1/*
2 * net/tipc/link.c: TIPC link code
3 *
4 * Copyright (c) 1996-2007, 2012-2014, 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 "link.h"
39#include "socket.h"
40#include "name_distr.h"
41#include "discover.h"
42#include "config.h"
43
44#include <linux/pkt_sched.h>
45
46/*
47 * Error message prefixes
48 */
49static const char *link_co_err = "Link changeover error, ";
50static const char *link_rst_msg = "Resetting link ";
51static const char *link_unk_evt = "Unknown link event ";
52
53/*
54 * Out-of-range value for link session numbers
55 */
56#define INVALID_SESSION 0x10000
57
58/*
59 * Link state events:
60 */
61#define  STARTING_EVT    856384768	/* link processing trigger */
62#define  TRAFFIC_MSG_EVT 560815u	/* rx'd ??? */
63#define  TIMEOUT_EVT     560817u	/* link timer expired */
64
65/*
66 * The following two 'message types' is really just implementation
67 * data conveniently stored in the message header.
68 * They must not be considered part of the protocol
69 */
70#define OPEN_MSG   0
71#define CLOSED_MSG 1
72
73/*
74 * State value stored in 'exp_msg_count'
75 */
76#define START_CHANGEOVER 100000u
77
78static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
79				       struct sk_buff *buf);
80static void tipc_link_proto_rcv(struct tipc_link *l_ptr, struct sk_buff *buf);
81static int  tipc_link_tunnel_rcv(struct tipc_node *n_ptr,
82				 struct sk_buff **buf);
83static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance);
84static void link_state_event(struct tipc_link *l_ptr, u32 event);
85static void link_reset_statistics(struct tipc_link *l_ptr);
86static void link_print(struct tipc_link *l_ptr, const char *str);
87static void tipc_link_sync_xmit(struct tipc_link *l);
88static void tipc_link_sync_rcv(struct tipc_node *n, struct sk_buff *buf);
89static int tipc_link_input(struct tipc_link *l, struct sk_buff *buf);
90static int tipc_link_prepare_input(struct tipc_link *l, struct sk_buff **buf);
91
92/*
93 *  Simple link routines
94 */
95static unsigned int align(unsigned int i)
96{
97	return (i + 3) & ~3u;
98}
99
100static void link_init_max_pkt(struct tipc_link *l_ptr)
101{
102	struct tipc_bearer *b_ptr;
103	u32 max_pkt;
104
105	rcu_read_lock();
106	b_ptr = rcu_dereference_rtnl(bearer_list[l_ptr->bearer_id]);
107	if (!b_ptr) {
108		rcu_read_unlock();
109		return;
110	}
111	max_pkt = (b_ptr->mtu & ~3);
112	rcu_read_unlock();
113
114	if (max_pkt > MAX_MSG_SIZE)
115		max_pkt = MAX_MSG_SIZE;
116
117	l_ptr->max_pkt_target = max_pkt;
118	if (l_ptr->max_pkt_target < MAX_PKT_DEFAULT)
119		l_ptr->max_pkt = l_ptr->max_pkt_target;
120	else
121		l_ptr->max_pkt = MAX_PKT_DEFAULT;
122
123	l_ptr->max_pkt_probes = 0;
124}
125
126static u32 link_next_sent(struct tipc_link *l_ptr)
127{
128	if (l_ptr->next_out)
129		return buf_seqno(l_ptr->next_out);
130	return mod(l_ptr->next_out_no);
131}
132
133static u32 link_last_sent(struct tipc_link *l_ptr)
134{
135	return mod(link_next_sent(l_ptr) - 1);
136}
137
138/*
139 *  Simple non-static link routines (i.e. referenced outside this file)
140 */
141int tipc_link_is_up(struct tipc_link *l_ptr)
142{
143	if (!l_ptr)
144		return 0;
145	return link_working_working(l_ptr) || link_working_unknown(l_ptr);
146}
147
148int tipc_link_is_active(struct tipc_link *l_ptr)
149{
150	return	(l_ptr->owner->active_links[0] == l_ptr) ||
151		(l_ptr->owner->active_links[1] == l_ptr);
152}
153
154/**
155 * link_timeout - handle expiration of link timer
156 * @l_ptr: pointer to link
157 */
158static void link_timeout(struct tipc_link *l_ptr)
159{
160	tipc_node_lock(l_ptr->owner);
161
162	/* update counters used in statistical profiling of send traffic */
163	l_ptr->stats.accu_queue_sz += l_ptr->out_queue_size;
164	l_ptr->stats.queue_sz_counts++;
165
166	if (l_ptr->first_out) {
167		struct tipc_msg *msg = buf_msg(l_ptr->first_out);
168		u32 length = msg_size(msg);
169
170		if ((msg_user(msg) == MSG_FRAGMENTER) &&
171		    (msg_type(msg) == FIRST_FRAGMENT)) {
172			length = msg_size(msg_get_wrapped(msg));
173		}
174		if (length) {
175			l_ptr->stats.msg_lengths_total += length;
176			l_ptr->stats.msg_length_counts++;
177			if (length <= 64)
178				l_ptr->stats.msg_length_profile[0]++;
179			else if (length <= 256)
180				l_ptr->stats.msg_length_profile[1]++;
181			else if (length <= 1024)
182				l_ptr->stats.msg_length_profile[2]++;
183			else if (length <= 4096)
184				l_ptr->stats.msg_length_profile[3]++;
185			else if (length <= 16384)
186				l_ptr->stats.msg_length_profile[4]++;
187			else if (length <= 32768)
188				l_ptr->stats.msg_length_profile[5]++;
189			else
190				l_ptr->stats.msg_length_profile[6]++;
191		}
192	}
193
194	/* do all other link processing performed on a periodic basis */
195
196	link_state_event(l_ptr, TIMEOUT_EVT);
197
198	if (l_ptr->next_out)
199		tipc_link_push_queue(l_ptr);
200
201	tipc_node_unlock(l_ptr->owner);
202}
203
204static void link_set_timer(struct tipc_link *l_ptr, u32 time)
205{
206	k_start_timer(&l_ptr->timer, time);
207}
208
209/**
210 * tipc_link_create - create a new link
211 * @n_ptr: pointer to associated node
212 * @b_ptr: pointer to associated bearer
213 * @media_addr: media address to use when sending messages over link
214 *
215 * Returns pointer to link.
216 */
217struct tipc_link *tipc_link_create(struct tipc_node *n_ptr,
218				   struct tipc_bearer *b_ptr,
219				   const struct tipc_media_addr *media_addr)
220{
221	struct tipc_link *l_ptr;
222	struct tipc_msg *msg;
223	char *if_name;
224	char addr_string[16];
225	u32 peer = n_ptr->addr;
226
227	if (n_ptr->link_cnt >= 2) {
228		tipc_addr_string_fill(addr_string, n_ptr->addr);
229		pr_err("Attempt to establish third link to %s\n", addr_string);
230		return NULL;
231	}
232
233	if (n_ptr->links[b_ptr->identity]) {
234		tipc_addr_string_fill(addr_string, n_ptr->addr);
235		pr_err("Attempt to establish second link on <%s> to %s\n",
236		       b_ptr->name, addr_string);
237		return NULL;
238	}
239
240	l_ptr = kzalloc(sizeof(*l_ptr), GFP_ATOMIC);
241	if (!l_ptr) {
242		pr_warn("Link creation failed, no memory\n");
243		return NULL;
244	}
245
246	l_ptr->addr = peer;
247	if_name = strchr(b_ptr->name, ':') + 1;
248	sprintf(l_ptr->name, "%u.%u.%u:%s-%u.%u.%u:unknown",
249		tipc_zone(tipc_own_addr), tipc_cluster(tipc_own_addr),
250		tipc_node(tipc_own_addr),
251		if_name,
252		tipc_zone(peer), tipc_cluster(peer), tipc_node(peer));
253		/* note: peer i/f name is updated by reset/activate message */
254	memcpy(&l_ptr->media_addr, media_addr, sizeof(*media_addr));
255	l_ptr->owner = n_ptr;
256	l_ptr->checkpoint = 1;
257	l_ptr->peer_session = INVALID_SESSION;
258	l_ptr->bearer_id = b_ptr->identity;
259	link_set_supervision_props(l_ptr, b_ptr->tolerance);
260	l_ptr->state = RESET_UNKNOWN;
261
262	l_ptr->pmsg = (struct tipc_msg *)&l_ptr->proto_msg;
263	msg = l_ptr->pmsg;
264	tipc_msg_init(msg, LINK_PROTOCOL, RESET_MSG, INT_H_SIZE, l_ptr->addr);
265	msg_set_size(msg, sizeof(l_ptr->proto_msg));
266	msg_set_session(msg, (tipc_random & 0xffff));
267	msg_set_bearer_id(msg, b_ptr->identity);
268	strcpy((char *)msg_data(msg), if_name);
269
270	l_ptr->priority = b_ptr->priority;
271	tipc_link_set_queue_limits(l_ptr, b_ptr->window);
272
273	l_ptr->net_plane = b_ptr->net_plane;
274	link_init_max_pkt(l_ptr);
275
276	l_ptr->next_out_no = 1;
277	__skb_queue_head_init(&l_ptr->waiting_sks);
278
279	link_reset_statistics(l_ptr);
280
281	tipc_node_attach_link(n_ptr, l_ptr);
282
283	k_init_timer(&l_ptr->timer, (Handler)link_timeout,
284		     (unsigned long)l_ptr);
285
286	link_state_event(l_ptr, STARTING_EVT);
287
288	return l_ptr;
289}
290
291void tipc_link_delete_list(unsigned int bearer_id, bool shutting_down)
292{
293	struct tipc_link *l_ptr;
294	struct tipc_node *n_ptr;
295
296	rcu_read_lock();
297	list_for_each_entry_rcu(n_ptr, &tipc_node_list, list) {
298		tipc_node_lock(n_ptr);
299		l_ptr = n_ptr->links[bearer_id];
300		if (l_ptr) {
301			tipc_link_reset(l_ptr);
302			if (shutting_down || !tipc_node_is_up(n_ptr)) {
303				tipc_node_detach_link(l_ptr->owner, l_ptr);
304				tipc_link_reset_fragments(l_ptr);
305				tipc_node_unlock(n_ptr);
306
307				/* Nobody else can access this link now: */
308				del_timer_sync(&l_ptr->timer);
309				kfree(l_ptr);
310			} else {
311				/* Detach/delete when failover is finished: */
312				l_ptr->flags |= LINK_STOPPED;
313				tipc_node_unlock(n_ptr);
314				del_timer_sync(&l_ptr->timer);
315			}
316			continue;
317		}
318		tipc_node_unlock(n_ptr);
319	}
320	rcu_read_unlock();
321}
322
323/**
324 * link_schedule_user - schedule user for wakeup after congestion
325 * @link: congested link
326 * @oport: sending port
327 * @chain_sz: size of buffer chain that was attempted sent
328 * @imp: importance of message attempted sent
329 * Create pseudo msg to send back to user when congestion abates
330 */
331static bool link_schedule_user(struct tipc_link *link, u32 oport,
332			       uint chain_sz, uint imp)
333{
334	struct sk_buff *buf;
335
336	buf = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0, tipc_own_addr,
337			      tipc_own_addr, oport, 0, 0);
338	if (!buf)
339		return false;
340	TIPC_SKB_CB(buf)->chain_sz = chain_sz;
341	TIPC_SKB_CB(buf)->chain_imp = imp;
342	__skb_queue_tail(&link->waiting_sks, buf);
343	link->stats.link_congs++;
344	return true;
345}
346
347/**
348 * link_prepare_wakeup - prepare users for wakeup after congestion
349 * @link: congested link
350 * Move a number of waiting users, as permitted by available space in
351 * the send queue, from link wait queue to node wait queue for wakeup
