xref: /openbmc/linux/net/tipc/bcast.c (revision ff6defa6)
1 /*
2  * net/tipc/bcast.c: TIPC broadcast code
3  *
4  * Copyright (c) 2004-2006, 2014, Ericsson AB
5  * Copyright (c) 2004, Intel Corporation.
6  * Copyright (c) 2005, 2010-2011, Wind River Systems
7  * All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions are met:
11  *
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. Neither the names of the copyright holders nor the names of its
18  *    contributors may be used to endorse or promote products derived from
19  *    this software without specific prior written permission.
20  *
21  * Alternatively, this software may be distributed under the terms of the
22  * GNU General Public License ("GPL") version 2 as published by the Free
23  * Software Foundation.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
26  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
29  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35  * POSSIBILITY OF SUCH DAMAGE.
36  */
37 
38 #include "core.h"
39 #include "link.h"
40 #include "socket.h"
41 #include "msg.h"
42 #include "bcast.h"
43 #include "name_distr.h"
44 
45 #define	MAX_PKT_DEFAULT_MCAST	1500	/* bcast link max packet size (fixed) */
46 #define	BCLINK_WIN_DEFAULT	20	/* bcast link window size (default) */
47 #define	BCBEARER		MAX_BEARERS
48 
49 /**
50  * struct tipc_bcbearer_pair - a pair of bearers used by broadcast link
51  * @primary: pointer to primary bearer
52  * @secondary: pointer to secondary bearer
53  *
54  * Bearers must have same priority and same set of reachable destinations
55  * to be paired.
56  */
57 
58 struct tipc_bcbearer_pair {
59 	struct tipc_bearer *primary;
60 	struct tipc_bearer *secondary;
61 };
62 
63 /**
64  * struct tipc_bcbearer - bearer used by broadcast link
65  * @bearer: (non-standard) broadcast bearer structure
66  * @media: (non-standard) broadcast media structure
67  * @bpairs: array of bearer pairs
68  * @bpairs_temp: temporary array of bearer pairs used by tipc_bcbearer_sort()
69  * @remains: temporary node map used by tipc_bcbearer_send()
70  * @remains_new: temporary node map used tipc_bcbearer_send()
71  *
72  * Note: The fields labelled "temporary" are incorporated into the bearer
73  * to avoid consuming potentially limited stack space through the use of
74  * large local variables within multicast routines.  Concurrent access is
75  * prevented through use of the spinlock "bclink_lock".
76  */
77 struct tipc_bcbearer {
78 	struct tipc_bearer bearer;
79 	struct tipc_media media;
80 	struct tipc_bcbearer_pair bpairs[MAX_BEARERS];
81 	struct tipc_bcbearer_pair bpairs_temp[TIPC_MAX_LINK_PRI + 1];
82 	struct tipc_node_map remains;
83 	struct tipc_node_map remains_new;
84 };
85 
86 /**
87  * struct tipc_bclink - link used for broadcast messages
88  * @lock: spinlock governing access to structure
89  * @link: (non-standard) broadcast link structure
90  * @node: (non-standard) node structure representing b'cast link's peer node
91  * @flags: represent bclink states
92  * @bcast_nodes: map of broadcast-capable nodes
93  * @retransmit_to: node that most recently requested a retransmit
94  *
95  * Handles sequence numbering, fragmentation, bundling, etc.
96  */
97 struct tipc_bclink {
98 	spinlock_t lock;
99 	struct tipc_link link;
100 	struct tipc_node node;
101 	unsigned int flags;
102 	struct tipc_node_map bcast_nodes;
103 	struct tipc_node *retransmit_to;
104 };
105 
106 static struct tipc_bcbearer *bcbearer;
107 static struct tipc_bclink *bclink;
108 static struct tipc_link *bcl;
109 
110 const char tipc_bclink_name[] = "broadcast-link";
111 
112 static void tipc_nmap_diff(struct tipc_node_map *nm_a,
113 			   struct tipc_node_map *nm_b,
114 			   struct tipc_node_map *nm_diff);
