xref: /linux/net/tipc/link.c (revision fe8ecccc10b3adc071de05ca7af728ca1a4ac9aa)
1 /*
2  * net/tipc/link.c: TIPC link code
3  *
4  * Copyright (c) 1996-2007, 2012-2016, Ericsson AB
5  * Copyright (c) 2004-2007, 2010-2013, Wind River Systems
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions are met:
10  *
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the names of the copyright holders nor the names of its
17  *    contributors may be used to endorse or promote products derived from
18  *    this software without specific prior written permission.
19  *
20  * Alternatively, this software may be distributed under the terms of the
21  * GNU General Public License ("GPL") version 2 as published by the Free
22  * Software Foundation.
23  *
24  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34  * POSSIBILITY OF SUCH DAMAGE.
35  */
36 
37 #include "core.h"
38 #include "subscr.h"
39 #include "link.h"
40 #include "bcast.h"
41 #include "socket.h"
42 #include "name_distr.h"
43 #include "discover.h"
44 #include "netlink.h"
45 #include "monitor.h"
46 
47 #include <linux/pkt_sched.h>
48 
49 struct tipc_stats {
50 	u32 sent_pkts;
51 	u32 recv_pkts;
52 	u32 sent_states;
53 	u32 recv_states;
54 	u32 sent_probes;
55 	u32 recv_probes;
56 	u32 sent_nacks;
57 	u32 recv_nacks;
58 	u32 sent_acks;
59 	u32 sent_bundled;
60 	u32 sent_bundles;
61 	u32 recv_bundled;
62 	u32 recv_bundles;
63 	u32 retransmitted;
64 	u32 sent_fragmented;
65 	u32 sent_fragments;
66 	u32 recv_fragmented;
67 	u32 recv_fragments;
68 	u32 link_congs;		/* # port sends blocked by congestion */
69 	u32 deferred_recv;
70 	u32 duplicates;
71 	u32 max_queue_sz;	/* send queue size high water mark */
72 	u32 accu_queue_sz;	/* used for send queue size profiling */
73 	u32 queue_sz_counts;	/* used for send queue size profiling */
74 	u32 msg_length_counts;	/* used for message length profiling */
75 	u32 msg_lengths_total;	/* used for message length profiling */
76 	u32 msg_length_profile[7]; /* used for msg. length profiling */
77 };
78 
79 /**
80  * struct tipc_link - TIPC link data structure
81  * @addr: network address of link's peer node
82  * @name: link name character string
83  * @media_addr: media address to use when sending messages over link
84  * @timer: link timer
85  * @net: pointer to namespace struct
86  * @refcnt: reference counter for permanent references (owner node & timer)
87  * @peer_session: link session # being used by peer end of link
88  * @peer_bearer_id: bearer id used by link's peer endpoint
89  * @bearer_id: local bearer id used by link
90  * @tolerance: minimum link continuity loss needed to reset link [in ms]
91  * @abort_limit: # of unacknowledged continuity probes needed to reset link
92  * @state: current state of link FSM
93  * @peer_caps: bitmap describing capabilities of peer node
94  * @silent_intv_cnt: # of timer intervals without any reception from peer
95  * @proto_msg: template for control messages generated by link
96  * @pmsg: convenience pointer to "proto_msg" field
97  * @priority: current link priority
98  * @net_plane: current link network plane ('A' through 'H')
99  * @mon_state: cookie with information needed by link monitor
100  * @backlog_limit: backlog queue congestion thresholds (indexed by importance)
101  * @exp_msg_count: # of tunnelled messages expected during link changeover
102  * @reset_rcv_checkpt: seq # of last acknowledged message at time of link reset
103  * @mtu: current maximum packet size for this link
104  * @advertised_mtu: advertised own mtu when link is being established
105  * @transmitq: queue for sent, non-acked messages
106  * @backlogq: queue for messages waiting to be sent
107  * @snt_nxt: next sequence number to use for outbound messages
108  * @last_retransmitted: sequence number of most recently retransmitted message
109  * @stale_cnt: counter for number of identical retransmit attempts
110  * @stale_limit: time when repeated identical retransmits must force link reset
111  * @ackers: # of peers that needs to ack each packet before it can be released
112  * @acked: # last packet acked by a certain peer. Used for broadcast.
113  * @rcv_nxt: next sequence number to expect for inbound messages
114  * @deferred_queue: deferred queue saved OOS b'cast message received from node
115  * @unacked_window: # of inbound messages rx'd without ack'ing back to peer
116  * @inputq: buffer queue for messages to be delivered upwards
117  * @namedq: buffer queue for name table messages to be delivered upwards
118  * @next_out: ptr to first unsent outbound message in queue
119  * @wakeupq: linked list of wakeup msgs waiting for link congestion to abate
120  * @long_msg_seq_no: next identifier to use for outbound fragmented messages
121  * @reasm_buf: head of partially reassembled inbound message fragments
122  * @bc_rcvr: marks that this is a broadcast receiver link
123  * @stats: collects statistics regarding link activity
124  */
125 struct tipc_link {
126 	u32 addr;
127 	char name[TIPC_MAX_LINK_NAME];
128 	struct net *net;
129 
130 	/* Management and link supervision data */
131 	u16 peer_session;
132 	u16 session;
133 	u16 snd_nxt_state;
134 	u16 rcv_nxt_state;
135 	u32 peer_bearer_id;
136 	u32 bearer_id;
137 	u32 tolerance;
138 	u32 abort_limit;
139 	u32 state;
140 	u16 peer_caps;
141 	bool in_session;
142 	bool active;
143 	u32 silent_intv_cnt;
144 	char if_name[TIPC_MAX_IF_NAME];
145 	u32 priority;
146 	char net_plane;
147 	struct tipc_mon_state mon_state;
148 	u16 rst_cnt;
149 
150 	/* Failover/synch */
151 	u16 drop_point;
152 	struct sk_buff *failover_reasm_skb;
153 
154 	/* Max packet negotiation */
155 	u16 mtu;
156 	u16 advertised_mtu;
157 
158 	/* Sending */
159 	struct sk_buff_head transmq;
160 	struct sk_buff_head backlogq;
161 	struct {
162 		u16 len;
163 		u16 limit;
164 	} backlog[5];
165 	u16 snd_nxt;
166 	u16 last_retransm;
167 	u16 window;
168 	u16 stale_cnt;
169 	unsigned long stale_limit;
170 
171 	/* Reception */
172 	u16 rcv_nxt;
173 	u32 rcv_unacked;
174 	struct sk_buff_head deferdq;
175 	struct sk_buff_head *inputq;
176 	struct sk_buff_head *namedq;
177 
178 	/* Congestion handling */
179 	struct sk_buff_head wakeupq;
180 
181 	/* Fragmentation/reassembly */
182 	struct sk_buff *reasm_buf;
183 
184 	/* Broadcast */
185 	u16 ackers;
186 	u16 acked;
187 	struct tipc_link *bc_rcvlink;
188 	struct tipc_link *bc_sndlink;
189 	unsigned long prev_retr;
190 	u16 prev_from;
191 	u16 prev_to;
192 	u8 nack_state;
193 	bool bc_peer_is_up;
194 
195 	/* Statistics */
196 	struct tipc_stats stats;
197 };
198 
199 /*
200  * Error message prefixes
201  */
202 static const char *link_co_err = "Link tunneling error, ";
203 static const char *link_rst_msg = "Resetting link ";
204 
205 /* Send states for broadcast NACKs
206  */
207 enum {
208 	BC_NACK_SND_CONDITIONAL,
209 	BC_NACK_SND_UNCONDITIONAL,
210 	BC_NACK_SND_SUPPRESS,
211 };
212 
213 #define TIPC_BC_RETR_LIMIT 10   /* [ms] */
214 
215 /*
216  * Interval between NACKs when packets arrive out of order
217  */
218 #define TIPC_NACK_INTV (TIPC_MIN_LINK_WIN * 2)
219 
220 /* Link FSM states:
221  */
222 enum {
223 	LINK_ESTABLISHED     = 0xe,
224 	LINK_ESTABLISHING    = 0xe  << 4,
225 	LINK_RESET           = 0x1  << 8,
226 	LINK_RESETTING       = 0x2  << 12,
227 	LINK_PEER_RESET      = 0xd  << 16,
228 	LINK_FAILINGOVER     = 0xf  << 20,
229 	LINK_SYNCHING        = 0xc  << 24
230 };
231 
232 /* Link FSM state checking routines
233  */
234 static int link_is_up(struct tipc_link *l)
235 {
236 	return l->state & (LINK_ESTABLISHED | LINK_SYNCHING);
237 }
238 
239 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
240 			       struct sk_buff_head *xmitq);
241 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
242 				      bool probe_reply, u16 rcvgap,
243 				      int tolerance, int priority,
244 				      struct sk_buff_head *xmitq);
245 static void link_print(struct tipc_link *l, const char *str);
246 static int tipc_link_build_nack_msg(struct tipc_link *l,
247 				    struct sk_buff_head *xmitq);
248 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
249 					struct sk_buff_head *xmitq);
250 static bool tipc_link_release_pkts(struct tipc_link *l, u16 to);
251 
252 /*
253  *  Simple non-static link routines (i.e. referenced outside this file)
254  */
255 bool tipc_link_is_up(struct tipc_link *l)
256 {
257 	return link_is_up(l);
258 }
259 
260 bool tipc_link_peer_is_down(struct tipc_link *l)
261 {
262 	return l->state == LINK_PEER_RESET;
263 }
264 
265 bool tipc_link_is_reset(struct tipc_link *l)
266 {
267 	return l->state & (LINK_RESET | LINK_FAILINGOVER | LINK_ESTABLISHING);
268 }
269 
270 bool tipc_link_is_establishing(struct tipc_link *l)
271 {
272 	return l->state == LINK_ESTABLISHING;