352 */
353static void link_prepare_wakeup(struct tipc_link *link)
354{
355	struct sk_buff_head *wq = &link->waiting_sks;
356	struct sk_buff *buf;
357	uint pend_qsz = link->out_queue_size;
358
359	for (buf = skb_peek(wq); buf; buf = skb_peek(wq)) {
360		if (pend_qsz >= link->queue_limit[TIPC_SKB_CB(buf)->chain_imp])
361			break;
362		pend_qsz += TIPC_SKB_CB(buf)->chain_sz;
363		__skb_queue_tail(&link->owner->waiting_sks, __skb_dequeue(wq));
364	}
365}
366
367/**
368 * link_release_outqueue - purge link's outbound message queue
369 * @l_ptr: pointer to link
370 */
371static void link_release_outqueue(struct tipc_link *l_ptr)
372{
373	kfree_skb_list(l_ptr->first_out);
374	l_ptr->first_out = NULL;
375	l_ptr->out_queue_size = 0;
376}
377
378/**
379 * tipc_link_reset_fragments - purge link's inbound message fragments queue
380 * @l_ptr: pointer to link
381 */
382void tipc_link_reset_fragments(struct tipc_link *l_ptr)
383{
384	kfree_skb(l_ptr->reasm_buf);
385	l_ptr->reasm_buf = NULL;
386}
387
388/**
389 * tipc_link_purge_queues - purge all pkt queues associated with link
390 * @l_ptr: pointer to link
391 */
392void tipc_link_purge_queues(struct tipc_link *l_ptr)
393{
394	kfree_skb_list(l_ptr->oldest_deferred_in);
395	kfree_skb_list(l_ptr->first_out);
396	tipc_link_reset_fragments(l_ptr);
397	kfree_skb(l_ptr->proto_msg_queue);
398	l_ptr->proto_msg_queue = NULL;
399}
400
401void tipc_link_reset(struct tipc_link *l_ptr)
402{
403	u32 prev_state = l_ptr->state;
404	u32 checkpoint = l_ptr->next_in_no;
405	int was_active_link = tipc_link_is_active(l_ptr);
406	struct tipc_node *owner = l_ptr->owner;
407
408	msg_set_session(l_ptr->pmsg, ((msg_session(l_ptr->pmsg) + 1) & 0xffff));
409
410	/* Link is down, accept any session */
411	l_ptr->peer_session = INVALID_SESSION;
412
413	/* Prepare for max packet size negotiation */
414	link_init_max_pkt(l_ptr);
415
416	l_ptr->state = RESET_UNKNOWN;
417
418	if ((prev_state == RESET_UNKNOWN) || (prev_state == RESET_RESET))
419		return;
420
421	tipc_node_link_down(l_ptr->owner, l_ptr);
422	tipc_bearer_remove_dest(l_ptr->bearer_id, l_ptr->addr);
423
424	if (was_active_link && tipc_node_active_links(l_ptr->owner)) {
425		l_ptr->reset_checkpoint = checkpoint;
426		l_ptr->exp_msg_count = START_CHANGEOVER;
427	}
428
429	/* Clean up all queues: */
430	link_release_outqueue(l_ptr);
431	kfree_skb(l_ptr->proto_msg_queue);
432	l_ptr->proto_msg_queue = NULL;
433	kfree_skb_list(l_ptr->oldest_deferred_in);
434	if (!skb_queue_empty(&l_ptr->waiting_sks)) {
435		skb_queue_splice_init(&l_ptr->waiting_sks, &owner->waiting_sks);
436		owner->action_flags |= TIPC_WAKEUP_USERS;
437	}
438	l_ptr->retransm_queue_head = 0;
439	l_ptr->retransm_queue_size = 0;
440	l_ptr->last_out = NULL;
441	l_ptr->first_out = NULL;
442	l_ptr->next_out = NULL;
443	l_ptr->unacked_window = 0;
444	l_ptr->checkpoint = 1;
445	l_ptr->next_out_no = 1;
446	l_ptr->deferred_inqueue_sz = 0;
447	l_ptr->oldest_deferred_in = NULL;
448	l_ptr->newest_deferred_in = NULL;
449	l_ptr->fsm_msg_cnt = 0;
450	l_ptr->stale_count = 0;
451	link_reset_statistics(l_ptr);
452}
453
454void tipc_link_reset_list(unsigned int bearer_id)
455{
456	struct tipc_link *l_ptr;
457	struct tipc_node *n_ptr;
458
459	rcu_read_lock();
460	list_for_each_entry_rcu(n_ptr, &tipc_node_list, list) {
461		tipc_node_lock(n_ptr);
462		l_ptr = n_ptr->links[bearer_id];
463		if (l_ptr)
464			tipc_link_reset(l_ptr);
465		tipc_node_unlock(n_ptr);
466	}
467	rcu_read_unlock();
468}
469
470static void link_activate(struct tipc_link *l_ptr)
471{
472	l_ptr->next_in_no = l_ptr->stats.recv_info = 1;
473	tipc_node_link_up(l_ptr->owner, l_ptr);
474	tipc_bearer_add_dest(l_ptr->bearer_id, l_ptr->addr);
475}
476
477/**
478 * link_state_event - link finite state machine
479 * @l_ptr: pointer to link
480 * @event: state machine event to process
481 */
482static void link_state_event(struct tipc_link *l_ptr, unsigned int event)
483{
484	struct tipc_link *other;
485	u32 cont_intv = l_ptr->continuity_interval;
486
487	if (l_ptr->flags & LINK_STOPPED)
488		return;
489
490	if (!(l_ptr->flags & LINK_STARTED) && (event != STARTING_EVT))
491		return;		/* Not yet. */
492
493	/* Check whether changeover is going on */
494	if (l_ptr->exp_msg_count) {
495		if (event == TIMEOUT_EVT)
496			link_set_timer(l_ptr, cont_intv);
497		return;
498	}
499
500	switch (l_ptr->state) {
501	case WORKING_WORKING:
502		switch (event) {
503		case TRAFFIC_MSG_EVT:
504		case ACTIVATE_MSG:
505			break;
506		case TIMEOUT_EVT:
507			if (l_ptr->next_in_no != l_ptr->checkpoint) {
508				l_ptr->checkpoint = l_ptr->next_in_no;
509				if (tipc_bclink_acks_missing(l_ptr->owner)) {
510					tipc_link_proto_xmit(l_ptr, STATE_MSG,
511							     0, 0, 0, 0, 0);
512					l_ptr->fsm_msg_cnt++;
513				} else if (l_ptr->max_pkt < l_ptr->max_pkt_target) {
514					tipc_link_proto_xmit(l_ptr, STATE_MSG,
515							     1, 0, 0, 0, 0);
516					l_ptr->fsm_msg_cnt++;
517				}
518				link_set_timer(l_ptr, cont_intv);
519				break;
520			}
521			l_ptr->state = WORKING_UNKNOWN;
522			l_ptr->fsm_msg_cnt = 0;
523			tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
524			l_ptr->fsm_msg_cnt++;
525			link_set_timer(l_ptr, cont_intv / 4);
526			break;
527		case RESET_MSG:
528			pr_info("%s<%s>, requested by peer\n", link_rst_msg,
529				l_ptr->name);
530			tipc_link_reset(l_ptr);
531			l_ptr->state = RESET_RESET;
532			l_ptr->fsm_msg_cnt = 0;
533			tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
534					     0, 0, 0, 0, 0);
535			l_ptr->fsm_msg_cnt++;
536			link_set_timer(l_ptr, cont_intv);
537			break;
538		default:
539			pr_err("%s%u in WW state\n", link_unk_evt, event);
540		}
541		break;
542	case WORKING_UNKNOWN:
543		switch (event) {
544		case TRAFFIC_MSG_EVT:
545		case ACTIVATE_MSG:
546			l_ptr->state = WORKING_WORKING;
547			l_ptr->fsm_msg_cnt = 0;
548			link_set_timer(l_ptr, cont_intv);
549			break;
550		case RESET_MSG:
551			pr_info("%s<%s>, requested by peer while probing\n",
552				link_rst_msg, l_ptr->name);
553			tipc_link_reset(l_ptr);
554			l_ptr->state = RESET_RESET;
555			l_ptr->fsm_msg_cnt = 0;
556			tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
557					     0, 0, 0, 0, 0);
558			l_ptr->fsm_msg_cnt++;
559			link_set_timer(l_ptr, cont_intv);
560			break;
561		case TIMEOUT_EVT:
562			if (l_ptr->next_in_no != l_ptr->checkpoint) {
563				l_ptr->state = WORKING_WORKING;
564				l_ptr->fsm_msg_cnt = 0;
565				l_ptr->checkpoint = l_ptr->next_in_no;
566				if (tipc_bclink_acks_missing(l_ptr->owner)) {
567					tipc_link_proto_xmit(l_ptr, STATE_MSG,
568							     0, 0, 0, 0, 0);
569					l_ptr->fsm_msg_cnt++;
570				}
571				link_set_timer(l_ptr, cont_intv);
572			} else if (l_ptr->fsm_msg_cnt < l_ptr->abort_limit) {
573				tipc_link_proto_xmit(l_ptr, STATE_MSG,
574						     1, 0, 0, 0, 0);
575				l_ptr->fsm_msg_cnt++;
576				link_set_timer(l_ptr, cont_intv / 4);
577			} else {	/* Link has failed */
578				pr_warn("%s<%s>, peer not responding\n",
579					link_rst_msg, l_ptr->name);
580				tipc_link_reset(l_ptr);
581				l_ptr->state = RESET_UNKNOWN;
582				l_ptr->fsm_msg_cnt = 0;
583				tipc_link_proto_xmit(l_ptr, RESET_MSG,
584						     0, 0, 0, 0, 0);
585				l_ptr->fsm_msg_cnt++;
586				link_set_timer(l_ptr, cont_intv);
587			}
588			break;
589		default:
590			pr_err("%s%u in WU state\n", link_unk_evt, event);
591		}
592		break;
593	case RESET_UNKNOWN:
594		switch (event) {
595		case TRAFFIC_MSG_EVT:
596			break;
597		case ACTIVATE_MSG:
598			other = l_ptr->owner->active_links[0];
599			if (other && link_working_unknown(other))
600				break;
601			l_ptr->state = WORKING_WORKING;
602			l_ptr->fsm_msg_cnt = 0;
603			link_activate(l_ptr);
604			tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
605			l_ptr->fsm_msg_cnt++;
606			if (l_ptr->owner->working_links == 1)
607				tipc_link_sync_xmit(l_ptr);
608			link_set_timer(l_ptr, cont_intv);
609			break;
610		case RESET_MSG:
611			l_ptr->state = RESET_RESET;
612			l_ptr->fsm_msg_cnt = 0;
613			tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
614					     1, 0, 0, 0, 0);
615			l_ptr->fsm_msg_cnt++;
616			link_set_timer(l_ptr, cont_intv);
617			break;
618		case STARTING_EVT:
619			l_ptr->flags |= LINK_STARTED;
620			/* fall through */
621		case TIMEOUT_EVT:
622			tipc_link_proto_xmit(l_ptr, RESET_MSG, 0, 0, 0, 0, 0);
623			l_ptr->fsm_msg_cnt++;
624			link_set_timer(l_ptr, cont_intv);
625			break;
626		default:
627			pr_err("%s%u in RU state\n", link_unk_evt, event);
628		}
629		break;
630	case RESET_RESET:
631		switch (event) {
632		case TRAFFIC_MSG_EVT:
633		case ACTIVATE_MSG:
634			other = l_ptr->owner->active_links[0];
635			if (other && link_working_unknown(other))
636				break;
637			l_ptr->state = WORKING_WORKING;
638			l_ptr->fsm_msg_cnt = 0;
639			link_activate(l_ptr);
640			tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
641			l_ptr->fsm_msg_cnt++;
642			if (l_ptr->owner->working_links == 1)
643				tipc_link_sync_xmit(l_ptr);
644			link_set_timer(l_ptr, cont_intv);
645			break;
646		case RESET_MSG:
647			break;
648		case TIMEOUT_EVT:
649			tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
650					     0, 0, 0, 0, 0);
651			l_ptr->fsm_msg_cnt++;
652			link_set_timer(l_ptr, cont_intv);
653			break;
654		default:
655			pr_err("%s%u in RR state\n", link_unk_evt, event);
656		}
657		break;
658	default:
659		pr_err("Unknown link state %u/%u\n", l_ptr->state, event);
660	}
661}
662
663/* tipc_link_cong: determine return value and how to treat the
664 * sent buffer during link congestion.
