xref: /linux/net/tipc/bcast.c (revision ca55b2fef3a9373fcfc30f82fd26bc7fccbda732)
1 /*
2  * net/tipc/bcast.c: TIPC broadcast code
3  *
4  * Copyright (c) 2004-2006, 2014-2015, 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 "socket.h"
39 #include "msg.h"
40 #include "bcast.h"
41 #include "name_distr.h"
42 #include "core.h"
43 
44 #define	MAX_PKT_DEFAULT_MCAST	1500	/* bcast link max packet size (fixed) */
45 #define	BCLINK_WIN_DEFAULT	50	/* bcast link window size (default) */
46 #define	BCLINK_WIN_MIN	        32	/* bcast minimum link window size */
47 
48 const char tipc_bclink_name[] = "broadcast-link";
49 
50 static void tipc_nmap_diff(struct tipc_node_map *nm_a,
51 			   struct tipc_node_map *nm_b,
52 			   struct tipc_node_map *nm_diff);
53 static void tipc_nmap_add(struct tipc_node_map *nm_ptr, u32 node);
54 static void tipc_nmap_remove(struct tipc_node_map *nm_ptr, u32 node);
55 
56 static void tipc_bclink_lock(struct net *net)
57 {
58 	struct tipc_net *tn = net_generic(net, tipc_net_id);
59 
60 	spin_lock_bh(&tn->bclink->lock);
61 }
62 
63 static void tipc_bclink_unlock(struct net *net)
64 {
65 	struct tipc_net *tn = net_generic(net, tipc_net_id);
66 
67 	spin_unlock_bh(&tn->bclink->lock);
68 }
69 
70 void tipc_bclink_input(struct net *net)
71 {
72 	struct tipc_net *tn = net_generic(net, tipc_net_id);
73 
74 	tipc_sk_mcast_rcv(net, &tn->bclink->arrvq, &tn->bclink->inputq);
75 }
76 
77 uint  tipc_bclink_get_mtu(void)
78 {
79 	return MAX_PKT_DEFAULT_MCAST;
80 }
81 
82 static u32 bcbuf_acks(struct sk_buff *buf)
83 {
84 	return (u32)(unsigned long)TIPC_SKB_CB(buf)->handle;
85 }
86 
87 static void bcbuf_set_acks(struct sk_buff *buf, u32 acks)
88 {
89 	TIPC_SKB_CB(buf)->handle = (void *)(unsigned long)acks;
90 }
91 
92 static void bcbuf_decr_acks(struct sk_buff *buf)
93 {
94 	bcbuf_set_acks(buf, bcbuf_acks(buf) - 1);
95 }
96 
97 void tipc_bclink_add_node(struct net *net, u32 addr)
98 {
99 	struct tipc_net *tn = net_generic(net, tipc_net_id);
100 
101 	tipc_bclink_lock(net);
102 	tipc_nmap_add(&tn->bclink->bcast_nodes, addr);
103 	tipc_bclink_unlock(net);
104 }
105 
106 void tipc_bclink_remove_node(struct net *net, u32 addr)
107 {
108 	struct tipc_net *tn = net_generic(net, tipc_net_id);
109 
110 	tipc_bclink_lock(net);
111 	tipc_nmap_remove(&tn->bclink->bcast_nodes, addr);
112 
113 	/* Last node? => reset backlog queue */
114 	if (!tn->bclink->bcast_nodes.count)
115 		tipc_link_purge_backlog(&tn->bclink->link);
116 
117 	tipc_bclink_unlock(net);
118 }
119 
120 static void bclink_set_last_sent(struct net *net)
121 {
122 	struct tipc_net *tn = net_generic(net, tipc_net_id);
123 	struct tipc_link *bcl = tn->bcl;
124 
125 	bcl->silent_intv_cnt = mod(bcl->snd_nxt - 1);
126 }
127 
128 u32 tipc_bclink_get_last_sent(struct net *net)
129 {
130 	struct tipc_net *tn = net_generic(net, tipc_net_id);
131 
132 	return tn->bcl->silent_intv_cnt;
133 }
134 
135 static void bclink_update_last_sent(struct tipc_node *node, u32 seqno)
136 {
137 	node->bclink.last_sent = less_eq(node->bclink.last_sent, seqno) ?
