xref: /openbmc/linux/net/tipc/link.c (revision 8795a739)
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 #include "trace.h"
47 
48 #include <linux/pkt_sched.h>
49 
50 struct tipc_stats {
51 	u32 sent_pkts;
52 	u32 recv_pkts;
53 	u32 sent_states;
54 	u32 recv_states;
55 	u32 sent_probes;
56 	u32 recv_probes;
57 	u32 sent_nacks;
58 	u32 recv_nacks;
59 	u32 sent_acks;
60 	u32 sent_bundled;
61 	u32 sent_bundles;
62 	u32 recv_bundled;
63 	u32 recv_bundles;
64 	u32 retransmitted;
65 	u32 sent_fragmented;
66 	u32 sent_fragments;
67 	u32 recv_fragmented;
68 	u32 recv_fragments;
69 	u32 link_congs;		/* # port sends blocked by congestion */
70 	u32 deferred_recv;
71 	u32 duplicates;
72 	u32 max_queue_sz;	/* send queue size high water mark */
73 	u32 accu_queue_sz;	/* used for send queue size profiling */
74 	u32 queue_sz_counts;	/* used for send queue size profiling */
75 	u32 msg_length_counts;	/* used for message length profiling */
76 	u32 msg_lengths_total;	/* used for message length profiling */
77 	u32 msg_length_profile[7]; /* used for msg. length profiling */
78 };
79 
80 /**
81  * struct tipc_link - TIPC link data structure
82  * @addr: network address of link's peer node
83  * @name: link name character string
84  * @media_addr: media address to use when sending messages over link
85  * @timer: link timer
86  * @net: pointer to namespace struct
87  * @refcnt: reference counter for permanent references (owner node & timer)
88  * @peer_session: link session # being used by peer end of link
89  * @peer_bearer_id: bearer id used by link's peer endpoint
90  * @bearer_id: local bearer id used by link
91  * @tolerance: minimum link continuity loss needed to reset link [in ms]
92  * @abort_limit: # of unacknowledged continuity probes needed to reset link
93  * @state: current state of link FSM
94  * @peer_caps: bitmap describing capabilities of peer node
95  * @silent_intv_cnt: # of timer intervals without any reception from peer
96  * @proto_msg: template for control messages generated by link
97  * @pmsg: convenience pointer to "proto_msg" field
98  * @priority: current link priority
99  * @net_plane: current link network plane ('A' through 'H')
100  * @mon_state: cookie with information needed by link monitor
101  * @backlog_limit: backlog queue congestion thresholds (indexed by importance)
102  * @exp_msg_count: # of tunnelled messages expected during link changeover
103  * @reset_rcv_checkpt: seq # of last acknowledged message at time of link reset
104  * @mtu: current maximum packet size for this link
105  * @advertised_mtu: advertised own mtu when link is being established
106  * @transmitq: queue for sent, non-acked messages
107  * @backlogq: queue for messages waiting to be sent
108  * @snt_nxt: next sequence number to use for outbound messages
109  * @ackers: # of peers that needs to ack each packet before it can be released
110  * @acked: # last packet acked by a certain peer. Used for broadcast.
111  * @rcv_nxt: next sequence number to expect for inbound messages
112  * @deferred_queue: deferred queue saved OOS b'cast message received from node
113  * @unacked_window: # of inbound messages rx'd without ack'ing back to peer
114  * @inputq: buffer queue for messages to be delivered upwards
115  * @namedq: buffer queue for name table messages to be delivered upwards
116  * @next_out: ptr to first unsent outbound message in queue
117  * @wakeupq: linked list of wakeup msgs waiting for link congestion to abate
118  * @long_msg_seq_no: next identifier to use for outbound fragmented messages
119  * @reasm_buf: head of partially reassembled inbound message fragments
120  * @bc_rcvr: marks that this is a broadcast receiver link
121  * @stats: collects statistics regarding link activity
122  */
123 struct tipc_link {
124 	u32 addr;
125 	char name[TIPC_MAX_LINK_NAME];
126 	struct net *net;
127 
128 	/* Management and link supervision data */
129 	u16 peer_session;
130 	u16 session;
131 	u16 snd_nxt_state;
132 	u16 rcv_nxt_state;
133 	u32 peer_bearer_id;
134 	u32 bearer_id;
135 	u32 tolerance;
136 	u32 abort_limit;
137 	u32 state;
138 	u16 peer_caps;
139 	bool in_session;
140 	bool active;
141 	u32 silent_intv_cnt;
142 	char if_name[TIPC_MAX_IF_NAME];
143 	u32 priority;
144 	char net_plane;
145 	struct tipc_mon_state mon_state;
146 	u16 rst_cnt;
147 
148 	/* Failover/synch */
149 	u16 drop_point;
150 	struct sk_buff *failover_reasm_skb;
151 	struct sk_buff_head failover_deferdq;
152 
153 	/* Max packet negotiation */
154 	u16 mtu;
155 	u16 advertised_mtu;
156 
157 	/* Sending */
158 	struct sk_buff_head transmq;
159 	struct sk_buff_head backlogq;
160 	struct {
161 		u16 len;
162 		u16 limit;
163 		struct sk_buff *target_bskb;
164 	} backlog[5];
165 	u16 snd_nxt;
166 	u16 window;
167 
168 	/* Reception */
169 	u16 rcv_nxt;
170 	u32 rcv_unacked;
171 	struct sk_buff_head deferdq;
172 	struct sk_buff_head *inputq;
173 	struct sk_buff_head *namedq;
174 
175 	/* Congestion handling */
176 	struct sk_buff_head wakeupq;
177 
178 	/* Fragmentation/reassembly */
179 	struct sk_buff *reasm_buf;
180 	struct sk_buff *reasm_tnlmsg;
181 
182 	/* Broadcast */
183 	u16 ackers;
184 	u16 acked;
185 	struct tipc_link *bc_rcvlink;
186 	struct tipc_link *bc_sndlink;
187 	u8 nack_state;
188 	bool bc_peer_is_up;
189 
190 	/* Statistics */
191 	struct tipc_stats stats;
192 };
193 
194 /*
195  * Error message prefixes
196  */
197 static const char *link_co_err = "Link tunneling error, ";
198 static const char *link_rst_msg = "Resetting link ";
199 
200 /* Send states for broadcast NACKs
201  */
202 enum {
203 	BC_NACK_SND_CONDITIONAL,
204 	BC_NACK_SND_UNCONDITIONAL,
205 	BC_NACK_SND_SUPPRESS,
206 };
207 
208 #define TIPC_BC_RETR_LIM  (jiffies + msecs_to_jiffies(10))
209 #define TIPC_UC_RETR_TIME (jiffies + msecs_to_jiffies(1))
210 
211 /*
212  * Interval between NACKs when packets arrive out of order
213  */
214 #define TIPC_NACK_INTV (TIPC_MIN_LINK_WIN * 2)
215 
216 /* Link FSM states:
217  */
218 enum {
219 	LINK_ESTABLISHED     = 0xe,
220 	LINK_ESTABLISHING    = 0xe  << 4,
221 	LINK_RESET           = 0x1  << 8,
222 	LINK_RESETTING       = 0x2  << 12,
223 	LINK_PEER_RESET      = 0xd  << 16,
224 	LINK_FAILINGOVER     = 0xf  << 20,
225 	LINK_SYNCHING        = 0xc  << 24
226 };
227 
228 /* Link FSM state checking routines
229  */
230 static int link_is_up(struct tipc_link *l)
231 {
232 	return l->state & (LINK_ESTABLISHED | LINK_SYNCHING);
233 }
234 
235 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
236 			       struct sk_buff_head *xmitq);
237 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
238 				      bool probe_reply, u16 rcvgap,
239 				      int tolerance, int priority,
240 				      struct sk_buff_head *xmitq);
241 static void link_print(struct tipc_link *l, const char *str);
242 static int tipc_link_build_nack_msg(struct tipc_link *l,
243 				    struct sk_buff_head *xmitq);
244 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
245 					struct sk_buff_head *xmitq);
246 static bool tipc_link_release_pkts(struct tipc_link *l, u16 to);
247 static u16 tipc_build_gap_ack_blks(struct tipc_link *l, void *data);
248 static int tipc_link_advance_transmq(struct tipc_link *l, u16 acked, u16 gap,
249 				     struct tipc_gap_ack_blks *ga,
250 				     struct sk_buff_head *xmitq);
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 	trace_tipc_link_reset(rcv_l, TIPC_DUMP_ALL, "bclink removed!");
363 	tipc_link_reset(rcv_l);
364 	rcv_l->state = LINK_RESET;
365 	if (!snd_l->ackers) {
366 		trace_tipc_link_reset(snd_l, TIPC_DUMP_ALL, "zero ackers!");
367 		tipc_link_reset(snd_l);
368 		snd_l->state = LINK_RESET;
369 		__skb_queue_purge(xmitq);
370 	}
371 }
372 
373 int tipc_link_bc_peers(struct tipc_link *l)
374 {
375 	return l->ackers;
376 }
377 
378 static u16 link_bc_rcv_gap(struct tipc_link *l)
379 {
380 	struct sk_buff *skb = skb_peek(&l->deferdq);
381 	u16 gap = 0;
382 
383 	if (more(l->snd_nxt, l->rcv_nxt))
384 		gap = l->snd_nxt - l->rcv_nxt;
385 	if (skb)
386 		gap = buf_seqno(skb) - l->rcv_nxt;
387 	return gap;
388 }
389 
390 void tipc_link_set_mtu(struct tipc_link *l, int mtu)
391 {
392 	l->mtu = mtu;
393 }
394 
395 int tipc_link_mtu(struct tipc_link *l)
396 {
397 	return l->mtu;
398 }
399 
400 u16 tipc_link_rcv_nxt(struct tipc_link *l)
401 {
402 	return l->rcv_nxt;
403 }
404 
405 u16 tipc_link_acked(struct tipc_link *l)
406 {
407 	return l->acked;
408 }
409 
410 char *tipc_link_name(struct tipc_link *l)
411 {
412 	return l->name;
413 }
414 
415 u32 tipc_link_state(struct tipc_link *l)
416 {
417 	return l->state;
418 }
419 
420 /**
421  * tipc_link_create - create a new link
422  * @n: pointer to associated node
423  * @if_name: associated interface name
424  * @bearer_id: id (index) of associated bearer
425  * @tolerance: link tolerance to be used by link
426  * @net_plane: network plane (A,B,c..) this link belongs to
427  * @mtu: mtu to be advertised by link
428  * @priority: priority to be used by link
429  * @window: send window to be used by link
430  * @session: session to be used by link
431  * @ownnode: identity of own node
432  * @peer: node id of peer node
433  * @peer_caps: bitmap describing peer node capabilities
434  * @bc_sndlink: the namespace global link used for broadcast sending
435  * @bc_rcvlink: the peer specific link used for broadcast reception
436  * @inputq: queue to put messages ready for delivery
437  * @namedq: queue to put binding table update messages ready for delivery
438  * @link: return value, pointer to put the created link
439  *
440  * Returns true if link was created, otherwise false
441  */
442 bool tipc_link_create(struct net *net, char *if_name, int bearer_id,
443 		      int tolerance, char net_plane, u32 mtu, int priority,
444 		      int window, u32 session, u32 self,
445 		      u32 peer, u8 *peer_id, u16 peer_caps,
446 		      struct tipc_link *bc_sndlink,
447 		      struct tipc_link *bc_rcvlink,
448 		      struct sk_buff_head *inputq,
449 		      struct sk_buff_head *namedq,
450 		      struct tipc_link **link)
451 {
452 	char peer_str[NODE_ID_STR_LEN] = {0,};
453 	char self_str[NODE_ID_STR_LEN] = {0,};
454 	struct tipc_link *l;
455 
456 	l = kzalloc(sizeof(*l), GFP_ATOMIC);
457 	if (!l)
458 		return false;
459 	*link = l;
460 	l->session = session;
461 
462 	/* Set link name for unicast links only */
463 	if (peer_id) {
464 		tipc_nodeid2string(self_str, tipc_own_id(net));
465 		if (strlen(self_str) > 16)
466 			sprintf(self_str, "%x", self);
467 		tipc_nodeid2string(peer_str, peer_id);
468 		if (strlen(peer_str) > 16)
469 			sprintf(peer_str, "%x", peer);
470 	}
471 	/* Peer i/f name will be completed by reset/activate message */
472 	snprintf(l->name, sizeof(l->name), "%s:%s-%s:unknown",
473 		 self_str, if_name, peer_str);
474 
475 	strcpy(l->if_name, if_name);
476 	l->addr = peer;
477 	l->peer_caps = peer_caps;
478 	l->net = net;
479 	l->in_session = false;
480 	l->bearer_id = bearer_id;
481 	l->tolerance = tolerance;
482 	if (bc_rcvlink)
483 		bc_rcvlink->tolerance = tolerance;
484 	l->net_plane = net_plane;
485 	l->advertised_mtu = mtu;
486 	l->mtu = mtu;
487 	l->priority = priority;
488 	tipc_link_set_queue_limits(l, window);
489 	l->ackers = 1;
490 	l->bc_sndlink = bc_sndlink;
491 	l->bc_rcvlink = bc_rcvlink;
492 	l->inputq = inputq;
493 	l->namedq = namedq;
494 	l->state = LINK_RESETTING;
495 	__skb_queue_head_init(&l->transmq);
496 	__skb_queue_head_init(&l->backlogq);
497 	__skb_queue_head_init(&l->deferdq);
498 	__skb_queue_head_init(&l->failover_deferdq);
499 	skb_queue_head_init(&l->wakeupq);
500 	skb_queue_head_init(l->inputq);
501 	return true;
502 }
503 
504 /**
505  * tipc_link_bc_create - create new link to be used for broadcast
506  * @n: pointer to associated node
507  * @mtu: mtu to be used initially if no peers
508  * @window: send window to be used
509  * @inputq: queue to put messages ready for delivery
510  * @namedq: queue to put binding table update messages ready for delivery
511  * @link: return value, pointer to put the created link
512  *