665 * - For plain, errorless user data messages we keep the buffer and
666 *   return -ELINKONG.
667 * - For all other messages we discard the buffer and return -EHOSTUNREACH
668 * - For TIPC internal messages we also reset the link
669 */
670static int tipc_link_cong(struct tipc_link *link, struct sk_buff *buf)
671{
672	struct tipc_msg *msg = buf_msg(buf);
673	uint imp = tipc_msg_tot_importance(msg);
674	u32 oport = msg_tot_origport(msg);
675
676	if (unlikely(imp > TIPC_CRITICAL_IMPORTANCE)) {
677		pr_warn("%s<%s>, send queue full", link_rst_msg, link->name);
678		tipc_link_reset(link);
679		goto drop;
680	}
681	if (unlikely(msg_errcode(msg)))
682		goto drop;
683	if (unlikely(msg_reroute_cnt(msg)))
684		goto drop;
685	if (TIPC_SKB_CB(buf)->wakeup_pending)
686		return -ELINKCONG;
687	if (link_schedule_user(link, oport, TIPC_SKB_CB(buf)->chain_sz, imp))
688		return -ELINKCONG;
689drop:
690	kfree_skb_list(buf);
691	return -EHOSTUNREACH;
692}
693
694/**
695 * __tipc_link_xmit(): same as tipc_link_xmit, but destlink is known & locked
696 * @link: link to use
697 * @buf: chain of buffers containing message
698 * Consumes the buffer chain, except when returning -ELINKCONG
699 * Returns 0 if success, otherwise errno: -ELINKCONG, -EMSGSIZE (plain socket
700 * user data messages) or -EHOSTUNREACH (all other messages/senders)
701 * Only the socket functions tipc_send_stream() and tipc_send_packet() need
702 * to act on the return value, since they may need to do more send attempts.
703 */
704int __tipc_link_xmit(struct tipc_link *link, struct sk_buff *buf)
705{
706	struct tipc_msg *msg = buf_msg(buf);
707	uint psz = msg_size(msg);
708	uint qsz = link->out_queue_size;
709	uint sndlim = link->queue_limit[0];
710	uint imp = tipc_msg_tot_importance(msg);
711	uint mtu = link->max_pkt;
712	uint ack = mod(link->next_in_no - 1);
713	uint seqno = link->next_out_no;
714	uint bc_last_in = link->owner->bclink.last_in;
715	struct tipc_media_addr *addr = &link->media_addr;
716	struct sk_buff *next = buf->next;
717
718	/* Match queue limits against msg importance: */
719	if (unlikely(qsz >= link->queue_limit[imp]))
720		return tipc_link_cong(link, buf);
721
722	/* Has valid packet limit been used ? */
723	if (unlikely(psz > mtu)) {
724		kfree_skb_list(buf);
725		return -EMSGSIZE;
726	}
727
728	/* Prepare each packet for sending, and add to outqueue: */
729	while (buf) {
730		next = buf->next;
731		msg = buf_msg(buf);
732		msg_set_word(msg, 2, ((ack << 16) | mod(seqno)));
733		msg_set_bcast_ack(msg, bc_last_in);
734
735		if (!link->first_out) {
736			link->first_out = buf;
737		} else if (qsz < sndlim) {
738			link->last_out->next = buf;
739		} else if (tipc_msg_bundle(link->last_out, buf, mtu)) {
740			link->stats.sent_bundled++;
741			buf = next;
742			next = buf->next;
743			continue;
744		} else if (tipc_msg_make_bundle(&buf, mtu, link->addr)) {
745			link->stats.sent_bundled++;
746			link->stats.sent_bundles++;
747			link->last_out->next = buf;
748			if (!link->next_out)
749				link->next_out = buf;
750		} else {
751			link->last_out->next = buf;
752			if (!link->next_out)
753				link->next_out = buf;
754		}
755
756		/* Send packet if possible: */
757		if (likely(++qsz <= sndlim)) {
758			tipc_bearer_send(link->bearer_id, buf, addr);
759			link->next_out = next;
760			link->unacked_window = 0;
761		}
762		seqno++;
763		link->last_out = buf;
764		buf = next;
765	}
766	link->next_out_no = seqno;
767	link->out_queue_size = qsz;
768	return 0;
769}
770
771/**
772 * tipc_link_xmit() is the general link level function for message sending
773 * @buf: chain of buffers containing message
774 * @dsz: amount of user data to be sent
775 * @dnode: address of destination node
776 * @selector: a number used for deterministic link selection
777 * Consumes the buffer chain, except when returning -ELINKCONG
778 * Returns 0 if success, otherwise errno: -ELINKCONG,-EHOSTUNREACH,-EMSGSIZE
779 */
780int tipc_link_xmit(struct sk_buff *buf, u32 dnode, u32 selector)
781{
782	struct tipc_link *link = NULL;
783	struct tipc_node *node;
784	int rc = -EHOSTUNREACH;
785
786	node = tipc_node_find(dnode);
787	if (node) {
788		tipc_node_lock(node);
789		link = node->active_links[selector & 1];
790		if (link)
791			rc = __tipc_link_xmit(link, buf);
792		tipc_node_unlock(node);
793	}
794
795	if (link)
796		return rc;
797
798	if (likely(in_own_node(dnode)))
799		return tipc_sk_rcv(buf);
800
801	kfree_skb_list(buf);
802	return rc;
803}
804
805/*
806 * tipc_link_sync_xmit - synchronize broadcast link endpoints.
807 *
808 * Give a newly added peer node the sequence number where it should
809 * start receiving and acking broadcast packets.
810 *
811 * Called with node locked
812 */
813static void tipc_link_sync_xmit(struct tipc_link *link)
814{
815	struct sk_buff *buf;
816	struct tipc_msg *msg;
817
818	buf = tipc_buf_acquire(INT_H_SIZE);
819	if (!buf)
820		return;
821
822	msg = buf_msg(buf);
823	tipc_msg_init(msg, BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE, link->addr);
824	msg_set_last_bcast(msg, link->owner->bclink.acked);
825	__tipc_link_xmit(link, buf);
826}
827
828/*
829 * tipc_link_sync_rcv - synchronize broadcast link endpoints.
830 * Receive the sequence number where we should start receiving and
831 * acking broadcast packets from a newly added peer node, and open
832 * up for reception of such packets.