138 						seqno : node->bclink.last_sent;
139 }
140 
141 /**
142  * tipc_bclink_retransmit_to - get most recent node to request retransmission
143  *
144  * Called with bclink_lock locked
145  */
146 struct tipc_node *tipc_bclink_retransmit_to(struct net *net)
147 {
148 	struct tipc_net *tn = net_generic(net, tipc_net_id);
149 
150 	return tn->bclink->retransmit_to;
151 }
152 
153 /**
154  * bclink_retransmit_pkt - retransmit broadcast packets
155  * @after: sequence number of last packet to *not* retransmit
156  * @to: sequence number of last packet to retransmit
157  *
158  * Called with bclink_lock locked
159  */
160 static void bclink_retransmit_pkt(struct tipc_net *tn, u32 after, u32 to)
161 {
162 	struct sk_buff *skb;
163 	struct tipc_link *bcl = tn->bcl;
164 
165 	skb_queue_walk(&bcl->transmq, skb) {
166 		if (more(buf_seqno(skb), after)) {
167 			tipc_link_retransmit(bcl, skb, mod(to - after));
168 			break;
169 		}
170 	}
171 }
172 
173 /**
174  * bclink_prepare_wakeup - prepare users for wakeup after congestion
175  * @bcl: broadcast link
176  * @resultq: queue for users which can be woken up
177  * Move a number of waiting users, as permitted by available space in
178  * the send queue, from link wait queue to specified queue for wakeup
179  */
180 static void bclink_prepare_wakeup(struct tipc_link *bcl, struct sk_buff_head *resultq)
181 {
182 	int pnd[TIPC_SYSTEM_IMPORTANCE + 1] = {0,};
183 	int imp, lim;
184 	struct sk_buff *skb, *tmp;
185 
186 	skb_queue_walk_safe(&bcl->wakeupq, skb, tmp) {
187 		imp = TIPC_SKB_CB(skb)->chain_imp;
188 		lim = bcl->window + bcl->backlog[imp].limit;
189 		pnd[imp] += TIPC_SKB_CB(skb)->chain_sz;
190 		if ((pnd[imp] + bcl->backlog[imp].len) >= lim)
191 			continue;
192 		skb_unlink(skb, &bcl->wakeupq);
193 		skb_queue_tail(resultq, skb);
194 	}
195 }
196 
197 /**
198  * tipc_bclink_wakeup_users - wake up pending users
199  *
200  * Called with no locks taken
201  */
202 void tipc_bclink_wakeup_users(struct net *net)
203 {
204 	struct tipc_net *tn = net_generic(net, tipc_net_id);
205 	struct tipc_link *bcl = tn->bcl;
206 	struct sk_buff_head resultq;
207 
208 	skb_queue_head_init(&resultq);
209 	bclink_prepare_wakeup(bcl, &resultq);
210 	tipc_sk_rcv(net, &resultq);
211 }
212 
213 /**
214  * tipc_bclink_acknowledge - handle acknowledgement of broadcast packets
215  * @n_ptr: node that sent acknowledgement info
216  * @acked: broadcast sequence # that has been acknowledged
217  *
218  * Node is locked, bclink_lock unlocked.
219  */
220 void tipc_bclink_acknowledge(struct tipc_node *n_ptr, u32 acked)
221 {
222 	struct sk_buff *skb, *tmp;
223 	unsigned int released = 0;
224 	struct net *net = n_ptr->net;
225 	struct tipc_net *tn = net_generic(net, tipc_net_id);
226 
227 	if (unlikely(!n_ptr->bclink.recv_permitted))
228 		return;
229 
230 	tipc_bclink_lock(net);
231 
232 	/* Bail out if tx queue is empty (no clean up is required) */
233 	skb = skb_peek(&tn->bcl->transmq);
234 	if (!skb)
235 		goto exit;
236 
237 	/* Determine which messages need to be acknowledged */
238 	if (acked == INVALID_LINK_SEQ) {
239 		/*
240 		 * Contact with specified node has been lost, so need to
241 		 * acknowledge sent messages only (if other nodes still exist)
242 		 * or both sent and unsent messages (otherwise)
243 		 */
244 		if (tn->bclink->bcast_nodes.count)
245 			acked = tn->bcl->silent_intv_cnt;
246 		else
247 			acked = tn->bcl->snd_nxt;
248 	} else {
249 		/*
250 		 * Bail out if specified sequence number does not correspond
251 		 * to a message that has been sent and not yet acknowledged
252 		 */
253 		if (less(acked, buf_seqno(skb)) ||
254 		    less(tn->bcl->silent_intv_cnt, acked) ||
255 		    less_eq(acked, n_ptr->bclink.acked))