115 static void tipc_nmap_add(struct tipc_node_map *nm_ptr, u32 node);
116 static void tipc_nmap_remove(struct tipc_node_map *nm_ptr, u32 node);
117 
118 static void tipc_bclink_lock(void)
119 {
120 	spin_lock_bh(&bclink->lock);
121 }
122 
123 static void tipc_bclink_unlock(void)
124 {
125 	struct tipc_node *node = NULL;
126 
127 	if (likely(!bclink->flags)) {
128 		spin_unlock_bh(&bclink->lock);
129 		return;
130 	}
131 
132 	if (bclink->flags & TIPC_BCLINK_RESET) {
133 		bclink->flags &= ~TIPC_BCLINK_RESET;
134 		node = tipc_bclink_retransmit_to();
135 	}
136 	spin_unlock_bh(&bclink->lock);
137 
138 	if (node)
139 		tipc_link_reset_all(node);
140 }
141 
142 uint  tipc_bclink_get_mtu(void)
143 {
144 	return MAX_PKT_DEFAULT_MCAST;
145 }
146 
147 void tipc_bclink_set_flags(unsigned int flags)
148 {
149 	bclink->flags |= flags;
150 }
151 
152 static u32 bcbuf_acks(struct sk_buff *buf)
153 {
154 	return (u32)(unsigned long)TIPC_SKB_CB(buf)->handle;
155 }
156 
157 static void bcbuf_set_acks(struct sk_buff *buf, u32 acks)
158 {
159 	TIPC_SKB_CB(buf)->handle = (void *)(unsigned long)acks;
160 }
161 
162 static void bcbuf_decr_acks(struct sk_buff *buf)
163 {
164 	bcbuf_set_acks(buf, bcbuf_acks(buf) - 1);
165 }
166 
167 void tipc_bclink_add_node(u32 addr)
168 {
169 	tipc_bclink_lock();
170 	tipc_nmap_add(&bclink->bcast_nodes, addr);
171 	tipc_bclink_unlock();
172 }
173 
174 void tipc_bclink_remove_node(u32 addr)
175 {
176 	tipc_bclink_lock();
177 	tipc_nmap_remove(&bclink->bcast_nodes, addr);
178 	tipc_bclink_unlock();
179 }
180 
181 static void bclink_set_last_sent(void)
182 {
183 	if (bcl->next_out)
184 		bcl->fsm_msg_cnt = mod(buf_seqno(bcl->next_out) - 1);
185 	else
186 		bcl->fsm_msg_cnt = mod(bcl->next_out_no - 1);
187 }
188 
189 u32 tipc_bclink_get_last_sent(void)
190 {
191 	return bcl->fsm_msg_cnt;
192 }
193 
194 static void bclink_update_last_sent(struct tipc_node *node, u32 seqno)
195 {
196 	node->bclink.last_sent = less_eq(node->bclink.last_sent, seqno) ?
197 						seqno : node->bclink.last_sent;
198 }
199 
200 
201 /**
202  * tipc_bclink_retransmit_to - get most recent node to request retransmission
203  *
204  * Called with bclink_lock locked
205  */
206 struct tipc_node *tipc_bclink_retransmit_to(void)
207 {
208 	return bclink->retransmit_to;
209 }
210 
211 /**
212  * bclink_retransmit_pkt - retransmit broadcast packets
213  * @after: sequence number of last packet to *not* retransmit
214  * @to: sequence number of last packet to retransmit
215  *
216  * Called with bclink_lock locked
217  */
218 static void bclink_retransmit_pkt(u32 after, u32 to)
219 {
220 	struct sk_buff *skb;
221 
222 	skb_queue_walk(&bcl->outqueue, skb) {
223 		if (more(buf_seqno(skb), after))
224 			break;
225 	}
226 	tipc_link_retransmit(bcl, skb, mod(to - after));
227 }
228 
229 /**
230  * tipc_bclink_wakeup_users - wake up pending users
231  *
232  * Called with no locks taken
233  */
234 void tipc_bclink_wakeup_users(void)
235 {
236 	struct sk_buff *skb;
237 
238 	while ((skb = skb_dequeue(&bclink->link.waiting_sks)))
239 		tipc_sk_rcv(skb);
240 
241 }
242 
243 /**
244  * tipc_bclink_acknowledge - handle acknowledgement of broadcast packets
245  * @n_ptr: node that sent acknowledgement info
246  * @acked: broadcast sequence # that has been acknowledged
247  *
248  * Node is locked, bclink_lock unlocked.
249  */
250 void tipc_bclink_acknowledge(struct tipc_node *n_ptr, u32 acked)
251 {
252 	struct sk_buff *skb, *tmp;
253 	struct sk_buff *next;
254 	unsigned int released = 0;
255 
256 	tipc_bclink_lock();
257 	/* Bail out if tx queue is empty (no clean up is required) */
258 	skb = skb_peek(&bcl->outqueue);
259 	if (!skb)
260 		goto exit;
261 
262 	/* Determine which messages need to be acknowledged */