833 *
834 * Called with node locked
835 */
836static void tipc_link_sync_rcv(struct tipc_node *n, struct sk_buff *buf)
837{
838	struct tipc_msg *msg = buf_msg(buf);
839
840	n->bclink.last_sent = n->bclink.last_in = msg_last_bcast(msg);
841	n->bclink.recv_permitted = true;
842	kfree_skb(buf);
843}
844
845/*
846 * tipc_link_push_packet: Push one unsent packet to the media
847 */
848static u32 tipc_link_push_packet(struct tipc_link *l_ptr)
849{
850	struct sk_buff *buf = l_ptr->first_out;
851	u32 r_q_size = l_ptr->retransm_queue_size;
852	u32 r_q_head = l_ptr->retransm_queue_head;
853
854	/* Step to position where retransmission failed, if any,    */
855	/* consider that buffers may have been released in meantime */
856	if (r_q_size && buf) {
857		u32 last = lesser(mod(r_q_head + r_q_size),
858				  link_last_sent(l_ptr));
859		u32 first = buf_seqno(buf);
860
861		while (buf && less(first, r_q_head)) {
862			first = mod(first + 1);
863			buf = buf->next;
864		}
865		l_ptr->retransm_queue_head = r_q_head = first;
866		l_ptr->retransm_queue_size = r_q_size = mod(last - first);
867	}
868
869	/* Continue retransmission now, if there is anything: */
870	if (r_q_size && buf) {
871		msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
872		msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
873		tipc_bearer_send(l_ptr->bearer_id, buf, &l_ptr->media_addr);
874		l_ptr->retransm_queue_head = mod(++r_q_head);
875		l_ptr->retransm_queue_size = --r_q_size;
876		l_ptr->stats.retransmitted++;
877		return 0;
878	}
879
880	/* Send deferred protocol message, if any: */
881	buf = l_ptr->proto_msg_queue;
882	if (buf) {
883		msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
884		msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
885		tipc_bearer_send(l_ptr->bearer_id, buf, &l_ptr->media_addr);
886		l_ptr->unacked_window = 0;
887		kfree_skb(buf);
888		l_ptr->proto_msg_queue = NULL;
889		return 0;
890	}
891
892	/* Send one deferred data message, if send window not full: */
893	buf = l_ptr->next_out;
894	if (buf) {
895		struct tipc_msg *msg = buf_msg(buf);
896		u32 next = msg_seqno(msg);
897		u32 first = buf_seqno(l_ptr->first_out);
898
899		if (mod(next - first) < l_ptr->queue_limit[0]) {
900			msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
901			msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
902			tipc_bearer_send(l_ptr->bearer_id, buf,
903					 &l_ptr->media_addr);
904			if (msg_user(msg) == MSG_BUNDLER)
905				msg_set_type(msg, BUNDLE_CLOSED);
906			l_ptr->next_out = buf->next;
907			return 0;
908		}
909	}
910	return 1;
911}
912
913/*
914 * push_queue(): push out the unsent messages of a link where
915 *               congestion has abated. Node is locked
916 */
917void tipc_link_push_queue(struct tipc_link *l_ptr)
918{
919	u32 res;
920
921	do {
922		res = tipc_link_push_packet(l_ptr);
923	} while (!res);
924}
925
926void tipc_link_reset_all(struct tipc_node *node)
927{
928	char addr_string[16];
929	u32 i;
930
931	tipc_node_lock(node);
932
933	pr_warn("Resetting all links to %s\n",
934		tipc_addr_string_fill(addr_string, node->addr));
935
936	for (i = 0; i < MAX_BEARERS; i++) {
937		if (node->links[i]) {
938			link_print(node->links[i], "Resetting link\n");
939			tipc_link_reset(node->links[i]);
940		}
941	}
942
943	tipc_node_unlock(node);
944}
945
946static void link_retransmit_failure(struct tipc_link *l_ptr,
947				    struct sk_buff *buf)
948{
949	struct tipc_msg *msg = buf_msg(buf);
950
951	pr_warn("Retransmission failure on link <%s>\n", l_ptr->name);
952
953	if (l_ptr->addr) {
954		/* Handle failure on standard link */
955		link_print(l_ptr, "Resetting link\n");
956		tipc_link_reset(l_ptr);
957
958	} else {
959		/* Handle failure on broadcast link */
960		struct tipc_node *n_ptr;
961		char addr_string[16];
962
963		pr_info("Msg seq number: %u,  ", msg_seqno(msg));
964		pr_cont("Outstanding acks: %lu\n",
965			(unsigned long) TIPC_SKB_CB(buf)->handle);
966
967		n_ptr = tipc_bclink_retransmit_to();
968		tipc_node_lock(n_ptr);
969
970		tipc_addr_string_fill(addr_string, n_ptr->addr);
971		pr_info("Broadcast link info for %s\n", addr_string);
972		pr_info("Reception permitted: %d,  Acked: %u\n",
973			n_ptr->bclink.recv_permitted,
974			n_ptr->bclink.acked);
975		pr_info("Last in: %u,  Oos state: %u,  Last sent: %u\n",
976			n_ptr->bclink.last_in,
977			n_ptr->bclink.oos_state,
978			n_ptr->bclink.last_sent);
979
980		tipc_node_unlock(n_ptr);
981
982		tipc_bclink_set_flags(TIPC_BCLINK_RESET);
983		l_ptr->stale_count = 0;
984	}
985}
986
987void tipc_link_retransmit(struct tipc_link *l_ptr, struct sk_buff *buf,
988			  u32 retransmits)
989{
990	struct tipc_msg *msg;
991
992	if (!buf)
993		return;
994
995	msg = buf_msg(buf);
996
997	/* Detect repeated retransmit failures */
998	if (l_ptr->last_retransmitted == msg_seqno(msg)) {
999		if (++l_ptr->stale_count > 100) {
1000			link_retransmit_failure(l_ptr, buf);
1001			return;
1002		}
1003	} else {
1004		l_ptr->last_retransmitted = msg_seqno(msg);
1005		l_ptr->stale_count = 1;
1006	}
1007
1008	while (retransmits && (buf != l_ptr->next_out) && buf) {
1009		msg = buf_msg(buf);
1010		msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1011		msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1012		tipc_bearer_send(l_ptr->bearer_id, buf, &l_ptr->media_addr);
1013		buf = buf->next;
1014		retransmits--;
1015		l_ptr->stats.retransmitted++;
1016	}
1017
1018	l_ptr->retransm_queue_head = l_ptr->retransm_queue_size = 0;
1019}
1020
1021/**
1022 * link_insert_deferred_queue - insert deferred messages back into receive chain
1023 */
1024static struct sk_buff *link_insert_deferred_queue(struct tipc_link *l_ptr,
1025						  struct sk_buff *buf)
1026{
1027	u32 seq_no;
1028
1029	if (l_ptr->oldest_deferred_in == NULL)
1030		return buf;
1031
1032	seq_no = buf_seqno(l_ptr->oldest_deferred_in);
1033	if (seq_no == mod(l_ptr->next_in_no)) {
1034		l_ptr->newest_deferred_in->next = buf;
1035		buf = l_ptr->oldest_deferred_in;
1036		l_ptr->oldest_deferred_in = NULL;
1037		l_ptr->deferred_inqueue_sz = 0;
1038	}
1039	return buf;
1040}
1041
1042/**
1043 * link_recv_buf_validate - validate basic format of received message
1044 *
1045 * This routine ensures a TIPC message has an acceptable header, and at least
1046 * as much data as the header indicates it should.  The routine also ensures
1047 * that the entire message header is stored in the main fragment of the message
1048 * buffer, to simplify future access to message header fields.
1049 *
1050 * Note: Having extra info present in the message header or data areas is OK.
1051 * TIPC will ignore the excess, under the assumption that it is optional info
1052 * introduced by a later release of the protocol.
1053 */
1054static int link_recv_buf_validate(struct sk_buff *buf)
1055{
1056	static u32 min_data_hdr_size[8] = {
1057		SHORT_H_SIZE, MCAST_H_SIZE, NAMED_H_SIZE, BASIC_H_SIZE,
1058		MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE
1059		};
1060
1061	struct tipc_msg *msg;
1062	u32 tipc_hdr[2];
1063	u32 size;
1064	u32 hdr_size;
1065	u32 min_hdr_size;
1066
1067	/* If this packet comes from the defer queue, the skb has already
1068	 * been validated
1069	 */
1070	if (unlikely(TIPC_SKB_CB(buf)->deferred))
1071		return 1;
1072
1073	if (unlikely(buf->len < MIN_H_SIZE))
1074		return 0;
1075
1076	msg = skb_header_pointer(buf, 0, sizeof(tipc_hdr), tipc_hdr);
1077	if (msg == NULL)
1078		return 0;
1079
1080	if (unlikely(msg_version(msg) != TIPC_VERSION))
1081		return 0;
1082
1083	size = msg_size(msg);
1084	hdr_size = msg_hdr_sz(msg);
1085	min_hdr_size = msg_isdata(msg) ?
1086		min_data_hdr_size[msg_type(msg)] : INT_H_SIZE;
1087
1088	if (unlikely((hdr_size < min_hdr_size) ||
1089		     (size < hdr_size) ||
1090		     (buf->len < size) ||
1091		     (size - hdr_size > TIPC_MAX_USER_MSG_SIZE)))
1092		return 0;
1093
1094	return pskb_may_pull(buf, hdr_size);
1095}
1096
1097/**
1098 * tipc_rcv - process TIPC packets/messages arriving from off-node
1099 * @head: pointer to message buffer chain
1100 * @b_ptr: pointer to bearer message arrived on
1101 *
1102 * Invoked with no locks held.  Bearer pointer must point to a valid bearer
1103 * structure (i.e. cannot be NULL), but bearer can be inactive.
1104 */
1105void tipc_rcv(struct sk_buff *head, struct tipc_bearer *b_ptr)
1106{
1107	while (head) {
1108		struct tipc_node *n_ptr;
1109		struct tipc_link *l_ptr;
1110		struct sk_buff *crs;
1111		struct sk_buff *buf = head;
1112		struct tipc_msg *msg;
1113		u32 seq_no;
1114		u32 ackd;
1115		u32 released = 0;
1116
1117		head = head->next;
1118		buf->next = NULL;
1119
1120		/* Ensure message is well-formed */
1121		if (unlikely(!link_recv_buf_validate(buf)))
1122			goto discard;
1123
1124		/* Ensure message data is a single contiguous unit */
1125		if (unlikely(skb_linearize(buf)))
1126			goto discard;
1127
1128		/* Handle arrival of a non-unicast link message */
1129		msg = buf_msg(buf);
1130
1131		if (unlikely(msg_non_seq(msg))) {
1132			if (msg_user(msg) ==  LINK_CONFIG)
1133				tipc_disc_rcv(buf, b_ptr);
1134			else
1135				tipc_bclink_rcv(buf);
1136			continue;
1137		}
1138
1139		/* Discard unicast link messages destined for another node */
1140		if (unlikely(!msg_short(msg) &&
1141			     (msg_destnode(msg) != tipc_own_addr)))
1142			goto discard;
1143
1144		/* Locate neighboring node that sent message */
1145		n_ptr = tipc_node_find(msg_prevnode(msg));
1146		if (unlikely(!n_ptr))
1147			goto discard;
1148		tipc_node_lock(n_ptr);
1149
1150		/* Locate unicast link endpoint that should handle message */
1151		l_ptr = n_ptr->links[b_ptr->identity];
1152		if (unlikely(!l_ptr))
1153			goto unlock_discard;
1154
1155		/* Verify that communication with node is currently allowed */
1156		if ((n_ptr->action_flags & TIPC_WAIT_PEER_LINKS_DOWN) &&