273 }
274 
275 bool tipc_link_is_synching(struct tipc_link *l)
276 {
277 	return l->state == LINK_SYNCHING;
278 }
279 
280 bool tipc_link_is_failingover(struct tipc_link *l)
281 {
282 	return l->state == LINK_FAILINGOVER;
283 }
284 
285 bool tipc_link_is_blocked(struct tipc_link *l)
286 {
287 	return l->state & (LINK_RESETTING | LINK_PEER_RESET | LINK_FAILINGOVER);
288 }
289 
290 static bool link_is_bc_sndlink(struct tipc_link *l)
291 {
292 	return !l->bc_sndlink;
293 }
294 
295 static bool link_is_bc_rcvlink(struct tipc_link *l)
296 {
297 	return ((l->bc_rcvlink == l) && !link_is_bc_sndlink(l));
298 }
299 
300 void tipc_link_set_active(struct tipc_link *l, bool active)
301 {
302 	l->active = active;
303 }
304 
305 u32 tipc_link_id(struct tipc_link *l)
306 {
307 	return l->peer_bearer_id << 16 | l->bearer_id;
308 }
309 
310 int tipc_link_window(struct tipc_link *l)
311 {
312 	return l->window;
313 }
314 
315 int tipc_link_prio(struct tipc_link *l)
316 {
317 	return l->priority;
318 }
319 
320 unsigned long tipc_link_tolerance(struct tipc_link *l)
321 {
322 	return l->tolerance;
323 }
324 
325 struct sk_buff_head *tipc_link_inputq(struct tipc_link *l)
326 {
327 	return l->inputq;
328 }
329 
330 char tipc_link_plane(struct tipc_link *l)
331 {
332 	return l->net_plane;
333 }
334 
335 void tipc_link_update_caps(struct tipc_link *l, u16 capabilities)
336 {
337 	l->peer_caps = capabilities;
338 }
339 
340 void tipc_link_add_bc_peer(struct tipc_link *snd_l,
341 			   struct tipc_link *uc_l,
342 			   struct sk_buff_head *xmitq)
343 {
344 	struct tipc_link *rcv_l = uc_l->bc_rcvlink;
345 
346 	snd_l->ackers++;
347 	rcv_l->acked = snd_l->snd_nxt - 1;
348 	snd_l->state = LINK_ESTABLISHED;
349 	tipc_link_build_bc_init_msg(uc_l, xmitq);
350 }
351 
352 void tipc_link_remove_bc_peer(struct tipc_link *snd_l,
353 			      struct tipc_link *rcv_l,
354 			      struct sk_buff_head *xmitq)
355 {
356 	u16 ack = snd_l->snd_nxt - 1;
357 
358 	snd_l->ackers--;
359 	rcv_l->bc_peer_is_up = true;
360 	rcv_l->state = LINK_ESTABLISHED;
361 	tipc_link_bc_ack_rcv(rcv_l, ack, xmitq);
362 	tipc_link_reset(rcv_l);
363 	rcv_l->state = LINK_RESET;
364 	if (!snd_l->ackers) {
365 		tipc_link_reset(snd_l);
366 		snd_l->state = LINK_RESET;
367 		__skb_queue_purge(xmitq);
368 	}
369 }
370 
371 int tipc_link_bc_peers(struct tipc_link *l)
372 {
373 	return l->ackers;
374 }
375 
376 static u16 link_bc_rcv_gap(struct tipc_link *l)
377 {
378 	struct sk_buff *skb = skb_peek(&l->deferdq);
379 	u16 gap = 0;
380 
381 	if (more(l->snd_nxt, l->rcv_nxt))
382 		gap = l->snd_nxt - l->rcv_nxt;
383 	if (skb)
384 		gap = buf_seqno(skb) - l->rcv_nxt;
385 	return gap;
386 }
387 
388 void tipc_link_set_mtu(struct tipc_link *l, int mtu)
389 {
390 	l->mtu = mtu;
391 }
392 
393 int tipc_link_mtu(struct tipc_link *l)
394 {
395 	return l->mtu;
396 }
397 
398 u16 tipc_link_rcv_nxt(struct tipc_link *l)
399 {
400 	return l->rcv_nxt;
401 }
402 
403 u16 tipc_link_acked(struct tipc_link *l)
404 {
405 	return l->acked;
406 }
407 
408 char *tipc_link_name(struct tipc_link *l)
409 {
410 	return l->name;
411 }
412 
413 u32 tipc_link_state(struct tipc_link *l)
414 {
415 	return l->state;
416 }
417 
418 /**
419  * tipc_link_create - create a new link
420  * @n: pointer to associated node
421  * @if_name: associated interface name
422  * @bearer_id: id (index) of associated bearer
423  * @tolerance: link tolerance to be used by link
424  * @net_plane: network plane (A,B,c..) this link belongs to
425  * @mtu: mtu to be advertised by link
426  * @priority: priority to be used by link
427  * @window: send window to be used by link
428  * @session: session to be used by link
429  * @ownnode: identity of own node
430  * @peer: node id of peer node
431  * @peer_caps: bitmap describing peer node capabilities
432  * @bc_sndlink: the namespace global link used for broadcast sending
433  * @bc_rcvlink: the peer specific link used for broadcast reception
434  * @inputq: queue to put messages ready for delivery
435  * @namedq: queue to put binding table update messages ready for delivery
436  * @link: return value, pointer to put the created link
437  *
438  * Returns true if link was created, otherwise false
439  */
440 bool tipc_link_create(struct net *net, char *if_name, int bearer_id,
441 		      int tolerance, char net_plane, u32 mtu, int priority,
442 		      int window, u32 session, u32 self,
443 		      u32 peer, u8 *peer_id, u16 peer_caps,
444 		      struct tipc_link *bc_sndlink,
445 		      struct tipc_link *bc_rcvlink,
446 		      struct sk_buff_head *inputq,
447 		      struct sk_buff_head *namedq,
448 		      struct tipc_link **link)
449 {
450 	char peer_str[NODE_ID_STR_LEN] = {0,};
451 	char self_str[NODE_ID_STR_LEN] = {0,};
452 	struct tipc_link *l;
453 
454 	l = kzalloc(sizeof(*l), GFP_ATOMIC);
455 	if (!l)
456 		return false;
457 	*link = l;
458 	l->session = session;
459 
460 	/* Set link name for unicast links only */
461 	if (peer_id) {
462 		tipc_nodeid2string(self_str, tipc_own_id(net));
463 		if (strlen(self_str) > 16)
464 			sprintf(self_str, "%x", self);
465 		tipc_nodeid2string(peer_str, peer_id);
466 		if (strlen(peer_str) > 16)
467 			sprintf(peer_str, "%x", peer);
468 	}
469 	/* Peer i/f name will be completed by reset/activate message */
470 	snprintf(l->name, sizeof(l->name), "%s:%s-%s:unknown",
471 		 self_str, if_name, peer_str);
472 
473 	strcpy(l->if_name, if_name);
474 	l->addr = peer;
475 	l->peer_caps = peer_caps;
476 	l->net = net;
477 	l->in_session = false;
478 	l->bearer_id = bearer_id;
479 	l->tolerance = tolerance;
480 	if (bc_rcvlink)
481 		bc_rcvlink->tolerance = tolerance;
482 	l->net_plane = net_plane;
483 	l->advertised_mtu = mtu;
484 	l->mtu = mtu;
485 	l->priority = priority;
486 	tipc_link_set_queue_limits(l, window);
487 	l->ackers = 1;
488 	l->bc_sndlink = bc_sndlink;
489 	l->bc_rcvlink = bc_rcvlink;
490 	l->inputq = inputq;
491 	l->namedq = namedq;
492 	l->state = LINK_RESETTING;
493 	__skb_queue_head_init(&l->transmq);
494 	__skb_queue_head_init(&l->backlogq);
495 	__skb_queue_head_init(&l->deferdq);
496 	skb_queue_head_init(&l->wakeupq);
497 	skb_queue_head_init(l->inputq);
498 	return true;
499 }
500 
501 /**
502  * tipc_link_bc_create - create new link to be used for broadcast
503  * @n: pointer to associated node
504  * @mtu: mtu to be used initially if no peers
505  * @window: send window to be used
506  * @inputq: queue to put messages ready for delivery
507  * @namedq: queue to put binding table update messages ready for delivery
508  * @link: return value, pointer to put the created link
509  *
510  * Returns true if link was created, otherwise false
511  */
512 bool tipc_link_bc_create(struct net *net, u32 ownnode, u32 peer,
513 			 int mtu, int window, u16 peer_caps,
514 			 struct sk_buff_head *inputq,
515 			 struct sk_buff_head *namedq,
516 			 struct tipc_link *bc_sndlink,
517 			 struct tipc_link **link)
518 {
519 	struct tipc_link *l;
520 
521 	if (!tipc_link_create(net, "", MAX_BEARERS, 0, 'Z', mtu, 0, window,
522 			      0, ownnode, peer, NULL, peer_caps, bc_sndlink,
523 			      NULL, inputq, namedq, link))
524 		return false;
525 
526 	l = *link;
527 	strcpy(l->name, tipc_bclink_name);
528 	tipc_link_reset(l);
529 	l->state = LINK_RESET;
530 	l->ackers = 0;
531 	l->bc_rcvlink = l;
532 
533 	/* Broadcast send link is always up */
534 	if (link_is_bc_sndlink(l))
535 		l->state = LINK_ESTABLISHED;
536 
537 	/* Disable replicast if even a single peer doesn't support it */
538 	if (link_is_bc_rcvlink(l) && !(peer_caps & TIPC_BCAST_RCAST))
539 		tipc_bcast_disable_rcast(net);
540 
541 	return true;
542 }
543 
544 /**
545  * tipc_link_fsm_evt - link finite state machine
546  * @l: pointer to link
547  * @evt: state machine event to be processed
548  */
549 int tipc_link_fsm_evt(struct tipc_link *l, int evt)
550 {
551 	int rc = 0;
552 
553 	switch (l->state) {
554 	case LINK_RESETTING:
555 		switch (evt) {
556 		case LINK_PEER_RESET_EVT:
557 			l->state = LINK_PEER_RESET;
558 			break;
559 		case LINK_RESET_EVT:
560 			l->state = LINK_RESET;
561 			break;
562 		case LINK_FAILURE_EVT:
563 		case LINK_FAILOVER_BEGIN_EVT:
564 		case LINK_ESTABLISH_EVT:
565 		case LINK_FAILOVER_END_EVT:
566 		case LINK_SYNCH_BEGIN_EVT:
567 		case LINK_SYNCH_END_EVT:
568 		default:
569 			goto illegal_evt;
570 		}
571 		break;
572 	case LINK_RESET:
573 		switch (evt) {
574 		case LINK_PEER_RESET_EVT:
575 			l->state = LINK_ESTABLISHING;
576 			break;
577 		case LINK_FAILOVER_BEGIN_EVT:
578 			l->state = LINK_FAILINGOVER;
579 		case LINK_FAILURE_EVT:
580 		case LINK_RESET_EVT:
581 		case LINK_ESTABLISH_EVT:
582 		case LINK_FAILOVER_END_EVT:
583 			break;
584 		case LINK_SYNCH_BEGIN_EVT:
585 		case LINK_SYNCH_END_EVT:
586 		default:
587 			goto illegal_evt;
588 		}
589 		break;
590 	case LINK_PEER_RESET:
591 		switch (evt) {
592 		case LINK_RESET_EVT:
593 			l->state = LINK_ESTABLISHING;
594 			break;
595 		case LINK_PEER_RESET_EVT:
596 		case LINK_ESTABLISH_EVT:
597 		case LINK_FAILURE_EVT:
598 			break;
599 		case LINK_SYNCH_BEGIN_EVT:
600 		case LINK_SYNCH_END_EVT:
601 		case LINK_FAILOVER_BEGIN_EVT:
602 		case LINK_FAILOVER_END_EVT:
603 		default:
604 			goto illegal_evt;
605 		}
606 		break;
607 	case LINK_FAILINGOVER:
608 		switch (evt) {
609 		case LINK_FAILOVER_END_EVT:
610 			l->state = LINK_RESET;
611 			break;
612 		case LINK_PEER_RESET_EVT:
613 		case LINK_RESET_EVT:
614 		case LINK_ESTABLISH_EVT:
615 		case LINK_FAILURE_EVT:
616 			break;
617 		case LINK_FAILOVER_BEGIN_EVT:
618 		case LINK_SYNCH_BEGIN_EVT:
619 		case LINK_SYNCH_END_EVT:
620 		default:
621 			goto illegal_evt;
622 		}
623 		break;
624 	case LINK_ESTABLISHING:
625 		switch (evt) {
626 		case LINK_ESTABLISH_EVT:
627 			l->state = LINK_ESTABLISHED;
628 			break;
629 		case LINK_FAILOVER_BEGIN_EVT:
630 			l->state = LINK_FAILINGOVER;
631 			break;
632 		case LINK_RESET_EVT:
633 			l->state = LINK_RESET;
634 			break;
635 		case LINK_FAILURE_EVT:
636 		case LINK_PEER_RESET_EVT:
637 		case LINK_SYNCH_BEGIN_EVT:
638 		case LINK_FAILOVER_END_EVT:
639 			break;
640 		case LINK_SYNCH_END_EVT:
641 		default:
642 			goto illegal_evt;
643 		}
644 		break;
645 	case LINK_ESTABLISHED:
646 		switch (evt) {
647 		case LINK_PEER_RESET_EVT:
648 			l->state = LINK_PEER_RESET;
649 			rc |= TIPC_LINK_DOWN_EVT;
650 			break;
651 		case LINK_FAILURE_EVT:
652 			l->state = LINK_RESETTING;
653 			rc |= TIPC_LINK_DOWN_EVT;
654 			break;
655 		case LINK_RESET_EVT:
656 			l->state = LINK_RESET;
657 			break;
658 		case LINK_ESTABLISH_EVT:
659 		case LINK_SYNCH_END_EVT:
660 			break;
661 		case LINK_SYNCH_BEGIN_EVT:
662 			l->state = LINK_SYNCHING;
663 			break;
664 		case LINK_FAILOVER_BEGIN_EVT:
665 		case LINK_FAILOVER_END_EVT:
666 		default:
667 			goto illegal_evt;
668 		}
669 		break;
670 	case LINK_SYNCHING:
671 		switch (evt) {
672 		case LINK_PEER_RESET_EVT:
673 			l->state = LINK_PEER_RESET;
674 			rc |= TIPC_LINK_DOWN_EVT;
675 			break;
676 		case LINK_FAILURE_EVT:
677 			l->state = LINK_RESETTING;
678 			rc |= TIPC_LINK_DOWN_EVT;
679 			break;
680 		case LINK_RESET_EVT:
681 			l->state = LINK_RESET;
682 			break;
683 		case LINK_ESTABLISH_EVT:
684 		case LINK_SYNCH_BEGIN_EVT:
685 			break;
686 		case LINK_SYNCH_END_EVT:
687 			l->state = LINK_ESTABLISHED;
688 			break;
689 		case LINK_FAILOVER_BEGIN_EVT:
690 		case LINK_FAILOVER_END_EVT:
691 		default:
692 			goto illegal_evt;
693 		}
694 		break;
695 	default:
696 		pr_err("Unknown FSM state %x in %s\n", l->state, l->name);
697 	}
698 	return rc;
699 illegal_evt:
700 	pr_err("Illegal FSM event %x in state %x on link %s\n",
701 	       evt, l->state, l->name);
702 	return rc;
703 }
704 
705 /* link_profile_stats - update statistical profiling of traffic
706  */
707 static void link_profile_stats(struct tipc_link *l)
708 {
709 	struct sk_buff *skb;
710 	struct tipc_msg *msg;
711 	int length;
712 
713 	/* Update counters used in statistical profiling of send traffic */
714 	l->stats.accu_queue_sz += skb_queue_len(&l->transmq);
715 	l->stats.queue_sz_counts++;
716 
717 	skb = skb_peek(&l->transmq);
718 	if (!skb)
719 		return;
720 	msg = buf_msg(skb);
721 	length = msg_size(msg);
722 
723 	if (msg_user(msg) == MSG_FRAGMENTER) {
724 		if (msg_type(msg) != FIRST_FRAGMENT)
725 			return;
726 		length = msg_size(msg_get_wrapped(msg));
727 	}
728 	l->stats.msg_lengths_total += length;
729 	l->stats.msg_length_counts++;
730 	if (length <= 64)
731 		l->stats.msg_length_profile[0]++;
732 	else if (length <= 256)
733 		l->stats.msg_length_profile[1]++;
734 	else if (length <= 1024)
735 		l->stats.msg_length_profile[2]++;
736 	else if (length <= 4096)
737 		l->stats.msg_length_profile[3]++;
738 	else if (length <= 16384)
739 		l->stats.msg_length_profile[4]++;
740 	else if (length <= 32768)
741 		l->stats.msg_length_profile[5]++;
742 	else
743 		l->stats.msg_length_profile[6]++;
744 }
745 
746 /* tipc_link_timeout - perform periodic task as instructed from node timeout
747  */
748 int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq)
749 {
750 	int mtyp = 0;
751 	int rc = 0;
752 	bool state = false;
753 	bool probe = false;
754 	bool setup = false;
755 	u16 bc_snt = l->bc_sndlink->snd_nxt - 1;
756 	u16 bc_acked = l->bc_rcvlink->acked;
757 	struct tipc_mon_state *mstate = &l->mon_state;
758 
759 	switch (l->state) {
760 	case LINK_ESTABLISHED:
761 	case LINK_SYNCHING:
762 		mtyp = STATE_MSG;
763 		link_profile_stats(l);
764 		tipc_mon_get_state(l->net, l->addr, mstate, l->bearer_id);
765 		if (mstate->reset || (l->silent_intv_cnt > l->abort_limit))
766 			return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
767 		state = bc_acked != bc_snt;
768 		state |= l->bc_rcvlink->rcv_unacked;
769 		state |= l->rcv_unacked;
770 		state |= !skb_queue_empty(&l->transmq);
771 		state |= !skb_queue_empty(&l->deferdq);
772 		probe = mstate->probing;
773 		probe |= l->silent_intv_cnt;
774 		if (probe || mstate->monitoring)
775 			l->silent_intv_cnt++;
776 		break;
777 	case LINK_RESET:
778 		setup = l->rst_cnt++ <= 4;
779 		setup |= !(l->rst_cnt % 16);
780 		mtyp = RESET_MSG;
781 		break;
782 	case LINK_ESTABLISHING:
783 		setup = true;
784 		mtyp = ACTIVATE_MSG;
785 		break;
786 	case LINK_PEER_RESET:
787 	case LINK_RESETTING:
788 	case LINK_FAILINGOVER:
789 		break;
790 	default:
791 		break;
792 	}
793 
794 	if (state || probe || setup)
795 		tipc_link_build_proto_msg(l, mtyp, probe, 0, 0, 0, 0, xmitq);
796 
797 	return rc;
798 }
799 
800 /**
801  * link_schedule_user - schedule a message sender for wakeup after congestion
802  * @l: congested link
803  * @hdr: header of message that is being sent
804  * Create pseudo msg to send back to user when congestion abates
805  */
806 static int link_schedule_user(struct tipc_link *l, struct tipc_msg *hdr)
807 {
808 	u32 dnode = tipc_own_addr(l->net);
809 	u32 dport = msg_origport(hdr);
810 	struct sk_buff *skb;
811 
812 	/* Create and schedule wakeup pseudo message */
813 	skb = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
814 			      dnode, l->addr, dport, 0, 0);
815 	if (!skb)
816 		return -ENOBUFS;
817 	msg_set_dest_droppable(buf_msg(skb), true);
818 	TIPC_SKB_CB(skb)->chain_imp = msg_importance(hdr);
819 	skb_queue_tail(&l->wakeupq, skb);
820 	l->stats.link_congs++;
821 	return -ELINKCONG;
822 }
823 
824 /**
825  * link_prepare_wakeup - prepare users for wakeup after congestion
826  * @l: congested link
827  * Wake up a number of waiting users, as permitted by available space
828  * in the send queue
829  */
830 static void link_prepare_wakeup(struct tipc_link *l)
831 {
832 	struct sk_buff *skb, *tmp;
833 	int imp, i = 0;
834 
835 	skb_queue_walk_safe(&l->wakeupq, skb, tmp) {
836 		imp = TIPC_SKB_CB(skb)->chain_imp;
837 		if (l->backlog[imp].len < l->backlog[imp].limit) {
838 			skb_unlink(skb, &l->wakeupq);
839 			skb_queue_tail(l->inputq, skb);
840 		} else if (i++ > 10) {
841 			break;
842 		}
843 	}
844 }
845 
846 void tipc_link_reset(struct tipc_link *l)
847 {
848 	struct sk_buff_head list;
849 
850 	__skb_queue_head_init(&list);
851 
852 	l->in_session = false;
853 	l->session++;
854 	l->mtu = l->advertised_mtu;
855 
856 	spin_lock_bh(&l->wakeupq.lock);
857 	skb_queue_splice_init(&l->wakeupq, &list);
858 	spin_unlock_bh(&l->wakeupq.lock);
859 
860 	spin_lock_bh(&l->inputq->lock);
861 	skb_queue_splice_init(&list, l->inputq);
862 	spin_unlock_bh(&l->inputq->lock);
863 
864 	__skb_queue_purge(&l->transmq);
865 	__skb_queue_purge(&l->deferdq);
866 	__skb_queue_purge(&l->backlogq);
867 	l->backlog[TIPC_LOW_IMPORTANCE].len = 0;
868 	l->backlog[TIPC_MEDIUM_IMPORTANCE].len = 0;
869 	l->backlog[TIPC_HIGH_IMPORTANCE].len = 0;
870 	l->backlog[TIPC_CRITICAL_IMPORTANCE].len = 0;
871 	l->backlog[TIPC_SYSTEM_IMPORTANCE].len = 0;
872 	kfree_skb(l->reasm_buf);
873 	kfree_skb(l->failover_reasm_skb);
874 	l->reasm_buf = NULL;
875 	l->failover_reasm_skb = NULL;
876 	l->rcv_unacked = 0;
877 	l->snd_nxt = 1;
878 	l->rcv_nxt = 1;
879 	l->snd_nxt_state = 1;
880 	l->rcv_nxt_state = 1;
881 	l->acked = 0;
882 	l->silent_intv_cnt = 0;
883 	l->rst_cnt = 0;
884 	l->stale_cnt = 0;
885 	l->bc_peer_is_up = false;
886 	memset(&l->mon_state, 0, sizeof(l->mon_state));
887 	tipc_link_reset_stats(l);
888 }
889 
890 /**
891  * tipc_link_xmit(): enqueue buffer list according to queue situation
892  * @link: link to use
893  * @list: chain of buffers containing message
894  * @xmitq: returned list of packets to be sent by caller
895  *
896  * Consumes the buffer chain.