513  * Returns true if link was created, otherwise false
514  */
515 bool tipc_link_bc_create(struct net *net, u32 ownnode, u32 peer,
516 			 int mtu, int window, u16 peer_caps,
517 			 struct sk_buff_head *inputq,
518 			 struct sk_buff_head *namedq,
519 			 struct tipc_link *bc_sndlink,
520 			 struct tipc_link **link)
521 {
522 	struct tipc_link *l;
523 
524 	if (!tipc_link_create(net, "", MAX_BEARERS, 0, 'Z', mtu, 0, window,
525 			      0, ownnode, peer, NULL, peer_caps, bc_sndlink,
526 			      NULL, inputq, namedq, link))
527 		return false;
528 
529 	l = *link;
530 	strcpy(l->name, tipc_bclink_name);
531 	trace_tipc_link_reset(l, TIPC_DUMP_ALL, "bclink created!");
532 	tipc_link_reset(l);
533 	l->state = LINK_RESET;
534 	l->ackers = 0;
535 	l->bc_rcvlink = l;
536 
537 	/* Broadcast send link is always up */
538 	if (link_is_bc_sndlink(l))
539 		l->state = LINK_ESTABLISHED;
540 
541 	/* Disable replicast if even a single peer doesn't support it */
542 	if (link_is_bc_rcvlink(l) && !(peer_caps & TIPC_BCAST_RCAST))
543 		tipc_bcast_disable_rcast(net);
544 
545 	return true;
546 }
547 
548 /**
549  * tipc_link_fsm_evt - link finite state machine
550  * @l: pointer to link
551  * @evt: state machine event to be processed
552  */
553 int tipc_link_fsm_evt(struct tipc_link *l, int evt)
554 {
555 	int rc = 0;
556 	int old_state = l->state;
557 
558 	switch (l->state) {
559 	case LINK_RESETTING:
560 		switch (evt) {
561 		case LINK_PEER_RESET_EVT:
562 			l->state = LINK_PEER_RESET;
563 			break;
564 		case LINK_RESET_EVT:
565 			l->state = LINK_RESET;
566 			break;
567 		case LINK_FAILURE_EVT:
568 		case LINK_FAILOVER_BEGIN_EVT:
569 		case LINK_ESTABLISH_EVT:
570 		case LINK_FAILOVER_END_EVT:
571 		case LINK_SYNCH_BEGIN_EVT:
572 		case LINK_SYNCH_END_EVT:
573 		default:
574 			goto illegal_evt;
575 		}
576 		break;
577 	case LINK_RESET:
578 		switch (evt) {
579 		case LINK_PEER_RESET_EVT:
580 			l->state = LINK_ESTABLISHING;
581 			break;
582 		case LINK_FAILOVER_BEGIN_EVT:
583 			l->state = LINK_FAILINGOVER;
584 		case LINK_FAILURE_EVT:
585 		case LINK_RESET_EVT:
586 		case LINK_ESTABLISH_EVT:
587 		case LINK_FAILOVER_END_EVT:
588 			break;
589 		case LINK_SYNCH_BEGIN_EVT:
590 		case LINK_SYNCH_END_EVT:
591 		default:
592 			goto illegal_evt;
593 		}
594 		break;
595 	case LINK_PEER_RESET:
596 		switch (evt) {
597 		case LINK_RESET_EVT:
598 			l->state = LINK_ESTABLISHING;
599 			break;
600 		case LINK_PEER_RESET_EVT:
601 		case LINK_ESTABLISH_EVT:
602 		case LINK_FAILURE_EVT:
603 			break;
604 		case LINK_SYNCH_BEGIN_EVT:
605 		case LINK_SYNCH_END_EVT:
606 		case LINK_FAILOVER_BEGIN_EVT:
607 		case LINK_FAILOVER_END_EVT:
608 		default:
609 			goto illegal_evt;
610 		}
611 		break;
612 	case LINK_FAILINGOVER:
613 		switch (evt) {
614 		case LINK_FAILOVER_END_EVT:
615 			l->state = LINK_RESET;
616 			break;
617 		case LINK_PEER_RESET_EVT:
618 		case LINK_RESET_EVT:
619 		case LINK_ESTABLISH_EVT:
620 		case LINK_FAILURE_EVT:
621 			break;
622 		case LINK_FAILOVER_BEGIN_EVT:
623 		case LINK_SYNCH_BEGIN_EVT:
624 		case LINK_SYNCH_END_EVT:
625 		default:
626 			goto illegal_evt;
627 		}
628 		break;
629 	case LINK_ESTABLISHING:
630 		switch (evt) {
631 		case LINK_ESTABLISH_EVT:
632 			l->state = LINK_ESTABLISHED;
633 			break;
634 		case LINK_FAILOVER_BEGIN_EVT:
635 			l->state = LINK_FAILINGOVER;
636 			break;
637 		case LINK_RESET_EVT:
638 			l->state = LINK_RESET;
639 			break;
640 		case LINK_FAILURE_EVT:
641 		case LINK_PEER_RESET_EVT:
642 		case LINK_SYNCH_BEGIN_EVT:
643 		case LINK_FAILOVER_END_EVT:
644 			break;
645 		case LINK_SYNCH_END_EVT:
646 		default:
647 			goto illegal_evt;
648 		}
649 		break;
650 	case LINK_ESTABLISHED:
651 		switch (evt) {
652 		case LINK_PEER_RESET_EVT:
653 			l->state = LINK_PEER_RESET;
654 			rc |= TIPC_LINK_DOWN_EVT;
655 			break;
656 		case LINK_FAILURE_EVT:
657 			l->state = LINK_RESETTING;
658 			rc |= TIPC_LINK_DOWN_EVT;
659 			break;
660 		case LINK_RESET_EVT:
661 			l->state = LINK_RESET;
662 			break;
663 		case LINK_ESTABLISH_EVT:
664 		case LINK_SYNCH_END_EVT:
665 			break;
666 		case LINK_SYNCH_BEGIN_EVT:
667 			l->state = LINK_SYNCHING;
668 			break;
669 		case LINK_FAILOVER_BEGIN_EVT:
670 		case LINK_FAILOVER_END_EVT:
671 		default:
672 			goto illegal_evt;
673 		}
674 		break;
675 	case LINK_SYNCHING:
676 		switch (evt) {
677 		case LINK_PEER_RESET_EVT:
678 			l->state = LINK_PEER_RESET;
679 			rc |= TIPC_LINK_DOWN_EVT;
680 			break;
681 		case LINK_FAILURE_EVT:
682 			l->state = LINK_RESETTING;
683 			rc |= TIPC_LINK_DOWN_EVT;
684 			break;
685 		case LINK_RESET_EVT:
686 			l->state = LINK_RESET;
687 			break;
688 		case LINK_ESTABLISH_EVT:
689 		case LINK_SYNCH_BEGIN_EVT:
690 			break;
691 		case LINK_SYNCH_END_EVT:
692 			l->state = LINK_ESTABLISHED;
693 			break;
694 		case LINK_FAILOVER_BEGIN_EVT:
695 		case LINK_FAILOVER_END_EVT:
696 		default:
697 			goto illegal_evt;
698 		}
699 		break;
700 	default:
701 		pr_err("Unknown FSM state %x in %s\n", l->state, l->name);
702 	}
703 	trace_tipc_link_fsm(l->name, old_state, l->state, evt);
704 	return rc;
705 illegal_evt:
706 	pr_err("Illegal FSM event %x in state %x on link %s\n",
707 	       evt, l->state, l->name);
708 	trace_tipc_link_fsm(l->name, old_state, l->state, evt);
709 	return rc;
710 }
711 
712 /* link_profile_stats - update statistical profiling of traffic
713  */
714 static void link_profile_stats(struct tipc_link *l)
715 {
716 	struct sk_buff *skb;
717 	struct tipc_msg *msg;
718 	int length;
719 
720 	/* Update counters used in statistical profiling of send traffic */
721 	l->stats.accu_queue_sz += skb_queue_len(&l->transmq);
722 	l->stats.queue_sz_counts++;
723 
724 	skb = skb_peek(&l->transmq);
725 	if (!skb)
726 		return;
727 	msg = buf_msg(skb);
728 	length = msg_size(msg);
729 
730 	if (msg_user(msg) == MSG_FRAGMENTER) {
731 		if (msg_type(msg) != FIRST_FRAGMENT)
732 			return;
733 		length = msg_size(msg_inner_hdr(msg));
734 	}
735 	l->stats.msg_lengths_total += length;
736 	l->stats.msg_length_counts++;
737 	if (length <= 64)
738 		l->stats.msg_length_profile[0]++;
739 	else if (length <= 256)
740 		l->stats.msg_length_profile[1]++;
741 	else if (length <= 1024)
742 		l->stats.msg_length_profile[2]++;
743 	else if (length <= 4096)
744 		l->stats.msg_length_profile[3]++;
745 	else if (length <= 16384)
746 		l->stats.msg_length_profile[4]++;
747 	else if (length <= 32768)
748 		l->stats.msg_length_profile[5]++;
749 	else
750 		l->stats.msg_length_profile[6]++;
751 }
752 
753 /**
754  * tipc_link_too_silent - check if link is "too silent"
755  * @l: tipc link to be checked
756  *
757  * Returns true if the link 'silent_intv_cnt' is about to reach the
758  * 'abort_limit' value, otherwise false
759  */
760 bool tipc_link_too_silent(struct tipc_link *l)
761 {
762 	return (l->silent_intv_cnt + 2 > l->abort_limit);
763 }
764 
765 /* tipc_link_timeout - perform periodic task as instructed from node timeout
766  */
767 int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq)
768 {
769 	int mtyp = 0;
770 	int rc = 0;
771 	bool state = false;
772 	bool probe = false;
773 	bool setup = false;
774 	u16 bc_snt = l->bc_sndlink->snd_nxt - 1;
775 	u16 bc_acked = l->bc_rcvlink->acked;
776 	struct tipc_mon_state *mstate = &l->mon_state;
777 
778 	trace_tipc_link_timeout(l, TIPC_DUMP_NONE, " ");
779 	trace_tipc_link_too_silent(l, TIPC_DUMP_ALL, " ");
780 	switch (l->state) {
781 	case LINK_ESTABLISHED:
782 	case LINK_SYNCHING:
783 		mtyp = STATE_MSG;
784 		link_profile_stats(l);
785 		tipc_mon_get_state(l->net, l->addr, mstate, l->bearer_id);
786 		if (mstate->reset || (l->silent_intv_cnt > l->abort_limit))
787 			return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
788 		state = bc_acked != bc_snt;
789 		state |= l->bc_rcvlink->rcv_unacked;
790 		state |= l->rcv_unacked;
791 		state |= !skb_queue_empty(&l->transmq);
792 		state |= !skb_queue_empty(&l->deferdq);
793 		probe = mstate->probing;
794 		probe |= l->silent_intv_cnt;
795 		if (probe || mstate->monitoring)
796 			l->silent_intv_cnt++;
797 		break;
798 	case LINK_RESET:
799 		setup = l->rst_cnt++ <= 4;
800 		setup |= !(l->rst_cnt % 16);
801 		mtyp = RESET_MSG;
802 		break;
803 	case LINK_ESTABLISHING:
804 		setup = true;
805 		mtyp = ACTIVATE_MSG;
806 		break;
807 	case LINK_PEER_RESET:
808 	case LINK_RESETTING:
809 	case LINK_FAILINGOVER:
810 		break;
811 	default:
812 		break;
813 	}
814 
815 	if (state || probe || setup)
816 		tipc_link_build_proto_msg(l, mtyp, probe, 0, 0, 0, 0, xmitq);
817 
818 	return rc;
819 }
820 
821 /**
822  * link_schedule_user - schedule a message sender for wakeup after congestion
823  * @l: congested link
824  * @hdr: header of message that is being sent
825  * Create pseudo msg to send back to user when congestion abates
826  */
827 static int link_schedule_user(struct tipc_link *l, struct tipc_msg *hdr)
828 {
829 	u32 dnode = tipc_own_addr(l->net);
830 	u32 dport = msg_origport(hdr);
831 	struct sk_buff *skb;
832 
833 	/* Create and schedule wakeup pseudo message */
834 	skb = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
835 			      dnode, l->addr, dport, 0, 0);
836 	if (!skb)
837 		return -ENOBUFS;
838 	msg_set_dest_droppable(buf_msg(skb), true);
839 	TIPC_SKB_CB(skb)->chain_imp = msg_importance(hdr);
840 	skb_queue_tail(&l->wakeupq, skb);
841 	l->stats.link_congs++;
842 	trace_tipc_link_conges(l, TIPC_DUMP_ALL, "wakeup scheduled!");
843 	return -ELINKCONG;
844 }
845 
846 /**
847  * link_prepare_wakeup - prepare users for wakeup after congestion
848  * @l: congested link
849  * Wake up a number of waiting users, as permitted by available space
850  * in the send queue
851  */
852 static void link_prepare_wakeup(struct tipc_link *l)
853 {
854 	struct sk_buff_head *wakeupq = &l->wakeupq;
855 	struct sk_buff_head *inputq = l->inputq;
856 	struct sk_buff *skb, *tmp;
857 	struct sk_buff_head tmpq;
858 	int avail[5] = {0,};
859 	int imp = 0;
860 
861 	__skb_queue_head_init(&tmpq);
862 
863 	for (; imp <= TIPC_SYSTEM_IMPORTANCE; imp++)
864 		avail[imp] = l->backlog[imp].limit - l->backlog[imp].len;
865 
866 	skb_queue_walk_safe(wakeupq, skb, tmp) {
867 		imp = TIPC_SKB_CB(skb)->chain_imp;
868 		if (avail[imp] <= 0)
869 			continue;
870 		avail[imp]--;
871 		__skb_unlink(skb, wakeupq);
872 		__skb_queue_tail(&tmpq, skb);
873 	}
874 
875 	spin_lock_bh(&inputq->lock);
876 	skb_queue_splice_tail(&tmpq, inputq);
877 	spin_unlock_bh(&inputq->lock);
878 
879 }
880 
881 void tipc_link_reset(struct tipc_link *l)
882 {
883 	struct sk_buff_head list;
884 	u32 imp;
885 
886 	__skb_queue_head_init(&list);
887 
888 	l->in_session = false;
889 	/* Force re-synch of peer session number before establishing */
890 	l->peer_session--;
891 	l->session++;
892 	l->mtu = l->advertised_mtu;
893 
894 	spin_lock_bh(&l->wakeupq.lock);
895 	skb_queue_splice_init(&l->wakeupq, &list);
896 	spin_unlock_bh(&l->wakeupq.lock);
897 
898 	spin_lock_bh(&l->inputq->lock);
899 	skb_queue_splice_init(&list, l->inputq);
900 	spin_unlock_bh(&l->inputq->lock);
901 
902 	__skb_queue_purge(&l->transmq);
903 	__skb_queue_purge(&l->deferdq);
904 	__skb_queue_purge(&l->backlogq);
905 	__skb_queue_purge(&l->failover_deferdq);
906 	for (imp = 0; imp <= TIPC_SYSTEM_IMPORTANCE; imp++) {
907 		l->backlog[imp].len = 0;
908 		l->backlog[imp].target_bskb = NULL;
909 	}
910 	kfree_skb(l->reasm_buf);
911 	kfree_skb(l->reasm_tnlmsg);
912 	kfree_skb(l->failover_reasm_skb);
913 	l->reasm_buf = NULL;
914 	l->reasm_tnlmsg = NULL;
915 	l->failover_reasm_skb = NULL;
916 	l->rcv_unacked = 0;
917 	l->snd_nxt = 1;
918 	l->rcv_nxt = 1;
919 	l->snd_nxt_state = 1;
920 	l->rcv_nxt_state = 1;
921 	l->acked = 0;
922 	l->silent_intv_cnt = 0;
923 	l->rst_cnt = 0;
924 	l->bc_peer_is_up = false;
925 	memset(&l->mon_state, 0, sizeof(l->mon_state));
926 	tipc_link_reset_stats(l);
927 }
928 
929 /**
930  * tipc_link_xmit(): enqueue buffer list according to queue situation
931  * @link: link to use
932  * @list: chain of buffers containing message
933  * @xmitq: returned list of packets to be sent by caller
934  *
935  * Consumes the buffer chain.