1157		    msg_user(msg) == LINK_PROTOCOL &&
1158		    (msg_type(msg) == RESET_MSG ||
1159		    msg_type(msg) == ACTIVATE_MSG) &&
1160		    !msg_redundant_link(msg))
1161			n_ptr->action_flags &= ~TIPC_WAIT_PEER_LINKS_DOWN;
1162
1163		if (tipc_node_blocked(n_ptr))
1164			goto unlock_discard;
1165
1166		/* Validate message sequence number info */
1167		seq_no = msg_seqno(msg);
1168		ackd = msg_ack(msg);
1169
1170		/* Release acked messages */
1171		if (n_ptr->bclink.recv_permitted)
1172			tipc_bclink_acknowledge(n_ptr, msg_bcast_ack(msg));
1173
1174		crs = l_ptr->first_out;
1175		while ((crs != l_ptr->next_out) &&
1176		       less_eq(buf_seqno(crs), ackd)) {
1177			struct sk_buff *next = crs->next;
1178			kfree_skb(crs);
1179			crs = next;
1180			released++;
1181		}
1182		if (released) {
1183			l_ptr->first_out = crs;
1184			l_ptr->out_queue_size -= released;
1185		}
1186
1187		/* Try sending any messages link endpoint has pending */
1188		if (unlikely(l_ptr->next_out))
1189			tipc_link_push_queue(l_ptr);
1190
1191		if (released && !skb_queue_empty(&l_ptr->waiting_sks)) {
1192			link_prepare_wakeup(l_ptr);
1193			l_ptr->owner->action_flags |= TIPC_WAKEUP_USERS;
1194		}
1195
1196		/* Process the incoming packet */
1197		if (unlikely(!link_working_working(l_ptr))) {
1198			if (msg_user(msg) == LINK_PROTOCOL) {
1199				tipc_link_proto_rcv(l_ptr, buf);
1200				head = link_insert_deferred_queue(l_ptr, head);
1201				tipc_node_unlock(n_ptr);
1202				continue;
1203			}
1204
1205			/* Traffic message. Conditionally activate link */
1206			link_state_event(l_ptr, TRAFFIC_MSG_EVT);
1207
1208			if (link_working_working(l_ptr)) {
1209				/* Re-insert buffer in front of queue */
1210				buf->next = head;
1211				head = buf;
1212				tipc_node_unlock(n_ptr);
1213				continue;
1214			}
1215			goto unlock_discard;
1216		}
1217
1218		/* Link is now in state WORKING_WORKING */
1219		if (unlikely(seq_no != mod(l_ptr->next_in_no))) {
1220			link_handle_out_of_seq_msg(l_ptr, buf);
1221			head = link_insert_deferred_queue(l_ptr, head);
1222			tipc_node_unlock(n_ptr);
1223			continue;
1224		}
1225		l_ptr->next_in_no++;
1226		if (unlikely(l_ptr->oldest_deferred_in))
1227			head = link_insert_deferred_queue(l_ptr, head);
1228
1229		if (unlikely(++l_ptr->unacked_window >= TIPC_MIN_LINK_WIN)) {
1230			l_ptr->stats.sent_acks++;
1231			tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1232		}
1233
1234		if (tipc_link_prepare_input(l_ptr, &buf)) {
1235			tipc_node_unlock(n_ptr);
1236			continue;
1237		}
1238		tipc_node_unlock(n_ptr);
1239		msg = buf_msg(buf);
1240		if (tipc_link_input(l_ptr, buf) != 0)
1241			goto discard;
1242		continue;
1243unlock_discard:
1244		tipc_node_unlock(n_ptr);
1245discard:
1246		kfree_skb(buf);
1247	}
1248}
1249
1250/**
1251 * tipc_link_prepare_input - process TIPC link messages
1252 *
1253 * returns nonzero if the message was consumed
1254 *
1255 * Node lock must be held
1256 */
1257static int tipc_link_prepare_input(struct tipc_link *l, struct sk_buff **buf)
1258{
1259	struct tipc_node *n;
1260	struct tipc_msg *msg;
1261	int res = -EINVAL;
1262
1263	n = l->owner;
1264	msg = buf_msg(*buf);
1265	switch (msg_user(msg)) {
1266	case CHANGEOVER_PROTOCOL:
1267		if (tipc_link_tunnel_rcv(n, buf))
1268			res = 0;
1269		break;
1270	case MSG_FRAGMENTER:
1271		l->stats.recv_fragments++;
1272		if (tipc_buf_append(&l->reasm_buf, buf)) {
1273			l->stats.recv_fragmented++;
1274			res = 0;
1275		} else if (!l->reasm_buf) {
1276			tipc_link_reset(l);
1277		}
1278		break;
1279	case MSG_BUNDLER:
1280		l->stats.recv_bundles++;
1281		l->stats.recv_bundled += msg_msgcnt(msg);
1282		res = 0;
1283		break;
1284	case NAME_DISTRIBUTOR:
1285		n->bclink.recv_permitted = true;
1286		res = 0;
1287		break;
1288	case BCAST_PROTOCOL:
1289		tipc_link_sync_rcv(n, *buf);
1290		break;
1291	default:
1292		res = 0;
1293	}
1294	return res;
1295}
1296/**
1297 * tipc_link_input - Deliver message too higher layers
1298 */
1299static int tipc_link_input(struct tipc_link *l, struct sk_buff *buf)
1300{
1301	struct tipc_msg *msg = buf_msg(buf);
1302	int res = 0;
1303
1304	switch (msg_user(msg)) {
1305	case TIPC_LOW_IMPORTANCE:
1306	case TIPC_MEDIUM_IMPORTANCE:
1307	case TIPC_HIGH_IMPORTANCE:
1308	case TIPC_CRITICAL_IMPORTANCE:
1309	case CONN_MANAGER:
1310		tipc_sk_rcv(buf);
1311		break;
1312	case NAME_DISTRIBUTOR:
1313		tipc_named_rcv(buf);
1314		break;
1315	case MSG_BUNDLER:
1316		tipc_link_bundle_rcv(buf);
1317		break;
1318	default:
1319		res = -EINVAL;
1320	}
1321	return res;
1322}
1323
1324/**
1325 * tipc_link_defer_pkt - Add out-of-sequence message to deferred reception queue
1326 *
1327 * Returns increase in queue length (i.e. 0 or 1)
1328 */
1329u32 tipc_link_defer_pkt(struct sk_buff **head, struct sk_buff **tail,
1330			struct sk_buff *buf)
1331{
1332	struct sk_buff *queue_buf;
1333	struct sk_buff **prev;
1334	u32 seq_no = buf_seqno(buf);
1335
1336	buf->next = NULL;
1337
1338	/* Empty queue ? */
1339	if (*head == NULL) {
1340		*head = *tail = buf;
1341		return 1;
1342	}
1343
1344	/* Last ? */
1345	if (less(buf_seqno(*tail), seq_no)) {
1346		(*tail)->next = buf;
1347		*tail = buf;
1348		return 1;
1349	}
1350
1351	/* Locate insertion point in queue, then insert; discard if duplicate */
1352	prev = head;
1353	queue_buf = *head;
1354	for (;;) {
1355		u32 curr_seqno = buf_seqno(queue_buf);
1356
1357		if (seq_no == curr_seqno) {
1358			kfree_skb(buf);
1359			return 0;
1360		}
1361
1362		if (less(seq_no, curr_seqno))
1363			break;
1364
1365		prev = &queue_buf->next;
1366		queue_buf = queue_buf->next;
1367	}
1368
1369	buf->next = queue_buf;
1370	*prev = buf;
1371	return 1;
1372}
1373
1374/*
1375 * link_handle_out_of_seq_msg - handle arrival of out-of-sequence packet
1376 */
1377static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
1378				       struct sk_buff *buf)
1379{
1380	u32 seq_no = buf_seqno(buf);
1381
1382	if (likely(msg_user(buf_msg(buf)) == LINK_PROTOCOL)) {
1383		tipc_link_proto_rcv(l_ptr, buf);
1384		return;
1385	}
1386
1387	/* Record OOS packet arrival (force mismatch on next timeout) */
1388	l_ptr->checkpoint--;
1389
1390	/*
1391	 * Discard packet if a duplicate; otherwise add it to deferred queue
1392	 * and notify peer of gap as per protocol specification
1393	 */
1394	if (less(seq_no, mod(l_ptr->next_in_no))) {
1395		l_ptr->stats.duplicates++;
1396		kfree_skb(buf);
1397		return;
1398	}
1399
1400	if (tipc_link_defer_pkt(&l_ptr->oldest_deferred_in,
1401				&l_ptr->newest_deferred_in, buf)) {
1402		l_ptr->deferred_inqueue_sz++;
1403		l_ptr->stats.deferred_recv++;
1404		TIPC_SKB_CB(buf)->deferred = true;
1405		if ((l_ptr->deferred_inqueue_sz % 16) == 1)
1406			tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1407	} else
1408		l_ptr->stats.duplicates++;
1409}
1410
1411/*
1412 * Send protocol message to the other endpoint.
1413 */
1414void tipc_link_proto_xmit(struct tipc_link *l_ptr, u32 msg_typ, int probe_msg,
1415			  u32 gap, u32 tolerance, u32 priority, u32 ack_mtu)
1416{
1417	struct sk_buff *buf = NULL;
1418	struct tipc_msg *msg = l_ptr->pmsg;
1419	u32 msg_size = sizeof(l_ptr->proto_msg);
1420	int r_flag;
1421
1422	/* Discard any previous message that was deferred due to congestion */
1423	if (l_ptr->proto_msg_queue) {
1424		kfree_skb(l_ptr->proto_msg_queue);
1425		l_ptr->proto_msg_queue = NULL;
1426	}
1427
1428	/* Don't send protocol message during link changeover */
1429	if (l_ptr->exp_msg_count)
1430		return;
1431
1432	/* Abort non-RESET send if communication with node is prohibited */
1433	if ((tipc_node_blocked(l_ptr->owner)) && (msg_typ != RESET_MSG))
1434		return;
1435
1436	/* Create protocol message with "out-of-sequence" sequence number */
1437	msg_set_type(msg, msg_typ);
1438	msg_set_net_plane(msg, l_ptr->net_plane);
1439	msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1440	msg_set_last_bcast(msg, tipc_bclink_get_last_sent());
1441
1442	if (msg_typ == STATE_MSG) {
1443		u32 next_sent = mod(l_ptr->next_out_no);
1444
1445		if (!tipc_link_is_up(l_ptr))
1446			return;
1447		if (l_ptr->next_out)
1448			next_sent = buf_seqno(l_ptr->next_out);
1449		msg_set_next_sent(msg, next_sent);
1450		if (l_ptr->oldest_deferred_in) {
1451			u32 rec = buf_seqno(l_ptr->oldest_deferred_in);
1452			gap = mod(rec - mod(l_ptr->next_in_no));
1453		}
1454		msg_set_seq_gap(msg, gap);
1455		if (gap)
1456			l_ptr->stats.sent_nacks++;
1457		msg_set_link_tolerance(msg, tolerance);
1458		msg_set_linkprio(msg, priority);
1459		msg_set_max_pkt(msg, ack_mtu);
1460		msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1461		msg_set_probe(msg, probe_msg != 0);
1462		if (probe_msg) {
1463			u32 mtu = l_ptr->max_pkt;
1464
1465			if ((mtu < l_ptr->max_pkt_target) &&
1466			    link_working_working(l_ptr) &&
1467			    l_ptr->fsm_msg_cnt) {
1468				msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1469				if (l_ptr->max_pkt_probes == 10) {
1470					l_ptr->max_pkt_target = (msg_size - 4);
1471					l_ptr->max_pkt_probes = 0;
1472					msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1473				}
1474				l_ptr->max_pkt_probes++;
1475			}
1476
1477			l_ptr->stats.sent_probes++;
1478		}
1479		l_ptr->stats.sent_states++;
1480	} else {		/* RESET_MSG or ACTIVATE_MSG */
1481		msg_set_ack(msg, mod(l_ptr->reset_checkpoint - 1));
1482		msg_set_seq_gap(msg, 0);
1483		msg_set_next_sent(msg, 1);
1484		msg_set_probe(msg, 0);
1485		msg_set_link_tolerance(msg, l_ptr->tolerance);
1486		msg_set_linkprio(msg, l_ptr->priority);
1487		msg_set_max_pkt(msg, l_ptr->max_pkt_target);
1488	}
1489
1490	r_flag = (l_ptr->owner->working_links > tipc_link_is_up(l_ptr));
1491	msg_set_redundant_link(msg, r_flag);
1492	msg_set_linkprio(msg, l_ptr->priority);
1493	msg_set_size(msg, msg_size);
1494
1495	msg_set_seqno(msg, mod(l_ptr->next_out_no + (0xffff/2)));
1496
1497	buf = tipc_buf_acquire(msg_size);