263 	if (acked == INVALID_LINK_SEQ) {
264 		/*
265 		 * Contact with specified node has been lost, so need to
266 		 * acknowledge sent messages only (if other nodes still exist)
267 		 * or both sent and unsent messages (otherwise)
268 		 */
269 		if (bclink->bcast_nodes.count)
270 			acked = bcl->fsm_msg_cnt;
271 		else
272 			acked = bcl->next_out_no;
273 	} else {
274 		/*
275 		 * Bail out if specified sequence number does not correspond
276 		 * to a message that has been sent and not yet acknowledged
277 		 */
278 		if (less(acked, buf_seqno(skb)) ||
279 		    less(bcl->fsm_msg_cnt, acked) ||
280 		    less_eq(acked, n_ptr->bclink.acked))
281 			goto exit;
282 	}
283 
284 	/* Skip over packets that node has previously acknowledged */
285 	skb_queue_walk(&bcl->outqueue, skb) {
286 		if (more(buf_seqno(skb), n_ptr->bclink.acked))
287 			break;
288 	}
289 
290 	/* Update packets that node is now acknowledging */
291 	skb_queue_walk_from_safe(&bcl->outqueue, skb, tmp) {
292 		if (more(buf_seqno(skb), acked))
293 			break;
294 
295 		next = tipc_skb_queue_next(&bcl->outqueue, skb);
296 		if (skb != bcl->next_out) {
297 			bcbuf_decr_acks(skb);
298 		} else {
299 			bcbuf_set_acks(skb, 0);
300 			bcl->next_out = next;
301 			bclink_set_last_sent();
302 		}
303 
304 		if (bcbuf_acks(skb) == 0) {
305 			__skb_unlink(skb, &bcl->outqueue);
306 			kfree_skb(skb);
307 			released = 1;
308 		}
309 	}
310 	n_ptr->bclink.acked = acked;
311 
312 	/* Try resolving broadcast link congestion, if necessary */
313 	if (unlikely(bcl->next_out)) {
314 		tipc_link_push_packets(bcl);
315 		bclink_set_last_sent();
316 	}
317 	if (unlikely(released && !skb_queue_empty(&bcl->waiting_sks)))
318 		n_ptr->action_flags |= TIPC_WAKEUP_BCAST_USERS;
319 
320 exit:
321 	tipc_bclink_unlock();
322 }
323 
324 /**
325  * tipc_bclink_update_link_state - update broadcast link state
326  *
327  * RCU and node lock set
328  */
329 void tipc_bclink_update_link_state(struct tipc_node *n_ptr, u32 last_sent)
330 {
331 	struct sk_buff *buf;
332 
333 	/* Ignore "stale" link state info */
334 	if (less_eq(last_sent, n_ptr->bclink.last_in))
335 		return;
336 
337 	/* Update link synchronization state; quit if in sync */
338 	bclink_update_last_sent(n_ptr, last_sent);
339 
340 	if (n_ptr->bclink.last_sent == n_ptr->bclink.last_in)
341 		return;
342 
343 	/* Update out-of-sync state; quit if loss is still unconfirmed */
344 	if ((++n_ptr->bclink.oos_state) == 1) {
345 		if (n_ptr->bclink.deferred_size < (TIPC_MIN_LINK_WIN / 2))
346 			return;
347 		n_ptr->bclink.oos_state++;
348 	}
349 
350 	/* Don't NACK if one has been recently sent (or seen) */
351 	if (n_ptr->bclink.oos_state & 0x1)
352 		return;
353 
354 	/* Send NACK */
355 	buf = tipc_buf_acquire(INT_H_SIZE);
356 	if (buf) {
357 		struct tipc_msg *msg = buf_msg(buf);
358 		struct sk_buff *skb = skb_peek(&n_ptr->bclink.deferred_queue);
359 		u32 to = skb ? buf_seqno(skb) - 1 : n_ptr->bclink.last_sent;
360 
361 		tipc_msg_init(msg, BCAST_PROTOCOL, STATE_MSG,
362 			      INT_H_SIZE, n_ptr->addr);
363 		msg_set_non_seq(msg, 1);
364 		msg_set_mc_netid(msg, tipc_net_id);
365 		msg_set_bcast_ack(msg, n_ptr->bclink.last_in);
366 		msg_set_bcgap_after(msg, n_ptr->bclink.last_in);
367 		msg_set_bcgap_to(msg, to);
368 
369 		tipc_bclink_lock();
370 		tipc_bearer_send(MAX_BEARERS, buf, NULL);
371 		bcl->stats.sent_nacks++;
372 		tipc_bclink_unlock();
373 		kfree_skb(buf);
374 
375 		n_ptr->bclink.oos_state++;
376 	}
377 }
378 
379 /**
380  * bclink_peek_nack - monitor retransmission requests sent by other nodes
381  *
382  * Delay any upcoming NACK by this node if another node has already
383  * requested the first message this node is going to ask for.