256 			goto exit;
257 	}
258 
259 	/* Skip over packets that node has previously acknowledged */
260 	skb_queue_walk(&tn->bcl->transmq, skb) {
261 		if (more(buf_seqno(skb), n_ptr->bclink.acked))
262 			break;
263 	}
264 
265 	/* Update packets that node is now acknowledging */
266 	skb_queue_walk_from_safe(&tn->bcl->transmq, skb, tmp) {
267 		if (more(buf_seqno(skb), acked))
268 			break;
269 		bcbuf_decr_acks(skb);
270 		bclink_set_last_sent(net);
271 		if (bcbuf_acks(skb) == 0) {
272 			__skb_unlink(skb, &tn->bcl->transmq);
273 			kfree_skb(skb);
274 			released = 1;
275 		}
276 	}
277 	n_ptr->bclink.acked = acked;
278 
279 	/* Try resolving broadcast link congestion, if necessary */
280 	if (unlikely(skb_peek(&tn->bcl->backlogq))) {
281 		tipc_link_push_packets(tn->bcl);
282 		bclink_set_last_sent(net);
283 	}
284 	if (unlikely(released && !skb_queue_empty(&tn->bcl->wakeupq)))
285 		n_ptr->action_flags |= TIPC_WAKEUP_BCAST_USERS;
286 exit:
287 	tipc_bclink_unlock(net);
288 }
289 
290 /**
291  * tipc_bclink_update_link_state - update broadcast link state
292  *
293  * RCU and node lock set
294  */
295 void tipc_bclink_update_link_state(struct tipc_node *n_ptr,
296 				   u32 last_sent)
297 {
298 	struct sk_buff *buf;
299 	struct net *net = n_ptr->net;
300 	struct tipc_net *tn = net_generic(net, tipc_net_id);
301 
302 	/* Ignore "stale" link state info */
303 	if (less_eq(last_sent, n_ptr->bclink.last_in))
304 		return;
305 
306 	/* Update link synchronization state; quit if in sync */
307 	bclink_update_last_sent(n_ptr, last_sent);
308 
309 	if (n_ptr->bclink.last_sent == n_ptr->bclink.last_in)
310 		return;
311 
312 	/* Update out-of-sync state; quit if loss is still unconfirmed */
313 	if ((++n_ptr->bclink.oos_state) == 1) {
314 		if (n_ptr->bclink.deferred_size < (TIPC_MIN_LINK_WIN / 2))
315 			return;
316 		n_ptr->bclink.oos_state++;
317 	}
318 
319 	/* Don't NACK if one has been recently sent (or seen) */
320 	if (n_ptr->bclink.oos_state & 0x1)
321 		return;
322 
323 	/* Send NACK */
324 	buf = tipc_buf_acquire(INT_H_SIZE);
325 	if (buf) {
326 		struct tipc_msg *msg = buf_msg(buf);
327 		struct sk_buff *skb = skb_peek(&n_ptr->bclink.deferdq);
328 		u32 to = skb ? buf_seqno(skb) - 1 : n_ptr->bclink.last_sent;
329 
330 		tipc_msg_init(tn->own_addr, msg, BCAST_PROTOCOL, STATE_MSG,
331 			      INT_H_SIZE, n_ptr->addr);
332 		msg_set_non_seq(msg, 1);
333 		msg_set_mc_netid(msg, tn->net_id);
334 		msg_set_bcast_ack(msg, n_ptr->bclink.last_in);
335 		msg_set_bcgap_after(msg, n_ptr->bclink.last_in);
336 		msg_set_bcgap_to(msg, to);
337 
338 		tipc_bclink_lock(net);
339 		tipc_bearer_send(net, MAX_BEARERS, buf, NULL);
340 		tn->bcl->stats.sent_nacks++;
341 		tipc_bclink_unlock(net);
342 		kfree_skb(buf);
343 
344 		n_ptr->bclink.oos_state++;
345 	}
346 }
347 
348 void tipc_bclink_sync_state(struct tipc_node *n, struct tipc_msg *hdr)
349 {
350 	u16 last = msg_last_bcast(hdr);
351 	int mtyp = msg_type(hdr);
352 
353 	if (unlikely(msg_user(hdr) != LINK_PROTOCOL))
354 		return;
355 	if (mtyp == STATE_MSG) {
356 		tipc_bclink_update_link_state(n, last);
357 		return;
358 	}
359 	/* Compatibility: older nodes don't know BCAST_PROTOCOL synchronization,
360 	 * and transfer synch info in LINK_PROTOCOL messages.
361 	 */
362 	if (tipc_node_is_up(n))
363 		return;
364 	if ((mtyp != RESET_MSG) && (mtyp != ACTIVATE_MSG))
365 		return;
366 	n->bclink.last_sent = last;
367 	n->bclink.last_in = last;
368 	n->bclink.oos_state = 0;
369 }
370 
371 /**
372  * bclink_peek_nack - monitor retransmission requests sent by other nodes
373  *
374  * Delay any upcoming NACK by this node if another node has already
375  * requested the first message this node is going to ask for.