897  * Returns 0 if success, or errno: -ELINKCONG, -EMSGSIZE or -ENOBUFS
898  * Messages at TIPC_SYSTEM_IMPORTANCE are always accepted
899  */
900 int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list,
901 		   struct sk_buff_head *xmitq)
902 {
903 	struct tipc_msg *hdr = buf_msg(skb_peek(list));
904 	unsigned int maxwin = l->window;
905 	int imp = msg_importance(hdr);
906 	unsigned int mtu = l->mtu;
907 	u16 ack = l->rcv_nxt - 1;
908 	u16 seqno = l->snd_nxt;
909 	u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
910 	struct sk_buff_head *transmq = &l->transmq;
911 	struct sk_buff_head *backlogq = &l->backlogq;
912 	struct sk_buff *skb, *_skb, *bskb;
913 	int pkt_cnt = skb_queue_len(list);
914 	int rc = 0;
915 
916 	if (unlikely(msg_size(hdr) > mtu)) {
917 		skb_queue_purge(list);
918 		return -EMSGSIZE;
919 	}
920 
921 	/* Allow oversubscription of one data msg per source at congestion */
922 	if (unlikely(l->backlog[imp].len >= l->backlog[imp].limit)) {
923 		if (imp == TIPC_SYSTEM_IMPORTANCE) {
924 			pr_warn("%s<%s>, link overflow", link_rst_msg, l->name);
925 			return -ENOBUFS;
926 		}
927 		rc = link_schedule_user(l, hdr);
928 	}
929 
930 	if (pkt_cnt > 1) {
931 		l->stats.sent_fragmented++;
932 		l->stats.sent_fragments += pkt_cnt;
933 	}
934 
935 	/* Prepare each packet for sending, and add to relevant queue: */
936 	while (skb_queue_len(list)) {
937 		skb = skb_peek(list);
938 		hdr = buf_msg(skb);
939 		msg_set_seqno(hdr, seqno);
940 		msg_set_ack(hdr, ack);
941 		msg_set_bcast_ack(hdr, bc_ack);
942 
943 		if (likely(skb_queue_len(transmq) < maxwin)) {
944 			_skb = skb_clone(skb, GFP_ATOMIC);
945 			if (!_skb) {
946 				skb_queue_purge(list);
947 				return -ENOBUFS;
948 			}
949 			__skb_dequeue(list);
950 			__skb_queue_tail(transmq, skb);
951 			__skb_queue_tail(xmitq, _skb);
952 			TIPC_SKB_CB(skb)->ackers = l->ackers;
953 			l->rcv_unacked = 0;
954 			l->stats.sent_pkts++;
955 			seqno++;
956 			continue;
957 		}
958 		if (tipc_msg_bundle(skb_peek_tail(backlogq), hdr, mtu)) {
959 			kfree_skb(__skb_dequeue(list));
960 			l->stats.sent_bundled++;
961 			continue;
962 		}
963 		if (tipc_msg_make_bundle(&bskb, hdr, mtu, l->addr)) {
964 			kfree_skb(__skb_dequeue(list));
965 			__skb_queue_tail(backlogq, bskb);
966 			l->backlog[msg_importance(buf_msg(bskb))].len++;
967 			l->stats.sent_bundled++;
968 			l->stats.sent_bundles++;
969 			continue;
970 		}
971 		l->backlog[imp].len += skb_queue_len(list);
972 		skb_queue_splice_tail_init(list, backlogq);
973 	}
974 	l->snd_nxt = seqno;
975 	return rc;
976 }
977 
978 static void tipc_link_advance_backlog(struct tipc_link *l,
979 				      struct sk_buff_head *xmitq)
980 {
981 	struct sk_buff *skb, *_skb;
982 	struct tipc_msg *hdr;
983 	u16 seqno = l->snd_nxt;
984 	u16 ack = l->rcv_nxt - 1;
985 	u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
986 
987 	while (skb_queue_len(&l->transmq) < l->window) {
988 		skb = skb_peek(&l->backlogq);
989 		if (!skb)
990 			break;
991 		_skb = skb_clone(skb, GFP_ATOMIC);
992 		if (!_skb)
993 			break;
994 		__skb_dequeue(&l->backlogq);
995 		hdr = buf_msg(skb);
996 		l->backlog[msg_importance(hdr)].len--;
997 		__skb_queue_tail(&l->transmq, skb);
998 		__skb_queue_tail(xmitq, _skb);
999 		TIPC_SKB_CB(skb)->ackers = l->ackers;
1000 		msg_set_seqno(hdr, seqno);
1001 		msg_set_ack(hdr, ack);
1002 		msg_set_bcast_ack(hdr, bc_ack);
1003 		l->rcv_unacked = 0;
1004 		l->stats.sent_pkts++;
1005 		seqno++;
1006 	}
1007 	l->snd_nxt = seqno;
1008 }
1009 
1010 static void link_retransmit_failure(struct tipc_link *l, struct sk_buff *skb)
1011 {
1012 	struct tipc_msg *hdr = buf_msg(skb);
1013 
1014 	pr_warn("Retransmission failure on link <%s>\n", l->name);
1015 	link_print(l, "State of link ");
1016 	pr_info("Failed msg: usr %u, typ %u, len %u, err %u\n",
1017 		msg_user(hdr), msg_type(hdr), msg_size(hdr), msg_errcode(hdr));
1018 	pr_info("sqno %u, prev: %x, src: %x\n",
1019 		msg_seqno(hdr), msg_prevnode(hdr), msg_orignode(hdr));
1020 }
1021 
1022 /* tipc_link_retrans() - retransmit one or more packets
1023  * @l: the link to transmit on
1024  * @r: the receiving link ordering the retransmit. Same as l if unicast
1025  * @from: retransmit from (inclusive) this sequence number
1026  * @to: retransmit to (inclusive) this sequence number
1027  * xmitq: queue for accumulating the retransmitted packets
1028  */
1029 static int tipc_link_retrans(struct tipc_link *l, struct tipc_link *r,
1030 			     u16 from, u16 to, struct sk_buff_head *xmitq)
1031 {
1032 	struct sk_buff *_skb, *skb = skb_peek(&l->transmq);
1033 	u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1034 	u16 ack = l->rcv_nxt - 1;
1035 	struct tipc_msg *hdr;
1036 
1037 	if (!skb)
1038 		return 0;
1039 
1040 	/* Detect repeated retransmit failures on same packet */
1041 	if (r->last_retransm != buf_seqno(skb)) {
1042 		r->last_retransm = buf_seqno(skb);
1043 		r->stale_limit = jiffies + msecs_to_jiffies(r->tolerance);
1044 	} else if (++r->stale_cnt > 99 && time_after(jiffies, r->stale_limit)) {
1045 		link_retransmit_failure(l, skb);
1046 		if (link_is_bc_sndlink(l))
1047 			return TIPC_LINK_DOWN_EVT;
1048 		return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1049 	}
1050 
1051 	skb_queue_walk(&l->transmq, skb) {
1052 		hdr = buf_msg(skb);
1053 		if (less(msg_seqno(hdr), from))
1054 			continue;
1055 		if (more(msg_seqno(hdr), to))
1056 			break;
1057 		_skb = __pskb_copy(skb, MIN_H_SIZE, GFP_ATOMIC);
1058 		if (!_skb)
1059 			return 0;
1060 		hdr = buf_msg(_skb);
1061 		msg_set_ack(hdr, ack);
1062 		msg_set_bcast_ack(hdr, bc_ack);
1063 		_skb->priority = TC_PRIO_CONTROL;
1064 		__skb_queue_tail(xmitq, _skb);
1065 		l->stats.retransmitted++;
1066 	}
1067 	return 0;
1068 }
1069 
1070 /* tipc_data_input - deliver data and name distr msgs to upper layer
1071  *
1072  * Consumes buffer if message is of right type
1073  * Node lock must be held
1074  */
1075 static bool tipc_data_input(struct tipc_link *l, struct sk_buff *skb,
1076 			    struct sk_buff_head *inputq)
1077 {
1078 	struct sk_buff_head *mc_inputq = l->bc_rcvlink->inputq;
1079 	struct tipc_msg *hdr = buf_msg(skb);
1080 
1081 	switch (msg_user(hdr)) {
1082 	case TIPC_LOW_IMPORTANCE:
1083 	case TIPC_MEDIUM_IMPORTANCE:
1084 	case TIPC_HIGH_IMPORTANCE:
1085 	case TIPC_CRITICAL_IMPORTANCE:
1086 		if (unlikely(msg_in_group(hdr) || msg_mcast(hdr))) {
1087 			skb_queue_tail(mc_inputq, skb);
1088 			return true;
1089 		}
1090 		/* else: fall through */
1091 	case CONN_MANAGER:
1092 		skb_queue_tail(inputq, skb);
1093 		return true;
1094 	case GROUP_PROTOCOL:
1095 		skb_queue_tail(mc_inputq, skb);
1096 		return true;
1097 	case NAME_DISTRIBUTOR:
1098 		l->bc_rcvlink->state = LINK_ESTABLISHED;
1099 		skb_queue_tail(l->namedq, skb);
1100 		return true;
1101 	case MSG_BUNDLER:
1102 	case TUNNEL_PROTOCOL:
1103 	case MSG_FRAGMENTER:
1104 	case BCAST_PROTOCOL:
1105 		return false;
1106 	default:
1107 		pr_warn("Dropping received illegal msg type\n");
1108 		kfree_skb(skb);
1109 		return false;
1110 	};
1111 }
1112 
1113 /* tipc_link_input - process packet that has passed link protocol check
1114  *
1115  * Consumes buffer
1116  */
1117 static int tipc_link_input(struct tipc_link *l, struct sk_buff *skb,
1118 			   struct sk_buff_head *inputq)
1119 {
1120 	struct tipc_msg *hdr = buf_msg(skb);
1121 	struct sk_buff **reasm_skb = &l->reasm_buf;
1122 	struct sk_buff *iskb;
1123 	struct sk_buff_head tmpq;
1124 	int usr = msg_user(hdr);
1125 	int rc = 0;
1126 	int pos = 0;
1127 	int ipos = 0;
1128 
1129 	if (unlikely(usr == TUNNEL_PROTOCOL)) {
1130 		if (msg_type(hdr) == SYNCH_MSG) {
1131 			__skb_queue_purge(&l->deferdq);
1132 			goto drop;
1133 		}
1134 		if (!tipc_msg_extract(skb, &iskb, &ipos))
1135 			return rc;
1136 		kfree_skb(skb);
1137 		skb = iskb;
1138 		hdr = buf_msg(skb);
1139 		if (less(msg_seqno(hdr), l->drop_point))
1140 			goto drop;
1141 		if (tipc_data_input(l, skb, inputq))
1142 			return rc;
1143 		usr = msg_user(hdr);
1144 		reasm_skb = &l->failover_reasm_skb;
1145 	}
1146 
1147 	if (usr == MSG_BUNDLER) {
1148 		skb_queue_head_init(&tmpq);
1149 		l->stats.recv_bundles++;
1150 		l->stats.recv_bundled += msg_msgcnt(hdr);
1151 		while (tipc_msg_extract(skb, &iskb, &pos))
1152 			tipc_data_input(l, iskb, &tmpq);
1153 		tipc_skb_queue_splice_tail(&tmpq, inputq);
1154 		return 0;
1155 	} else if (usr == MSG_FRAGMENTER) {
1156 		l->stats.recv_fragments++;
1157 		if (tipc_buf_append(reasm_skb, &skb)) {
1158 			l->stats.recv_fragmented++;
1159 			tipc_data_input(l, skb, inputq);
1160 		} else if (!*reasm_skb && !link_is_bc_rcvlink(l)) {
1161 			pr_warn_ratelimited("Unable to build fragment list\n");
1162 			return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1163 		}