936  * Returns 0 if success, or errno: -ELINKCONG, -EMSGSIZE or -ENOBUFS
937  * Messages at TIPC_SYSTEM_IMPORTANCE are always accepted
938  */
939 int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list,
940 		   struct sk_buff_head *xmitq)
941 {
942 	struct tipc_msg *hdr = buf_msg(skb_peek(list));
943 	unsigned int maxwin = l->window;
944 	int imp = msg_importance(hdr);
945 	unsigned int mtu = l->mtu;
946 	u16 ack = l->rcv_nxt - 1;
947 	u16 seqno = l->snd_nxt;
948 	u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
949 	struct sk_buff_head *transmq = &l->transmq;
950 	struct sk_buff_head *backlogq = &l->backlogq;
951 	struct sk_buff *skb, *_skb, **tskb;
952 	int pkt_cnt = skb_queue_len(list);
953 	int rc = 0;
954 
955 	if (unlikely(msg_size(hdr) > mtu)) {
956 		pr_warn("Too large msg, purging xmit list %d %d %d %d %d!\n",
957 			skb_queue_len(list), msg_user(hdr),
958 			msg_type(hdr), msg_size(hdr), mtu);
959 		__skb_queue_purge(list);
960 		return -EMSGSIZE;
961 	}
962 
963 	/* Allow oversubscription of one data msg per source at congestion */
964 	if (unlikely(l->backlog[imp].len >= l->backlog[imp].limit)) {
965 		if (imp == TIPC_SYSTEM_IMPORTANCE) {
966 			pr_warn("%s<%s>, link overflow", link_rst_msg, l->name);
967 			return -ENOBUFS;
968 		}
969 		rc = link_schedule_user(l, hdr);
970 	}
971 
972 	if (pkt_cnt > 1) {
973 		l->stats.sent_fragmented++;
974 		l->stats.sent_fragments += pkt_cnt;
975 	}
976 
977 	/* Prepare each packet for sending, and add to relevant queue: */
978 	while (skb_queue_len(list)) {
979 		skb = skb_peek(list);
980 		hdr = buf_msg(skb);
981 		msg_set_seqno(hdr, seqno);
982 		msg_set_ack(hdr, ack);
983 		msg_set_bcast_ack(hdr, bc_ack);
984 
985 		if (likely(skb_queue_len(transmq) < maxwin)) {
986 			_skb = skb_clone(skb, GFP_ATOMIC);
987 			if (!_skb) {
988 				__skb_queue_purge(list);
989 				return -ENOBUFS;
990 			}
991 			__skb_dequeue(list);
992 			__skb_queue_tail(transmq, skb);
993 			/* next retransmit attempt */
994 			if (link_is_bc_sndlink(l))
995 				TIPC_SKB_CB(skb)->nxt_retr = TIPC_BC_RETR_LIM;
996 			__skb_queue_tail(xmitq, _skb);
997 			TIPC_SKB_CB(skb)->ackers = l->ackers;
998 			l->rcv_unacked = 0;
999 			l->stats.sent_pkts++;
1000 			seqno++;
1001 			continue;
1002 		}
1003 		tskb = &l->backlog[imp].target_bskb;
1004 		if (tipc_msg_bundle(*tskb, hdr, mtu)) {
1005 			kfree_skb(__skb_dequeue(list));
1006 			l->stats.sent_bundled++;
1007 			continue;
1008 		}
1009 		if (tipc_msg_make_bundle(tskb, hdr, mtu, l->addr)) {
1010 			kfree_skb(__skb_dequeue(list));
1011 			__skb_queue_tail(backlogq, *tskb);
1012 			l->backlog[imp].len++;
1013 			l->stats.sent_bundled++;
1014 			l->stats.sent_bundles++;
1015 			continue;
1016 		}
1017 		l->backlog[imp].target_bskb = NULL;
1018 		l->backlog[imp].len += skb_queue_len(list);
1019 		skb_queue_splice_tail_init(list, backlogq);
1020 	}
1021 	l->snd_nxt = seqno;
1022 	return rc;
1023 }
1024 
1025 static void tipc_link_advance_backlog(struct tipc_link *l,
1026 				      struct sk_buff_head *xmitq)
1027 {
1028 	struct sk_buff *skb, *_skb;
1029 	struct tipc_msg *hdr;
1030 	u16 seqno = l->snd_nxt;
1031 	u16 ack = l->rcv_nxt - 1;
1032 	u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1033 	u32 imp;
1034 
1035 	while (skb_queue_len(&l->transmq) < l->window) {
1036 		skb = skb_peek(&l->backlogq);
1037 		if (!skb)
1038 			break;
1039 		_skb = skb_clone(skb, GFP_ATOMIC);
1040 		if (!_skb)
1041 			break;
1042 		__skb_dequeue(&l->backlogq);
1043 		hdr = buf_msg(skb);
1044 		imp = msg_importance(hdr);
1045 		l->backlog[imp].len--;
1046 		if (unlikely(skb == l->backlog[imp].target_bskb))
1047 			l->backlog[imp].target_bskb = NULL;
1048 		__skb_queue_tail(&l->transmq, skb);
1049 		/* next retransmit attempt */
1050 		if (link_is_bc_sndlink(l))
1051 			TIPC_SKB_CB(skb)->nxt_retr = TIPC_BC_RETR_LIM;
1052 
1053 		__skb_queue_tail(xmitq, _skb);
1054 		TIPC_SKB_CB(skb)->ackers = l->ackers;
1055 		msg_set_seqno(hdr, seqno);
1056 		msg_set_ack(hdr, ack);
1057 		msg_set_bcast_ack(hdr, bc_ack);
1058 		l->rcv_unacked = 0;
1059 		l->stats.sent_pkts++;
1060 		seqno++;
1061 	}
1062 	l->snd_nxt = seqno;
1063 }
1064 
1065 /**
1066  * link_retransmit_failure() - Detect repeated retransmit failures
1067  * @l: tipc link sender
1068  * @r: tipc link receiver (= l in case of unicast)
1069  * @rc: returned code
1070  *
1071  * Return: true if the repeated retransmit failures happens, otherwise
1072  * false
1073  */
1074 static bool link_retransmit_failure(struct tipc_link *l, struct tipc_link *r,
1075 				    int *rc)
1076 {
1077 	struct sk_buff *skb = skb_peek(&l->transmq);
1078 	struct tipc_msg *hdr;
1079 
1080 	if (!skb)
1081 		return false;
1082 
1083 	if (!TIPC_SKB_CB(skb)->retr_cnt)
1084 		return false;
1085 
1086 	if (!time_after(jiffies, TIPC_SKB_CB(skb)->retr_stamp +
1087 			msecs_to_jiffies(r->tolerance)))
1088 		return false;
1089 
1090 	hdr = buf_msg(skb);
1091 	if (link_is_bc_sndlink(l) && !less(r->acked, msg_seqno(hdr)))
1092 		return false;
1093 
1094 	pr_warn("Retransmission failure on link <%s>\n", l->name);
1095 	link_print(l, "State of link ");
1096 	pr_info("Failed msg: usr %u, typ %u, len %u, err %u\n",
1097 		msg_user(hdr), msg_type(hdr), msg_size(hdr), msg_errcode(hdr));
1098 	pr_info("sqno %u, prev: %x, dest: %x\n",
1099 		msg_seqno(hdr), msg_prevnode(hdr), msg_destnode(hdr));
1100 	pr_info("retr_stamp %d, retr_cnt %d\n",
1101 		jiffies_to_msecs(TIPC_SKB_CB(skb)->retr_stamp),
1102 		TIPC_SKB_CB(skb)->retr_cnt);
1103 
1104 	trace_tipc_list_dump(&l->transmq, true, "retrans failure!");
1105 	trace_tipc_link_dump(l, TIPC_DUMP_NONE, "retrans failure!");
1106 	trace_tipc_link_dump(r, TIPC_DUMP_NONE, "retrans failure!");
1107 
1108 	if (link_is_bc_sndlink(l)) {
1109 		r->state = LINK_RESET;
1110 		*rc = TIPC_LINK_DOWN_EVT;
1111 	} else {
1112 		*rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1113 	}
1114 
1115 	return true;
1116 }
1117 
1118 /* tipc_link_bc_retrans() - retransmit zero or more packets
1119  * @l: the link to transmit on
1120  * @r: the receiving link ordering the retransmit. Same as l if unicast
1121  * @from: retransmit from (inclusive) this sequence number
1122  * @to: retransmit to (inclusive) this sequence number
1123  * xmitq: queue for accumulating the retransmitted packets
1124  */
1125 static int tipc_link_bc_retrans(struct tipc_link *l, struct tipc_link *r,
1126 				u16 from, u16 to, struct sk_buff_head *xmitq)
1127 {
1128 	struct sk_buff *_skb, *skb = skb_peek(&l->transmq);
1129 	u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1130 	u16 ack = l->rcv_nxt - 1;
1131 	struct tipc_msg *hdr;
1132 	int rc = 0;
1133 
1134 	if (!skb)
1135 		return 0;
1136 	if (less(to, from))
1137 		return 0;
1138 
1139 	trace_tipc_link_retrans(r, from, to, &l->transmq);
1140 
1141 	if (link_retransmit_failure(l, r, &rc))
1142 		return rc;
1143 
1144 	skb_queue_walk(&l->transmq, skb) {
1145 		hdr = buf_msg(skb);
1146 		if (less(msg_seqno(hdr), from))
1147 			continue;
1148 		if (more(msg_seqno(hdr), to))
1149 			break;
1150 
1151 		if (time_before(jiffies, TIPC_SKB_CB(skb)->nxt_retr))
1152 			continue;
1153 		TIPC_SKB_CB(skb)->nxt_retr = TIPC_BC_RETR_LIM;
1154 		_skb = __pskb_copy(skb, LL_MAX_HEADER + MIN_H_SIZE, GFP_ATOMIC);
1155 		if (!_skb)
1156 			return 0;
1157 		hdr = buf_msg(_skb);
1158 		msg_set_ack(hdr, ack);
1159 		msg_set_bcast_ack(hdr, bc_ack);
1160 		_skb->priority = TC_PRIO_CONTROL;
1161 		__skb_queue_tail(xmitq, _skb);
1162 		l->stats.retransmitted++;
1163 
1164 		/* Increase actual retrans counter & mark first time */
1165 		if (!TIPC_SKB_CB(skb)->retr_cnt++)
1166 			TIPC_SKB_CB(skb)->retr_stamp = jiffies;
1167 	}
1168 	return 0;
1169 }
1170 
1171 /* tipc_data_input - deliver data and name distr msgs to upper layer
1172  *
1173  * Consumes buffer if message is of right type
1174  * Node lock must be held
1175  */
1176 static bool tipc_data_input(struct tipc_link *l, struct sk_buff *skb,
1177 			    struct sk_buff_head *inputq)
1178 {
1179 	struct sk_buff_head *mc_inputq = l->bc_rcvlink->inputq;
1180 	struct tipc_msg *hdr = buf_msg(skb);
1181 
1182 	switch (msg_user(hdr)) {
1183 	case TIPC_LOW_IMPORTANCE:
1184 	case TIPC_MEDIUM_IMPORTANCE:
1185 	case TIPC_HIGH_IMPORTANCE:
1186 	case TIPC_CRITICAL_IMPORTANCE:
1187 		if (unlikely(msg_in_group(hdr) || msg_mcast(hdr))) {
1188 			skb_queue_tail(mc_inputq, skb);
1189 			return true;
1190 		}
1191 		/* fall through */
1192 	case CONN_MANAGER:
1193 		skb_queue_tail(inputq, skb);
1194 		return true;
1195 	case GROUP_PROTOCOL:
1196 		skb_queue_tail(mc_inputq, skb);
1197 		return true;
1198 	case NAME_DISTRIBUTOR:
1199 		l->bc_rcvlink->state = LINK_ESTABLISHED;
1200 		skb_queue_tail(l->namedq, skb);
1201 		return true;
1202 	case MSG_BUNDLER:
1203 	case TUNNEL_PROTOCOL:
1204 	case MSG_FRAGMENTER:
1205 	case BCAST_PROTOCOL:
1206 		return false;
1207 	default:
1208 		pr_warn("Dropping received illegal msg type\n");
1209 		kfree_skb(skb);
1210 		return true;
1211 	};
1212 }
1213 
1214 /* tipc_link_input - process packet that has passed link protocol check
1215  *
1216  * Consumes buffer
1217  */
1218 static int tipc_link_input(struct tipc_link *l, struct sk_buff *skb,
1219 			   struct sk_buff_head *inputq,
1220 			   struct sk_buff **reasm_skb)
1221 {
1222 	struct tipc_msg *hdr = buf_msg(skb);
1223 	struct sk_buff *iskb;
1224 	struct sk_buff_head tmpq;
1225 	int usr = msg_user(hdr);
1226 	int pos = 0;
1227 
1228 	if (usr == MSG_BUNDLER) {
1229 		skb_queue_head_init(&tmpq);
1230 		l->stats.recv_bundles++;
1231 		l->stats.recv_bundled += msg_msgcnt(hdr);
1232 		while (tipc_msg_extract(skb, &iskb, &pos))
1233 			tipc_data_input(l, iskb, &tmpq);