1498	if (!buf)
1499		return;
1500
1501	skb_copy_to_linear_data(buf, msg, sizeof(l_ptr->proto_msg));
1502	buf->priority = TC_PRIO_CONTROL;
1503
1504	tipc_bearer_send(l_ptr->bearer_id, buf, &l_ptr->media_addr);
1505	l_ptr->unacked_window = 0;
1506	kfree_skb(buf);
1507}
1508
1509/*
1510 * Receive protocol message :
1511 * Note that network plane id propagates through the network, and may
1512 * change at any time. The node with lowest address rules
1513 */
1514static void tipc_link_proto_rcv(struct tipc_link *l_ptr, struct sk_buff *buf)
1515{
1516	u32 rec_gap = 0;
1517	u32 max_pkt_info;
1518	u32 max_pkt_ack;
1519	u32 msg_tol;
1520	struct tipc_msg *msg = buf_msg(buf);
1521
1522	/* Discard protocol message during link changeover */
1523	if (l_ptr->exp_msg_count)
1524		goto exit;
1525
1526	if (l_ptr->net_plane != msg_net_plane(msg))
1527		if (tipc_own_addr > msg_prevnode(msg))
1528			l_ptr->net_plane = msg_net_plane(msg);
1529
1530	switch (msg_type(msg)) {
1531
1532	case RESET_MSG:
1533		if (!link_working_unknown(l_ptr) &&
1534		    (l_ptr->peer_session != INVALID_SESSION)) {
1535			if (less_eq(msg_session(msg), l_ptr->peer_session))
1536				break; /* duplicate or old reset: ignore */
1537		}
1538
1539		if (!msg_redundant_link(msg) && (link_working_working(l_ptr) ||
1540				link_working_unknown(l_ptr))) {
1541			/*
1542			 * peer has lost contact -- don't allow peer's links
1543			 * to reactivate before we recognize loss & clean up
1544			 */
1545			l_ptr->owner->action_flags |= TIPC_WAIT_OWN_LINKS_DOWN;
1546		}
1547
1548		link_state_event(l_ptr, RESET_MSG);
1549
1550		/* fall thru' */
1551	case ACTIVATE_MSG:
1552		/* Update link settings according other endpoint's values */
1553		strcpy((strrchr(l_ptr->name, ':') + 1), (char *)msg_data(msg));
1554
1555		msg_tol = msg_link_tolerance(msg);
1556		if (msg_tol > l_ptr->tolerance)
1557			link_set_supervision_props(l_ptr, msg_tol);
1558
1559		if (msg_linkprio(msg) > l_ptr->priority)
1560			l_ptr->priority = msg_linkprio(msg);
1561
1562		max_pkt_info = msg_max_pkt(msg);
1563		if (max_pkt_info) {
1564			if (max_pkt_info < l_ptr->max_pkt_target)
1565				l_ptr->max_pkt_target = max_pkt_info;
1566			if (l_ptr->max_pkt > l_ptr->max_pkt_target)
1567				l_ptr->max_pkt = l_ptr->max_pkt_target;
1568		} else {
1569			l_ptr->max_pkt = l_ptr->max_pkt_target;
1570		}
1571
1572		/* Synchronize broadcast link info, if not done previously */
1573		if (!tipc_node_is_up(l_ptr->owner)) {
1574			l_ptr->owner->bclink.last_sent =
1575				l_ptr->owner->bclink.last_in =
1576				msg_last_bcast(msg);
1577			l_ptr->owner->bclink.oos_state = 0;
1578		}
1579
1580		l_ptr->peer_session = msg_session(msg);
1581		l_ptr->peer_bearer_id = msg_bearer_id(msg);
1582
1583		if (msg_type(msg) == ACTIVATE_MSG)
1584			link_state_event(l_ptr, ACTIVATE_MSG);
1585		break;
1586	case STATE_MSG:
1587
1588		msg_tol = msg_link_tolerance(msg);
1589		if (msg_tol)
1590			link_set_supervision_props(l_ptr, msg_tol);
1591
1592		if (msg_linkprio(msg) &&
1593		    (msg_linkprio(msg) != l_ptr->priority)) {
1594			pr_warn("%s<%s>, priority change %u->%u\n",
1595				link_rst_msg, l_ptr->name, l_ptr->priority,
1596				msg_linkprio(msg));
1597			l_ptr->priority = msg_linkprio(msg);
1598			tipc_link_reset(l_ptr); /* Enforce change to take effect */
1599			break;
1600		}
1601
1602		/* Record reception; force mismatch at next timeout: */
1603		l_ptr->checkpoint--;
1604
1605		link_state_event(l_ptr, TRAFFIC_MSG_EVT);
1606		l_ptr->stats.recv_states++;
1607		if (link_reset_unknown(l_ptr))
1608			break;
1609
1610		if (less_eq(mod(l_ptr->next_in_no), msg_next_sent(msg))) {
1611			rec_gap = mod(msg_next_sent(msg) -
1612				      mod(l_ptr->next_in_no));
1613		}
1614
1615		max_pkt_ack = msg_max_pkt(msg);
1616		if (max_pkt_ack > l_ptr->max_pkt) {
1617			l_ptr->max_pkt = max_pkt_ack;
1618			l_ptr->max_pkt_probes = 0;
1619		}
1620
1621		max_pkt_ack = 0;
1622		if (msg_probe(msg)) {
1623			l_ptr->stats.recv_probes++;
1624			if (msg_size(msg) > sizeof(l_ptr->proto_msg))
1625				max_pkt_ack = msg_size(msg);
1626		}
1627
1628		/* Protocol message before retransmits, reduce loss risk */
1629		if (l_ptr->owner->bclink.recv_permitted)
1630			tipc_bclink_update_link_state(l_ptr->owner,
1631						      msg_last_bcast(msg));
1632
1633		if (rec_gap || (msg_probe(msg))) {
1634			tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, rec_gap, 0,
1635					     0, max_pkt_ack);
1636		}
1637		if (msg_seq_gap(msg)) {
1638			l_ptr->stats.recv_nacks++;
1639			tipc_link_retransmit(l_ptr, l_ptr->first_out,
1640					     msg_seq_gap(msg));
1641		}
1642		break;
1643	}
1644exit:
1645	kfree_skb(buf);
1646}
1647
1648
1649/* tipc_link_tunnel_xmit(): Tunnel one packet via a link belonging to
1650 * a different bearer. Owner node is locked.
1651 */
1652static void tipc_link_tunnel_xmit(struct tipc_link *l_ptr,
1653				  struct tipc_msg *tunnel_hdr,
1654				  struct tipc_msg *msg,
1655				  u32 selector)
1656{
1657	struct tipc_link *tunnel;
1658	struct sk_buff *buf;
1659	u32 length = msg_size(msg);
1660
1661	tunnel = l_ptr->owner->active_links[selector & 1];
1662	if (!tipc_link_is_up(tunnel)) {
1663		pr_warn("%stunnel link no longer available\n", link_co_err);
1664		return;
1665	}
1666	msg_set_size(tunnel_hdr, length + INT_H_SIZE);
1667	buf = tipc_buf_acquire(length + INT_H_SIZE);
1668	if (!buf) {
1669		pr_warn("%sunable to send tunnel msg\n", link_co_err);
1670		return;
1671	}
1672	skb_copy_to_linear_data(buf, tunnel_hdr, INT_H_SIZE);
1673	skb_copy_to_linear_data_offset(buf, INT_H_SIZE, msg, length);
1674	__tipc_link_xmit(tunnel, buf);
1675}
1676
1677
1678/* tipc_link_failover_send_queue(): A link has gone down, but a second
1679 * link is still active. We can do failover. Tunnel the failing link's
1680 * whole send queue via the remaining link. This way, we don't lose
1681 * any packets, and sequence order is preserved for subsequent traffic
1682 * sent over the remaining link. Owner node is locked.
1683 */
1684void tipc_link_failover_send_queue(struct tipc_link *l_ptr)
1685{
1686	u32 msgcount = l_ptr->out_queue_size;
1687	struct sk_buff *crs = l_ptr->first_out;
1688	struct tipc_link *tunnel = l_ptr->owner->active_links[0];
1689	struct tipc_msg tunnel_hdr;
1690	int split_bundles;
1691
1692	if (!tunnel)
1693		return;
1694
1695	tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
1696		 ORIGINAL_MSG, INT_H_SIZE, l_ptr->addr);
1697	msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
1698	msg_set_msgcnt(&tunnel_hdr, msgcount);
1699
1700	if (!l_ptr->first_out) {
1701		struct sk_buff *buf;
1702
1703		buf = tipc_buf_acquire(INT_H_SIZE);
1704		if (buf) {
1705			skb_copy_to_linear_data(buf, &tunnel_hdr, INT_H_SIZE);
1706			msg_set_size(&tunnel_hdr, INT_H_SIZE);
1707			__tipc_link_xmit(tunnel, buf);
1708		} else {
1709			pr_warn("%sunable to send changeover msg\n",
1710				link_co_err);
1711		}
1712		return;
1713	}
1714
1715	split_bundles = (l_ptr->owner->active_links[0] !=
1716			 l_ptr->owner->active_links[1]);
1717
1718	while (crs) {
1719		struct tipc_msg *msg = buf_msg(crs);
1720
1721		if ((msg_user(msg) == MSG_BUNDLER) && split_bundles) {
1722			struct tipc_msg *m = msg_get_wrapped(msg);
1723			unchar *pos = (unchar *)m;
1724
1725			msgcount = msg_msgcnt(msg);
1726			while (msgcount--) {
1727				msg_set_seqno(m, msg_seqno(msg));
1728				tipc_link_tunnel_xmit(l_ptr, &tunnel_hdr, m,
1729						      msg_link_selector(m));
1730				pos += align(msg_size(m));
1731				m = (struct tipc_msg *)pos;
1732			}
1733		} else {
1734			tipc_link_tunnel_xmit(l_ptr, &tunnel_hdr, msg,
1735					      msg_link_selector(msg));
1736		}
1737		crs = crs->next;
1738	}
1739}
1740
1741/* tipc_link_dup_queue_xmit(): A second link has become active. Tunnel a
1742 * duplicate of the first link's send queue via the new link. This way, we
1743 * are guaranteed that currently queued packets from a socket are delivered
1744 * before future traffic from the same socket, even if this is using the
1745 * new link. The last arriving copy of each duplicate packet is dropped at
1746 * the receiving end by the regular protocol check, so packet cardinality
1747 * and sequence order is preserved per sender/receiver socket pair.
1748 * Owner node is locked.
1749 */
1750void tipc_link_dup_queue_xmit(struct tipc_link *l_ptr,
1751			      struct tipc_link *tunnel)
1752{
1753	struct sk_buff *iter;
1754	struct tipc_msg tunnel_hdr;
1755
1756	tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
1757		 DUPLICATE_MSG, INT_H_SIZE, l_ptr->addr);
1758	msg_set_msgcnt(&tunnel_hdr, l_ptr->out_queue_size);
1759	msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
1760	iter = l_ptr->first_out;
1761	while (iter) {
1762		struct sk_buff *outbuf;
1763		struct tipc_msg *msg = buf_msg(iter);
1764		u32 length = msg_size(msg);
1765
1766		if (msg_user(msg) == MSG_BUNDLER)
1767			msg_set_type(msg, CLOSED_MSG);
1768		msg_set_ack(msg, mod(l_ptr->next_in_no - 1));	/* Update */
1769		msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1770		msg_set_size(&tunnel_hdr, length + INT_H_SIZE);
1771		outbuf = tipc_buf_acquire(length + INT_H_SIZE);
1772		if (outbuf == NULL) {
1773			pr_warn("%sunable to send duplicate msg\n",
1774				link_co_err);
1775			return;
1776		}
1777		skb_copy_to_linear_data(outbuf, &tunnel_hdr, INT_H_SIZE);
1778		skb_copy_to_linear_data_offset(outbuf, INT_H_SIZE, iter->data,
1779					       length);
1780		__tipc_link_xmit(tunnel, outbuf);
1781		if (!tipc_link_is_up(l_ptr))
1782			return;
1783		iter = iter->next;
1784	}
1785}
1786
1787/**
1788 * buf_extract - extracts embedded TIPC message from another message
1789 * @skb: encapsulating message buffer
1790 * @from_pos: offset to extract from
1791 *
1792 * Returns a new message buffer containing an embedded message.  The
1793 * encapsulating message itself is left unchanged.