376  */
377 static void bclink_peek_nack(struct net *net, struct tipc_msg *msg)
378 {
379 	struct tipc_node *n_ptr = tipc_node_find(net, msg_destnode(msg));
380 
381 	if (unlikely(!n_ptr))
382 		return;
383 
384 	tipc_node_lock(n_ptr);
385 	if (n_ptr->bclink.recv_permitted &&
386 	    (n_ptr->bclink.last_in != n_ptr->bclink.last_sent) &&
387 	    (n_ptr->bclink.last_in == msg_bcgap_after(msg)))
388 		n_ptr->bclink.oos_state = 2;
389 	tipc_node_unlock(n_ptr);
390 	tipc_node_put(n_ptr);
391 }
392 
393 /* tipc_bclink_xmit - deliver buffer chain to all nodes in cluster
394  *                    and to identified node local sockets
395  * @net: the applicable net namespace
396  * @list: chain of buffers containing message
397  * Consumes the buffer chain, except when returning -ELINKCONG
398  * Returns 0 if success, otherwise errno: -ELINKCONG,-EHOSTUNREACH,-EMSGSIZE
399  */
400 int tipc_bclink_xmit(struct net *net, struct sk_buff_head *list)
401 {
402 	struct tipc_net *tn = net_generic(net, tipc_net_id);
403 	struct tipc_link *bcl = tn->bcl;
404 	struct tipc_bclink *bclink = tn->bclink;
405 	int rc = 0;
406 	int bc = 0;
407 	struct sk_buff *skb;
408 	struct sk_buff_head arrvq;
409 	struct sk_buff_head inputq;
410 
411 	/* Prepare clone of message for local node */
412 	skb = tipc_msg_reassemble(list);
413 	if (unlikely(!skb))
414 		return -EHOSTUNREACH;
415 
416 	/* Broadcast to all nodes */
417 	if (likely(bclink)) {
418 		tipc_bclink_lock(net);
419 		if (likely(bclink->bcast_nodes.count)) {
420 			rc = __tipc_link_xmit(net, bcl, list);
421 			if (likely(!rc)) {
422 				u32 len = skb_queue_len(&bcl->transmq);
423 
424 				bclink_set_last_sent(net);
425 				bcl->stats.queue_sz_counts++;
426 				bcl->stats.accu_queue_sz += len;
427 			}
428 			bc = 1;
429 		}
430 		tipc_bclink_unlock(net);
431 	}
432 
433 	if (unlikely(!bc))
434 		__skb_queue_purge(list);
435 
436 	if (unlikely(rc)) {
437 		kfree_skb(skb);
438 		return rc;
439 	}
440 	/* Deliver message clone */
441 	__skb_queue_head_init(&arrvq);
442 	skb_queue_head_init(&inputq);
443 	__skb_queue_tail(&arrvq, skb);
444 	tipc_sk_mcast_rcv(net, &arrvq, &inputq);
445 	return rc;
446 }
447 
448 /**
449  * bclink_accept_pkt - accept an incoming, in-sequence broadcast packet
450  *
451  * Called with both sending node's lock and bclink_lock taken.
452  */
453 static void bclink_accept_pkt(struct tipc_node *node, u32 seqno)
454 {
455 	struct tipc_net *tn = net_generic(node->net, tipc_net_id);
456 
457 	bclink_update_last_sent(node, seqno);
458 	node->bclink.last_in = seqno;
459 	node->bclink.oos_state = 0;
460 	tn->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 - tn->own_addr) % TIPC_MIN_LINK_WIN) == 0) {
467 		tipc_link_proto_xmit(node_active_link(node, node->addr),
468 				     STATE_MSG, 0, 0, 0, 0);
469 		tn->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 net *net, struct sk_buff *buf)
479 {
480 	struct tipc_net *tn = net_generic(net, tipc_net_id);
481 	struct tipc_link *bcl = tn->bcl;
482 	struct tipc_msg *msg = buf_msg(buf);
483 	struct tipc_node *node;
484 	u32 next_in;
485 	u32 seqno;
486 	int deferred = 0;
487 	int pos = 0;
488 	struct sk_buff *iskb;
489 	struct sk_buff_head *arrvq, *inputq;
490 
491 	/* Screen out unwanted broadcast messages */
492 	if (msg_mc_netid(msg) != tn->net_id)
493 		goto exit;
494 
495 	node = tipc_node_find(net, msg_prevnode(msg));
496 	if (unlikely(!node))
497 		goto exit;
498 
499 	tipc_node_lock(node);
500 	if (unlikely(!node->bclink.recv_permitted))
501 		goto unlock;
502 
503 	/* Handle broadcast protocol message */
504 	if (unlikely(msg_user(msg) == BCAST_PROTOCOL)) {
505 		if (msg_type(msg) != STATE_MSG)
506 			goto unlock;
507 		if (msg_destnode(msg) == tn->own_addr) {
508 			tipc_bclink_acknowledge(node, msg_bcast_ack(msg));
509 			tipc_bclink_lock(net);