1234 		tipc_skb_queue_splice_tail(&tmpq, inputq);
1235 		return 0;
1236 	} else if (usr == MSG_FRAGMENTER) {
1237 		l->stats.recv_fragments++;
1238 		if (tipc_buf_append(reasm_skb, &skb)) {
1239 			l->stats.recv_fragmented++;
1240 			tipc_data_input(l, skb, inputq);
1241 		} else if (!*reasm_skb && !link_is_bc_rcvlink(l)) {
1242 			pr_warn_ratelimited("Unable to build fragment list\n");
1243 			return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1244 		}
1245 		return 0;
1246 	} else if (usr == BCAST_PROTOCOL) {
1247 		tipc_bcast_lock(l->net);
1248 		tipc_link_bc_init_rcv(l->bc_rcvlink, hdr);
1249 		tipc_bcast_unlock(l->net);
1250 	}
1251 
1252 	kfree_skb(skb);
1253 	return 0;
1254 }
1255 
1256 /* tipc_link_tnl_rcv() - receive TUNNEL_PROTOCOL message, drop or process the
1257  *			 inner message along with the ones in the old link's
1258  *			 deferdq
1259  * @l: tunnel link
1260  * @skb: TUNNEL_PROTOCOL message
1261  * @inputq: queue to put messages ready for delivery
1262  */
1263 static int tipc_link_tnl_rcv(struct tipc_link *l, struct sk_buff *skb,
1264 			     struct sk_buff_head *inputq)
1265 {
1266 	struct sk_buff **reasm_skb = &l->failover_reasm_skb;
1267 	struct sk_buff **reasm_tnlmsg = &l->reasm_tnlmsg;
1268 	struct sk_buff_head *fdefq = &l->failover_deferdq;
1269 	struct tipc_msg *hdr = buf_msg(skb);
1270 	struct sk_buff *iskb;
1271 	int ipos = 0;
1272 	int rc = 0;
1273 	u16 seqno;
1274 
1275 	if (msg_type(hdr) == SYNCH_MSG) {
1276 		kfree_skb(skb);
1277 		return 0;
1278 	}
1279 
1280 	/* Not a fragment? */
1281 	if (likely(!msg_nof_fragms(hdr))) {
1282 		if (unlikely(!tipc_msg_extract(skb, &iskb, &ipos))) {
1283 			pr_warn_ratelimited("Unable to extract msg, defq: %d\n",
1284 					    skb_queue_len(fdefq));
1285 			return 0;
1286 		}
1287 		kfree_skb(skb);
1288 	} else {
1289 		/* Set fragment type for buf_append */
1290 		if (msg_fragm_no(hdr) == 1)
1291 			msg_set_type(hdr, FIRST_FRAGMENT);
1292 		else if (msg_fragm_no(hdr) < msg_nof_fragms(hdr))
1293 			msg_set_type(hdr, FRAGMENT);
1294 		else
1295 			msg_set_type(hdr, LAST_FRAGMENT);
1296 
1297 		if (!tipc_buf_append(reasm_tnlmsg, &skb)) {
1298 			/* Successful but non-complete reassembly? */
1299 			if (*reasm_tnlmsg || link_is_bc_rcvlink(l))
1300 				return 0;
1301 			pr_warn_ratelimited("Unable to reassemble tunnel msg\n");
1302 			return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1303 		}
1304 		iskb = skb;
1305 	}
1306 
1307 	do {
1308 		seqno = buf_seqno(iskb);
1309 		if (unlikely(less(seqno, l->drop_point))) {
1310 			kfree_skb(iskb);
1311 			continue;
1312 		}
1313 		if (unlikely(seqno != l->drop_point)) {
1314 			__tipc_skb_queue_sorted(fdefq, seqno, iskb);
1315 			continue;
1316 		}
1317 
1318 		l->drop_point++;
1319 		if (!tipc_data_input(l, iskb, inputq))
1320 			rc |= tipc_link_input(l, iskb, inputq, reasm_skb);
1321 		if (unlikely(rc))
1322 			break;
1323 	} while ((iskb = __tipc_skb_dequeue(fdefq, l->drop_point)));
1324 
1325 	return rc;
1326 }
1327 
1328 static bool tipc_link_release_pkts(struct tipc_link *l, u16 acked)
1329 {
1330 	bool released = false;
1331 	struct sk_buff *skb, *tmp;
1332 
1333 	skb_queue_walk_safe(&l->transmq, skb, tmp) {
1334 		if (more(buf_seqno(skb), acked))
1335 			break;
1336 		__skb_unlink(skb, &l->transmq);
1337 		kfree_skb(skb);
1338 		released = true;
1339 	}
1340 	return released;
1341 }
1342 
1343 /* tipc_build_gap_ack_blks - build Gap ACK blocks
1344  * @l: tipc link that data have come with gaps in sequence if any
1345  * @data: data buffer to store the Gap ACK blocks after built
1346  *
1347  * returns the actual allocated memory size
1348  */
1349 static u16 tipc_build_gap_ack_blks(struct tipc_link *l, void *data)
1350 {
1351 	struct sk_buff *skb = skb_peek(&l->deferdq);
1352 	struct tipc_gap_ack_blks *ga = data;
1353 	u16 len, expect, seqno = 0;
1354 	u8 n = 0;
1355 
1356 	if (!skb)
1357 		goto exit;
1358 
1359 	expect = buf_seqno(skb);
1360 	skb_queue_walk(&l->deferdq, skb) {
1361 		seqno = buf_seqno(skb);
1362 		if (unlikely(more(seqno, expect))) {
1363 			ga->gacks[n].ack = htons(expect - 1);
1364 			ga->gacks[n].gap = htons(seqno - expect);
1365 			if (++n >= MAX_GAP_ACK_BLKS) {
1366 				pr_info_ratelimited("Too few Gap ACK blocks!\n");
1367 				goto exit;
1368 			}
1369 		} else if (unlikely(less(seqno, expect))) {
1370 			pr_warn("Unexpected skb in deferdq!\n");
1371 			continue;
1372 		}
1373 		expect = seqno + 1;
1374 	}
1375 
1376 	/* last block */
1377 	ga->gacks[n].ack = htons(seqno);
1378 	ga->gacks[n].gap = 0;
1379 	n++;
1380 
1381 exit:
1382 	len = tipc_gap_ack_blks_sz(n);
1383 	ga->len = htons(len);
1384 	ga->gack_cnt = n;
1385 	return len;
1386 }
1387 
1388 /* tipc_link_advance_transmq - advance TIPC link transmq queue by releasing
1389  *			       acked packets, also doing retransmissions if
1390  *			       gaps found
1391  * @l: tipc link with transmq queue to be advanced
1392  * @acked: seqno of last packet acked by peer without any gaps before
1393  * @gap: # of gap packets
1394  * @ga: buffer pointer to Gap ACK blocks from peer
1395  * @xmitq: queue for accumulating the retransmitted packets if any
1396  *
1397  * In case of a repeated retransmit failures, the call will return shortly
1398  * with a returned code (e.g. TIPC_LINK_DOWN_EVT)
1399  */
1400 static int tipc_link_advance_transmq(struct tipc_link *l, u16 acked, u16 gap,
1401 				     struct tipc_gap_ack_blks *ga,
1402 				     struct sk_buff_head *xmitq)
1403 {
1404 	struct sk_buff *skb, *_skb, *tmp;
1405 	struct tipc_msg *hdr;
1406 	u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1407 	u16 ack = l->rcv_nxt - 1;
1408 	bool passed = false;
1409 	u16 seqno, n = 0;
1410 	int rc = 0;
1411 
1412 	skb_queue_walk_safe(&l->transmq, skb, tmp) {
1413 		seqno = buf_seqno(skb);
1414 
1415 next_gap_ack:
1416 		if (less_eq(seqno, acked)) {
1417 			/* release skb */
1418 			__skb_unlink(skb, &l->transmq);
1419 			kfree_skb(skb);
1420 		} else if (less_eq(seqno, acked + gap)) {
1421 			/* First, check if repeated retrans failures occurs? */
1422 			if (!passed && link_retransmit_failure(l, l, &rc))
1423 				return rc;
1424 			passed = true;
1425 
1426 			/* retransmit skb if unrestricted*/
1427 			if (time_before(jiffies, TIPC_SKB_CB(skb)->nxt_retr))
1428 				continue;
1429 			TIPC_SKB_CB(skb)->nxt_retr = TIPC_UC_RETR_TIME;
1430 			_skb = __pskb_copy(skb, LL_MAX_HEADER + MIN_H_SIZE,
1431 					   GFP_ATOMIC);
1432 			if (!_skb)
1433 				continue;
1434 			hdr = buf_msg(_skb);
1435 			msg_set_ack(hdr, ack);
1436 			msg_set_bcast_ack(hdr, bc_ack);
1437 			_skb->priority = TC_PRIO_CONTROL;
1438 			__skb_queue_tail(xmitq, _skb);
1439 			l->stats.retransmitted++;
1440 
1441 			/* Increase actual retrans counter & mark first time */
1442 			if (!TIPC_SKB_CB(skb)->retr_cnt++)
1443 				TIPC_SKB_CB(skb)->retr_stamp = jiffies;
1444 		} else {
1445 			/* retry with Gap ACK blocks if any */
1446 			if (!ga || n >= ga->gack_cnt)
1447 				break;
1448 			acked = ntohs(ga->gacks[n].ack);
1449 			gap = ntohs(ga->gacks[n].gap);
1450 			n++;
1451 			goto next_gap_ack;
1452 		}
1453 	}
1454 
1455 	return 0;
1456 }
1457 
1458 /* tipc_link_build_state_msg: prepare link state message for transmission
1459  *
1460  * Note that sending of broadcast ack is coordinated among nodes, to reduce
1461  * risk of ack storms towards the sender
1462  */
1463 int tipc_link_build_state_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1464 {
1465 	if (!l)
1466 		return 0;
1467 
1468 	/* Broadcast ACK must be sent via a unicast link => defer to caller */
1469 	if (link_is_bc_rcvlink(l)) {
1470 		if (((l->rcv_nxt ^ tipc_own_addr(l->net)) & 0xf) != 0xf)
1471 			return 0;
1472 		l->rcv_unacked = 0;
1473 
1474 		/* Use snd_nxt to store peer's snd_nxt in broadcast rcv link */
1475 		l->snd_nxt = l->rcv_nxt;
1476 		return TIPC_LINK_SND_STATE;
1477 	}
1478 
1479 	/* Unicast ACK */
1480 	l->rcv_unacked = 0;
1481 	l->stats.sent_acks++;
1482 	tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, 0, xmitq);
1483 	return 0;
1484 }
1485 
1486 /* tipc_link_build_reset_msg: prepare link RESET or ACTIVATE message
1487  */
1488 void tipc_link_build_reset_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1489 {
1490 	int mtyp = RESET_MSG;
1491 	struct sk_buff *skb;
1492 
1493 	if (l->state == LINK_ESTABLISHING)
1494 		mtyp = ACTIVATE_MSG;
1495 
1496 	tipc_link_build_proto_msg(l, mtyp, 0, 0, 0, 0, 0, xmitq);
1497 
1498 	/* Inform peer that this endpoint is going down if applicable */
1499 	skb = skb_peek_tail(xmitq);
1500 	if (skb && (l->state == LINK_RESET))
1501 		msg_set_peer_stopping(buf_msg(skb), 1);
1502 }
1503 
1504 /* tipc_link_build_nack_msg: prepare link nack message for transmission
1505  * Note that sending of broadcast NACK is coordinated among nodes, to
1506  * reduce the risk of NACK storms towards the sender
1507  */
1508 static int tipc_link_build_nack_msg(struct tipc_link *l,
1509 				    struct sk_buff_head *xmitq)
1510 {
1511 	u32 def_cnt = ++l->stats.deferred_recv;
1512 	u32 defq_len = skb_queue_len(&l->deferdq);
1513 	int match1, match2;
1514 
1515 	if (link_is_bc_rcvlink(l)) {
1516 		match1 = def_cnt & 0xf;
1517 		match2 = tipc_own_addr(l->net) & 0xf;
1518 		if (match1 == match2)
1519 			return TIPC_LINK_SND_STATE;
1520 		return 0;
1521 	}
1522 
1523 	if (defq_len >= 3 && !((defq_len - 3) % 16))
1524 		tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, 0, xmitq);
1525 	return 0;
1526 }
1527 
1528 /* tipc_link_rcv - process TIPC packets/messages arriving from off-node
1529  * @l: the link that should handle the message
1530  * @skb: TIPC packet
1531  * @xmitq: queue to place packets to be sent after this call
1532  */