1164 		return 0;
1165 	} else if (usr == BCAST_PROTOCOL) {
1166 		tipc_bcast_lock(l->net);
1167 		tipc_link_bc_init_rcv(l->bc_rcvlink, hdr);
1168 		tipc_bcast_unlock(l->net);
1169 	}
1170 drop:
1171 	kfree_skb(skb);
1172 	return 0;
1173 }
1174 
1175 static bool tipc_link_release_pkts(struct tipc_link *l, u16 acked)
1176 {
1177 	bool released = false;
1178 	struct sk_buff *skb, *tmp;
1179 
1180 	skb_queue_walk_safe(&l->transmq, skb, tmp) {
1181 		if (more(buf_seqno(skb), acked))
1182 			break;
1183 		__skb_unlink(skb, &l->transmq);
1184 		kfree_skb(skb);
1185 		released = true;
1186 	}
1187 	return released;
1188 }
1189 
1190 /* tipc_link_build_state_msg: prepare link state message for transmission
1191  *
1192  * Note that sending of broadcast ack is coordinated among nodes, to reduce
1193  * risk of ack storms towards the sender
1194  */
1195 int tipc_link_build_state_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1196 {
1197 	if (!l)
1198 		return 0;
1199 
1200 	/* Broadcast ACK must be sent via a unicast link => defer to caller */
1201 	if (link_is_bc_rcvlink(l)) {
1202 		if (((l->rcv_nxt ^ tipc_own_addr(l->net)) & 0xf) != 0xf)
1203 			return 0;
1204 		l->rcv_unacked = 0;
1205 
1206 		/* Use snd_nxt to store peer's snd_nxt in broadcast rcv link */
1207 		l->snd_nxt = l->rcv_nxt;
1208 		return TIPC_LINK_SND_STATE;
1209 	}
1210 
1211 	/* Unicast ACK */
1212 	l->rcv_unacked = 0;
1213 	l->stats.sent_acks++;
1214 	tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, 0, xmitq);
1215 	return 0;
1216 }
1217 
1218 /* tipc_link_build_reset_msg: prepare link RESET or ACTIVATE message
1219  */
1220 void tipc_link_build_reset_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1221 {
1222 	int mtyp = RESET_MSG;
1223 	struct sk_buff *skb;
1224 
1225 	if (l->state == LINK_ESTABLISHING)
1226 		mtyp = ACTIVATE_MSG;
1227 
1228 	tipc_link_build_proto_msg(l, mtyp, 0, 0, 0, 0, 0, xmitq);
1229 
1230 	/* Inform peer that this endpoint is going down if applicable */
1231 	skb = skb_peek_tail(xmitq);
1232 	if (skb && (l->state == LINK_RESET))
1233 		msg_set_peer_stopping(buf_msg(skb), 1);
1234 }
1235 
1236 /* tipc_link_build_nack_msg: prepare link nack message for transmission
1237  * Note that sending of broadcast NACK is coordinated among nodes, to
1238  * reduce the risk of NACK storms towards the sender
1239  */
1240 static int tipc_link_build_nack_msg(struct tipc_link *l,
1241 				    struct sk_buff_head *xmitq)
1242 {
1243 	u32 def_cnt = ++l->stats.deferred_recv;
1244 	int match1, match2;
1245 
1246 	if (link_is_bc_rcvlink(l)) {
1247 		match1 = def_cnt & 0xf;
1248 		match2 = tipc_own_addr(l->net) & 0xf;
1249 		if (match1 == match2)
1250 			return TIPC_LINK_SND_STATE;
1251 		return 0;
1252 	}
1253 
1254 	if ((skb_queue_len(&l->deferdq) == 1) || !(def_cnt % TIPC_NACK_INTV))
1255 		tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, 0, xmitq);
1256 	return 0;
1257 }
1258 
1259 /* tipc_link_rcv - process TIPC packets/messages arriving from off-node
1260  * @l: the link that should handle the message
1261  * @skb: TIPC packet
1262  * @xmitq: queue to place packets to be sent after this call
1263  */
1264 int tipc_link_rcv(struct tipc_link *l, struct sk_buff *skb,
1265 		  struct sk_buff_head *xmitq)
1266 {
1267 	struct sk_buff_head *defq = &l->deferdq;
1268 	struct tipc_msg *hdr;
1269 	u16 seqno, rcv_nxt, win_lim;
1270 	int rc = 0;
1271 
1272 	do {
1273 		hdr = buf_msg(skb);
1274 		seqno = msg_seqno(hdr);
1275 		rcv_nxt = l->rcv_nxt;
1276 		win_lim = rcv_nxt + TIPC_MAX_LINK_WIN;
1277 
1278 		/* Verify and update link state */
1279 		if (unlikely(msg_user(hdr) == LINK_PROTOCOL))
1280 			return tipc_link_proto_rcv(l, skb, xmitq);
1281 
1282 		if (unlikely(!link_is_up(l))) {
1283 			if (l->state == LINK_ESTABLISHING)
1284 				rc = TIPC_LINK_UP_EVT;
1285 			goto drop;
1286 		}
1287 
1288 		/* Don't send probe at next timeout expiration */
1289 		l->silent_intv_cnt = 0;
1290 
1291 		/* Drop if outside receive window */
1292 		if (unlikely(less(seqno, rcv_nxt) || more(seqno, win_lim))) {
1293 			l->stats.duplicates++;
1294 			goto drop;
1295 		}
1296 
1297 		/* Forward queues and wake up waiting users */
1298 		if (likely(tipc_link_release_pkts(l, msg_ack(hdr)))) {
1299 			l->stale_cnt = 0;
1300 			tipc_link_advance_backlog(l, xmitq);
1301 			if (unlikely(!skb_queue_empty(&l->wakeupq)))
1302 				link_prepare_wakeup(l);
1303 		}
1304 
1305 		/* Defer delivery if sequence gap */
1306 		if (unlikely(seqno != rcv_nxt)) {
1307 			__tipc_skb_queue_sorted(defq, seqno, skb);
1308 			rc |= tipc_link_build_nack_msg(l, xmitq);
1309 			break;
1310 		}
1311 
1312 		/* Deliver packet */
1313 		l->rcv_nxt++;
1314 		l->stats.recv_pkts++;
1315 		if (!tipc_data_input(l, skb, l->inputq))
1316 			rc |= tipc_link_input(l, skb, l->inputq);
1317 		if (unlikely(++l->rcv_unacked >= TIPC_MIN_LINK_WIN))
1318 			rc |= tipc_link_build_state_msg(l, xmitq);
1319 		if (unlikely(rc & ~TIPC_LINK_SND_STATE))
1320 			break;
1321 	} while ((skb = __skb_dequeue(defq)));
1322 
1323 	return rc;
1324 drop:
1325 	kfree_skb(skb);
1326 	return rc;
1327 }
1328 
1329 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
1330 				      bool probe_reply, u16 rcvgap,
1331 				      int tolerance, int priority,
1332 				      struct sk_buff_head *xmitq)
1333 {
1334 	struct tipc_link *bcl = l->bc_rcvlink;
1335 	struct sk_buff *skb;
1336 	struct tipc_msg *hdr;
1337 	struct sk_buff_head *dfq = &l->deferdq;
1338 	bool node_up = link_is_up(bcl);
1339 	struct tipc_mon_state *mstate = &l->mon_state;
1340 	int dlen = 0;
1341 	void *data;
1342 
1343 	/* Don't send protocol message during reset or link failover */
1344 	if (tipc_link_is_blocked(l))
1345 		return;
1346 
1347 	if (!tipc_link_is_up(l) && (mtyp == STATE_MSG))
1348 		return;
1349 
1350 	if (!skb_queue_empty(dfq))
1351 		rcvgap = buf_seqno(skb_peek(dfq)) - l->rcv_nxt;
1352 
1353 	skb = tipc_msg_create(LINK_PROTOCOL, mtyp, INT_H_SIZE,
1354 			      tipc_max_domain_size, l->addr,
1355 			      tipc_own_addr(l->net), 0, 0, 0);
1356 	if (!skb)
1357 		return;
1358 
1359 	hdr = buf_msg(skb);
1360 	data = msg_data(hdr);
1361 	msg_set_session(hdr, l->session);
1362 	msg_set_bearer_id(hdr, l->bearer_id);
1363 	msg_set_net_plane(hdr, l->net_plane);
1364 	msg_set_next_sent(hdr, l->snd_nxt);
1365 	msg_set_ack(hdr, l->rcv_nxt - 1);
1366 	msg_set_bcast_ack(hdr, bcl->rcv_nxt - 1);
1367 	msg_set_bc_ack_invalid(hdr, !node_up);
1368 	msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1369 	msg_set_link_tolerance(hdr, tolerance);
1370 	msg_set_linkprio(hdr, priority);
1371 	msg_set_redundant_link(hdr, node_up);
1372 	msg_set_seq_gap(hdr, 0);
1373 	msg_set_seqno(hdr, l->snd_nxt + U16_MAX / 2);
1374 
1375 	if (mtyp == STATE_MSG) {
1376 		if (l->peer_caps & TIPC_LINK_PROTO_SEQNO)
1377 			msg_set_seqno(hdr, l->snd_nxt_state++);
1378 		msg_set_seq_gap(hdr, rcvgap);
1379 		msg_set_bc_gap(hdr, link_bc_rcv_gap(bcl));
1380 		msg_set_probe(hdr, probe);
1381 		msg_set_is_keepalive(hdr, probe || probe_reply);
1382 		tipc_mon_prep(l->net, data, &dlen, mstate, l->bearer_id);
1383 		msg_set_size(hdr, INT_H_SIZE + dlen);
1384 		skb_trim(skb, INT_H_SIZE + dlen);
1385 		l->stats.sent_states++;
1386 		l->rcv_unacked = 0;
1387 	} else {
1388 		/* RESET_MSG or ACTIVATE_MSG */
1389 		msg_set_max_pkt(hdr, l->advertised_mtu);
1390 		strcpy(data, l->if_name);
1391 		msg_set_size(hdr, INT_H_SIZE + TIPC_MAX_IF_NAME);
1392 		skb_trim(skb, INT_H_SIZE + TIPC_MAX_IF_NAME);
1393 	}
1394 	if (probe)
1395 		l->stats.sent_probes++;
1396 	if (rcvgap)
1397 		l->stats.sent_nacks++;
1398 	skb->priority = TC_PRIO_CONTROL;
1399 	__skb_queue_tail(xmitq, skb);
1400 }
1401 
1402 void tipc_link_create_dummy_tnl_msg(struct tipc_link *l,
1403 				    struct sk_buff_head *xmitq)
1404 {
1405 	u32 onode = tipc_own_addr(l->net);
1406 	struct tipc_msg *hdr, *ihdr;
1407 	struct sk_buff_head tnlq;
1408 	struct sk_buff *skb;
1409 	u32 dnode = l->addr;
1410 
1411 	skb_queue_head_init(&tnlq);
1412 	skb = tipc_msg_create(TUNNEL_PROTOCOL, FAILOVER_MSG,
1413 			      INT_H_SIZE, BASIC_H_SIZE,
1414 			      dnode, onode, 0, 0, 0);
1415 	if (!skb) {
1416 		pr_warn("%sunable to create tunnel packet\n", link_co_err);
1417 		return;
1418 	}
1419 
1420 	hdr = buf_msg(skb);
1421 	msg_set_msgcnt(hdr, 1);
1422 	msg_set_bearer_id(hdr, l->peer_bearer_id);
1423 
1424 	ihdr = (struct tipc_msg *)msg_data(hdr);
1425 	tipc_msg_init(onode, ihdr, TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1426 		      BASIC_H_SIZE, dnode);
1427 	msg_set_errcode(ihdr, TIPC_ERR_NO_PORT);
1428 	__skb_queue_tail(&tnlq, skb);
1429 	tipc_link_xmit(l, &tnlq, xmitq);
1430 }
1431 
1432 /* tipc_link_tnl_prepare(): prepare and return a list of tunnel packets
1433  * with contents of the link's transmit and backlog queues.