510 			bcl->stats.recv_nacks++;
511 			tn->bclink->retransmit_to = node;
512 			bclink_retransmit_pkt(tn, msg_bcgap_after(msg),
513 					      msg_bcgap_to(msg));
514 			tipc_bclink_unlock(net);
515 			tipc_node_unlock(node);
516 		} else {
517 			tipc_node_unlock(node);
518 			bclink_peek_nack(net, msg);
519 		}
520 		tipc_node_put(node);
521 		goto exit;
522 	}
523 
524 	/* Handle in-sequence broadcast message */
525 	seqno = msg_seqno(msg);
526 	next_in = mod(node->bclink.last_in + 1);
527 	arrvq = &tn->bclink->arrvq;
528 	inputq = &tn->bclink->inputq;
529 
530 	if (likely(seqno == next_in)) {
531 receive:
532 		/* Deliver message to destination */
533 		if (likely(msg_isdata(msg))) {
534 			tipc_bclink_lock(net);
535 			bclink_accept_pkt(node, seqno);
536 			spin_lock_bh(&inputq->lock);
537 			__skb_queue_tail(arrvq, buf);
538 			spin_unlock_bh(&inputq->lock);
539 			node->action_flags |= TIPC_BCAST_MSG_EVT;
540 			tipc_bclink_unlock(net);
541 			tipc_node_unlock(node);
542 		} else if (msg_user(msg) == MSG_BUNDLER) {
543 			tipc_bclink_lock(net);
544 			bclink_accept_pkt(node, seqno);
545 			bcl->stats.recv_bundles++;
546 			bcl->stats.recv_bundled += msg_msgcnt(msg);
547 			pos = 0;
548 			while (tipc_msg_extract(buf, &iskb, &pos)) {
549 				spin_lock_bh(&inputq->lock);
550 				__skb_queue_tail(arrvq, iskb);
551 				spin_unlock_bh(&inputq->lock);
552 			}
553 			node->action_flags |= TIPC_BCAST_MSG_EVT;
554 			tipc_bclink_unlock(net);
555 			tipc_node_unlock(node);
556 		} else if (msg_user(msg) == MSG_FRAGMENTER) {
557 			tipc_bclink_lock(net);
558 			bclink_accept_pkt(node, seqno);
559 			tipc_buf_append(&node->bclink.reasm_buf, &buf);
560 			if (unlikely(!buf && !node->bclink.reasm_buf)) {
561 				tipc_bclink_unlock(net);
562 				goto unlock;
563 			}
564 			bcl->stats.recv_fragments++;
565 			if (buf) {
566 				bcl->stats.recv_fragmented++;
567 				msg = buf_msg(buf);
568 				tipc_bclink_unlock(net);
569 				goto receive;
570 			}
571 			tipc_bclink_unlock(net);
572 			tipc_node_unlock(node);
573 		} else {
574 			tipc_bclink_lock(net);
575 			bclink_accept_pkt(node, seqno);
576 			tipc_bclink_unlock(net);
577 			tipc_node_unlock(node);
578 			kfree_skb(buf);
579 		}
580 		buf = NULL;
581 
582 		/* Determine new synchronization state */
583 		tipc_node_lock(node);
584 		if (unlikely(!tipc_node_is_up(node)))
585 			goto unlock;
586 
587 		if (node->bclink.last_in == node->bclink.last_sent)
588 			goto unlock;
589 
590 		if (skb_queue_empty(&node->bclink.deferdq)) {
591 			node->bclink.oos_state = 1;
592 			goto unlock;
593 		}
594 
595 		msg = buf_msg(skb_peek(&node->bclink.deferdq));
596 		seqno = msg_seqno(msg);
597 		next_in = mod(next_in + 1);
598 		if (seqno != next_in)
599 			goto unlock;
600 
601 		/* Take in-sequence message from deferred queue & deliver it */
602 		buf = __skb_dequeue(&node->bclink.deferdq);
603 		goto receive;
604 	}
605 
606 	/* Handle out-of-sequence broadcast message */
607 	if (less(next_in, seqno)) {
608 		deferred = tipc_link_defer_pkt(&node->bclink.deferdq,
609 					       buf);
610 		bclink_update_last_sent(node, seqno);
611 		buf = NULL;
612 	}
613 
614 	tipc_bclink_lock(net);
615 
616 	if (deferred)
617 		bcl->stats.deferred_recv++;
618 	else
619 		bcl->stats.duplicates++;
620 
621 	tipc_bclink_unlock(net);
622 
623 unlock:
624 	tipc_node_unlock(node);
625 	tipc_node_put(node);
626 exit:
627 	kfree_skb(buf);
628 }
629 
630 u32 tipc_bclink_acks_missing(struct tipc_node *n_ptr)
631 {
632 	return (n_ptr->bclink.recv_permitted &&
633 		(tipc_bclink_get_last_sent(n_ptr->net) != n_ptr->bclink.acked));
634 }
635 
636 
637 /**
638  * tipc_bcbearer_send - send a packet through the broadcast pseudo-bearer
639  *
640  * Send packet over as many bearers as necessary to reach all nodes
641  * that have joined the broadcast link.