1533 int tipc_link_rcv(struct tipc_link *l, struct sk_buff *skb,
1534 		  struct sk_buff_head *xmitq)
1535 {
1536 	struct sk_buff_head *defq = &l->deferdq;
1537 	struct tipc_msg *hdr = buf_msg(skb);
1538 	u16 seqno, rcv_nxt, win_lim;
1539 	int rc = 0;
1540 
1541 	/* Verify and update link state */
1542 	if (unlikely(msg_user(hdr) == LINK_PROTOCOL))
1543 		return tipc_link_proto_rcv(l, skb, xmitq);
1544 
1545 	/* Don't send probe at next timeout expiration */
1546 	l->silent_intv_cnt = 0;
1547 
1548 	do {
1549 		hdr = buf_msg(skb);
1550 		seqno = msg_seqno(hdr);
1551 		rcv_nxt = l->rcv_nxt;
1552 		win_lim = rcv_nxt + TIPC_MAX_LINK_WIN;
1553 
1554 		if (unlikely(!link_is_up(l))) {
1555 			if (l->state == LINK_ESTABLISHING)
1556 				rc = TIPC_LINK_UP_EVT;
1557 			goto drop;
1558 		}
1559 
1560 		/* Drop if outside receive window */
1561 		if (unlikely(less(seqno, rcv_nxt) || more(seqno, win_lim))) {
1562 			l->stats.duplicates++;
1563 			goto drop;
1564 		}
1565 
1566 		/* Forward queues and wake up waiting users */
1567 		if (likely(tipc_link_release_pkts(l, msg_ack(hdr)))) {
1568 			tipc_link_advance_backlog(l, xmitq);
1569 			if (unlikely(!skb_queue_empty(&l->wakeupq)))
1570 				link_prepare_wakeup(l);
1571 		}
1572 
1573 		/* Defer delivery if sequence gap */
1574 		if (unlikely(seqno != rcv_nxt)) {
1575 			__tipc_skb_queue_sorted(defq, seqno, skb);
1576 			rc |= tipc_link_build_nack_msg(l, xmitq);
1577 			break;
1578 		}
1579 
1580 		/* Deliver packet */
1581 		l->rcv_nxt++;
1582 		l->stats.recv_pkts++;
1583 
1584 		if (unlikely(msg_user(hdr) == TUNNEL_PROTOCOL))
1585 			rc |= tipc_link_tnl_rcv(l, skb, l->inputq);
1586 		else if (!tipc_data_input(l, skb, l->inputq))
1587 			rc |= tipc_link_input(l, skb, l->inputq, &l->reasm_buf);
1588 		if (unlikely(++l->rcv_unacked >= TIPC_MIN_LINK_WIN))
1589 			rc |= tipc_link_build_state_msg(l, xmitq);
1590 		if (unlikely(rc & ~TIPC_LINK_SND_STATE))
1591 			break;
1592 	} while ((skb = __tipc_skb_dequeue(defq, l->rcv_nxt)));
1593 
1594 	return rc;
1595 drop:
1596 	kfree_skb(skb);
1597 	return rc;
1598 }
1599 
1600 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
1601 				      bool probe_reply, u16 rcvgap,
1602 				      int tolerance, int priority,
1603 				      struct sk_buff_head *xmitq)
1604 {
1605 	struct tipc_link *bcl = l->bc_rcvlink;
1606 	struct sk_buff *skb;
1607 	struct tipc_msg *hdr;
1608 	struct sk_buff_head *dfq = &l->deferdq;
1609 	bool node_up = link_is_up(bcl);
1610 	struct tipc_mon_state *mstate = &l->mon_state;
1611 	int dlen = 0;
1612 	void *data;
1613 	u16 glen = 0;
1614 
1615 	/* Don't send protocol message during reset or link failover */
1616 	if (tipc_link_is_blocked(l))
1617 		return;
1618 
1619 	if (!tipc_link_is_up(l) && (mtyp == STATE_MSG))
1620 		return;
1621 
1622 	if (!skb_queue_empty(dfq))
1623 		rcvgap = buf_seqno(skb_peek(dfq)) - l->rcv_nxt;
1624 
1625 	skb = tipc_msg_create(LINK_PROTOCOL, mtyp, INT_H_SIZE,
1626 			      tipc_max_domain_size + MAX_GAP_ACK_BLKS_SZ,
1627 			      l->addr, tipc_own_addr(l->net), 0, 0, 0);
1628 	if (!skb)
1629 		return;
1630 
1631 	hdr = buf_msg(skb);
1632 	data = msg_data(hdr);
1633 	msg_set_session(hdr, l->session);
1634 	msg_set_bearer_id(hdr, l->bearer_id);
1635 	msg_set_net_plane(hdr, l->net_plane);
1636 	msg_set_next_sent(hdr, l->snd_nxt);
1637 	msg_set_ack(hdr, l->rcv_nxt - 1);
1638 	msg_set_bcast_ack(hdr, bcl->rcv_nxt - 1);
1639 	msg_set_bc_ack_invalid(hdr, !node_up);
1640 	msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1641 	msg_set_link_tolerance(hdr, tolerance);
1642 	msg_set_linkprio(hdr, priority);
1643 	msg_set_redundant_link(hdr, node_up);
1644 	msg_set_seq_gap(hdr, 0);
1645 	msg_set_seqno(hdr, l->snd_nxt + U16_MAX / 2);
1646 
1647 	if (mtyp == STATE_MSG) {
1648 		if (l->peer_caps & TIPC_LINK_PROTO_SEQNO)
1649 			msg_set_seqno(hdr, l->snd_nxt_state++);
1650 		msg_set_seq_gap(hdr, rcvgap);
1651 		msg_set_bc_gap(hdr, link_bc_rcv_gap(bcl));
1652 		msg_set_probe(hdr, probe);
1653 		msg_set_is_keepalive(hdr, probe || probe_reply);
1654 		if (l->peer_caps & TIPC_GAP_ACK_BLOCK)
1655 			glen = tipc_build_gap_ack_blks(l, data);
1656 		tipc_mon_prep(l->net, data + glen, &dlen, mstate, l->bearer_id);
1657 		msg_set_size(hdr, INT_H_SIZE + glen + dlen);
1658 		skb_trim(skb, INT_H_SIZE + glen + dlen);
1659 		l->stats.sent_states++;
1660 		l->rcv_unacked = 0;
1661 	} else {
1662 		/* RESET_MSG or ACTIVATE_MSG */
1663 		if (mtyp == ACTIVATE_MSG) {
1664 			msg_set_dest_session_valid(hdr, 1);
1665 			msg_set_dest_session(hdr, l->peer_session);
1666 		}
1667 		msg_set_max_pkt(hdr, l->advertised_mtu);
1668 		strcpy(data, l->if_name);
1669 		msg_set_size(hdr, INT_H_SIZE + TIPC_MAX_IF_NAME);
1670 		skb_trim(skb, INT_H_SIZE + TIPC_MAX_IF_NAME);
1671 	}
1672 	if (probe)
1673 		l->stats.sent_probes++;
1674 	if (rcvgap)
1675 		l->stats.sent_nacks++;
1676 	skb->priority = TC_PRIO_CONTROL;
1677 	__skb_queue_tail(xmitq, skb);
1678 	trace_tipc_proto_build(skb, false, l->name);
1679 }
1680 
1681 void tipc_link_create_dummy_tnl_msg(struct tipc_link *l,
1682 				    struct sk_buff_head *xmitq)
1683 {
1684 	u32 onode = tipc_own_addr(l->net);
1685 	struct tipc_msg *hdr, *ihdr;
1686 	struct sk_buff_head tnlq;
1687 	struct sk_buff *skb;
1688 	u32 dnode = l->addr;
1689 
1690 	__skb_queue_head_init(&tnlq);
1691 	skb = tipc_msg_create(TUNNEL_PROTOCOL, FAILOVER_MSG,
1692 			      INT_H_SIZE, BASIC_H_SIZE,
1693 			      dnode, onode, 0, 0, 0);
1694 	if (!skb) {
1695 		pr_warn("%sunable to create tunnel packet\n", link_co_err);
1696 		return;
1697 	}
1698 
1699 	hdr = buf_msg(skb);
1700 	msg_set_msgcnt(hdr, 1);
1701 	msg_set_bearer_id(hdr, l->peer_bearer_id);
1702 
1703 	ihdr = (struct tipc_msg *)msg_data(hdr);
1704 	tipc_msg_init(onode, ihdr, TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1705 		      BASIC_H_SIZE, dnode);
1706 	msg_set_errcode(ihdr, TIPC_ERR_NO_PORT);
1707 	__skb_queue_tail(&tnlq, skb);
1708 	tipc_link_xmit(l, &tnlq, xmitq);
1709 }
1710 
1711 /* tipc_link_tnl_prepare(): prepare and return a list of tunnel packets
1712  * with contents of the link's transmit and backlog queues.
1713  */
1714 void tipc_link_tnl_prepare(struct tipc_link *l, struct tipc_link *tnl,
1715 			   int mtyp, struct sk_buff_head *xmitq)
1716 {
1717 	struct sk_buff_head *fdefq = &tnl->failover_deferdq;
1718 	struct sk_buff *skb, *tnlskb;
1719 	struct tipc_msg *hdr, tnlhdr;
1720 	struct sk_buff_head *queue = &l->transmq;
1721 	struct sk_buff_head tmpxq, tnlq, frags;
1722 	u16 pktlen, pktcnt, seqno = l->snd_nxt;
1723 	bool pktcnt_need_update = false;
1724 	u16 syncpt;
1725 	int rc;
1726 
1727 	if (!tnl)
1728 		return;
1729 
1730 	__skb_queue_head_init(&tnlq);
1731 	__skb_queue_head_init(&tmpxq);
1732 	__skb_queue_head_init(&frags);
1733 
1734 	/* At least one packet required for safe algorithm => add dummy */
1735 	skb = tipc_msg_create(TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1736 			      BASIC_H_SIZE, 0, l->addr, tipc_own_addr(l->net),
1737 			      0, 0, TIPC_ERR_NO_PORT);
1738 	if (!skb) {
1739 		pr_warn("%sunable to create tunnel packet\n", link_co_err);
1740 		return;
1741 	}
1742 	__skb_queue_tail(&tnlq, skb);
1743 	tipc_link_xmit(l, &tnlq, &tmpxq);
1744 	__skb_queue_purge(&tmpxq);
1745 
1746 	/* Link Synching:
1747 	 * From now on, send only one single ("dummy") SYNCH message
1748 	 * to peer. The SYNCH message does not contain any data, just
1749 	 * a header conveying the synch point to the peer.
1750 	 */
1751 	if (mtyp == SYNCH_MSG && (tnl->peer_caps & TIPC_TUNNEL_ENHANCED)) {
1752 		tnlskb = tipc_msg_create(TUNNEL_PROTOCOL, SYNCH_MSG,
1753 					 INT_H_SIZE, 0, l->addr,
1754 					 tipc_own_addr(l->net),
1755 					 0, 0, 0);
1756 		if (!tnlskb) {
1757 			pr_warn("%sunable to create dummy SYNCH_MSG\n",
1758 				link_co_err);
1759 			return;
1760 		}
1761 
1762 		hdr = buf_msg(tnlskb);
1763 		syncpt = l->snd_nxt + skb_queue_len(&l->backlogq) - 1;
1764 		msg_set_syncpt(hdr, syncpt);
1765 		msg_set_bearer_id(hdr, l->peer_bearer_id);
1766 		__skb_queue_tail(&tnlq, tnlskb);
1767 		tipc_link_xmit(tnl, &tnlq, xmitq);
1768 		return;
1769 	}
1770 
1771 	/* Initialize reusable tunnel packet header */
1772 	tipc_msg_init(tipc_own_addr(l->net), &tnlhdr, TUNNEL_PROTOCOL,
1773 		      mtyp, INT_H_SIZE, l->addr);
1774 	if (mtyp == SYNCH_MSG)
1775 		pktcnt = l->snd_nxt - buf_seqno(skb_peek(&l->transmq));
1776 	else
1777 		pktcnt = skb_queue_len(&l->transmq);
1778 	pktcnt += skb_queue_len(&l->backlogq);
1779 	msg_set_msgcnt(&tnlhdr, pktcnt);
1780 	msg_set_bearer_id(&tnlhdr, l->peer_bearer_id);
1781 tnl:
1782 	/* Wrap each packet into a tunnel packet */
1783 	skb_queue_walk(queue, skb) {
1784 		hdr = buf_msg(skb);
1785 		if (queue == &l->backlogq)
1786 			msg_set_seqno(hdr, seqno++);
1787 		pktlen = msg_size(hdr);
1788 
1789 		/* Tunnel link MTU is not large enough? This could be
1790 		 * due to:
1791 		 * 1) Link MTU has just changed or set differently;
1792 		 * 2) Or FAILOVER on the top of a SYNCH message
1793 		 *
1794 		 * The 2nd case should not happen if peer supports
1795 		 * TIPC_TUNNEL_ENHANCED
1796 		 */
1797 		if (pktlen > tnl->mtu - INT_H_SIZE) {
1798 			if (mtyp == FAILOVER_MSG &&
1799 			    (tnl->peer_caps & TIPC_TUNNEL_ENHANCED)) {
1800 				rc = tipc_msg_fragment(skb, &tnlhdr, tnl->mtu,
1801 						       &frags);
1802 				if (rc) {
1803 					pr_warn("%sunable to frag msg: rc %d\n",
1804 						link_co_err, rc);
1805 					return;
1806 				}
1807 				pktcnt += skb_queue_len(&frags) - 1;
1808 				pktcnt_need_update = true;
1809 				skb_queue_splice_tail_init(&frags, &tnlq);
1810 				continue;
1811 			}
1812 			/* Unluckily, peer doesn't have TIPC_TUNNEL_ENHANCED
1813 			 * => Just warn it and return!
