED_H_SIZE, BASIC_H_SIZE,
-		MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE
-		};
-
-	struct tipc_msg *msg;
-	u32 tipc_hdr[2];
-	u32 size;
-	u32 hdr_size;
-	u32 min_hdr_size;
-
-	/* If this packet comes from the defer queue, the skb has already
-	 * been validated
-	 */
-	if (unlikely(TIPC_SKB_CB(buf)->deferred))
-		return 1;
+	unsigned int post_synch;
+	struct tipc_link *pl;
 
-	if (unlikely(buf->len < MIN_H_SIZE))
-		return 0;
+	pl  = tipc_parallel_link(l);
+	if (pl == l)
+		goto synched;
 
-	msg = skb_header_pointer(buf, 0, sizeof(tipc_hdr), tipc_hdr);
-	if (msg == NULL)
-		return 0;
+	/* Was last pre-synch packet added to input queue ? */
+	if (less_eq(pl->next_in_no, l->synch_point))
+		return false;
 
-	if (unlikely(msg_version(msg) != TIPC_VERSION))
-		return 0;
+	/* Is it still in the input queue ? */
+	post_synch = mod(pl->next_in_no - l->synch_point) - 1;
+	if (skb_queue_len(&pl->inputq) > post_synch)
+		return false;
+synched:
+	l->flags &= ~LINK_SYNCHING;
+	return true;
+}
 
-	size = msg_size(msg);
-	hdr_size = msg_hdr_sz(msg);
-	min_hdr_size = msg_isdata(msg) ?
-		min_data_hdr_size[msg_type(msg)] : INT_H_SIZE;
+static void link_retrieve_defq(struct tipc_link *link,
+			       struct sk_buff_head *list)
+{
+	u32 seq_no;
 
-	if (unlikely((hdr_size < min_hdr_size) ||
-		     (size < hdr_size) ||
-		     (buf->len < size) ||
-		     (size - hdr_size > TIPC_MAX_USER_MSG_SIZE)))
-		return 0;
+	if (skb_queue_empty(&link->deferdq))
+		return;
 
-	return pskb_may_pull(buf, hdr_size);
+	seq_no = buf_seqno(skb_peek(&link->deferdq));
+	if (seq_no == mod(link->next_in_no))
+		skb_queue_splice_tail_init(&link->deferdq, list);
 }
 
 /**
@@ -1125,16 +1095,11 @@ void tipc_rcv(struct net *net, struct sk_buff *skb, struct tipc_bearer *b_ptr)
 
 	while ((skb = __skb_dequeue(&head))) {
 		/* Ensure message is well-formed */
-		if (unlikely(!link_recv_buf_validate(skb)))
-			goto discard;
-
-		/* Ensure message data is a single contiguous unit */
-		if (unlikely(skb_linearize(skb)))
+		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);
@@ -1152,8 +1117,8 @@ void tipc_rcv(struct net *net, struct sk_buff *skb, struct tipc_bearer *b_ptr)
 		n_ptr = tipc_node_find(net, msg_prevnode(msg));
 		if (unlikely(!n_ptr))
 			goto discard;
-		tipc_node_lock(n_ptr);
 
+		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))
@@ -1175,21 +1140,20 @@ void tipc_rcv(struct net *net, struct sk_buff *skb, struct tipc_bearer *b_ptr)
 		ackd = msg_ack(msg);
 
 		/* Release acked messages */
-		if (n_ptr->bclink.recv_permitted)
+		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->outqueue, skb1, tmp) {
-			if (skb1 == l_ptr->next_out ||
-			    more(buf_seqno(skb1), ackd))
+		skb_queue_walk_safe(&l_ptr->transmq, skb1, tmp) {
+			if (more(buf_seqno(skb1), ackd))
 				break;
-			 __skb_unlink(skb1, &l_ptr->outqueue);
+			 __skb_unlink(skb1, &l_ptr->transmq);
 			 kfree_skb(skb1);
 			 released = 1;
 		}
 