642  *
643  * Returns 0 (packet sent successfully) under all circumstances,
644  * since the broadcast link's pseudo-bearer never blocks
645  */
646 static int tipc_bcbearer_send(struct net *net, struct sk_buff *buf,
647 			      struct tipc_bearer *unused1,
648 			      struct tipc_media_addr *unused2)
649 {
650 	int bp_index;
651 	struct tipc_msg *msg = buf_msg(buf);
652 	struct tipc_net *tn = net_generic(net, tipc_net_id);
653 	struct tipc_bcbearer *bcbearer = tn->bcbearer;
654 	struct tipc_bclink *bclink = tn->bclink;
655 
656 	/* Prepare broadcast link message for reliable transmission,
657 	 * if first time trying to send it;
658 	 * preparation is skipped for broadcast link protocol messages
659 	 * since they are sent in an unreliable manner and don't need it
660 	 */
661 	if (likely(!msg_non_seq(buf_msg(buf)))) {
662 		bcbuf_set_acks(buf, bclink->bcast_nodes.count);
663 		msg_set_non_seq(msg, 1);
664 		msg_set_mc_netid(msg, tn->net_id);
665 		tn->bcl->stats.sent_info++;
666 		if (WARN_ON(!bclink->bcast_nodes.count)) {
667 			dump_stack();
668 			return 0;
669 		}
670 	}
671 
672 	/* Send buffer over bearers until all targets reached */
673 	bcbearer->remains = bclink->bcast_nodes;
674 
675 	for (bp_index = 0; bp_index < MAX_BEARERS; bp_index++) {
676 		struct tipc_bearer *p = bcbearer->bpairs[bp_index].primary;
677 		struct tipc_bearer *s = bcbearer->bpairs[bp_index].secondary;
678 		struct tipc_bearer *bp[2] = {p, s};
679 		struct tipc_bearer *b = bp[msg_link_selector(msg)];
680 		struct sk_buff *tbuf;
681 
682 		if (!p)
683 			break; /* No more bearers to try */
684 		if (!b)
685 			b = p;
686 		tipc_nmap_diff(&bcbearer->remains, &b->nodes,
687 			       &bcbearer->remains_new);
688 		if (bcbearer->remains_new.count == bcbearer->remains.count)
689 			continue; /* Nothing added by bearer pair */
690 
691 		if (bp_index == 0) {
692 			/* Use original buffer for first bearer */
693 			tipc_bearer_send(net, b->identity, buf, &b->bcast_addr);
694 		} else {
695 			/* Avoid concurrent buffer access */
696 			tbuf = pskb_copy_for_clone(buf, GFP_ATOMIC);
697 			if (!tbuf)
698 				break;
699 			tipc_bearer_send(net, b->identity, tbuf,
700 					 &b->bcast_addr);
701 			kfree_skb(tbuf); /* Bearer keeps a clone */
702 		}
703 		if (bcbearer->remains_new.count == 0)
704 			break; /* All targets reached */
705 
706 		bcbearer->remains = bcbearer->remains_new;
707 	}
708 
709 	return 0;
710 }
711 
712 /**
713  * tipc_bcbearer_sort - create sets of bearer pairs used by broadcast bearer
714  */
715 void tipc_bcbearer_sort(struct net *net, struct tipc_node_map *nm_ptr,
716 			u32 node, bool action)
717 {
718 	struct tipc_net *tn = net_generic(net, tipc_net_id);
719 	struct tipc_bcbearer *bcbearer = tn->bcbearer;
720 	struct tipc_bcbearer_pair *bp_temp = bcbearer->bpairs_temp;
721 	struct tipc_bcbearer_pair *bp_curr;
722 	struct tipc_bearer *b;
723 	int b_index;
724 	int pri;
725 
726 	tipc_bclink_lock(net);
727 
728 	if (action)
729 		tipc_nmap_add(nm_ptr, node);
730 	else
731 		tipc_nmap_remove(nm_ptr, node);
732 
733 	/* Group bearers by priority (can assume max of two per priority) */
734 	memset(bp_temp, 0, sizeof(bcbearer->bpairs_temp));
735 
736 	rcu_read_lock();
737 	for (b_index = 0; b_index < MAX_BEARERS; b_index++) {
738 		b = rcu_dereference_rtnl(tn->bearer_list[b_index]);
739 		if (!b || !b->nodes.count)
740 			continue;
741 
742 		if (!bp_temp[b->priority].primary)
743 			bp_temp[b->priority].primary = b;
744 		else
745 			bp_temp[b->priority].secondary = b;
746 	}
747 	rcu_read_unlock();
748 
749 	/* Create array of bearer pairs for broadcasting */
750 	bp_curr = bcbearer->bpairs;
751 	memset(bcbearer->bpairs, 0, sizeof(bcbearer->bpairs));
752 
753 	for (pri = TIPC_MAX_LINK_PRI; pri >= 0; pri--) {
754 
755 		if (!bp_temp[pri].primary)
756 			continue;
757 
758 		bp_curr->primary = bp_temp[pri].primary;
759 
760 		if (bp_temp[pri].secondary) {
761 			if (tipc_nmap_equal(&bp_temp[pri].primary->nodes,
762 					    &bp_temp[pri].secondary->nodes)) {
763 				bp_curr->secondary = bp_temp[pri].secondary;
764 			} else {
765 				bp_curr++;
766 				bp_curr->primary = bp_temp[pri].secondary;
767 			}
768 		}
769 