1814 			 */
1815 			pr_warn_ratelimited("%stoo large msg <%d, %d>: %d!\n",
1816 					    link_co_err, msg_user(hdr),
1817 					    msg_type(hdr), msg_size(hdr));
1818 			return;
1819 		}
1820 
1821 		msg_set_size(&tnlhdr, pktlen + INT_H_SIZE);
1822 		tnlskb = tipc_buf_acquire(pktlen + INT_H_SIZE, GFP_ATOMIC);
1823 		if (!tnlskb) {
1824 			pr_warn("%sunable to send packet\n", link_co_err);
1825 			return;
1826 		}
1827 		skb_copy_to_linear_data(tnlskb, &tnlhdr, INT_H_SIZE);
1828 		skb_copy_to_linear_data_offset(tnlskb, INT_H_SIZE, hdr, pktlen);
1829 		__skb_queue_tail(&tnlq, tnlskb);
1830 	}
1831 	if (queue != &l->backlogq) {
1832 		queue = &l->backlogq;
1833 		goto tnl;
1834 	}
1835 
1836 	if (pktcnt_need_update)
1837 		skb_queue_walk(&tnlq, skb) {
1838 			hdr = buf_msg(skb);
1839 			msg_set_msgcnt(hdr, pktcnt);
1840 		}
1841 
1842 	tipc_link_xmit(tnl, &tnlq, xmitq);
1843 
1844 	if (mtyp == FAILOVER_MSG) {
1845 		tnl->drop_point = l->rcv_nxt;
1846 		tnl->failover_reasm_skb = l->reasm_buf;
1847 		l->reasm_buf = NULL;
1848 
1849 		/* Failover the link's deferdq */
1850 		if (unlikely(!skb_queue_empty(fdefq))) {
1851 			pr_warn("Link failover deferdq not empty: %d!\n",
1852 				skb_queue_len(fdefq));
1853 			__skb_queue_purge(fdefq);
1854 		}
1855 		skb_queue_splice_init(&l->deferdq, fdefq);
1856 	}
1857 }
1858 
1859 /**
1860  * tipc_link_failover_prepare() - prepare tnl for link failover
1861  *
1862  * This is a special version of the precursor - tipc_link_tnl_prepare(),
1863  * see the tipc_node_link_failover() for details
1864  *
1865  * @l: failover link
1866  * @tnl: tunnel link
1867  * @xmitq: queue for messages to be xmited
1868  */
1869 void tipc_link_failover_prepare(struct tipc_link *l, struct tipc_link *tnl,
1870 				struct sk_buff_head *xmitq)
1871 {
1872 	struct sk_buff_head *fdefq = &tnl->failover_deferdq;
1873 
1874 	tipc_link_create_dummy_tnl_msg(tnl, xmitq);
1875 
1876 	/* This failover link enpoint was never established before,
1877 	 * so it has not received anything from peer.
1878 	 * Otherwise, it must be a normal failover situation or the
1879 	 * node has entered SELF_DOWN_PEER_LEAVING and both peer nodes
1880 	 * would have to start over from scratch instead.
1881 	 */
1882 	tnl->drop_point = 1;
1883 	tnl->failover_reasm_skb = NULL;
1884 
1885 	/* Initiate the link's failover deferdq */
1886 	if (unlikely(!skb_queue_empty(fdefq))) {
1887 		pr_warn("Link failover deferdq not empty: %d!\n",
1888 			skb_queue_len(fdefq));
1889 		__skb_queue_purge(fdefq);
1890 	}
1891 }
1892 
1893 /* tipc_link_validate_msg(): validate message against current link state
1894  * Returns true if message should be accepted, otherwise false
1895  */
1896 bool tipc_link_validate_msg(struct tipc_link *l, struct tipc_msg *hdr)
1897 {
1898 	u16 curr_session = l->peer_session;
1899 	u16 session = msg_session(hdr);
1900 	int mtyp = msg_type(hdr);
1901 
1902 	if (msg_user(hdr) != LINK_PROTOCOL)
1903 		return true;
1904 
1905 	switch (mtyp) {
1906 	case RESET_MSG:
1907 		if (!l->in_session)
1908 			return true;
1909 		/* Accept only RESET with new session number */
1910 		return more(session, curr_session);
1911 	case ACTIVATE_MSG:
1912 		if (!l->in_session)
1913 			return true;
1914 		/* Accept only ACTIVATE with new or current session number */
1915 		return !less(session, curr_session);
1916 	case STATE_MSG:
1917 		/* Accept only STATE with current session number */
1918 		if (!l->in_session)
1919 			return false;
1920 		if (session != curr_session)
1921 			return false;
1922 		/* Extra sanity check */
1923 		if (!link_is_up(l) && msg_ack(hdr))
1924 			return false;
1925 		if (!(l->peer_caps & TIPC_LINK_PROTO_SEQNO))
1926 			return true;
1927 		/* Accept only STATE with new sequence number */
1928 		return !less(msg_seqno(hdr), l->rcv_nxt_state);
1929 	default:
1930 		return false;
1931 	}
1932 }
1933 
1934 /* tipc_link_proto_rcv(): receive link level protocol message :
1935  * Note that network plane id propagates through the network, and may
1936  * change at any time. The node with lowest numerical id determines
1937  * network plane
1938  */
1939 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
1940 			       struct sk_buff_head *xmitq)
1941 {
1942 	struct tipc_msg *hdr = buf_msg(skb);
1943 	struct tipc_gap_ack_blks *ga = NULL;
1944 	u16 rcvgap = 0;
1945 	u16 ack = msg_ack(hdr);
1946 	u16 gap = msg_seq_gap(hdr);
1947 	u16 peers_snd_nxt =  msg_next_sent(hdr);
1948 	u16 peers_tol = msg_link_tolerance(hdr);
1949 	u16 peers_prio = msg_linkprio(hdr);
1950 	u16 rcv_nxt = l->rcv_nxt;
1951 	u16 dlen = msg_data_sz(hdr);
1952 	int mtyp = msg_type(hdr);
1953 	bool reply = msg_probe(hdr);
1954 	u16 glen = 0;
1955 	void *data;
1956 	char *if_name;
1957 	int rc = 0;
1958 
1959 	trace_tipc_proto_rcv(skb, false, l->name);
1960 	if (tipc_link_is_blocked(l) || !xmitq)
1961 		goto exit;
1962 
1963 	if (tipc_own_addr(l->net) > msg_prevnode(hdr))
1964 		l->net_plane = msg_net_plane(hdr);
1965 
1966 	skb_linearize(skb);
1967 	hdr = buf_msg(skb);
1968 	data = msg_data(hdr);
1969 
1970 	if (!tipc_link_validate_msg(l, hdr)) {
1971 		trace_tipc_skb_dump(skb, false, "PROTO invalid (1)!");
1972 		trace_tipc_link_dump(l, TIPC_DUMP_NONE, "PROTO invalid (1)!");
1973 		goto exit;
1974 	}
1975 
1976 	switch (mtyp) {
1977 	case RESET_MSG:
1978 	case ACTIVATE_MSG:
1979 		/* Complete own link name with peer's interface name */
1980 		if_name =  strrchr(l->name, ':') + 1;
1981 		if (sizeof(l->name) - (if_name - l->name) <= TIPC_MAX_IF_NAME)
1982 			break;
1983 		if (msg_data_sz(hdr) < TIPC_MAX_IF_NAME)
1984 			break;
1985 		strncpy(if_name, data, TIPC_MAX_IF_NAME);
1986 
1987 		/* Update own tolerance if peer indicates a non-zero value */
1988 		if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
1989 			l->tolerance = peers_tol;
1990 			l->bc_rcvlink->tolerance = peers_tol;
1991 		}
1992 		/* Update own priority if peer's priority is higher */
1993 		if (in_range(peers_prio, l->priority + 1, TIPC_MAX_LINK_PRI))
1994 			l->priority = peers_prio;
1995 
1996 		/* If peer is going down we want full re-establish cycle */
1997 		if (msg_peer_stopping(hdr)) {
1998 			rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1999 			break;
2000 		}
2001 
2002 		/* If this endpoint was re-created while peer was ESTABLISHING
2003 		 * it doesn't know current session number. Force re-synch.
2004 		 */
2005 		if (mtyp == ACTIVATE_MSG && msg_dest_session_valid(hdr) &&
2006 		    l->session != msg_dest_session(hdr)) {
2007 			if (less(l->session, msg_dest_session(hdr)))
2008 				l->session = msg_dest_session(hdr) + 1;
2009 			break;
2010 		}
2011 
2012 		/* ACTIVATE_MSG serves as PEER_RESET if link is already down */
2013 		if (mtyp == RESET_MSG || !link_is_up(l))
2014 			rc = tipc_link_fsm_evt(l, LINK_PEER_RESET_EVT);
2015 
2016 		/* ACTIVATE_MSG takes up link if it was already locally reset */
2017 		if (mtyp == ACTIVATE_MSG && l->state == LINK_ESTABLISHING)
2018 			rc = TIPC_LINK_UP_EVT;
2019 
2020 		l->peer_session = msg_session(hdr);
2021 		l->in_session = true;
2022 		l->peer_bearer_id = msg_bearer_id(hdr);
2023 		if (l->mtu > msg_max_pkt(hdr))
2024 			l->mtu = msg_max_pkt(hdr);
2025 		break;
2026 
2027 	case STATE_MSG:
2028 		l->rcv_nxt_state = msg_seqno(hdr) + 1;
2029 
2030 		/* Update own tolerance if peer indicates a non-zero value */
2031 		if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
2032 			l->tolerance = peers_tol;
2033 			l->bc_rcvlink->tolerance = peers_tol;
2034 		}
2035 		/* Update own prio if peer indicates a different value */
2036 		if ((peers_prio != l->priority) &&
2037 		    in_range(peers_prio, 1, TIPC_MAX_LINK_PRI)) {
2038 			l->priority = peers_prio;
2039 			rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
2040 		}
2041 
2042 		l->silent_intv_cnt = 0;
2043 		l->stats.recv_states++;
2044 		if (msg_probe(hdr))
2045 			l->stats.recv_probes++;
2046 
2047 		if (!link_is_up(l)) {
2048 			if (l->state == LINK_ESTABLISHING)
2049 				rc = TIPC_LINK_UP_EVT;
2050 			break;
2051 		}
2052 
2053 		/* Receive Gap ACK blocks from peer if any */
2054 		if (l->peer_caps & TIPC_GAP_ACK_BLOCK) {
2055 			ga = (struct tipc_gap_ack_blks *)data;
2056 			glen = ntohs(ga->len);
2057 			/* sanity check: if failed, ignore Gap ACK blocks */
2058 			if (glen != tipc_gap_ack_blks_sz(ga->gack_cnt))
2059 				ga = NULL;
2060 		}
2061 
2062 		tipc_mon_rcv(l->net, data + glen, dlen - glen, l->addr,
2063 			     &l->mon_state, l->bearer_id);
2064 
2065 		/* Send NACK if peer has sent pkts we haven't received yet */
2066 		if (more(peers_snd_nxt, rcv_nxt) && !tipc_link_is_synching(l))
2067 			rcvgap = peers_snd_nxt - l->rcv_nxt;
2068 		if (rcvgap || reply)
2069 			tipc_link_build_proto_msg(l, STATE_MSG, 0, reply,
2070 						  rcvgap, 0, 0, xmitq);
2071 
2072 		rc |= tipc_link_advance_transmq(l, ack, gap, ga, xmitq);
2073 
2074 		/* If NACK, retransmit will now start at right position */
2075 		if (gap)
2076 			l->stats.recv_nacks++;
2077 
2078 		tipc_link_advance_backlog(l, xmitq);
2079 		if (unlikely(!skb_queue_empty(&l->wakeupq)))
2080 			link_prepare_wakeup(l);
2081 	}
2082 exit:
2083 	kfree_skb(skb);
2084 	return rc;
2085 }
2086 
2087 /* tipc_link_build_bc_proto_msg() - create broadcast protocol message
2088  */
2089 static bool tipc_link_build_bc_proto_msg(struct tipc_link *l, bool bcast,
2090 					 u16 peers_snd_nxt,
2091 					 struct sk_buff_head *xmitq)
2092 {
2093 	struct sk_buff *skb;
2094 	struct tipc_msg *hdr;
2095 	struct sk_buff *dfrd_skb = skb_peek(&l->deferdq);
2096 	u16 ack = l->rcv_nxt - 1;
2097 	u16 gap_to = peers_snd_nxt - 1;
2098 
2099 	skb = tipc_msg_create(BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE,
2100 			      0, l->addr, tipc_own_addr(l->net), 0, 0, 0);
2101 	if (!skb)
2102 		return false;
2103 	hdr = buf_msg(skb);
2104 	msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
2105 	msg_set_bcast_ack(hdr, ack);
2106 	msg_set_bcgap_after(hdr, ack);
2107 	if (dfrd_skb)
2108 		gap_to = buf_seqno(dfrd_skb) - 1;
2109 	msg_set_bcgap_to(hdr, gap_to);
2110 	msg_set_non_seq(hdr, bcast);
2111 	__skb_queue_tail(xmitq, skb);
2112 	return true;
2113 }
2114 
2115 /* tipc_link_build_bc_init_msg() - synchronize broadcast link endpoints.
2116  *
2117  * Give a newly added peer node the sequence number where it should
2118  * start receiving and acking broadcast packets.