1794 */
1795static struct sk_buff *buf_extract(struct sk_buff *skb, u32 from_pos)
1796{
1797	struct tipc_msg *msg = (struct tipc_msg *)(skb->data + from_pos);
1798	u32 size = msg_size(msg);
1799	struct sk_buff *eb;
1800
1801	eb = tipc_buf_acquire(size);
1802	if (eb)
1803		skb_copy_to_linear_data(eb, msg, size);
1804	return eb;
1805}
1806
1807
1808
1809/* tipc_link_dup_rcv(): Receive a tunnelled DUPLICATE_MSG packet.
1810 * Owner node is locked.
1811 */
1812static void tipc_link_dup_rcv(struct tipc_link *l_ptr,
1813			      struct sk_buff *t_buf)
1814{
1815	struct sk_buff *buf;
1816
1817	if (!tipc_link_is_up(l_ptr))
1818		return;
1819
1820	buf = buf_extract(t_buf, INT_H_SIZE);
1821	if (buf == NULL) {
1822		pr_warn("%sfailed to extract inner dup pkt\n", link_co_err);
1823		return;
1824	}
1825
1826	/* Add buffer to deferred queue, if applicable: */
1827	link_handle_out_of_seq_msg(l_ptr, buf);
1828}
1829
1830/*  tipc_link_failover_rcv(): Receive a tunnelled ORIGINAL_MSG packet
1831 *  Owner node is locked.
1832 */
1833static struct sk_buff *tipc_link_failover_rcv(struct tipc_link *l_ptr,
1834					      struct sk_buff *t_buf)
1835{
1836	struct tipc_msg *t_msg = buf_msg(t_buf);
1837	struct sk_buff *buf = NULL;
1838	struct tipc_msg *msg;
1839
1840	if (tipc_link_is_up(l_ptr))
1841		tipc_link_reset(l_ptr);
1842
1843	/* First failover packet? */
1844	if (l_ptr->exp_msg_count == START_CHANGEOVER)
1845		l_ptr->exp_msg_count = msg_msgcnt(t_msg);
1846
1847	/* Should there be an inner packet? */
1848	if (l_ptr->exp_msg_count) {
1849		l_ptr->exp_msg_count--;
1850		buf = buf_extract(t_buf, INT_H_SIZE);
1851		if (buf == NULL) {
1852			pr_warn("%sno inner failover pkt\n", link_co_err);
1853			goto exit;
1854		}
1855		msg = buf_msg(buf);
1856
1857		if (less(msg_seqno(msg), l_ptr->reset_checkpoint)) {
1858			kfree_skb(buf);
1859			buf = NULL;
1860			goto exit;
1861		}
1862		if (msg_user(msg) == MSG_FRAGMENTER) {
1863			l_ptr->stats.recv_fragments++;
1864			tipc_buf_append(&l_ptr->reasm_buf, &buf);
1865		}
1866	}
1867exit:
1868	if ((l_ptr->exp_msg_count == 0) && (l_ptr->flags & LINK_STOPPED)) {
1869		tipc_node_detach_link(l_ptr->owner, l_ptr);
1870		kfree(l_ptr);
1871	}
1872	return buf;
1873}
1874
1875/*  tipc_link_tunnel_rcv(): Receive a tunnelled packet, sent
1876 *  via other link as result of a failover (ORIGINAL_MSG) or
1877 *  a new active link (DUPLICATE_MSG). Failover packets are
1878 *  returned to the active link for delivery upwards.
1879 *  Owner node is locked.
1880 */
1881static int tipc_link_tunnel_rcv(struct tipc_node *n_ptr,
1882				struct sk_buff **buf)
1883{
1884	struct sk_buff *t_buf = *buf;
1885	struct tipc_link *l_ptr;
1886	struct tipc_msg *t_msg = buf_msg(t_buf);
1887	u32 bearer_id = msg_bearer_id(t_msg);
1888
1889	*buf = NULL;
1890
1891	if (bearer_id >= MAX_BEARERS)
1892		goto exit;
1893
1894	l_ptr = n_ptr->links[bearer_id];
1895	if (!l_ptr)
1896		goto exit;
1897
1898	if (msg_type(t_msg) == DUPLICATE_MSG)
1899		tipc_link_dup_rcv(l_ptr, t_buf);
1900	else if (msg_type(t_msg) == ORIGINAL_MSG)
1901		*buf = tipc_link_failover_rcv(l_ptr, t_buf);
1902	else
1903		pr_warn("%sunknown tunnel pkt received\n", link_co_err);
1904exit:
1905	kfree_skb(t_buf);
1906	return *buf != NULL;
1907}
1908
1909/*
1910 *  Bundler functionality:
1911 */
1912void tipc_link_bundle_rcv(struct sk_buff *buf)
1913{
1914	u32 msgcount = msg_msgcnt(buf_msg(buf));
1915	u32 pos = INT_H_SIZE;
1916	struct sk_buff *obuf;
1917	struct tipc_msg *omsg;
1918
1919	while (msgcount--) {
1920		obuf = buf_extract(buf, pos);
1921		if (obuf == NULL) {
1922			pr_warn("Link unable to unbundle message(s)\n");
1923			break;
1924		}
1925		omsg = buf_msg(obuf);
1926		pos += align(msg_size(omsg));
1927		if (msg_isdata(omsg)) {
1928			if (unlikely(msg_type(omsg) == TIPC_MCAST_MSG))
1929				tipc_sk_mcast_rcv(obuf);
1930			else
1931				tipc_sk_rcv(obuf);
1932		} else if (msg_user(omsg) == CONN_MANAGER) {
1933			tipc_sk_rcv(obuf);
1934		} else if (msg_user(omsg) == NAME_DISTRIBUTOR) {
1935			tipc_named_rcv(obuf);
1936		} else {
1937			pr_warn("Illegal bundled msg: %u\n", msg_user(omsg));
1938			kfree_skb(obuf);
1939		}
1940	}
1941	kfree_skb(buf);
1942}
1943
1944static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance)
1945{
1946	if ((tolerance < TIPC_MIN_LINK_TOL) || (tolerance > TIPC_MAX_LINK_TOL))
1947		return;
1948
1949	l_ptr->tolerance = tolerance;
1950	l_ptr->continuity_interval =
1951		((tolerance / 4) > 500) ? 500 : tolerance / 4;
1952	l_ptr->abort_limit = tolerance / (l_ptr->continuity_interval / 4);
1953}
1954
1955void tipc_link_set_queue_limits(struct tipc_link *l_ptr, u32 window)
1956{
1957	/* Data messages from this node, inclusive FIRST_FRAGM */
1958	l_ptr->queue_limit[TIPC_LOW_IMPORTANCE] = window;
1959	l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE] = (window / 3) * 4;
1960	l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE] = (window / 3) * 5;
1961	l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE] = (window / 3) * 6;
1962	/* Transiting data messages,inclusive FIRST_FRAGM */
1963	l_ptr->queue_limit[TIPC_LOW_IMPORTANCE + 4] = 300;
1964	l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE + 4] = 600;
1965	l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE + 4] = 900;
1966	l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE + 4] = 1200;
1967	l_ptr->queue_limit[CONN_MANAGER] = 1200;
1968	l_ptr->queue_limit[CHANGEOVER_PROTOCOL] = 2500;
1969	l_ptr->queue_limit[NAME_DISTRIBUTOR] = 3000;
1970	/* FRAGMENT and LAST_FRAGMENT packets */
1971	l_ptr->queue_limit[MSG_FRAGMENTER] = 4000;
1972}
1973
1974/* tipc_link_find_owner - locate owner node of link by link's name
1975 * @name: pointer to link name string
1976 * @bearer_id: pointer to index in 'node->links' array where the link was found.
1977 *
1978 * Returns pointer to node owning the link, or 0 if no matching link is found.
1979 */
1980static struct tipc_node *tipc_link_find_owner(const char *link_name,
1981					      unsigned int *bearer_id)
1982{
1983	struct tipc_link *l_ptr;
1984	struct tipc_node *n_ptr;
1985	struct tipc_node *found_node = 0;
1986	int i;
1987
1988	*bearer_id = 0;
1989	rcu_read_lock();
1990	list_for_each_entry_rcu(n_ptr, &tipc_node_list, list) {
1991		tipc_node_lock(n_ptr);
1992		for (i = 0; i < MAX_BEARERS; i++) {
1993			l_ptr = n_ptr->links[i];
1994			if (l_ptr && !strcmp(l_ptr->name, link_name)) {
1995				*bearer_id = i;
1996				found_node = n_ptr;
1997				break;
1998			}
1999		}
2000		tipc_node_unlock(n_ptr);
2001		if (found_node)
2002			break;
2003	}
2004	rcu_read_unlock();
2005
2006	return found_node;
2007}
2008
2009/**
2010 * link_value_is_valid -- validate proposed link tolerance/priority/window
2011 *
2012 * @cmd: value type (TIPC_CMD_SET_LINK_*)
2013 * @new_value: the new value
2014 *
2015 * Returns 1 if value is within range, 0 if not.
2016 */
2017static int link_value_is_valid(u16 cmd, u32 new_value)
2018{
2019	switch (cmd) {
2020	case TIPC_CMD_SET_LINK_TOL:
2021		return (new_value >= TIPC_MIN_LINK_TOL) &&
2022			(new_value <= TIPC_MAX_LINK_TOL);
2023	case TIPC_CMD_SET_LINK_PRI:
2024		return (new_value <= TIPC_MAX_LINK_PRI);
2025	case TIPC_CMD_SET_LINK_WINDOW:
2026		return (new_value >= TIPC_MIN_LINK_WIN) &&
2027			(new_value <= TIPC_MAX_LINK_WIN);
2028	}
2029	return 0;
2030}
2031
2032/**
2033 * link_cmd_set_value - change priority/tolerance/window for link/bearer/media
2034 * @name: ptr to link, bearer, or media name
2035 * @new_value: new value of link, bearer, or media setting
2036 * @cmd: which link, bearer, or media attribute to set (TIPC_CMD_SET_LINK_*)
2037 *
2038 * Caller must hold RTNL lock to ensure link/bearer/media is not deleted.
2039 *
2040 * Returns 0 if value updated and negative value on error.