770 		bp_curr++;
771 	}
772 
773 	tipc_bclink_unlock(net);
774 }
775 
776 static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
777 				      struct tipc_stats *stats)
778 {
779 	int i;
780 	struct nlattr *nest;
781 
782 	struct nla_map {
783 		__u32 key;
784 		__u32 val;
785 	};
786 
787 	struct nla_map map[] = {
788 		{TIPC_NLA_STATS_RX_INFO, stats->recv_info},
789 		{TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
790 		{TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
791 		{TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
792 		{TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
793 		{TIPC_NLA_STATS_TX_INFO, stats->sent_info},
794 		{TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
795 		{TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
796 		{TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
797 		{TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
798 		{TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
799 		{TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
800 		{TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
801 		{TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
802 		{TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
803 		{TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
804 		{TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
805 		{TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
806 		{TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
807 			(stats->accu_queue_sz / stats->queue_sz_counts) : 0}
808 	};
809 
810 	nest = nla_nest_start(skb, TIPC_NLA_LINK_STATS);
811 	if (!nest)
812 		return -EMSGSIZE;
813 
814 	for (i = 0; i <  ARRAY_SIZE(map); i++)
815 		if (nla_put_u32(skb, map[i].key, map[i].val))
816 			goto msg_full;
817 
818 	nla_nest_end(skb, nest);
819 
820 	return 0;
821 msg_full:
822 	nla_nest_cancel(skb, nest);
823 
824 	return -EMSGSIZE;
825 }
826 
827 int tipc_nl_add_bc_link(struct net *net, struct tipc_nl_msg *msg)
828 {
829 	int err;
830 	void *hdr;
831 	struct nlattr *attrs;
832 	struct nlattr *prop;
833 	struct tipc_net *tn = net_generic(net, tipc_net_id);
834 	struct tipc_link *bcl = tn->bcl;
835 
836 	if (!bcl)
837 		return 0;
838 
839 	tipc_bclink_lock(net);
840 
841 	hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
842 			  NLM_F_MULTI, TIPC_NL_LINK_GET);
843 	if (!hdr)
844 		return -EMSGSIZE;
845 
846 	attrs = nla_nest_start(msg->skb, TIPC_NLA_LINK);
847 	if (!attrs)
848 		goto msg_full;
849 
850 	/* The broadcast link is always up */
851 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
852 		goto attr_msg_full;
853 
854 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
855 		goto attr_msg_full;
856 	if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
857 		goto attr_msg_full;
858 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, bcl->rcv_nxt))
859 		goto attr_msg_full;
860 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, bcl->snd_nxt))
861 		goto attr_msg_full;
862 
863 	prop = nla_nest_start(msg->skb, TIPC_NLA_LINK_PROP);
864 	if (!prop)
865 		goto attr_msg_full;
866 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->window))
867 		goto prop_msg_full;
868 	nla_nest_end(msg->skb, prop);
869 
870 	err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
871 	if (err)
872 		goto attr_msg_full;
873 
874 	tipc_bclink_unlock(net);
875 	nla_nest_end(msg->skb, attrs);
876 	genlmsg_end(msg->skb, hdr);
877 
878 	return 0;
879 
880 prop_msg_full:
881 	nla_nest_cancel(msg->skb, prop);
882 attr_msg_full:
883 	nla_nest_cancel(msg->skb, attrs);
884 msg_full:
885 	tipc_bclink_unlock(net);
886 	genlmsg_cancel(msg->skb, hdr);
887 
888 	return -EMSGSIZE;
889 }
890 
891 int tipc_bclink_reset_stats(struct net *net)
892 {
893 	struct tipc_net *tn = net_generic(net, tipc_net_id);
894 	struct tipc_link *bcl = tn->bcl;
895 
896 	if (!bcl)
897 		return -ENOPROTOOPT;
898 
899 	tipc_bclink_lock(net);
900 	memset(&bcl->stats, 0, sizeof(bcl->stats));
901 	tipc_bclink_unlock(net);
902 	return 0;
903 }
904 
905 int tipc_bclink_set_queue_limits(struct net *net, u32 limit)
906 {
907 	struct tipc_net *tn = net_generic(net, tipc_net_id);
908 	struct tipc_link *bcl = tn->bcl;
909 
910 	if (!bcl)