2119  */
2120 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
2121 					struct sk_buff_head *xmitq)
2122 {
2123 	struct sk_buff_head list;
2124 
2125 	__skb_queue_head_init(&list);
2126 	if (!tipc_link_build_bc_proto_msg(l->bc_rcvlink, false, 0, &list))
2127 		return;
2128 	msg_set_bc_ack_invalid(buf_msg(skb_peek(&list)), true);
2129 	tipc_link_xmit(l, &list, xmitq);
2130 }
2131 
2132 /* tipc_link_bc_init_rcv - receive initial broadcast synch data from peer
2133  */
2134 void tipc_link_bc_init_rcv(struct tipc_link *l, struct tipc_msg *hdr)
2135 {
2136 	int mtyp = msg_type(hdr);
2137 	u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
2138 
2139 	if (link_is_up(l))
2140 		return;
2141 
2142 	if (msg_user(hdr) == BCAST_PROTOCOL) {
2143 		l->rcv_nxt = peers_snd_nxt;
2144 		l->state = LINK_ESTABLISHED;
2145 		return;
2146 	}
2147 
2148 	if (l->peer_caps & TIPC_BCAST_SYNCH)
2149 		return;
2150 
2151 	if (msg_peer_node_is_up(hdr))
2152 		return;
2153 
2154 	/* Compatibility: accept older, less safe initial synch data */
2155 	if ((mtyp == RESET_MSG) || (mtyp == ACTIVATE_MSG))
2156 		l->rcv_nxt = peers_snd_nxt;
2157 }
2158 
2159 /* tipc_link_bc_sync_rcv - update rcv link according to peer's send state
2160  */
2161 int tipc_link_bc_sync_rcv(struct tipc_link *l, struct tipc_msg *hdr,
2162 			  struct sk_buff_head *xmitq)
2163 {
2164 	struct tipc_link *snd_l = l->bc_sndlink;
2165 	u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
2166 	u16 from = msg_bcast_ack(hdr) + 1;
2167 	u16 to = from + msg_bc_gap(hdr) - 1;
2168 	int rc = 0;
2169 
2170 	if (!link_is_up(l))
2171 		return rc;
2172 
2173 	if (!msg_peer_node_is_up(hdr))
2174 		return rc;
2175 
2176 	/* Open when peer ackowledges our bcast init msg (pkt #1) */
2177 	if (msg_ack(hdr))
2178 		l->bc_peer_is_up = true;
2179 
2180 	if (!l->bc_peer_is_up)
2181 		return rc;
2182 
2183 	l->stats.recv_nacks++;
2184 
2185 	/* Ignore if peers_snd_nxt goes beyond receive window */
2186 	if (more(peers_snd_nxt, l->rcv_nxt + l->window))
2187 		return rc;
2188 
2189 	rc = tipc_link_bc_retrans(snd_l, l, from, to, xmitq);
2190 
2191 	l->snd_nxt = peers_snd_nxt;
2192 	if (link_bc_rcv_gap(l))
2193 		rc |= TIPC_LINK_SND_STATE;
2194 
2195 	/* Return now if sender supports nack via STATE messages */
2196 	if (l->peer_caps & TIPC_BCAST_STATE_NACK)
2197 		return rc;
2198 
2199 	/* Otherwise, be backwards compatible */
2200 
2201 	if (!more(peers_snd_nxt, l->rcv_nxt)) {
2202 		l->nack_state = BC_NACK_SND_CONDITIONAL;
2203 		return 0;
2204 	}
2205 
2206 	/* Don't NACK if one was recently sent or peeked */
2207 	if (l->nack_state == BC_NACK_SND_SUPPRESS) {
2208 		l->nack_state = BC_NACK_SND_UNCONDITIONAL;
2209 		return 0;
2210 	}
2211 
2212 	/* Conditionally delay NACK sending until next synch rcv */
2213 	if (l->nack_state == BC_NACK_SND_CONDITIONAL) {
2214 		l->nack_state = BC_NACK_SND_UNCONDITIONAL;
2215 		if ((peers_snd_nxt - l->rcv_nxt) < TIPC_MIN_LINK_WIN)
2216 			return 0;
2217 	}
2218 
2219 	/* Send NACK now but suppress next one */
2220 	tipc_link_build_bc_proto_msg(l, true, peers_snd_nxt, xmitq);
2221 	l->nack_state = BC_NACK_SND_SUPPRESS;
2222 	return 0;
2223 }
2224 
2225 void tipc_link_bc_ack_rcv(struct tipc_link *l, u16 acked,
2226 			  struct sk_buff_head *xmitq)
2227 {
2228 	struct sk_buff *skb, *tmp;
2229 	struct tipc_link *snd_l = l->bc_sndlink;
2230 
2231 	if (!link_is_up(l) || !l->bc_peer_is_up)
2232 		return;
2233 
2234 	if (!more(acked, l->acked))
2235 		return;
2236 
2237 	trace_tipc_link_bc_ack(l, l->acked, acked, &snd_l->transmq);
2238 	/* Skip over packets peer has already acked */
2239 	skb_queue_walk(&snd_l->transmq, skb) {
2240 		if (more(buf_seqno(skb), l->acked))
2241 			break;
2242 	}
2243 
2244 	/* Update/release the packets peer is acking now */
2245 	skb_queue_walk_from_safe(&snd_l->transmq, skb, tmp) {
2246 		if (more(buf_seqno(skb), acked))
2247 			break;
2248 		if (!--TIPC_SKB_CB(skb)->ackers) {
2249 			__skb_unlink(skb, &snd_l->transmq);
2250 			kfree_skb(skb);
2251 		}
2252 	}
2253 	l->acked = acked;
2254 	tipc_link_advance_backlog(snd_l, xmitq);
2255 	if (unlikely(!skb_queue_empty(&snd_l->wakeupq)))
2256 		link_prepare_wakeup(snd_l);
2257 }
2258 
2259 /* tipc_link_bc_nack_rcv(): receive broadcast nack message
2260  * This function is here for backwards compatibility, since
2261  * no BCAST_PROTOCOL/STATE messages occur from TIPC v2.5.
2262  */
2263 int tipc_link_bc_nack_rcv(struct tipc_link *l, struct sk_buff *skb,
2264 			  struct sk_buff_head *xmitq)
2265 {
2266 	struct tipc_msg *hdr = buf_msg(skb);
2267 	u32 dnode = msg_destnode(hdr);
2268 	int mtyp = msg_type(hdr);
2269 	u16 acked = msg_bcast_ack(hdr);
2270 	u16 from = acked + 1;
2271 	u16 to = msg_bcgap_to(hdr);
2272 	u16 peers_snd_nxt = to + 1;
2273 	int rc = 0;
2274 
2275 	kfree_skb(skb);
2276 
2277 	if (!tipc_link_is_up(l) || !l->bc_peer_is_up)
2278 		return 0;
2279 
2280 	if (mtyp != STATE_MSG)
2281 		return 0;
2282 
2283 	if (dnode == tipc_own_addr(l->net)) {
2284 		tipc_link_bc_ack_rcv(l, acked, xmitq);
2285 		rc = tipc_link_bc_retrans(l->bc_sndlink, l, from, to, xmitq);
2286 		l->stats.recv_nacks++;
2287 		return rc;
2288 	}
2289 
2290 	/* Msg for other node => suppress own NACK at next sync if applicable */
2291 	if (more(peers_snd_nxt, l->rcv_nxt) && !less(l->rcv_nxt, from))
2292 		l->nack_state = BC_NACK_SND_SUPPRESS;
2293 
2294 	return 0;
2295 }
2296 
2297 void tipc_link_set_queue_limits(struct tipc_link *l, u32 win)
2298 {
2299 	int max_bulk = TIPC_MAX_PUBL / (l->mtu / ITEM_SIZE);
2300 
2301 	l->window = win;
2302 	l->backlog[TIPC_LOW_IMPORTANCE].limit      = max_t(u16, 50, win);
2303 	l->backlog[TIPC_MEDIUM_IMPORTANCE].limit   = max_t(u16, 100, win * 2);
2304 	l->backlog[TIPC_HIGH_IMPORTANCE].limit     = max_t(u16, 150, win * 3);
2305 	l->backlog[TIPC_CRITICAL_IMPORTANCE].limit = max_t(u16, 200, win * 4);
2306 	l->backlog[TIPC_SYSTEM_IMPORTANCE].limit   = max_bulk;
2307 }
2308 
2309 /**
2310  * link_reset_stats - reset link statistics
2311  * @l: pointer to link
2312  */
2313 void tipc_link_reset_stats(struct tipc_link *l)
2314 {
2315 	memset(&l->stats, 0, sizeof(l->stats));
2316 }
2317 
2318 static void link_print(struct tipc_link *l, const char *str)
2319 {
2320 	struct sk_buff *hskb = skb_peek(&l->transmq);
2321 	u16 head = hskb ? msg_seqno(buf_msg(hskb)) : l->snd_nxt - 1;
2322 	u16 tail = l->snd_nxt - 1;
2323 
2324 	pr_info("%s Link <%s> state %x\n", str, l->name, l->state);
2325 	pr_info("XMTQ: %u [%u-%u], BKLGQ: %u, SNDNX: %u, RCVNX: %u\n",
2326 		skb_queue_len(&l->transmq), head, tail,
2327 		skb_queue_len(&l->backlogq), l->snd_nxt, l->rcv_nxt);
2328 }
2329 
2330 /* Parse and validate nested (link) properties valid for media, bearer and link
2331  */
2332 int tipc_nl_parse_link_prop(struct nlattr *prop, struct nlattr *props[])
2333 {
2334 	int err;
2335 
2336 	err = nla_parse_nested_deprecated(props, TIPC_NLA_PROP_MAX, prop,
2337 					  tipc_nl_prop_policy, NULL);
2338 	if (err)
2339 		return err;
2340 
2341 	if (props[TIPC_NLA_PROP_PRIO]) {
2342 		u32 prio;
2343 
2344 		prio = nla_get_u32(props[TIPC_NLA_PROP_PRIO]);
2345 		if (prio > TIPC_MAX_LINK_PRI)
2346 			return -EINVAL;
2347 	}
2348 
2349 	if (props[TIPC_NLA_PROP_TOL]) {
2350 		u32 tol;
2351 
2352 		tol = nla_get_u32(props[TIPC_NLA_PROP_TOL]);
2353 		if ((tol < TIPC_MIN_LINK_TOL) || (tol > TIPC_MAX_LINK_TOL))
2354 			return -EINVAL;
2355 	}
2356 
2357 	if (props[TIPC_NLA_PROP_WIN]) {
2358 		u32 win;
2359 
2360 		win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
2361 		if ((win < TIPC_MIN_LINK_WIN) || (win > TIPC_MAX_LINK_WIN))
2362 			return -EINVAL;
2363 	}
2364 
2365 	return 0;
2366 }
2367 
2368 static int __tipc_nl_add_stats(struct sk_buff *skb, struct tipc_stats *s)
2369 {
2370 	int i;
2371 	struct nlattr *stats;
2372 
2373 	struct nla_map {
2374 		u32 key;
2375 		u32 val;
2376 	};
2377 
2378 	struct nla_map map[] = {
2379 		{TIPC_NLA_STATS_RX_INFO, 0},
2380 		{TIPC_NLA_STATS_RX_FRAGMENTS, s->recv_fragments},
2381 		{TIPC_NLA_STATS_RX_FRAGMENTED, s->recv_fragmented},
2382 		{TIPC_NLA_STATS_RX_BUNDLES, s->recv_bundles},
2383 		{TIPC_NLA_STATS_RX_BUNDLED, s->recv_bundled},
2384 		{TIPC_NLA_STATS_TX_INFO, 0},
2385 		{TIPC_NLA_STATS_TX_FRAGMENTS, s->sent_fragments},
2386 		{TIPC_NLA_STATS_TX_FRAGMENTED, s->sent_fragmented},
2387 		{TIPC_NLA_STATS_TX_BUNDLES, s->sent_bundles},
2388 		{TIPC_NLA_STATS_TX_BUNDLED, s->sent_bundled},
2389 		{TIPC_NLA_STATS_MSG_PROF_TOT, (s->msg_length_counts) ?
2390 			s->msg_length_counts : 1},
2391 		{TIPC_NLA_STATS_MSG_LEN_CNT, s->msg_length_counts},
2392 		{TIPC_NLA_STATS_MSG_LEN_TOT, s->msg_lengths_total},
2393 		{TIPC_NLA_STATS_MSG_LEN_P0, s->msg_length_profile[0]},
2394 		{TIPC_NLA_STATS_MSG_LEN_P1, s->msg_length_profile[1]},
2395 		{TIPC_NLA_STATS_MSG_LEN_P2, s->msg_length_profile[2]},
2396 		{TIPC_NLA_STATS_MSG_LEN_P3, s->msg_length_profile[3]},
2397 		{TIPC_NLA_STATS_MSG_LEN_P4, s->msg_length_profile[4]},
2398 		{TIPC_NLA_STATS_MSG_LEN_P5, s->msg_length_profile[5]},
2399 		{TIPC_NLA_STATS_MSG_LEN_P6, s->msg_length_profile[6]},
2400 		{TIPC_NLA_STATS_RX_STATES, s->recv_states},
2401 		{TIPC_NLA_STATS_RX_PROBES, s->recv_probes},
2402 		{TIPC_NLA_STATS_RX_NACKS, s->recv_nacks},
2403 		{TIPC_NLA_STATS_RX_DEFERRED, s->deferred_recv},
2404 		{TIPC_NLA_STATS_TX_STATES, s->sent_states},
2405 		{TIPC_NLA_STATS_TX_PROBES, s->sent_probes},
2406 		{TIPC_NLA_STATS_TX_NACKS, s->sent_nacks},
2407 		{TIPC_NLA_STATS_TX_ACKS, s->sent_acks},
2408 		{TIPC_NLA_STATS_RETRANSMITTED, s->retransmitted},
2409 		{TIPC_NLA_STATS_DUPLICATES, s->duplicates},
2410 		{TIPC_NLA_STATS_LINK_CONGS, s->link_congs},
2411 		{TIPC_NLA_STATS_MAX_QUEUE, s->max_queue_sz},
2412 		{TIPC_NLA_STATS_AVG_QUEUE, s->queue_sz_counts ?