1434  */
1435 void tipc_link_tnl_prepare(struct tipc_link *l, struct tipc_link *tnl,
1436 			   int mtyp, struct sk_buff_head *xmitq)
1437 {
1438 	struct sk_buff *skb, *tnlskb;
1439 	struct tipc_msg *hdr, tnlhdr;
1440 	struct sk_buff_head *queue = &l->transmq;
1441 	struct sk_buff_head tmpxq, tnlq;
1442 	u16 pktlen, pktcnt, seqno = l->snd_nxt;
1443 
1444 	if (!tnl)
1445 		return;
1446 
1447 	skb_queue_head_init(&tnlq);
1448 	skb_queue_head_init(&tmpxq);
1449 
1450 	/* At least one packet required for safe algorithm => add dummy */
1451 	skb = tipc_msg_create(TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1452 			      BASIC_H_SIZE, 0, l->addr, tipc_own_addr(l->net),
1453 			      0, 0, TIPC_ERR_NO_PORT);
1454 	if (!skb) {
1455 		pr_warn("%sunable to create tunnel packet\n", link_co_err);
1456 		return;
1457 	}
1458 	skb_queue_tail(&tnlq, skb);
1459 	tipc_link_xmit(l, &tnlq, &tmpxq);
1460 	__skb_queue_purge(&tmpxq);
1461 
1462 	/* Initialize reusable tunnel packet header */
1463 	tipc_msg_init(tipc_own_addr(l->net), &tnlhdr, TUNNEL_PROTOCOL,
1464 		      mtyp, INT_H_SIZE, l->addr);
1465 	pktcnt = skb_queue_len(&l->transmq) + skb_queue_len(&l->backlogq);
1466 	msg_set_msgcnt(&tnlhdr, pktcnt);
1467 	msg_set_bearer_id(&tnlhdr, l->peer_bearer_id);
1468 tnl:
1469 	/* Wrap each packet into a tunnel packet */
1470 	skb_queue_walk(queue, skb) {
1471 		hdr = buf_msg(skb);
1472 		if (queue == &l->backlogq)
1473 			msg_set_seqno(hdr, seqno++);
1474 		pktlen = msg_size(hdr);
1475 		msg_set_size(&tnlhdr, pktlen + INT_H_SIZE);
1476 		tnlskb = tipc_buf_acquire(pktlen + INT_H_SIZE, GFP_ATOMIC);
1477 		if (!tnlskb) {
1478 			pr_warn("%sunable to send packet\n", link_co_err);
1479 			return;
1480 		}
1481 		skb_copy_to_linear_data(tnlskb, &tnlhdr, INT_H_SIZE);
1482 		skb_copy_to_linear_data_offset(tnlskb, INT_H_SIZE, hdr, pktlen);
1483 		__skb_queue_tail(&tnlq, tnlskb);
1484 	}
1485 	if (queue != &l->backlogq) {
1486 		queue = &l->backlogq;
1487 		goto tnl;
1488 	}
1489 
1490 	tipc_link_xmit(tnl, &tnlq, xmitq);
1491 
1492 	if (mtyp == FAILOVER_MSG) {
1493 		tnl->drop_point = l->rcv_nxt;
1494 		tnl->failover_reasm_skb = l->reasm_buf;
1495 		l->reasm_buf = NULL;
1496 	}
1497 }
1498 
1499 /* tipc_link_validate_msg(): validate message against current link state
1500  * Returns true if message should be accepted, otherwise false
1501  */
1502 bool tipc_link_validate_msg(struct tipc_link *l, struct tipc_msg *hdr)
1503 {
1504 	u16 curr_session = l->peer_session;
1505 	u16 session = msg_session(hdr);
1506 	int mtyp = msg_type(hdr);
1507 
1508 	if (msg_user(hdr) != LINK_PROTOCOL)
1509 		return true;
1510 
1511 	switch (mtyp) {
1512 	case RESET_MSG:
1513 		if (!l->in_session)
1514 			return true;
1515 		/* Accept only RESET with new session number */
1516 		return more(session, curr_session);
1517 	case ACTIVATE_MSG:
1518 		if (!l->in_session)
1519 			return true;
1520 		/* Accept only ACTIVATE with new or current session number */
1521 		return !less(session, curr_session);
1522 	case STATE_MSG:
1523 		/* Accept only STATE with current session number */
1524 		if (!l->in_session)
1525 			return false;
1526 		if (session != curr_session)
1527 			return false;
1528 		/* Extra sanity check */
1529 		if (!link_is_up(l) && msg_ack(hdr))
1530 			return false;
1531 		if (!(l->peer_caps & TIPC_LINK_PROTO_SEQNO))
1532 			return true;
1533 		/* Accept only STATE with new sequence number */
1534 		return !less(msg_seqno(hdr), l->rcv_nxt_state);
1535 	default:
1536 		return false;
1537 	}
1538 }
1539 
1540 /* tipc_link_proto_rcv(): receive link level protocol message :
1541  * Note that network plane id propagates through the network, and may
1542  * change at any time. The node with lowest numerical id determines
1543  * network plane
1544  */
1545 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
1546 			       struct sk_buff_head *xmitq)
1547 {
1548 	struct tipc_msg *hdr = buf_msg(skb);
1549 	u16 rcvgap = 0;
1550 	u16 ack = msg_ack(hdr);
1551 	u16 gap = msg_seq_gap(hdr);
1552 	u16 peers_snd_nxt =  msg_next_sent(hdr);
1553 	u16 peers_tol = msg_link_tolerance(hdr);
1554 	u16 peers_prio = msg_linkprio(hdr);
1555 	u16 rcv_nxt = l->rcv_nxt;
1556 	u16 dlen = msg_data_sz(hdr);
1557 	int mtyp = msg_type(hdr);
1558 	bool reply = msg_probe(hdr);
1559 	void *data;
1560 	char *if_name;
1561 	int rc = 0;
1562 
1563 	if (tipc_link_is_blocked(l) || !xmitq)
1564 		goto exit;
1565 
1566 	if (tipc_own_addr(l->net) > msg_prevnode(hdr))
1567 		l->net_plane = msg_net_plane(hdr);
1568 
1569 	skb_linearize(skb);
1570 	hdr = buf_msg(skb);
1571 	data = msg_data(hdr);
1572 
1573 	if (!tipc_link_validate_msg(l, hdr))
1574 		goto exit;
1575 
1576 	switch (mtyp) {
1577 	case RESET_MSG:
1578 	case ACTIVATE_MSG:
1579 		/* Complete own link name with peer's interface name */
1580 		if_name =  strrchr(l->name, ':') + 1;
1581 		if (sizeof(l->name) - (if_name - l->name) <= TIPC_MAX_IF_NAME)
1582 			break;
1583 		if (msg_data_sz(hdr) < TIPC_MAX_IF_NAME)
1584 			break;
1585 		strncpy(if_name, data, TIPC_MAX_IF_NAME);
1586 
1587 		/* Update own tolerance if peer indicates a non-zero value */
1588 		if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
1589 			l->tolerance = peers_tol;
1590 			l->bc_rcvlink->tolerance = peers_tol;
1591 		}
1592 		/* Update own priority if peer's priority is higher */
1593 		if (in_range(peers_prio, l->priority + 1, TIPC_MAX_LINK_PRI))
1594 			l->priority = peers_prio;
1595 
1596 		/* ACTIVATE_MSG serves as PEER_RESET if link is already down */
1597 		if (msg_peer_stopping(hdr))
1598 			rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1599 		else if ((mtyp == RESET_MSG) || !link_is_up(l))
1600 			rc = tipc_link_fsm_evt(l, LINK_PEER_RESET_EVT);
1601 
1602 		/* ACTIVATE_MSG takes up link if it was already locally reset */
1603 		if ((mtyp == ACTIVATE_MSG) && (l->state == LINK_ESTABLISHING))
1604 			rc = TIPC_LINK_UP_EVT;
1605 
1606 		l->peer_session = msg_session(hdr);
1607 		l->in_session = true;
1608 		l->peer_bearer_id = msg_bearer_id(hdr);
1609 		if (l->mtu > msg_max_pkt(hdr))
1610 			l->mtu = msg_max_pkt(hdr);
1611 		break;
1612 
1613 	case STATE_MSG:
1614 		l->rcv_nxt_state = msg_seqno(hdr) + 1;
1615 
1616 		/* Update own tolerance if peer indicates a non-zero value */
1617 		if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
1618 			l->tolerance = peers_tol;
1619 			l->bc_rcvlink->tolerance = peers_tol;
1620 		}
1621 		/* Update own prio if peer indicates a different value */
1622 		if ((peers_prio != l->priority) &&
1623 		    in_range(peers_prio, 1, TIPC_MAX_LINK_PRI)) {
1624 			l->priority = peers_prio;
1625 			rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1626 		}
1627 
1628 		l->silent_intv_cnt = 0;
1629 		l->stats.recv_states++;
1630 		if (msg_probe(hdr))
1631 			l->stats.recv_probes++;
1632 
1633 		if (!link_is_up(l)) {
1634 			if (l->state == LINK_ESTABLISHING)
1635 				rc = TIPC_LINK_UP_EVT;
1636 			break;
1637 		}
1638 		tipc_mon_rcv(l->net, data, dlen, l->addr,
1639 			     &l->mon_state, l->bearer_id);
1640 
1641 		/* Send NACK if peer has sent pkts we haven't received yet */
1642 		if (more(peers_snd_nxt, rcv_nxt) && !tipc_link_is_synching(l))
1643 			rcvgap = peers_snd_nxt - l->rcv_nxt;
1644 		if (rcvgap || reply)
1645 			tipc_link_build_proto_msg(l, STATE_MSG, 0, reply,
1646 						  rcvgap, 0, 0, xmitq);
1647 		tipc_link_release_pkts(l, ack);
1648 
1649 		/* If NACK, retransmit will now start at right position */
1650 		if (gap) {
1651 			rc = tipc_link_retrans(l, l, ack + 1, ack + gap, xmitq);
1652 			l->stats.recv_nacks++;
1653 		}
1654 
1655 		tipc_link_advance_backlog(l, xmitq);
1656 		if (unlikely(!skb_queue_empty(&l->wakeupq)))
1657 			link_prepare_wakeup(l);
1658 	}
1659 exit:
1660 	kfree_skb(skb);
1661 	return rc;
1662 }
1663 
1664 /* tipc_link_build_bc_proto_msg() - create broadcast protocol message
1665  */
1666 static bool tipc_link_build_bc_proto_msg(struct tipc_link *l, bool bcast,
1667 					 u16 peers_snd_nxt,
1668 					 struct sk_buff_head *xmitq)
1669 {
1670 	struct sk_buff *skb;
1671 	struct tipc_msg *hdr;
1672 	struct sk_buff *dfrd_skb = skb_peek(&l->deferdq);
1673 	u16 ack = l->rcv_nxt - 1;
1674 	u16 gap_to = peers_snd_nxt - 1;
1675 
1676 	skb = tipc_msg_create(BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE,
1677 			      0, l->addr, tipc_own_addr(l->net), 0, 0, 0);
1678 	if (!skb)
1679 		return false;
1680 	hdr = buf_msg(skb);
1681 	msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1682 	msg_set_bcast_ack(hdr, ack);
1683 	msg_set_bcgap_after(hdr, ack);
1684 	if (dfrd_skb)
1685 		gap_to = buf_seqno(dfrd_skb) - 1;
1686 	msg_set_bcgap_to(hdr, gap_to);
1687 	msg_set_non_seq(hdr, bcast);
1688 	__skb_queue_tail(xmitq, skb);
1689 	return true;
1690 }
1691 
1692 /* tipc_link_build_bc_init_msg() - synchronize broadcast link endpoints.
1693  *
1694  * Give a newly added peer node the sequence number where it should
1695  * start receiving and acking broadcast packets.
