2 * net/tipc/link.c: TIPC link code
4 * Copyright (c) 1996-2007, 2012-2014, Ericsson AB
5 * Copyright (c) 2004-2007, 2010-2013, Wind River Systems
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions are met:
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.
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.
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.
40 #include "name_distr.h"
44 #include <linux/pkt_sched.h>
47 * Error message prefixes
49 static const char *link_co_err = "Link changeover error, ";
50 static const char *link_rst_msg = "Resetting link ";
51 static const char *link_unk_evt = "Unknown link event ";
54 * Out-of-range value for link session numbers
56 #define INVALID_SESSION 0x10000
61 #define STARTING_EVT 856384768 /* link processing trigger */
62 #define TRAFFIC_MSG_EVT 560815u /* rx'd ??? */
63 #define TIMEOUT_EVT 560817u /* link timer expired */
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
74 * State value stored in 'exp_msg_count'
76 #define START_CHANGEOVER 100000u
78 static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
80 static void tipc_link_proto_rcv(struct tipc_link *l_ptr, struct sk_buff *buf);
81 static int tipc_link_tunnel_rcv(struct tipc_node *n_ptr,
82 struct sk_buff **buf);
83 static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance);
84 static int tipc_link_iovec_long_xmit(struct tipc_port *sender,
85 struct iovec const *msg_sect,
86 unsigned int len, u32 destnode);
87 static void link_state_event(struct tipc_link *l_ptr, u32 event);
88 static void link_reset_statistics(struct tipc_link *l_ptr);
89 static void link_print(struct tipc_link *l_ptr, const char *str);
90 static int tipc_link_frag_xmit(struct tipc_link *l_ptr, struct sk_buff *buf);
91 static void tipc_link_sync_xmit(struct tipc_link *l);
92 static void tipc_link_sync_rcv(struct tipc_node *n, struct sk_buff *buf);
95 * Simple link routines
97 static unsigned int align(unsigned int i)
102 static void link_init_max_pkt(struct tipc_link *l_ptr)
106 max_pkt = (l_ptr->b_ptr->mtu & ~3);
107 if (max_pkt > MAX_MSG_SIZE)
108 max_pkt = MAX_MSG_SIZE;
110 l_ptr->max_pkt_target = max_pkt;
111 if (l_ptr->max_pkt_target < MAX_PKT_DEFAULT)
112 l_ptr->max_pkt = l_ptr->max_pkt_target;
114 l_ptr->max_pkt = MAX_PKT_DEFAULT;
116 l_ptr->max_pkt_probes = 0;
119 static u32 link_next_sent(struct tipc_link *l_ptr)
122 return buf_seqno(l_ptr->next_out);
123 return mod(l_ptr->next_out_no);
126 static u32 link_last_sent(struct tipc_link *l_ptr)
128 return mod(link_next_sent(l_ptr) - 1);
132 * Simple non-static link routines (i.e. referenced outside this file)
134 int tipc_link_is_up(struct tipc_link *l_ptr)
138 return link_working_working(l_ptr) || link_working_unknown(l_ptr);
141 int tipc_link_is_active(struct tipc_link *l_ptr)
143 return (l_ptr->owner->active_links[0] == l_ptr) ||
144 (l_ptr->owner->active_links[1] == l_ptr);
148 * link_timeout - handle expiration of link timer
149 * @l_ptr: pointer to link
151 static void link_timeout(struct tipc_link *l_ptr)
153 tipc_node_lock(l_ptr->owner);
155 /* update counters used in statistical profiling of send traffic */
156 l_ptr->stats.accu_queue_sz += l_ptr->out_queue_size;
157 l_ptr->stats.queue_sz_counts++;
159 if (l_ptr->first_out) {
160 struct tipc_msg *msg = buf_msg(l_ptr->first_out);
161 u32 length = msg_size(msg);
163 if ((msg_user(msg) == MSG_FRAGMENTER) &&
164 (msg_type(msg) == FIRST_FRAGMENT)) {
165 length = msg_size(msg_get_wrapped(msg));
168 l_ptr->stats.msg_lengths_total += length;
169 l_ptr->stats.msg_length_counts++;
171 l_ptr->stats.msg_length_profile[0]++;
172 else if (length <= 256)
173 l_ptr->stats.msg_length_profile[1]++;
174 else if (length <= 1024)
175 l_ptr->stats.msg_length_profile[2]++;
176 else if (length <= 4096)
177 l_ptr->stats.msg_length_profile[3]++;
178 else if (length <= 16384)
179 l_ptr->stats.msg_length_profile[4]++;
180 else if (length <= 32768)
181 l_ptr->stats.msg_length_profile[5]++;
183 l_ptr->stats.msg_length_profile[6]++;
187 /* do all other link processing performed on a periodic basis */
189 link_state_event(l_ptr, TIMEOUT_EVT);
192 tipc_link_push_queue(l_ptr);
194 tipc_node_unlock(l_ptr->owner);
197 static void link_set_timer(struct tipc_link *l_ptr, u32 time)
199 k_start_timer(&l_ptr->timer, time);
203 * tipc_link_create - create a new link
204 * @n_ptr: pointer to associated node
205 * @b_ptr: pointer to associated bearer
206 * @media_addr: media address to use when sending messages over link
208 * Returns pointer to link.