2413 			(s->accu_queue_sz / s->queue_sz_counts) : 0}
2414 	};
2415 
2416 	stats = nla_nest_start_noflag(skb, TIPC_NLA_LINK_STATS);
2417 	if (!stats)
2418 		return -EMSGSIZE;
2419 
2420 	for (i = 0; i <  ARRAY_SIZE(map); i++)
2421 		if (nla_put_u32(skb, map[i].key, map[i].val))
2422 			goto msg_full;
2423 
2424 	nla_nest_end(skb, stats);
2425 
2426 	return 0;
2427 msg_full:
2428 	nla_nest_cancel(skb, stats);
2429 
2430 	return -EMSGSIZE;
2431 }
2432 
2433 /* Caller should hold appropriate locks to protect the link */
2434 int __tipc_nl_add_link(struct net *net, struct tipc_nl_msg *msg,
2435 		       struct tipc_link *link, int nlflags)
2436 {
2437 	u32 self = tipc_own_addr(net);
2438 	struct nlattr *attrs;
2439 	struct nlattr *prop;
2440 	void *hdr;
2441 	int err;
2442 
2443 	hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2444 			  nlflags, TIPC_NL_LINK_GET);
2445 	if (!hdr)
2446 		return -EMSGSIZE;
2447 
2448 	attrs = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK);
2449 	if (!attrs)
2450 		goto msg_full;
2451 
2452 	if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, link->name))
2453 		goto attr_msg_full;
2454 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_DEST, tipc_cluster_mask(self)))
2455 		goto attr_msg_full;
2456 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_MTU, link->mtu))
2457 		goto attr_msg_full;
2458 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, link->stats.recv_pkts))
2459 		goto attr_msg_full;
2460 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, link->stats.sent_pkts))
2461 		goto attr_msg_full;
2462 
2463 	if (tipc_link_is_up(link))
2464 		if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2465 			goto attr_msg_full;
2466 	if (link->active)
2467 		if (nla_put_flag(msg->skb, TIPC_NLA_LINK_ACTIVE))
2468 			goto attr_msg_full;
2469 
2470 	prop = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK_PROP);
2471 	if (!prop)
2472 		goto attr_msg_full;
2473 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2474 		goto prop_msg_full;
2475 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_TOL, link->tolerance))
2476 		goto prop_msg_full;
2477 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN,
2478 			link->window))
2479 		goto prop_msg_full;
2480 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2481 		goto prop_msg_full;
2482 	nla_nest_end(msg->skb, prop);
2483 
2484 	err = __tipc_nl_add_stats(msg->skb, &link->stats);
2485 	if (err)
2486 		goto attr_msg_full;
2487 
2488 	nla_nest_end(msg->skb, attrs);
2489 	genlmsg_end(msg->skb, hdr);
2490 
2491 	return 0;
2492 
2493 prop_msg_full:
2494 	nla_nest_cancel(msg->skb, prop);
2495 attr_msg_full:
2496 	nla_nest_cancel(msg->skb, attrs);
2497 msg_full:
2498 	genlmsg_cancel(msg->skb, hdr);
2499 
2500 	return -EMSGSIZE;
2501 }
2502 
2503 static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
2504 				      struct tipc_stats *stats)
2505 {
2506 	int i;
2507 	struct nlattr *nest;
2508 
2509 	struct nla_map {
2510 		__u32 key;
2511 		__u32 val;
2512 	};
2513 
2514 	struct nla_map map[] = {
2515 		{TIPC_NLA_STATS_RX_INFO, stats->recv_pkts},
2516 		{TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
2517 		{TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
2518 		{TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
2519 		{TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
2520 		{TIPC_NLA_STATS_TX_INFO, stats->sent_pkts},
2521 		{TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
2522 		{TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
2523 		{TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
2524 		{TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
2525 		{TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
2526 		{TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
2527 		{TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
2528 		{TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
2529 		{TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
2530 		{TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
2531 		{TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
2532 		{TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
2533 		{TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
2534 			(stats->accu_queue_sz / stats->queue_sz_counts) : 0}
2535 	};
2536 
2537 	nest = nla_nest_start_noflag(skb, TIPC_NLA_LINK_STATS);
2538 	if (!nest)
2539 		return -EMSGSIZE;
2540 
2541 	for (i = 0; i <  ARRAY_SIZE(map); i++)
2542 		if (nla_put_u32(skb, map[i].key, map[i].val))
2543 			goto msg_full;
2544 
2545 	nla_nest_end(skb, nest);
2546 
2547 	return 0;
2548 msg_full:
2549 	nla_nest_cancel(skb, nest);
2550 
2551 	return -EMSGSIZE;
2552 }
2553 
2554 int tipc_nl_add_bc_link(struct net *net, struct tipc_nl_msg *msg)
2555 {
2556 	int err;
2557 	void *hdr;
2558 	struct nlattr *attrs;
2559 	struct nlattr *prop;
2560 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2561 	u32 bc_mode = tipc_bcast_get_broadcast_mode(net);
2562 	u32 bc_ratio = tipc_bcast_get_broadcast_ratio(net);
2563 	struct tipc_link *bcl = tn->bcl;
2564 
2565 	if (!bcl)
2566 		return 0;
2567 
2568 	tipc_bcast_lock(net);
2569 
2570 	hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2571 			  NLM_F_MULTI, TIPC_NL_LINK_GET);
2572 	if (!hdr) {
2573 		tipc_bcast_unlock(net);
2574 		return -EMSGSIZE;
2575 	}
2576 
2577 	attrs = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK);
2578 	if (!attrs)
2579 		goto msg_full;
2580 
2581 	/* The broadcast link is always up */
2582 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2583 		goto attr_msg_full;
2584 
2585 	if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
2586 		goto attr_msg_full;
2587 	if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
2588 		goto attr_msg_full;
2589 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, 0))
2590 		goto attr_msg_full;
2591 	if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, 0))
2592 		goto attr_msg_full;
2593 
2594 	prop = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK_PROP);
2595 	if (!prop)
2596 		goto attr_msg_full;
2597 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->window))
2598 		goto prop_msg_full;
2599 	if (nla_put_u32(msg->skb, TIPC_NLA_PROP_BROADCAST, bc_mode))
2600 		goto prop_msg_full;
2601 	if (bc_mode & BCLINK_MODE_SEL)
2602 		if (nla_put_u32(msg->skb, TIPC_NLA_PROP_BROADCAST_RATIO,
2603 				bc_ratio))
2604 			goto prop_msg_full;
2605 	nla_nest_end(msg->skb, prop);
2606 
2607 	err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
2608 	if (err)
2609 		goto attr_msg_full;
2610 
2611 	tipc_bcast_unlock(net);
2612 	nla_nest_end(msg->skb, attrs);
2613 	genlmsg_end(msg->skb, hdr);
2614 
2615 	return 0;
2616 
2617 prop_msg_full:
2618 	nla_nest_cancel(msg->skb, prop);
2619 attr_msg_full:
2620 	nla_nest_cancel(msg->skb, attrs);
2621 msg_full:
2622 	tipc_bcast_unlock(net);
2623 	genlmsg_cancel(msg->skb, hdr);
2624 
2625 	return -EMSGSIZE;
2626 }
2627 
2628 void tipc_link_set_tolerance(struct tipc_link *l, u32 tol,
2629 			     struct sk_buff_head *xmitq)
2630 {
2631 	l->tolerance = tol;
2632 	if (l->bc_rcvlink)
2633 		l->bc_rcvlink->tolerance = tol;
2634 	if (link_is_up(l))
2635 		tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, tol, 0, xmitq);
2636 }
2637 
2638 void tipc_link_set_prio(struct tipc_link *l, u32 prio,
2639 			struct sk_buff_head *xmitq)
2640 {
2641 	l->priority = prio;
2642 	tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, prio, xmitq);
2643 }
2644 
2645 void tipc_link_set_abort_limit(struct tipc_link *l, u32 limit)
2646 {
2647 	l->abort_limit = limit;
2648 }
2649 
2650 char *tipc_link_name_ext(struct tipc_link *l, char *buf)
2651 {
2652 	if (!l)
2653 		scnprintf(buf, TIPC_MAX_LINK_NAME, "null");
2654 	else if (link_is_bc_sndlink(l))
2655 		scnprintf(buf, TIPC_MAX_LINK_NAME, "broadcast-sender");
2656 	else if (link_is_bc_rcvlink(l))
2657 		scnprintf(buf, TIPC_MAX_LINK_NAME,
2658 			  "broadcast-receiver, peer %x", l->addr);
2659 	else
2660 		memcpy(buf, l->name, TIPC_MAX_LINK_NAME);
2661 
2662 	return buf;
2663 }
2664 
2665 /**
2666  * tipc_link_dump - dump TIPC link data
2667  * @l: tipc link to be dumped
2668  * @dqueues: bitmask to decide if any link queue to be dumped?
2669  *           - TIPC_DUMP_NONE: don't dump link queues
2670  *           - TIPC_DUMP_TRANSMQ: dump link transmq queue
2671  *           - TIPC_DUMP_BACKLOGQ: dump link backlog queue
2672  *           - TIPC_DUMP_DEFERDQ: dump link deferd queue
2673  *           - TIPC_DUMP_INPUTQ: dump link input queue
2674  *           - TIPC_DUMP_WAKEUP: dump link wakeup queue
2675  *           - TIPC_DUMP_ALL: dump all the link queues above
2676  * @buf: returned buffer of dump data in format
2677  */
2678 int tipc_link_dump(struct tipc_link *l, u16 dqueues, char *buf)
2679 {
2680 	int i = 0;
2681 	size_t sz = (dqueues) ? LINK_LMAX : LINK_LMIN;
2682 	struct sk_buff_head *list;
2683 	struct sk_buff *hskb, *tskb;
2684 	u32 len;
2685 
2686 	if (!l) {
2687 		i += scnprintf(buf, sz, "link data: (null)\n");
2688 		return i;
2689 	}
2690 
2691 	i += scnprintf(buf, sz, "link data: %x", l->addr);
2692 	i += scnprintf(buf + i, sz - i, " %x", l->state);
2693 	i += scnprintf(buf + i, sz - i, " %u", l->in_session);
2694 	i += scnprintf(buf + i, sz - i, " %u", l->session);
2695 	i += scnprintf(buf + i, sz - i, " %u", l->peer_session);
2696 	i += scnprintf(buf + i, sz - i, " %u", l->snd_nxt);
2697 	i += scnprintf(buf + i, sz - i, " %u", l->rcv_nxt);
2698 	i += scnprintf(buf + i, sz - i, " %u", l->snd_nxt_state);
2699 	i += scnprintf(buf + i, sz - i, " %u", l->rcv_nxt_state);
2700 	i += scnprintf(buf + i, sz - i, " %x", l->peer_caps);
2701 	i += scnprintf(buf + i, sz - i, " %u", l->silent_intv_cnt);
2702 	i += scnprintf(buf + i, sz - i, " %u", l->rst_cnt);
2703 	i += scnprintf(buf + i, sz - i, " %u", 0);
2704 	i += scnprintf(buf + i, sz - i, " %u", 0);
2705 	i += scnprintf(buf + i, sz - i, " %u", l->acked);
2706 
2707 	list = &l->transmq;
2708 	len = skb_queue_len(list);
2709 	hskb = skb_peek(list);
2710 	tskb = skb_peek_tail(list);
2711 	i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2712 		       (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2713 		       (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2714 
2715 	list = &l->deferdq;
2716 	len = skb_queue_len(list);
2717 	hskb = skb_peek(list);
2718 	tskb = skb_peek_tail(list);
2719 	i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2720 		       (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2721 		       (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2722 
2723 	list = &l->backlogq;
2724 	len = skb_queue_len(list);
2725 	hskb = skb_peek(list);
2726 	tskb = skb_peek_tail(list);
2727 	i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2728 		       (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2729 		       (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2730 
2731 	list = l->inputq;
2732 	len = skb_queue_len(list);
2733 	hskb = skb_peek(list);
2734 	tskb = skb_peek_tail(list);
2735 	i += scnprintf(buf + i, sz - i, " | %u %u %u\n", len,
2736 		       (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2737 		       (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2738 
2739 	if (dqueues & TIPC_DUMP_TRANSMQ) {
2740 		i += scnprintf(buf + i, sz - i, "transmq: ");
2741 		i += tipc_list_dump(&l->transmq, false, buf + i);
2742 	}
2743 	if (dqueues & TIPC_DUMP_BACKLOGQ) {
2744 		i += scnprintf(buf + i, sz - i,
2745 			       "backlogq: <%u %u %u %u %u>, ",
2746 			       l->backlog[TIPC_LOW_IMPORTANCE].len,
2747 			       l->backlog[TIPC_MEDIUM_IMPORTANCE].len,
2748 			       l->backlog[TIPC_HIGH_IMPORTANCE].len,
2749 			       l->backlog[TIPC_CRITICAL_IMPORTANCE].len,
2750 			       l->backlog[TIPC_SYSTEM_IMPORTANCE].len);
2751 		i += tipc_list_dump(&l->backlogq, false, buf + i);
2752 	}
2753 	if (dqueues & TIPC_DUMP_DEFERDQ) {
2754 		i += scnprintf(buf + i, sz - i, "deferdq: ");
2755 		i += tipc_list_dump(&l->deferdq, false, buf + i);
2756 	}
2757 	if (dqueues & TIPC_DUMP_INPUTQ) {
2758 		i += scnprintf(buf + i, sz - i, "inputq: ");
2759 		i += tipc_list_dump(l->inputq, false, buf + i);
2760 	}
2761 	if (dqueues & TIPC_DUMP_WAKEUP) {
2762 		i += scnprintf(buf + i, sz - i, "wakeup: ");
2763 		i += tipc_list_dump(&l->wakeupq, false, buf + i);
2764 	}
2765 
2766 	return i;
2767 }
2768