384  */
385 static void bclink_peek_nack(struct tipc_msg *msg)
386 {
387 	struct tipc_node *n_ptr = tipc_node_find(msg_destnode(msg));
388 
389 	if (unlikely(!n_ptr))
390 		return;
391 
392 	tipc_node_lock(n_ptr);
393 
394 	if (n_ptr->bclink.recv_permitted &&
395 	    (n_ptr->bclink.last_in != n_ptr->bclink.last_sent) &&
396 	    (n_ptr->bclink.last_in == msg_bcgap_after(msg)))
397 		n_ptr->bclink.oos_state = 2;
398 
399 	tipc_node_unlock(n_ptr);
400 }
401 
402 /* tipc_bclink_xmit - broadcast buffer chain to all nodes in cluster
403  *                    and to identified node local sockets
404  * @list: chain of buffers containing message
405  * Consumes the buffer chain, except when returning -ELINKCONG
406  * Returns 0 if success, otherwise errno: -ELINKCONG,-EHOSTUNREACH,-EMSGSIZE
407  */
408 int tipc_bclink_xmit(struct sk_buff_head *list)
409 {
410 	int rc = 0;
411 	int bc = 0;
412 	struct sk_buff *skb;
413 
414 	/* Prepare clone of message for local node */
415 	skb = tipc_msg_reassemble(list);
416 	if (unlikely(!skb)) {
417 		__skb_queue_purge(list);
418 		return -EHOSTUNREACH;
419 	}
420 
421 	/* Broadcast to all other nodes */
422 	if (likely(bclink)) {
423 		tipc_bclink_lock();
424 		if (likely(bclink->bcast_nodes.count)) {
425 			rc = __tipc_link_xmit(bcl, list);
426 			if (likely(!rc)) {
427 				u32 len = skb_queue_len(&bcl->outqueue);
428 
429 				bclink_set_last_sent();
430 				bcl->stats.queue_sz_counts++;
431 				bcl->stats.accu_queue_sz += len;
432 			}
433 			bc = 1;
434 		}
435 		tipc_bclink_unlock();
436 	}
437 
438 	if (unlikely(!bc))
439 		__skb_queue_purge(list);
440 
441 	/* Deliver message clone */
442 	if (likely(!rc))
443 		tipc_sk_mcast_rcv(skb);
444 	else
445 		kfree_skb(skb);
446 
447 	return rc;
448 }
449 
450 /**
451  * bclink_accept_pkt - accept an incoming, in-sequence broadcast packet
452  *
453  * Called with both sending node's lock and bclink_lock taken.
454  */
455 static void bclink_accept_pkt(struct tipc_node *node, u32 seqno)
456 {
457 	bclink_update_last_sent(node, seqno);
458 	node->bclink.last_in = seqno;
459 	node->bclink.oos_state = 0;
460 	bcl->stats.recv_info++;
461 
462 	/*
463 	 * Unicast an ACK periodically, ensuring that
464 	 * all nodes in the cluster don't ACK at the same time
465 	 */
466 	if (((seqno - tipc_own_addr) % TIPC_MIN_LINK_WIN) == 0) {
467 		tipc_link_proto_xmit(node->active_links[node->addr & 1],
468 				     STATE_MSG, 0, 0, 0, 0, 0);
469 		bcl->stats.sent_acks++;
470 	}
471 }
472 
473 /**
474  * tipc_bclink_rcv - receive a broadcast packet, and deliver upwards
475  *
476  * RCU is locked, no other locks set
477  */
478 void tipc_bclink_rcv(struct sk_buff *buf)
479 {
480 	struct tipc_msg *msg = buf_msg(buf);
481 	struct tipc_node *node;
482 	u32 next_in;
483 	u32 seqno;
484 	int deferred = 0;
485 
486 	/* Screen out unwanted broadcast messages */
487 	if (msg_mc_netid(msg) != tipc_net_id)
488 		goto exit;
489 
490 	node = tipc_node_find(msg_prevnode(msg));
491 	if (unlikely(!node))
492 		goto exit;
493 
494 	tipc_node_lock(node);
495 	if (unlikely(!node->bclink.recv_permitted))
496 		goto unlock;
497 
498 	/* Handle broadcast protocol message */
499 	if (unlikely(msg_user(msg) == BCAST_PROTOCOL)) {
500 		if (msg_type(msg) != STATE_MSG)
501 			goto unlock;
502 		if (msg_destnode(msg) == tipc_own_addr) {
503 			tipc_bclink_acknowledge(node, msg_bcast_ack(msg));
504 			tipc_node_unlock(node);
505 			tipc_bclink_lock();
506 			bcl->stats.recv_nacks++;
507 			bclink->retransmit_to = node;
508 			bclink_retransmit_pkt(msg_bcgap_after(msg),
509 					      msg_bcgap_to(msg));
510 			tipc_bclink_unlock();
511 		} else {
512 			tipc_node_unlock(node);
513 			bclink_peek_nack(msg);
514 		}
515 		goto exit;
516 	}
517 
518 	/* Handle in-sequence broadcast message */
519 	seqno = msg_seqno(msg);
520 	next_in = mod(node->bclink.last_in + 1);
521 
522 	if (likely(seqno == next_in)) {
523 receive:
524 		/* Deliver message to destination */
525 		if (likely(msg_isdata(msg))) {
526 			tipc_bclink_lock();
527 			bclink_accept_pkt(node, seqno);
528 			tipc_bclink_unlock();