1696  */
1697 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
1698 					struct sk_buff_head *xmitq)
1699 {
1700 	struct sk_buff_head list;
1701 
1702 	__skb_queue_head_init(&list);
1703 	if (!tipc_link_build_bc_proto_msg(l->bc_rcvlink, false, 0, &list))
1704 		return;
1705 	msg_set_bc_ack_invalid(buf_msg(skb_peek(&list)), true);
1706 	tipc_link_xmit(l, &list, xmitq);
1707 }
1708 
1709 /* tipc_link_bc_init_rcv - receive initial broadcast synch data from peer
1710  */
1711 void tipc_link_bc_init_rcv(struct tipc_link *l, struct tipc_msg *hdr)
1712 {
1713 	int mtyp = msg_type(hdr);
1714 	u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
1715 
1716 	if (link_is_up(l))
1717 		return;
1718 
1719 	if (msg_user(hdr) == BCAST_PROTOCOL) {
1720 		l->rcv_nxt = peers_snd_nxt;
1721 		l->state = LINK_ESTABLISHED;
1722 		return;
1723 	}
1724 
1725 	if (l->peer_caps & TIPC_BCAST_SYNCH)
1726 		return;
1727 
1728 	if (msg_peer_node_is_up(hdr))
1729 		return;
1730 
1731 	/* Compatibility: accept older, less safe initial synch data */
1732 	if ((mtyp == RESET_MSG) || (mtyp == ACTIVATE_MSG))
1733 		l->rcv_nxt = peers_snd_nxt;
1734 }
1735 
1736 /* link_bc_retr eval()- check if the indicated range can be retransmitted now
1737  * - Adjust permitted range if there is overlap with previous retransmission
1738  */
1739 static bool link_bc_retr_eval(struct tipc_link *l, u16 *from, u16 *to)
1740 {
1741 	unsigned long elapsed = jiffies_to_msecs(jiffies - l->prev_retr);
1742 
1743 	if (less(*to, *from))
1744 		return false;
1745 
1746 	/* New retransmission request */
1747 	if ((elapsed > TIPC_BC_RETR_LIMIT) ||
1748 	    less(*to, l->prev_from) || more(*from, l->prev_to)) {
1749 		l->prev_from = *from;
1750 		l->prev_to = *to;
1751 		l->prev_retr = jiffies;
1752 		return true;
1753 	}
1754 
1755 	/* Inside range of previous retransmit */
1756 	if (!less(*from, l->prev_from) && !more(*to, l->prev_to))
1757 		return false;
1758 
1759 	/* Fully or partially outside previous range => exclude overlap */
1760 	if (less(*from, l->prev_from)) {
1761 		*to = l->prev_from - 1;
1762 		l->prev_from = *from;
1763 	}
1764 	if (more(*to, l->prev_to)) {
1765 		*from = l->prev_to + 1;
1766 		l->prev_to = *to;
1767 	}
1768 	l->prev_retr = jiffies;
1769 	return true;
1770 }
1771 
1772 /* tipc_link_bc_sync_rcv - update rcv link according to peer's send state
1773  */
1774 int tipc_link_bc_sync_rcv(struct tipc_link *l, struct tipc_msg *hdr,
1775 			  struct sk_buff_head *xmitq)
1776 {
1777 	struct tipc_link *snd_l = l->bc_sndlink;
1778 	u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
1779 	u16 from = msg_bcast_ack(hdr) + 1;
1780 	u16 to = from + msg_bc_gap(hdr) - 1;
1781 	int rc = 0;
1782 
1783 	if (!link_is_up(l))
1784 		return rc;
1785 
1786 	if (!msg_peer_node_is_up(hdr))
1787 		return rc;
1788 
1789 	/* Open when peer ackowledges our bcast init msg (pkt #1) */
1790 	if (msg_ack(hdr))
1791 		l->bc_peer_is_up = true;
1792 
1793 	if (!l->bc_peer_is_up)
1794 		return rc;
1795 
1796 	l->stats.recv_nacks++;
1797 
1798 	/* Ignore if peers_snd_nxt goes beyond receive window */
1799 	if (more(peers_snd_nxt, l->rcv_nxt + l->window))
1800 		return rc;
1801 
1802 	if (link_bc_retr_eval(snd_l, &from, &to))
1803 		rc = tipc_link_retrans(snd_l, l, from, to, xmitq);
1804 
1805 	l->snd_nxt = peers_snd_nxt;
1806 	if (link_bc_rcv_gap(l))
1807 		rc |= TIPC_LINK_SND_STATE;
1808 
1809 	/* Return now if sender supports nack via STATE messages */
1810 	if (l->peer_caps & TIPC_BCAST_STATE_NACK)
1811 		return rc;
1812 
1813 	/* Otherwise, be backwards compatible */
1814 
1815 	if (!more(peers_snd_nxt, l->rcv_nxt)) {
1816 		l->nack_state = BC_NACK_SND_CONDITIONAL;
1817 		return 0;
1818 	}
1819 
1820 	/* Don't NACK if one was recently sent or peeked */
1821 	if (l->nack_state == BC_NACK_SND_SUPPRESS) {
1822 		l->nack_state = BC_NACK_SND_UNCONDITIONAL;
1823 		return 0;
1824 	}
1825 
1826 	/* Conditionally delay NACK sending until next synch rcv */
1827 	if (l->nack_state == BC_NACK_SND_CONDITIONAL) {
1828 		l->nack_state = BC_NACK_SND_UNCONDITIONAL;
1829 		if ((peers_snd_nxt - l->rcv_nxt) < TIPC_MIN_LINK_WIN)
1830 			return 0;
1831 	}
1832 
1833 	/* Send NACK now but suppress next one */
1834 	tipc_link_build_bc_proto_msg(l, true, peers_snd_nxt, xmitq);
1835 	l->nack_state = BC_NACK_SND_SUPPRESS;
1836 	return 0;
1837 }
1838 
1839 void tipc_link_bc_ack_rcv(struct tipc_link *l, u16 acked,
1840 			  struct sk_buff_head *xmitq)
1841 {
1842 	struct sk_buff *skb, *tmp;
1843 	struct tipc_link *snd_l = l->bc_sndlink;
1844 
1845 	if (!link_is_up(l) || !l->bc_peer_is_up)
1846 		return;
1847 
1848 	if (!more(acked, l->acked))
1849 		return;
1850 
1851 	/* Skip over packets peer has already acked */
1852 	skb_queue_walk(&snd_l->transmq, skb) {
1853 		if (more(buf_seqno(skb), l->acked))
1854 			break;
1855 	}
1856 
1857 	/* Update/release the packets peer is acking now */
1858 	skb_queue_walk_from_safe(&snd_l->transmq, skb, tmp) {
1859 		if (more(buf_seqno(skb), acked))
1860 			break;
1861 		if (!--TIPC_SKB_CB(skb)->ackers) {
1862 			__skb_unlink(skb, &snd_l->transmq);
1863 			kfree_skb(skb);
1864 		}
1865 	}
1866 	l->acked = acked;
1867 	tipc_link_advance_backlog(snd_l, xmitq);
1868 	if (unlikely(!skb_queue_empty(&snd_l->wakeupq)))
1869 		link_prepare_wakeup(snd_l);
1870 }
1871 
1872 /* tipc_link_bc_nack_rcv(): receive broadcast nack message
1873  * This function is here for backwards compatibility, since
1874  * no BCAST_PROTOCOL/STATE messages occur from TIPC v2.5.
1875  */
1876 int tipc_link_bc_nack_rcv(struct tipc_link *l, struct sk_buff *skb,
1877 			  struct sk_buff_head *xmitq)
1878 {
1879 	struct tipc_msg *hdr = buf_msg(skb);
1880 	u32 dnode = msg_destnode(hdr);
1881 	int mtyp = msg_type(hdr);
1882 	u16 acked = msg_bcast_ack(hdr);
1883 	u16 from = acked + 1;
1884 	u16 to = msg_bcgap_to(hdr);
1885 	u16 peers_snd_nxt = to + 1;
1886 	int rc = 0;
1887 
1888 	kfree_skb(skb);
1889 
1890 	if (!tipc_link_is_up(l) || !l->bc_peer_is_up)
1891 		return 0;
1892 
1893 	if (mtyp != STATE_MSG)
1894 		return 0;
1895 
1896 	if (dnode == tipc_own_addr(l->net)) {
1897 		tipc_link_bc_ack_rcv(l, acked, xmitq);
1898 		rc = tipc_link_retrans(l->bc_sndlink, l, from, to, xmitq);
1899 		l->stats.recv_nacks++;
1900 		return rc;
1901 	}
1902 
1903 	/* Msg for other node => suppress own NACK at next sync if applicable */
1904 	if (more(peers_snd_nxt, l->rcv_nxt) && !less(l->rcv_nxt, from))
1905 		l->nack_state = BC_NACK_SND_SUPPRESS;
1906 
1907 	return 0;
1908 }
1909 
1910 void tipc_link_set_queue_limits(struct tipc_link *l, u32 win)
1911 {
1912 	int max_bulk = TIPC_MAX_PUBL / (l->mtu / ITEM_SIZE);
1913 
1914 	l->window = win;
1915 	l->backlog[TIPC_LOW_IMPORTANCE].limit      = max_t(u16, 50, win);
1916 	l->backlog[TIPC_MEDIUM_IMPORTANCE].limit   = max_t(u16, 100, win * 2);
1917 	l->backlog[TIPC_HIGH_IMPORTANCE].limit     = max_t(u16, 150, win * 3);
1918 	l->backlog[TIPC_CRITICAL_IMPORTANCE].limit = max_t(u16, 200, win * 4);
1919 	l->backlog[TIPC_SYSTEM_IMPORTANCE].limit   = max_bulk;
1920 }
1921 
1922 /**
1923  * link_reset_stats - reset link statistics
1924  * @l: pointer to link
1925  */
1926 void tipc_link_reset_stats(struct tipc_link *l)
1927 {
1928 	memset(&l->stats, 0, sizeof(l->stats));
1929 }
1930 
1931 static void link_print(struct tipc_link *l, const char *str)
1932 {
1933 	struct sk_buff *hskb = skb_peek(&l->transmq);
1934 	u16 head = hskb ? msg_seqno(buf_msg(hskb)) : l->snd_nxt - 1;
1935 	u16 tail = l->snd_nxt - 1;
1936 
1937 	pr_info("%s Link <%s> state %x\n", str, l->name, l->state);
1938 	pr_info("XMTQ: %u [%u-%u], BKLGQ: %u, SNDNX: %u, RCVNX: %u\n",
1939 		skb_queue_len(&l->transmq), head, tail,
1940 		skb_queue_len(&l->backlogq), l->snd_nxt, l->rcv_nxt);
1941 }
1942 
1943 /* Parse and validate nested (link) properties valid for media, bearer and link
1944  */
1945 int tipc_nl_parse_link_prop(struct nlattr *prop, struct nlattr *props[])
1946 {
1947 	int err;
1948 
1949 	err = nla_parse_nested(props, TIPC_NLA_PROP_MAX, prop,
1950 			       tipc_nl_prop_policy, NULL);
1951 	if (err)
1952 		return err;
1953 
1954 	if (props[TIPC_NLA_PROP_PRIO]) {
1955 		u32 prio;
1956 
1957 		prio = nla_get_u32(props[TIPC_NLA_PROP_PRIO]);
1958 		if (prio > TIPC_MAX_LINK_PRI)
1959 			return -EINVAL;
1960 	}
1961 
1962 	if (props[TIPC_NLA_PROP_TOL]) {
1963 		u32 tol;
1964 
1965 		tol = nla_get_u32(props[TIPC_NLA_PROP_TOL]);
1966 		if ((tol < TIPC_MIN_LINK_TOL) || (tol > TIPC_MAX_LINK_TOL))
1967 			return -EINVAL;
1968 	}
1969 
1970 	if (props[TIPC_NLA_PROP_WIN]) {
1971 		u32 win;
1972 
1973 		win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
1974 		if ((win < TIPC_MIN_LINK_WIN) || (win > TIPC_MAX_LINK_WIN))
1975 			return -EINVAL;
1976 	}
1977 
1978 	return 0;
1979 }
1980 
1981 static int __tipc_nl_add_stats(struct sk_buff *skb, struct tipc_stats *s)
1982 {
1983 	int i;
1984 	struct nlattr *stats;
1985 
1986 	struct nla_map {
1987 		u32 key;
1988 		u32 val;
1989 	};
1990 
1991 	struct nla_map map[] = {
1992 		{TIPC_NLA_STATS_RX_INFO, 0},
1993 		{TIPC_NLA_STATS_RX_FRAGMENTS, s->recv_fragments},
1994 		{TIPC_NLA_STATS_RX_FRAGMENTED, s->recv_fragmented},
1995 		{TIPC_NLA_STATS_RX_BUNDLES, s->recv_bundles},
1996 		{TIPC_NLA_STATS_RX_BUNDLED, s->recv_bundled},
1997 		{TIPC_NLA_STATS_TX_INFO, 0},
1998 		{TIPC_NLA_STATS_TX_FRAGMENTS, s->sent_fragments},
1999 		{TIPC_NLA_STATS_TX_FRAGMENTED, s->sent_fragmented},
2000 		{TIPC_NLA_STATS_TX_BUNDLES, s->sent_bundles},
2001 		{TIPC_NLA_STATS_TX_BUNDLED, s->sent_bundled},
2002 		{TIPC_NLA_STATS_MSG_PROF_TOT, (s->msg_length_counts) ?