 		/* Try sending any messages link endpoint has pending */
-		if (unlikely(l_ptr->next_out))
+		if (unlikely(skb_queue_len(&l_ptr->backlogq)))
 			tipc_link_push_packets(l_ptr);
 
 		if (released && !skb_queue_empty(&l_ptr->wakeupq))
@@ -1223,11 +1187,18 @@ void tipc_rcv(struct net *net, struct sk_buff *skb, struct tipc_bearer *b_ptr)
 			skb = NULL;
 			goto unlock;
 		}
+		/* Synchronize with parallel link if applicable */
+		if (unlikely((l_ptr->flags & LINK_SYNCHING) && !msg_dup(msg))) {
+			link_handle_out_of_seq_msg(l_ptr, skb);
+			if (link_synch(l_ptr))
+				link_retrieve_defq(l_ptr, &head);
+			skb = NULL;
+			goto unlock;
+		}
 		l_ptr->next_in_no++;
-		if (unlikely(!skb_queue_empty(&l_ptr->deferred_queue)))
+		if (unlikely(!skb_queue_empty(&l_ptr->deferdq)))
 			link_retrieve_defq(l_ptr, &head);
-
-		if (unlikely(++l_ptr->unacked_window >= TIPC_MIN_LINK_WIN)) {
+		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, 0);
 		}
@@ -1235,6 +1206,7 @@ void tipc_rcv(struct net *net, struct sk_buff *skb, struct tipc_bearer *b_ptr)
 		skb = NULL;
 unlock:
 		tipc_node_unlock(n_ptr);
+		tipc_node_put(n_ptr);
 discard:
 		if (unlikely(skb))
 			kfree_skb(skb);
@@ -1299,6 +1271,10 @@ static void tipc_link_input(struct tipc_link *link, struct sk_buff *skb)
 
 	switch (msg_user(msg)) {
 	case CHANGEOVER_PROTOCOL:
+		if (msg_dup(msg)) {
+			link->flags |= LINK_SYNCHING;
+			link->synch_point = msg_seqno(msg_get_wrapped(msg));
+		}
 		if (!tipc_link_tunnel_rcv(node, &skb))
 			break;
 		if (msg_user(buf_msg(skb)) != MSG_BUNDLER) {
@@ -1394,10 +1370,9 @@ static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
 		return;
 	}
 
-	if (tipc_link_defer_pkt(&l_ptr->deferred_queue, buf)) {
+	if (tipc_link_defer_pkt(&l_ptr->deferdq, buf)) {
 		l_ptr->stats.deferred_recv++;
-		TIPC_SKB_CB(buf)->deferred = true;
-		if ((skb_queue_len(&l_ptr->deferred_queue) % 16) == 1)
+		if ((skb_queue_len(&l_ptr->deferdq) % TIPC_MIN_LINK_WIN) == 1)
 			tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
 	} else {
 		l_ptr->stats.duplicates++;
@@ -1434,11 +1409,11 @@ void tipc_link_proto_xmit(struct tipc_link *l_ptr, u32 msg_typ, int probe_msg,
 
 		if (!tipc_link_is_up(l_ptr))
 			return;
-		if (l_ptr->next_out)
-			next_sent = buf_seqno(l_ptr->next_out);
+		if (skb_queue_len(&l_ptr->backlogq))
+			next_sent = buf_seqno(skb_peek(&l_ptr->backlogq));
 		msg_set_next_sent(msg, next_sent);
-		if (!skb_queue_empty(&l_ptr->deferred_queue)) {
-			u32 rec = buf_seqno(skb_peek(&l_ptr->deferred_queue));
+		if (!skb_queue_empty(&l_ptr->deferdq)) {
+			u32 rec = buf_seqno(skb_peek(&l_ptr->deferdq));
 			gap = mod(rec - mod(l_ptr->next_in_no));
 		}
 		msg_set_seq_gap(msg, gap);
@@ -1490,10 +1465,9 @@ void tipc_link_proto_xmit(struct tipc_link *l_ptr, u32 msg_typ, int probe_msg,
 
 	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->unacked_window = 0;
+	l_ptr->rcv_unacked = 0;
 	kfree_skb(buf);
 }
 
@@ -1628,7 +1602,7 @@ static void tipc_link_proto_rcv(struct tipc_link *l_ptr,
 		}
 		if (msg_seq_gap(msg)) {
 			l_ptr->stats.recv_nacks++;
-			tipc_link_retransmit(l_ptr, skb_peek(&l_ptr->outqueue),
+			tipc_link_retransmit(l_ptr, skb_peek(&l_ptr->transmq),
 					     msg_seq_gap(msg));
 		}
 		break;
@@ -1675,7 +1649,7 @@ static void tipc_link_tunnel_xmit(struct tipc_link *l_ptr,
  */
 void tipc_link_failover_send_queue(struct tipc_link *l_ptr)
 {
-	u32 msgcount = skb_queue_len(&l_ptr->outqueue);
+	int msgcount;
 	struct tipc_link *tunnel = l_ptr->owner->active_links[0];
 	struct tipc_msg tunnel_hdr;
 	struct sk_buff *skb;
@@ -1686,10 +1660,13 @@ void tipc_link_failover_send_queue(struct tipc_link *l_ptr)
 
 	tipc_msg_init(link_own_addr(l_ptr), &tunnel_hdr, CHANGEOVER_PROTOCOL,
 		      ORIGINAL_MSG, INT_H_SIZE, l_ptr->addr);
+	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->outqueue)) {
+	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);
@@ -1705,7 +1682,7 @@ void tipc_link_failover_send_queue(struct tipc_link *l_ptr)
 	split_bundles = (l_ptr->owner->active_links[0] !=
 			 l_ptr->owner->active_links[1]);
 