529 			tipc_node_unlock(node);
530 			if (likely(msg_mcast(msg)))
531 				tipc_sk_mcast_rcv(buf);
532 			else
533 				kfree_skb(buf);
534 		} else if (msg_user(msg) == MSG_BUNDLER) {
535 			tipc_bclink_lock();
536 			bclink_accept_pkt(node, seqno);
537 			bcl->stats.recv_bundles++;
538 			bcl->stats.recv_bundled += msg_msgcnt(msg);
539 			tipc_bclink_unlock();
540 			tipc_node_unlock(node);
541 			tipc_link_bundle_rcv(buf);
542 		} else if (msg_user(msg) == MSG_FRAGMENTER) {
543 			tipc_buf_append(&node->bclink.reasm_buf, &buf);
544 			if (unlikely(!buf && !node->bclink.reasm_buf))
545 				goto unlock;
546 			tipc_bclink_lock();
547 			bclink_accept_pkt(node, seqno);
548 			bcl->stats.recv_fragments++;
549 			if (buf) {
550 				bcl->stats.recv_fragmented++;
551 				msg = buf_msg(buf);
552 				tipc_bclink_unlock();
553 				goto receive;
554 			}
555 			tipc_bclink_unlock();
556 			tipc_node_unlock(node);
557 		} else if (msg_user(msg) == NAME_DISTRIBUTOR) {
558 			tipc_bclink_lock();
559 			bclink_accept_pkt(node, seqno);
560 			tipc_bclink_unlock();
561 			tipc_node_unlock(node);
562 			tipc_named_rcv(buf);
563 		} else {
564 			tipc_bclink_lock();
565 			bclink_accept_pkt(node, seqno);
566 			tipc_bclink_unlock();
567 			tipc_node_unlock(node);
568 			kfree_skb(buf);
569 		}
570 		buf = NULL;
571 
572 		/* Determine new synchronization state */
573 		tipc_node_lock(node);
574 		if (unlikely(!tipc_node_is_up(node)))
575 			goto unlock;
576 
577 		if (node->bclink.last_in == node->bclink.last_sent)
578 			goto unlock;
579 
580 		if (skb_queue_empty(&node->bclink.deferred_queue)) {
581 			node->bclink.oos_state = 1;
582 			goto unlock;
583 		}
584 
585 		msg = buf_msg(skb_peek(&node->bclink.deferred_queue));
586 		seqno = msg_seqno(msg);
587 		next_in = mod(next_in + 1);
588 		if (seqno != next_in)
589 			goto unlock;
590 
591 		/* Take in-sequence message from deferred queue & deliver it */
592 		buf = __skb_dequeue(&node->bclink.deferred_queue);
593 		goto receive;
594 	}
595 
596 	/* Handle out-of-sequence broadcast message */
597 	if (less(next_in, seqno)) {
598 		deferred = tipc_link_defer_pkt(&node->bclink.deferred_queue,
599 					       buf);
600 		bclink_update_last_sent(node, seqno);
601 		buf = NULL;
602 	}
603 
604 	tipc_bclink_lock();
605 
606 	if (deferred)
607 		bcl->stats.deferred_recv++;
608 	else
609 		bcl->stats.duplicates++;
610 
611 	tipc_bclink_unlock();
612 
613 unlock:
614 	tipc_node_unlock(node);
615 exit:
616 	kfree_skb(buf);
617 }
618 
619 u32 tipc_bclink_acks_missing(struct tipc_node *n_ptr)
620 {
621 	return (n_ptr->bclink.recv_permitted &&
622 		(tipc_bclink_get_last_sent() != n_ptr->bclink.acked));
623 }
624 
625 
626 /**
627  * tipc_bcbearer_send - send a packet through the broadcast pseudo-bearer
628  *
629  * Send packet over as many bearers as necessary to reach all nodes
630  * that have joined the broadcast link.
631  *
632  * Returns 0 (packet sent successfully) under all circumstances,
633  * since the broadcast link's pseudo-bearer never blocks
634  */
635 static int tipc_bcbearer_send(struct sk_buff *buf, struct tipc_bearer *unused1,
636 			      struct tipc_media_addr *unused2)
637 {
638 	int bp_index;
639 	struct tipc_msg *msg = buf_msg(buf);
640 
641 	/* Prepare broadcast link message for reliable transmission,
642 	 * if first time trying to send it;
643 	 * preparation is skipped for broadcast link protocol messages
644 	 * since they are sent in an unreliable manner and don't need it
645 	 */
646 	if (likely(!msg_non_seq(buf_msg(buf)))) {
647 		bcbuf_set_acks(buf, bclink->bcast_nodes.count);
648 		msg_set_non_seq(msg, 1);
649 		msg_set_mc_netid(msg, tipc_net_id);
650 		bcl->stats.sent_info++;
651 
652 		if (WARN_ON(!bclink->bcast_nodes.count)) {
653 			dump_stack();
654 			return 0;
655 		}
656 	}
657 
658 	/* Send buffer over bearers until all targets reached */
659 	bcbearer->remains = bclink->bcast_nodes;
660 
661 	for (bp_index = 0; bp_index < MAX_BEARERS; bp_index++) {
662 		struct tipc_bearer *p = bcbearer->bpairs[bp_index].primary;
663 		struct tipc_bearer *s = bcbearer->bpairs[bp_index].secondary;