2003 			s->msg_length_counts : 1},
2004 		{TIPC_NLA_STATS_MSG_LEN_CNT, s->msg_length_counts},
2005 		{TIPC_NLA_STATS_MSG_LEN_TOT, s->msg_lengths_total},
2006 		{TIPC_NLA_STATS_MSG_LEN_P0, s->msg_length_profile[0]},
2007 		{TIPC_NLA_STATS_MSG_LEN_P1, s->msg_length_profile[1]},
2008 		{TIPC_NLA_STATS_MSG_LEN_P2, s->msg_length_profile[2]},
2009 		{TIPC_NLA_STATS_MSG_LEN_P3, s->msg_length_profile[3]},
2010 		{TIPC_NLA_STATS_MSG_LEN_P4, s->msg_length_profile[4]},
2011 		{TIPC_NLA_STATS_MSG_LEN_P5, s->msg_length_profile[5]},
2012 		{TIPC_NLA_STATS_MSG_LEN_P6, s->msg_length_profile[6]},
2013 		{TIPC_NLA_STATS_RX_STATES, s->recv_states},
2014 		{TIPC_NLA_STATS_RX_PROBES, s->recv_probes},
2015 		{TIPC_NLA_STATS_RX_NACKS, s->recv_nacks},
2016 		{TIPC_NLA_STATS_RX_DEFERRED, s->deferred_recv},
2017 		{TIPC_NLA_STATS_TX_STATES, s->sent_states},
2018 		{TIPC_NLA_STATS_TX_PROBES, s->sent_probes},
2019 		{TIPC_NLA_STATS_TX_NACKS, s->sent_nacks},
2020 		{TIPC_NLA_STATS_TX_ACKS, s->sent_acks},
2021 		{TIPC_NLA_STATS_RETRANSMITTED, s->retransmitted},
2022 		{TIPC_NLA_STATS_DUPLICATES, s->duplicates},
2023 		{TIPC_NLA_STATS_LINK_CONGS, s->link_congs},
2024 		{TIPC_NLA_STATS_MAX_QUEUE, s->max_queue_sz},
2025 		{TIPC_NLA_STATS_AVG_QUEUE, s->queue_sz_counts ?
2026 			(s->accu_queue_sz / s->queue_sz_counts) : 0}
2027 	};
2028 
2029 	stats = nla_nest_start(skb, TIPC_NLA_LINK_STATS);
2030 	if (!stats)
2031 		return -EMSGSIZE;
2032 
2033 	for (i = 0; i <  ARRAY_SIZE(map); i++)
2034 		if (nla_put_u32(skb, map[i].key, map[i].val))
2035 			goto msg_full;
2036 
2037 	nla_nest_end(skb, stats);
2038 
2039 	return 0;
2040 msg_full:
2041 	nla_nest_cancel(skb, stats);
2042 
2043 	return -EMSGSIZE;
2044 }
2045 
2046 /* Caller should hold appropriate locks to protect the link */
2047 int __tipc_nl_add_link(struct net *net, struct tipc_nl_msg *msg,
2048 		       struct tipc_link *link, int nlflags)
2049 {
2050 	u32 self = tipc_own_addr(net);
2051 	struct nlattr *attrs;
2052 	struct nlattr *prop;
2053 	void *hdr;
2054 	int err;
2055 
2056 	hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2057 			  nlflags, TIPC_NL_LINK_GET);
2058 	if (!hdr)
2059 		return -EMSGSIZE;
2060 
2061 	attrs = nla_nest_start(msg->skb, TIPC_NLA_LINK);
2062 	if (!attrs)
2063 		goto msg_full;
2064 
2065 	if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, link->name))
2066 		goto attr_msg_full;
2067 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_DEST, tipc_cluster_mask(self)))
2068 		goto attr_msg_full;
2069 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_MTU, link->mtu))
2070 		goto attr_msg_full;
2071 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, link->stats.recv_pkts))
2072 		goto attr_msg_full;
2073 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, link->stats.sent_pkts))
2074 		goto attr_msg_full;
2075 
2076 	if (tipc_link_is_up(link))
2077 		if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2078 			goto attr_msg_full;
2079 	if (link->active)
2080 		if (nla_put_flag(msg->skb, TIPC_NLA_LINK_ACTIVE))
2081 			goto attr_msg_full;
2082 
2083 	prop = nla_nest_start(msg->skb, TIPC_NLA_LINK_PROP);
2084 	if (!prop)
2085 		goto attr_msg_full;
2086 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2087 		goto prop_msg_full;
2088 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_TOL, link->tolerance))
2089 		goto prop_msg_full;
2090 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN,
2091 			link->window))
2092 		goto prop_msg_full;
2093 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2094 		goto prop_msg_full;
2095 	nla_nest_end(msg->skb, prop);
2096 
2097 	err = __tipc_nl_add_stats(msg->skb, &link->stats);
2098 	if (err)
2099 		goto attr_msg_full;
2100 
2101 	nla_nest_end(msg->skb, attrs);
2102 	genlmsg_end(msg->skb, hdr);
2103 
2104 	return 0;
2105 
2106 prop_msg_full:
2107 	nla_nest_cancel(msg->skb, prop);
2108 attr_msg_full:
2109 	nla_nest_cancel(msg->skb, attrs);
2110 msg_full:
2111 	genlmsg_cancel(msg->skb, hdr);
2112 
2113 	return -EMSGSIZE;
2114 }
2115 
2116 static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
2117 				      struct tipc_stats *stats)
2118 {
2119 	int i;
2120 	struct nlattr *nest;
2121 
2122 	struct nla_map {
2123 		__u32 key;
2124 		__u32 val;
2125 	};
2126 
2127 	struct nla_map map[] = {
2128 		{TIPC_NLA_STATS_RX_INFO, stats->recv_pkts},
2129 		{TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
2130 		{TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
2131 		{TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
2132 		{TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
2133 		{TIPC_NLA_STATS_TX_INFO, stats->sent_pkts},
2134 		{TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
2135 		{TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
2136 		{TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
2137 		{TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
2138 		{TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
2139 		{TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
2140 		{TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
2141 		{TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
2142 		{TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
2143 		{TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
2144 		{TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
2145 		{TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
2146 		{TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
2147 			(stats->accu_queue_sz / stats->queue_sz_counts) : 0}
2148 	};
2149 
2150 	nest = nla_nest_start(skb, TIPC_NLA_LINK_STATS);
2151 	if (!nest)
2152 		return -EMSGSIZE;
2153 
2154 	for (i = 0; i <  ARRAY_SIZE(map); i++)
2155 		if (nla_put_u32(skb, map[i].key, map[i].val))
2156 			goto msg_full;
2157 
2158 	nla_nest_end(skb, nest);
2159 
2160 	return 0;
2161 msg_full:
2162 	nla_nest_cancel(skb, nest);
2163 
2164 	return -EMSGSIZE;
2165 }
2166 
2167 int tipc_nl_add_bc_link(struct net *net, struct tipc_nl_msg *msg)
2168 {
2169 	int err;
2170 	void *hdr;
2171 	struct nlattr *attrs;
2172 	struct nlattr *prop;
2173 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2174 	struct tipc_link *bcl = tn->bcl;
2175 
2176 	if (!bcl)
2177 		return 0;
2178 
2179 	tipc_bcast_lock(net);
2180 
2181 	hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2182 			  NLM_F_MULTI, TIPC_NL_LINK_GET);
2183 	if (!hdr) {
2184 		tipc_bcast_unlock(net);
2185 		return -EMSGSIZE;
2186 	}
2187 
2188 	attrs = nla_nest_start(msg->skb, TIPC_NLA_LINK);
2189 	if (!attrs)
2190 		goto msg_full;
2191 
2192 	/* The broadcast link is always up */
2193 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2194 		goto attr_msg_full;
2195 
2196 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
2197 		goto attr_msg_full;
2198 	if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
2199 		goto attr_msg_full;
2200 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, 0))
2201 		goto attr_msg_full;
2202 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, 0))
2203 		goto attr_msg_full;
2204 
2205 	prop = nla_nest_start(msg->skb, TIPC_NLA_LINK_PROP);
2206 	if (!prop)
2207 		goto attr_msg_full;
2208 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->window))
2209 		goto prop_msg_full;
2210 	nla_nest_end(msg->skb, prop);
2211 
2212 	err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
2213 	if (err)
2214 		goto attr_msg_full;
2215 
2216 	tipc_bcast_unlock(net);
2217 	nla_nest_end(msg->skb, attrs);
2218 	genlmsg_end(msg->skb, hdr);
2219 
2220 	return 0;
2221 
2222 prop_msg_full:
2223 	nla_nest_cancel(msg->skb, prop);
2224 attr_msg_full:
2225 	nla_nest_cancel(msg->skb, attrs);
2226 msg_full:
2227 	tipc_bcast_unlock(net);
2228 	genlmsg_cancel(msg->skb, hdr);
2229 
2230 	return -EMSGSIZE;
2231 }
2232 
2233 void tipc_link_set_tolerance(struct tipc_link *l, u32 tol,
2234 			     struct sk_buff_head *xmitq)
2235 {
2236 	l->tolerance = tol;
2237 	if (l->bc_rcvlink)
2238 		l->bc_rcvlink->tolerance = tol;
2239 	if (link_is_up(l))
2240 		tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, tol, 0, xmitq);
2241 }
2242 
2243 void tipc_link_set_prio(struct tipc_link *l, u32 prio,
2244 			struct sk_buff_head *xmitq)
2245 {
2246 	l->priority = prio;
2247 	tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, prio, xmitq);
2248 }
2249 
2250 void tipc_link_set_abort_limit(struct tipc_link *l, u32 limit)
2251 {
2252 	l->abort_limit = limit;
2253 }
2254