-	skb_queue_walk(&l_ptr->outqueue, skb) {
+	skb_queue_walk(&l_ptr->transmq, skb) {
 		struct tipc_msg *msg = buf_msg(skb);
 
 		if ((msg_user(msg) == MSG_BUNDLER) && split_bundles) {
@@ -1736,80 +1713,66 @@ void tipc_link_failover_send_queue(struct tipc_link *l_ptr)
  * and sequence order is preserved per sender/receiver socket pair.
  * Owner node is locked.
  */
-void tipc_link_dup_queue_xmit(struct tipc_link *l_ptr,
-			      struct tipc_link *tunnel)
+void tipc_link_dup_queue_xmit(struct tipc_link *link,
+			      struct tipc_link *tnl)
 {
 	struct sk_buff *skb;
-	struct tipc_msg tunnel_hdr;
-
-	tipc_msg_init(link_own_addr(l_ptr), &tunnel_hdr, CHANGEOVER_PROTOCOL,
-		      DUPLICATE_MSG, INT_H_SIZE, l_ptr->addr);
-	msg_set_msgcnt(&tunnel_hdr, skb_queue_len(&l_ptr->outqueue));
-	msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
-	skb_queue_walk(&l_ptr->outqueue, skb) {
+	struct tipc_msg tnl_hdr;
+	struct sk_buff_head *queue = &link->transmq;
+	int mcnt;
+
+	tipc_msg_init(link_own_addr(link), &tnl_hdr, CHANGEOVER_PROTOCOL,
+		      DUPLICATE_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 length = msg_size(msg);
+		u32 len = msg_size(msg);
 
-		if (msg_user(msg) == MSG_BUNDLER)
-			msg_set_type(msg, CLOSED_MSG);
-		msg_set_ack(msg, mod(l_ptr->next_in_no - 1));	/* Update */
-		msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
-		msg_set_size(&tunnel_hdr, length + INT_H_SIZE);
-		outskb = tipc_buf_acquire(length + INT_H_SIZE);
+		msg_set_ack(msg, mod(link->next_in_no - 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;
 		}
-		skb_copy_to_linear_data(outskb, &tunnel_hdr, INT_H_SIZE);
-		skb_copy_to_linear_data_offset(outskb, INT_H_SIZE, skb->data,
-					       length);
-		__tipc_link_xmit_skb(tunnel, outskb);
-		if (!tipc_link_is_up(l_ptr))
+		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;
 	}
-}
-
-/**
- * buf_extract - extracts embedded TIPC message from another message
- * @skb: encapsulating message buffer
- * @from_pos: offset to extract from
- *
- * Returns a new message buffer containing an embedded message.  The
- * encapsulating buffer is left unchanged.
- */
-static struct sk_buff *buf_extract(struct sk_buff *skb, u32 from_pos)
-{
-	struct tipc_msg *msg = (struct tipc_msg *)(skb->data + from_pos);
-	u32 size = msg_size(msg);
-	struct sk_buff *eb;
-
-	eb = tipc_buf_acquire(size);
-	if (eb)
-		skb_copy_to_linear_data(eb, msg, size);
-	return eb;
+	if (queue == &link->backlogq)
+		return;
+	queue = &link->backlogq;
+	goto tunnel_queue;
 }
 
 /* tipc_link_dup_rcv(): Receive a tunnelled DUPLICATE_MSG packet.
  * Owner node is locked.
  */
-static void tipc_link_dup_rcv(struct tipc_link *l_ptr,
-			      struct sk_buff *t_buf)
+static void tipc_link_dup_rcv(struct tipc_link *link,
+			      struct sk_buff *skb)
 {
-	struct sk_buff *buf;
+	struct sk_buff *iskb;
+	int pos = 0;
 
-	if (!tipc_link_is_up(l_ptr))
+	if (!tipc_link_is_up(link))
 		return;
 
-	buf = buf_extract(t_buf, INT_H_SIZE);
-	if (buf == NULL) {
+	if (!tipc_msg_extract(skb, &iskb, &pos)) {
 		pr_warn("%sfailed to extract inner dup pkt\n", link_co_err);
 		return;
 	}
-
-	/* Add buffer to deferred queue, if applicable: */
-	link_handle_out_of_seq_msg(l_ptr, buf);
+	/* Append buffer to deferred queue, if applicable: */
+	link_handle_out_of_seq_msg(link, iskb);
 }
 