664 		struct tipc_bearer *bp[2] = {p, s};
665 		struct tipc_bearer *b = bp[msg_link_selector(msg)];
666 		struct sk_buff *tbuf;
667 
668 		if (!p)
669 			break; /* No more bearers to try */
670 		if (!b)
671 			b = p;
672 		tipc_nmap_diff(&bcbearer->remains, &b->nodes,
673 			       &bcbearer->remains_new);
674 		if (bcbearer->remains_new.count == bcbearer->remains.count)
675 			continue; /* Nothing added by bearer pair */
676 
677 		if (bp_index == 0) {
678 			/* Use original buffer for first bearer */
679 			tipc_bearer_send(b->identity, buf, &b->bcast_addr);
680 		} else {
681 			/* Avoid concurrent buffer access */
682 			tbuf = pskb_copy_for_clone(buf, GFP_ATOMIC);
683 			if (!tbuf)
684 				break;
685 			tipc_bearer_send(b->identity, tbuf, &b->bcast_addr);
686 			kfree_skb(tbuf); /* Bearer keeps a clone */
687 		}
688 		if (bcbearer->remains_new.count == 0)
689 			break; /* All targets reached */
690 
691 		bcbearer->remains = bcbearer->remains_new;
692 	}
693 
694 	return 0;
695 }
696 
697 /**
698  * tipc_bcbearer_sort - create sets of bearer pairs used by broadcast bearer
699  */
700 void tipc_bcbearer_sort(struct tipc_node_map *nm_ptr, u32 node, bool action)
701 {
702 	struct tipc_bcbearer_pair *bp_temp = bcbearer->bpairs_temp;
703 	struct tipc_bcbearer_pair *bp_curr;
704 	struct tipc_bearer *b;
705 	int b_index;
706 	int pri;
707 
708 	tipc_bclink_lock();
709 
710 	if (action)
711 		tipc_nmap_add(nm_ptr, node);
712 	else
713 		tipc_nmap_remove(nm_ptr, node);
714 
715 	/* Group bearers by priority (can assume max of two per priority) */
716 	memset(bp_temp, 0, sizeof(bcbearer->bpairs_temp));
717 
718 	rcu_read_lock();
719 	for (b_index = 0; b_index < MAX_BEARERS; b_index++) {
720 		b = rcu_dereference_rtnl(bearer_list[b_index]);
721 		if (!b || !b->nodes.count)
722 			continue;
723 
724 		if (!bp_temp[b->priority].primary)
725 			bp_temp[b->priority].primary = b;
726 		else
727 			bp_temp[b->priority].secondary = b;
728 	}
729 	rcu_read_unlock();
730 
731 	/* Create array of bearer pairs for broadcasting */
732 	bp_curr = bcbearer->bpairs;
733 	memset(bcbearer->bpairs, 0, sizeof(bcbearer->bpairs));
734 
735 	for (pri = TIPC_MAX_LINK_PRI; pri >= 0; pri--) {
736 
737 		if (!bp_temp[pri].primary)
738 			continue;
739 
740 		bp_curr->primary = bp_temp[pri].primary;
741 
742 		if (bp_temp[pri].secondary) {
743 			if (tipc_nmap_equal(&bp_temp[pri].primary->nodes,
744 					    &bp_temp[pri].secondary->nodes)) {
745 				bp_curr->secondary = bp_temp[pri].secondary;
746 			} else {
747 				bp_curr++;
748 				bp_curr->primary = bp_temp[pri].secondary;
749 			}
750 		}
751 
752 		bp_curr++;
753 	}
754 
755 	tipc_bclink_unlock();
756 }
757 
758 static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
759 				      struct tipc_stats *stats)
760 {
761 	int i;
762 	struct nlattr *nest;
763 
764 	struct nla_map {
765 		__u32 key;
766 		__u32 val;
767 	};
768 
769 	struct nla_map map[] = {
770 		{TIPC_NLA_STATS_RX_INFO, stats->recv_info},
771 		{TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
772 		{TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
773 		{TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
774 		{TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
775 		{TIPC_NLA_STATS_TX_INFO, stats->sent_info},
776 		{TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
777 		{TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
778 		{TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
779 		{TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
780 		{TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
781 		{TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
782 		{TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
783 		{TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
784 		{TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
785 		{TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
786 		{TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
787 		{TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
788 		{TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