2041 */
2042static int link_cmd_set_value(const char *name, u32 new_value, u16 cmd)
2043{
2044	struct tipc_node *node;
2045	struct tipc_link *l_ptr;
2046	struct tipc_bearer *b_ptr;
2047	struct tipc_media *m_ptr;
2048	int bearer_id;
2049	int res = 0;
2050
2051	node = tipc_link_find_owner(name, &bearer_id);
2052	if (node) {
2053		tipc_node_lock(node);
2054		l_ptr = node->links[bearer_id];
2055
2056		if (l_ptr) {
2057			switch (cmd) {
2058			case TIPC_CMD_SET_LINK_TOL:
2059				link_set_supervision_props(l_ptr, new_value);
2060				tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0,
2061						     new_value, 0, 0);
2062				break;
2063			case TIPC_CMD_SET_LINK_PRI:
2064				l_ptr->priority = new_value;
2065				tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0,
2066						     0, new_value, 0);
2067				break;
2068			case TIPC_CMD_SET_LINK_WINDOW:
2069				tipc_link_set_queue_limits(l_ptr, new_value);
2070				break;
2071			default:
2072				res = -EINVAL;
2073				break;
2074			}
2075		}
2076		tipc_node_unlock(node);
2077		return res;
2078	}
2079
2080	b_ptr = tipc_bearer_find(name);
2081	if (b_ptr) {
2082		switch (cmd) {
2083		case TIPC_CMD_SET_LINK_TOL:
2084			b_ptr->tolerance = new_value;
2085			break;
2086		case TIPC_CMD_SET_LINK_PRI:
2087			b_ptr->priority = new_value;
2088			break;
2089		case TIPC_CMD_SET_LINK_WINDOW:
2090			b_ptr->window = new_value;
2091			break;
2092		default:
2093			res = -EINVAL;
2094			break;
2095		}
2096		return res;
2097	}
2098
2099	m_ptr = tipc_media_find(name);
2100	if (!m_ptr)
2101		return -ENODEV;
2102	switch (cmd) {
2103	case TIPC_CMD_SET_LINK_TOL:
2104		m_ptr->tolerance = new_value;
2105		break;
2106	case TIPC_CMD_SET_LINK_PRI:
2107		m_ptr->priority = new_value;
2108		break;
2109	case TIPC_CMD_SET_LINK_WINDOW:
2110		m_ptr->window = new_value;
2111		break;
2112	default:
2113		res = -EINVAL;
2114		break;
2115	}
2116	return res;
2117}
2118
2119struct sk_buff *tipc_link_cmd_config(const void *req_tlv_area, int req_tlv_space,
2120				     u16 cmd)
2121{
2122	struct tipc_link_config *args;
2123	u32 new_value;
2124	int res;
2125
2126	if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_CONFIG))
2127		return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2128
2129	args = (struct tipc_link_config *)TLV_DATA(req_tlv_area);
2130	new_value = ntohl(args->value);
2131
2132	if (!link_value_is_valid(cmd, new_value))
2133		return tipc_cfg_reply_error_string(
2134			"cannot change, value invalid");
2135
2136	if (!strcmp(args->name, tipc_bclink_name)) {
2137		if ((cmd == TIPC_CMD_SET_LINK_WINDOW) &&
2138		    (tipc_bclink_set_queue_limits(new_value) == 0))
2139			return tipc_cfg_reply_none();
2140		return tipc_cfg_reply_error_string(TIPC_CFG_NOT_SUPPORTED
2141						   " (cannot change setting on broadcast link)");
2142	}
2143
2144	res = link_cmd_set_value(args->name, new_value, cmd);
2145	if (res)
2146		return tipc_cfg_reply_error_string("cannot change link setting");
2147
2148	return tipc_cfg_reply_none();
2149}
2150
2151/**
2152 * link_reset_statistics - reset link statistics
2153 * @l_ptr: pointer to link
2154 */
2155static void link_reset_statistics(struct tipc_link *l_ptr)
2156{
2157	memset(&l_ptr->stats, 0, sizeof(l_ptr->stats));
2158	l_ptr->stats.sent_info = l_ptr->next_out_no;
2159	l_ptr->stats.recv_info = l_ptr->next_in_no;
2160}
2161
2162struct sk_buff *tipc_link_cmd_reset_stats(const void *req_tlv_area, int req_tlv_space)
2163{
2164	char *link_name;
2165	struct tipc_link *l_ptr;
2166	struct tipc_node *node;
2167	unsigned int bearer_id;
2168
2169	if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2170		return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2171
2172	link_name = (char *)TLV_DATA(req_tlv_area);
2173	if (!strcmp(link_name, tipc_bclink_name)) {
2174		if (tipc_bclink_reset_stats())
2175			return tipc_cfg_reply_error_string("link not found");
2176		return tipc_cfg_reply_none();
2177	}
2178	node = tipc_link_find_owner(link_name, &bearer_id);
2179	if (!node)
2180		return tipc_cfg_reply_error_string("link not found");
2181
2182	tipc_node_lock(node);
2183	l_ptr = node->links[bearer_id];
2184	if (!l_ptr) {
2185		tipc_node_unlock(node);
2186		return tipc_cfg_reply_error_string("link not found");
2187	}
2188	link_reset_statistics(l_ptr);
2189	tipc_node_unlock(node);
2190	return tipc_cfg_reply_none();
2191}
2192
2193/**
2194 * percent - convert count to a percentage of total (rounding up or down)
2195 */
2196static u32 percent(u32 count, u32 total)
2197{
2198	return (count * 100 + (total / 2)) / total;
2199}
2200
2201/**
2202 * tipc_link_stats - print link statistics
2203 * @name: link name
2204 * @buf: print buffer area
2205 * @buf_size: size of print buffer area
2206 *
2207 * Returns length of print buffer data string (or 0 if error)
2208 */
2209static int tipc_link_stats(const char *name, char *buf, const u32 buf_size)
2210{
2211	struct tipc_link *l;
2212	struct tipc_stats *s;
2213	struct tipc_node *node;
2214	char *status;
2215	u32 profile_total = 0;
2216	unsigned int bearer_id;
2217	int ret;
2218
2219	if (!strcmp(name, tipc_bclink_name))
2220		return tipc_bclink_stats(buf, buf_size);
2221
2222	node = tipc_link_find_owner(name, &bearer_id);
2223	if (!node)
2224		return 0;
2225
2226	tipc_node_lock(node);
2227
2228	l = node->links[bearer_id];
2229	if (!l) {
2230		tipc_node_unlock(node);
2231		return 0;
2232	}
2233
2234	s = &l->stats;
2235
2236	if (tipc_link_is_active(l))
2237		status = "ACTIVE";
2238	else if (tipc_link_is_up(l))
2239		status = "STANDBY";
2240	else
2241		status = "DEFUNCT";
2242
2243	ret = tipc_snprintf(buf, buf_size, "Link <%s>\n"
2244			    "  %s  MTU:%u  Priority:%u  Tolerance:%u ms"
2245			    "  Window:%u packets\n",
2246			    l->name, status, l->max_pkt, l->priority,
2247			    l->tolerance, l->queue_limit[0]);
2248
2249	ret += tipc_snprintf(buf + ret, buf_size - ret,
2250			     "  RX packets:%u fragments:%u/%u bundles:%u/%u\n",
2251			     l->next_in_no - s->recv_info, s->recv_fragments,
2252			     s->recv_fragmented, s->recv_bundles,
2253			     s->recv_bundled);
2254
2255	ret += tipc_snprintf(buf + ret, buf_size - ret,
2256			     "  TX packets:%u fragments:%u/%u bundles:%u/%u\n",
2257			     l->next_out_no - s->sent_info, s->sent_fragments,
2258			     s->sent_fragmented, s->sent_bundles,
2259			     s->sent_bundled);
2260
2261	profile_total = s->msg_length_counts;
2262	if (!profile_total)
2263		profile_total = 1;
2264
2265	ret += tipc_snprintf(buf + ret, buf_size - ret,
2266			     "  TX profile sample:%u packets  average:%u octets\n"
2267			     "  0-64:%u%% -256:%u%% -1024:%u%% -4096:%u%% "
2268			     "-16384:%u%% -32768:%u%% -66000:%u%%\n",
2269			     s->msg_length_counts,
2270			     s->msg_lengths_total / profile_total,
2271			     percent(s->msg_length_profile[0], profile_total),
2272			     percent(s->msg_length_profile[1], profile_total),
2273			     percent(s->msg_length_profile[2], profile_total),
2274			     percent(s->msg_length_profile[3], profile_total),
2275			     percent(s->msg_length_profile[4], profile_total),
2276			     percent(s->msg_length_profile[5], profile_total),
2277			     percent(s->msg_length_profile[6], profile_total));
2278
2279	ret += tipc_snprintf(buf + ret, buf_size - ret,
2280			     "  RX states:%u probes:%u naks:%u defs:%u"
2281			     " dups:%u\n", s->recv_states, s->recv_probes,
2282			     s->recv_nacks, s->deferred_recv, s->duplicates);
2283
2284	ret += tipc_snprintf(buf + ret, buf_size - ret,
2285			     "  TX states:%u probes:%u naks:%u acks:%u"
2286			     " dups:%u\n", s->sent_states, s->sent_probes,
2287			     s->sent_nacks, s->sent_acks, s->retransmitted);
2288
2289	ret += tipc_snprintf(buf + ret, buf_size - ret,
2290			     "  Congestion link:%u  Send queue"
2291			     " max:%u avg:%u\n", s->link_congs,
2292			     s->max_queue_sz, s->queue_sz_counts ?
2293			     (s->accu_queue_sz / s->queue_sz_counts) : 0);
2294
2295	tipc_node_unlock(node);
2296	return ret;
2297}
2298
2299struct sk_buff *tipc_link_cmd_show_stats(const void *req_tlv_area, int req_tlv_space)
2300{
2301	struct sk_buff *buf;
2302	struct tlv_desc *rep_tlv;
2303	int str_len;
2304	int pb_len;
2305	char *pb;
2306
2307	if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2308		return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2309
2310	buf = tipc_cfg_reply_alloc(TLV_SPACE(ULTRA_STRING_MAX_LEN));
2311	if (!buf)
2312		return NULL;
2313
2314	rep_tlv = (struct tlv_desc *)buf->data;
2315	pb = TLV_DATA(rep_tlv);
2316	pb_len = ULTRA_STRING_MAX_LEN;
2317	str_len = tipc_link_stats((char *)TLV_DATA(req_tlv_area),
2318				  pb, pb_len);
2319	if (!str_len) {
2320		kfree_skb(buf);
2321		return tipc_cfg_reply_error_string("link not found");
2322	}
2323	str_len += 1;	/* for "\0" */
2324	skb_put(buf, TLV_SPACE(str_len));
2325	TLV_SET(rep_tlv, TIPC_TLV_ULTRA_STRING, NULL, str_len);
2326
2327	return buf;
2328}
2329
2330/**
2331 * tipc_link_get_max_pkt - get maximum packet size to use when sending to destination
2332 * @dest: network address of destination node
2333 * @selector: used to select from set of active links
2334 *
2335 * If no active link can be found, uses default maximum packet size.
2336 */
2337u32 tipc_link_get_max_pkt(u32 dest, u32 selector)
2338{
2339	struct tipc_node *n_ptr;
2340	struct tipc_link *l_ptr;
2341	u32 res = MAX_PKT_DEFAULT;
2342
2343	if (dest == tipc_own_addr)
2344		return MAX_MSG_SIZE;
2345
2346	n_ptr = tipc_node_find(dest);
2347	if (n_ptr) {
2348		tipc_node_lock(n_ptr);
2349		l_ptr = n_ptr->active_links[selector & 1];
2350		if (l_ptr)
2351			res = l_ptr->max_pkt;
2352		tipc_node_unlock(n_ptr);
2353	}
2354	return res;
2355}
2356
2357static void link_print(struct tipc_link *l_ptr, const char *str)
2358{
2359	struct tipc_bearer *b_ptr;
2360
2361	rcu_read_lock();
2362	b_ptr = rcu_dereference_rtnl(bearer_list[l_ptr->bearer_id]);
2363	if (b_ptr)
2364		pr_info("%s Link %x<%s>:", str, l_ptr->addr, b_ptr->name);
2365	rcu_read_unlock();
2366
2367	if (link_working_unknown(l_ptr))
2368		pr_cont(":WU\n");
2369	else if (link_reset_reset(l_ptr))
2370		pr_cont(":RR\n");
2371	else if (link_reset_unknown(l_ptr))
2372		pr_cont(":RU\n");
2373	else if (link_working_working(l_ptr))
2374		pr_cont(":WW\n");
2375	else
2376		pr_cont("\n");
2377}
2378