 /*  tipc_link_failover_rcv(): Receive a tunnelled ORIGINAL_MSG packet
@@ -1821,6 +1784,7 @@ static struct sk_buff *tipc_link_failover_rcv(struct tipc_link *l_ptr,
 	struct tipc_msg *t_msg = buf_msg(t_buf);
 	struct sk_buff *buf = NULL;
 	struct tipc_msg *msg;
+	int pos = 0;
 
 	if (tipc_link_is_up(l_ptr))
 		tipc_link_reset(l_ptr);
@@ -1832,8 +1796,7 @@ static struct sk_buff *tipc_link_failover_rcv(struct tipc_link *l_ptr,
 	/* Should there be an inner packet? */
 	if (l_ptr->exp_msg_count) {
 		l_ptr->exp_msg_count--;
-		buf = buf_extract(t_buf, INT_H_SIZE);
-		if (buf == NULL) {
+		if (!tipc_msg_extract(t_buf, &buf, &pos)) {
 			pr_warn("%sno inner failover pkt\n", link_co_err);
 			goto exit;
 		}
@@ -1901,23 +1864,16 @@ static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tol)
 	l_ptr->abort_limit = tol / (jiffies_to_msecs(l_ptr->cont_intv) / 4);
 }
 
-void tipc_link_set_queue_limits(struct tipc_link *l_ptr, u32 window)
+void tipc_link_set_queue_limits(struct tipc_link *l, u32 win)
 {
-	/* Data messages from this node, inclusive FIRST_FRAGM */
-	l_ptr->queue_limit[TIPC_LOW_IMPORTANCE] = window;
-	l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE] = (window / 3) * 4;
-	l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE] = (window / 3) * 5;
-	l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE] = (window / 3) * 6;
-	/* Transiting data messages,inclusive FIRST_FRAGM */
-	l_ptr->queue_limit[TIPC_LOW_IMPORTANCE + 4] = 300;
-	l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE + 4] = 600;
-	l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE + 4] = 900;
-	l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE + 4] = 1200;
-	l_ptr->queue_limit[CONN_MANAGER] = 1200;
-	l_ptr->queue_limit[CHANGEOVER_PROTOCOL] = 2500;
-	l_ptr->queue_limit[NAME_DISTRIBUTOR] = 3000;
-	/* FRAGMENT and LAST_FRAGMENT packets */
-	l_ptr->queue_limit[MSG_FRAGMENTER] = 4000;
+	int max_bulk = TIPC_MAX_PUBLICATIONS / (l->max_pkt / ITEM_SIZE);
+
+	l->window = win;
+	l->backlog[TIPC_LOW_IMPORTANCE].limit      = win / 2;
+	l->backlog[TIPC_MEDIUM_IMPORTANCE].limit   = win;
+	l->backlog[TIPC_HIGH_IMPORTANCE].limit     = win / 2 * 3;
+	l->backlog[TIPC_CRITICAL_IMPORTANCE].limit = win * 2;
+	l->backlog[TIPC_SYSTEM_IMPORTANCE].limit   = max_bulk;
 }
 
 /* tipc_link_find_owner - locate owner node of link by link's name
@@ -2216,7 +2172,7 @@ static int __tipc_nl_add_link(struct net *net, struct tipc_nl_msg *msg,
 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_TOL, link->tolerance))
 		goto prop_msg_full;
 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN,
-			link->queue_limit[TIPC_LOW_IMPORTANCE]))
+			link->window))
 		goto prop_msg_full;
 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
 		goto prop_msg_full;
@@ -2282,7 +2238,6 @@ int tipc_nl_link_dump(struct sk_buff *skb, struct netlink_callback *cb)
 	msg.seq = cb->nlh->nlmsg_seq;
 
 	rcu_read_lock();
-
 	if (prev_node) {
 		node = tipc_node_find(net, prev_node);
 		if (!node) {
@@ -2295,6 +2250,7 @@ int tipc_nl_link_dump(struct sk_buff *skb, struct netlink_callback *cb)
 			cb->prev_seq = 1;
 			goto out;
 		}
+		tipc_node_put(node);
 
 		list_for_each_entry_continue_rcu(node, &tn->node_list,
 						 list) {
@@ -2302,6 +2258,7 @@ int tipc_nl_link_dump(struct sk_buff *skb, struct netlink_callback *cb)
 			err = __tipc_nl_add_node_links(net, &msg, node,
 						       &prev_link);
 			tipc_node_unlock(node);
+			tipc_node_put(node);
 			if (err)
 				goto out;