789 			(stats->accu_queue_sz / stats->queue_sz_counts) : 0}
790 	};
791 
792 	nest = nla_nest_start(skb, TIPC_NLA_LINK_STATS);
793 	if (!nest)
794 		return -EMSGSIZE;
795 
796 	for (i = 0; i <  ARRAY_SIZE(map); i++)
797 		if (nla_put_u32(skb, map[i].key, map[i].val))
798 			goto msg_full;
799 
800 	nla_nest_end(skb, nest);
801 
802 	return 0;
803 msg_full:
804 	nla_nest_cancel(skb, nest);
805 
806 	return -EMSGSIZE;
807 }
808 
809 int tipc_nl_add_bc_link(struct tipc_nl_msg *msg)
810 {
811 	int err;
812 	void *hdr;
813 	struct nlattr *attrs;
814 	struct nlattr *prop;
815 
816 	if (!bcl)
817 		return 0;
818 
819 	tipc_bclink_lock();
820 
821 	hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_v2_family,
822 			  NLM_F_MULTI, TIPC_NL_LINK_GET);
823 	if (!hdr)
824 		return -EMSGSIZE;
825 
826 	attrs = nla_nest_start(msg->skb, TIPC_NLA_LINK);
827 	if (!attrs)
828 		goto msg_full;
829 
830 	/* The broadcast link is always up */
831 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
832 		goto attr_msg_full;
833 
834 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
835 		goto attr_msg_full;
836 	if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
837 		goto attr_msg_full;
838 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, bcl->next_in_no))
839 		goto attr_msg_full;
840 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, bcl->next_out_no))
841 		goto attr_msg_full;
842 
843 	prop = nla_nest_start(msg->skb, TIPC_NLA_LINK_PROP);
844 	if (!prop)
845 		goto attr_msg_full;
846 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->queue_limit[0]))
847 		goto prop_msg_full;
848 	nla_nest_end(msg->skb, prop);
849 
850 	err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
851 	if (err)
852 		goto attr_msg_full;
853 
854 	tipc_bclink_unlock();
855 	nla_nest_end(msg->skb, attrs);
856 	genlmsg_end(msg->skb, hdr);
857 
858 	return 0;
859 
860 prop_msg_full:
861 	nla_nest_cancel(msg->skb, prop);
862 attr_msg_full:
863 	nla_nest_cancel(msg->skb, attrs);
864 msg_full:
865 	tipc_bclink_unlock();
866 	genlmsg_cancel(msg->skb, hdr);
867 
868 	return -EMSGSIZE;
869 }
870 
871 int tipc_bclink_stats(char *buf, const u32 buf_size)
872 {
873 	int ret;
874 	struct tipc_stats *s;
875 
876 	if (!bcl)
877 		return 0;
878 
879 	tipc_bclink_lock();
880 
881 	s = &bcl->stats;
882 
883 	ret = tipc_snprintf(buf, buf_size, "Link <%s>\n"
884 			    "  Window:%u packets\n",
885 			    bcl->name, bcl->queue_limit[0]);
886 	ret += tipc_snprintf(buf + ret, buf_size - ret,
887 			     "  RX packets:%u fragments:%u/%u bundles:%u/%u\n",
888 			     s->recv_info, s->recv_fragments,
889 			     s->recv_fragmented, s->recv_bundles,
890 			     s->recv_bundled);
891 	ret += tipc_snprintf(buf + ret, buf_size - ret,
892 			     "  TX packets:%u fragments:%u/%u bundles:%u/%u\n",
893 			     s->sent_info, s->sent_fragments,
894 			     s->sent_fragmented, s->sent_bundles,
895 			     s->sent_bundled);
896 	ret += tipc_snprintf(buf + ret, buf_size - ret,
897 			     "  RX naks:%u defs:%u dups:%u\n",
898 			     s->recv_nacks, s->deferred_recv, s->duplicates);
899 	ret += tipc_snprintf(buf + ret, buf_size - ret,
900 			     "  TX naks:%u acks:%u dups:%u\n",
901 			     s->sent_nacks, s->sent_acks, s->retransmitted);
902 	ret += tipc_snprintf(buf + ret, buf_size - ret,
903 			     "  Congestion link:%u  Send queue max:%u avg:%u\n",
904 			     s->link_congs, s->max_queue_sz,
905 			     s->queue_sz_counts ?
906 			     (s->accu_queue_sz / s->queue_sz_counts) : 0);
907 
908 	tipc_bclink_unlock();
909 	return ret;
910 }
911 
912 int tipc_bclink_reset_stats(void)
913 {
914 	if (!bcl)
915 		return -ENOPROTOOPT;
916 
917 	tipc_bclink_lock();
918 	memset(&bcl->stats, 0, sizeof(bcl->stats));
919 	tipc_bclink_unlock();
920 	return 0;
921 }
922 
923 int tipc_bclink_set_queue_limits(u32 limit)
924 {
925 	if (!bcl)
926 		return -ENOPROTOOPT;
927 	if ((limit < TIPC_MIN_LINK_WIN) || (limit > TIPC_MAX_LINK_WIN))
928 		return -EINVAL;
929 
930 	tipc_bclink_lock();
931 	tipc_link_set_queue_limits(bcl, limit);
932 	tipc_bclink_unlock();
933 	return 0;
934 }