911 		return -ENOPROTOOPT;
912 	if (limit < BCLINK_WIN_MIN)
913 		limit = BCLINK_WIN_MIN;
914 	if (limit > TIPC_MAX_LINK_WIN)
915 		return -EINVAL;
916 	tipc_bclink_lock(net);
917 	tipc_link_set_queue_limits(bcl, limit);
918 	tipc_bclink_unlock(net);
919 	return 0;
920 }
921 
922 int tipc_nl_bc_link_set(struct net *net, struct nlattr *attrs[])
923 {
924 	int err;
925 	u32 win;
926 	struct nlattr *props[TIPC_NLA_PROP_MAX + 1];
927 
928 	if (!attrs[TIPC_NLA_LINK_PROP])
929 		return -EINVAL;
930 
931 	err = tipc_nl_parse_link_prop(attrs[TIPC_NLA_LINK_PROP], props);
932 	if (err)
933 		return err;
934 
935 	if (!props[TIPC_NLA_PROP_WIN])
936 		return -EOPNOTSUPP;
937 
938 	win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
939 
940 	return tipc_bclink_set_queue_limits(net, win);
941 }
942 
943 int tipc_bclink_init(struct net *net)
944 {
945 	struct tipc_net *tn = net_generic(net, tipc_net_id);
946 	struct tipc_bcbearer *bcbearer;
947 	struct tipc_bclink *bclink;
948 	struct tipc_link *bcl;
949 
950 	bcbearer = kzalloc(sizeof(*bcbearer), GFP_ATOMIC);
951 	if (!bcbearer)
952 		return -ENOMEM;
953 
954 	bclink = kzalloc(sizeof(*bclink), GFP_ATOMIC);
955 	if (!bclink) {
956 		kfree(bcbearer);
957 		return -ENOMEM;
958 	}
959 
960 	bcl = &bclink->link;
961 	bcbearer->bearer.media = &bcbearer->media;
962 	bcbearer->media.send_msg = tipc_bcbearer_send;
963 	sprintf(bcbearer->media.name, "tipc-broadcast");
964 
965 	spin_lock_init(&bclink->lock);
966 	__skb_queue_head_init(&bcl->transmq);
967 	__skb_queue_head_init(&bcl->backlogq);
968 	__skb_queue_head_init(&bcl->deferdq);
969 	skb_queue_head_init(&bcl->wakeupq);
970 	bcl->snd_nxt = 1;
971 	spin_lock_init(&bclink->node.lock);
972 	__skb_queue_head_init(&bclink->arrvq);
973 	skb_queue_head_init(&bclink->inputq);
974 	bcl->owner = &bclink->node;
975 	bcl->owner->net = net;
976 	bcl->mtu = MAX_PKT_DEFAULT_MCAST;
977 	tipc_link_set_queue_limits(bcl, BCLINK_WIN_DEFAULT);
978 	bcl->bearer_id = MAX_BEARERS;
979 	rcu_assign_pointer(tn->bearer_list[MAX_BEARERS], &bcbearer->bearer);
980 	bcl->pmsg = (struct tipc_msg *)&bcl->proto_msg;
981 	msg_set_prevnode(bcl->pmsg, tn->own_addr);
982 	strlcpy(bcl->name, tipc_bclink_name, TIPC_MAX_LINK_NAME);
983 	tn->bcbearer = bcbearer;
984 	tn->bclink = bclink;
985 	tn->bcl = bcl;
986 	return 0;
987 }
988 
989 void tipc_bclink_stop(struct net *net)
990 {
991 	struct tipc_net *tn = net_generic(net, tipc_net_id);
992 
993 	tipc_bclink_lock(net);
994 	tipc_link_purge_queues(tn->bcl);
995 	tipc_bclink_unlock(net);
996 
997 	RCU_INIT_POINTER(tn->bearer_list[BCBEARER], NULL);
998 	synchronize_net();
999 	kfree(tn->bcbearer);
1000 	kfree(tn->bclink);
1001 }
1002 
1003 /**
1004  * tipc_nmap_add - add a node to a node map
1005  */
1006 static void tipc_nmap_add(struct tipc_node_map *nm_ptr, u32 node)
1007 {
1008 	int n = tipc_node(node);
1009 	int w = n / WSIZE;
1010 	u32 mask = (1 << (n % WSIZE));
1011 
1012 	if ((nm_ptr->map[w] & mask) == 0) {
1013 		nm_ptr->count++;
1014 		nm_ptr->map[w] |= mask;
1015 	}
1016 }
1017 
1018 /**
1019  * tipc_nmap_remove - remove a node from a node map
1020  */
1021 static void tipc_nmap_remove(struct tipc_node_map *nm_ptr, u32 node)
1022 {
1023 	int n = tipc_node(node);
1024 	int w = n / WSIZE;
1025 	u32 mask = (1 << (n % WSIZE));
1026 
1027 	if ((nm_ptr->map[w] & mask) != 0) {
1028 		nm_ptr->map[w] &= ~mask;
1029 		nm_ptr->count--;
1030 	}
1031 }
1032 
1033 /**
1034  * tipc_nmap_diff - find differences between node maps
1035  * @nm_a: input node map A
1036  * @nm_b: input node map B
1037  * @nm_diff: output node map A-B (i.e. nodes of A that are not in B)
1038  */
1039 static void tipc_nmap_diff(struct tipc_node_map *nm_a,
1040 			   struct tipc_node_map *nm_b,
1041 			   struct tipc_node_map *nm_diff)
1042 {
1043 	int stop = ARRAY_SIZE(nm_a->map);
1044 	int w;
1045 	int b;
1046 	u32 map;
1047 
1048 	memset(nm_diff, 0, sizeof(*nm_diff));
1049 	for (w = 0; w < stop; w++) {
1050 		map = nm_a->map[w] ^ (nm_a->map[w] & nm_b->map[w]);
1051 		nm_diff->map[w] = map;
1052 		if (map != 0) {
1053 			for (b = 0 ; b < WSIZE; b++) {
1054 				if (map & (1 << b))
1055 					nm_diff->count++;
1056 			}
1057 		}
1058 	}
1059 }
1060