935 
936 int tipc_bclink_init(void)
937 {
938 	bcbearer = kzalloc(sizeof(*bcbearer), GFP_ATOMIC);
939 	if (!bcbearer)
940 		return -ENOMEM;
941 
942 	bclink = kzalloc(sizeof(*bclink), GFP_ATOMIC);
943 	if (!bclink) {
944 		kfree(bcbearer);
945 		return -ENOMEM;
946 	}
947 
948 	bcl = &bclink->link;
949 	bcbearer->bearer.media = &bcbearer->media;
950 	bcbearer->media.send_msg = tipc_bcbearer_send;
951 	sprintf(bcbearer->media.name, "tipc-broadcast");
952 
953 	spin_lock_init(&bclink->lock);
954 	__skb_queue_head_init(&bcl->outqueue);
955 	__skb_queue_head_init(&bcl->deferred_queue);
956 	skb_queue_head_init(&bcl->waiting_sks);
957 	bcl->next_out_no = 1;
958 	spin_lock_init(&bclink->node.lock);
959 	__skb_queue_head_init(&bclink->node.waiting_sks);
960 	bcl->owner = &bclink->node;
961 	bcl->max_pkt = MAX_PKT_DEFAULT_MCAST;
962 	tipc_link_set_queue_limits(bcl, BCLINK_WIN_DEFAULT);
963 	bcl->bearer_id = MAX_BEARERS;
964 	rcu_assign_pointer(bearer_list[MAX_BEARERS], &bcbearer->bearer);
965 	bcl->state = WORKING_WORKING;
966 	strlcpy(bcl->name, tipc_bclink_name, TIPC_MAX_LINK_NAME);
967 	return 0;
968 }
969 
970 void tipc_bclink_stop(void)
971 {
972 	tipc_bclink_lock();
973 	tipc_link_purge_queues(bcl);
974 	tipc_bclink_unlock();
975 
976 	RCU_INIT_POINTER(bearer_list[BCBEARER], NULL);
977 	synchronize_net();
978 	kfree(bcbearer);
979 	kfree(bclink);
980 }
981 
982 /**
983  * tipc_nmap_add - add a node to a node map
984  */
985 static void tipc_nmap_add(struct tipc_node_map *nm_ptr, u32 node)
986 {
987 	int n = tipc_node(node);
988 	int w = n / WSIZE;
989 	u32 mask = (1 << (n % WSIZE));
990 
991 	if ((nm_ptr->map[w] & mask) == 0) {
992 		nm_ptr->count++;
993 		nm_ptr->map[w] |= mask;
994 	}
995 }
996 
997 /**
998  * tipc_nmap_remove - remove a node from a node map
999  */
1000 static void tipc_nmap_remove(struct tipc_node_map *nm_ptr, u32 node)
1001 {
1002 	int n = tipc_node(node);
1003 	int w = n / WSIZE;
1004 	u32 mask = (1 << (n % WSIZE));
1005 
1006 	if ((nm_ptr->map[w] & mask) != 0) {
1007 		nm_ptr->map[w] &= ~mask;
1008 		nm_ptr->count--;
1009 	}
1010 }
1011 
1012 /**
1013  * tipc_nmap_diff - find differences between node maps
1014  * @nm_a: input node map A
1015  * @nm_b: input node map B
1016  * @nm_diff: output node map A-B (i.e. nodes of A that are not in B)
1017  */
1018 static void tipc_nmap_diff(struct tipc_node_map *nm_a,
1019 			   struct tipc_node_map *nm_b,
1020 			   struct tipc_node_map *nm_diff)
1021 {
1022 	int stop = ARRAY_SIZE(nm_a->map);
1023 	int w;
1024 	int b;
1025 	u32 map;
1026 
1027 	memset(nm_diff, 0, sizeof(*nm_diff));
1028 	for (w = 0; w < stop; w++) {
1029 		map = nm_a->map[w] ^ (nm_a->map[w] & nm_b->map[w]);
1030 		nm_diff->map[w] = map;
1031 		if (map != 0) {
1032 			for (b = 0 ; b < WSIZE; b++) {
1033 				if (map & (1 << b))
1034 					nm_diff->count++;
1035 			}
1036 		}
1037 	}
1038 }
1039 
1040 /**
1041  * tipc_port_list_add - add a port to a port list, ensuring no duplicates
1042  */
1043 void tipc_port_list_add(struct tipc_port_list *pl_ptr, u32 port)
1044 {
1045 	struct tipc_port_list *item = pl_ptr;
1046 	int i;
1047 	int item_sz = PLSIZE;
1048 	int cnt = pl_ptr->count;
1049 
1050 	for (; ; cnt -= item_sz, item = item->next) {
1051 		if (cnt < PLSIZE)
1052 			item_sz = cnt;
1053 		for (i = 0; i < item_sz; i++)
1054 			if (item->ports[i] == port)
1055 				return;
1056 		if (i < PLSIZE) {
1057 			item->ports[i] = port;
1058 			pl_ptr->count++;
1059 			return;
1060 		}
1061 		if (!item->next) {
1062 			item->next = kmalloc(sizeof(*item), GFP_ATOMIC);
1063 			if (!item->next) {
1064 				pr_warn("Incomplete multicast delivery, no memory\n");
1065 				return;
1066 			}
1067 			item->next->next = NULL;
1068 		}
1069 	}
1070 }
1071 
1072 /**
1073  * tipc_port_list_free - free dynamically created entries in port_list chain
1074  *
1075  */
1076 void tipc_port_list_free(struct tipc_port_list *pl_ptr)
1077 {
1078 	struct tipc_port_list *item;
1079 	struct tipc_port_list *next;
1080 
1081 	for (item = pl_ptr->next; item; item = next) {
1082 		next = item->next;
1083 		kfree(item);
1084 	}
1085 }
1086