210 struct tipc_link *tipc_link_create(struct tipc_node *n_ptr,
211 struct tipc_bearer *b_ptr,
212 const struct tipc_media_addr *media_addr)
214 struct tipc_link *l_ptr;
215 struct tipc_msg *msg;
217 char addr_string[16];
218 u32 peer = n_ptr->addr;
220 if (n_ptr->link_cnt >= 2) {
221 tipc_addr_string_fill(addr_string, n_ptr->addr);
222 pr_err("Attempt to establish third link to %s\n", addr_string);
226 if (n_ptr->links[b_ptr->identity]) {
227 tipc_addr_string_fill(addr_string, n_ptr->addr);
228 pr_err("Attempt to establish second link on <%s> to %s\n",
229 b_ptr->name, addr_string);
233 l_ptr = kzalloc(sizeof(*l_ptr), GFP_ATOMIC);
235 pr_warn("Link creation failed, no memory\n");
240 if_name = strchr(b_ptr->name, ':') + 1;
241 sprintf(l_ptr->name, "%u.%u.%u:%s-%u.%u.%u:unknown",
242 tipc_zone(tipc_own_addr), tipc_cluster(tipc_own_addr),
243 tipc_node(tipc_own_addr),
245 tipc_zone(peer), tipc_cluster(peer), tipc_node(peer));
246 /* note: peer i/f name is updated by reset/activate message */
247 memcpy(&l_ptr->media_addr, media_addr, sizeof(*media_addr));
248 l_ptr->owner = n_ptr;
249 l_ptr->checkpoint = 1;
250 l_ptr->peer_session = INVALID_SESSION;
251 l_ptr->b_ptr = b_ptr;
252 link_set_supervision_props(l_ptr, b_ptr->tolerance);
253 l_ptr->state = RESET_UNKNOWN;
255 l_ptr->pmsg = (struct tipc_msg *)&l_ptr->proto_msg;
257 tipc_msg_init(msg, LINK_PROTOCOL, RESET_MSG, INT_H_SIZE, l_ptr->addr);
258 msg_set_size(msg, sizeof(l_ptr->proto_msg));
259 msg_set_session(msg, (tipc_random & 0xffff));
260 msg_set_bearer_id(msg, b_ptr->identity);
261 strcpy((char *)msg_data(msg), if_name);
263 l_ptr->priority = b_ptr->priority;
264 tipc_link_set_queue_limits(l_ptr, b_ptr->window);
266 link_init_max_pkt(l_ptr);
268 l_ptr->next_out_no = 1;
269 INIT_LIST_HEAD(&l_ptr->waiting_ports);
271 link_reset_statistics(l_ptr);
273 tipc_node_attach_link(n_ptr, l_ptr);
275 k_init_timer(&l_ptr->timer, (Handler)link_timeout,
276 (unsigned long)l_ptr);
278 link_state_event(l_ptr, STARTING_EVT);
284 void tipc_link_delete_list(unsigned int bearer_id, bool shutting_down)
286 struct tipc_link *l_ptr;
287 struct tipc_node *n_ptr;
289 list_for_each_entry(n_ptr, &tipc_node_list, list) {
290 spin_lock_bh(&n_ptr->lock);
291 l_ptr = n_ptr->links[bearer_id];
293 tipc_link_reset(l_ptr);
294 if (shutting_down || !tipc_node_is_up(n_ptr)) {
295 tipc_node_detach_link(l_ptr->owner, l_ptr);
296 tipc_link_reset_fragments(l_ptr);
297 spin_unlock_bh(&n_ptr->lock);
299 /* Nobody else can access this link now: */
300 del_timer_sync(&l_ptr->timer);
303 /* Detach/delete when failover is finished: */
304 l_ptr->flags |= LINK_STOPPED;
305 spin_unlock_bh(&n_ptr->lock);
306 del_timer_sync(&l_ptr->timer);
310 spin_unlock_bh(&n_ptr->lock);
315 * link_schedule_port - schedule port for deferred sending
316 * @l_ptr: pointer to link
317 * @origport: reference to sending port
318 * @sz: amount of data to be sent
320 * Schedules port for renewed sending of messages after link congestion
323 static int link_schedule_port(struct tipc_link *l_ptr, u32 origport, u32 sz)
325 struct tipc_port *p_ptr;
327 spin_lock_bh(&tipc_port_list_lock);
328 p_ptr = tipc_port_lock(origport);
330 if (!list_empty(&p_ptr->wait_list))
332 p_ptr->congested = 1;
333 p_ptr->waiting_pkts = 1 + ((sz - 1) / l_ptr->max_pkt);
334 list_add_tail(&p_ptr->wait_list, &l_ptr->waiting_ports);
335 l_ptr->stats.link_congs++;
337 tipc_port_unlock(p_ptr);
339 spin_unlock_bh(&tipc_port_list_lock);
343 void tipc_link_wakeup_ports(struct tipc_link *l_ptr, int all)
345 struct tipc_port *p_ptr;
346 struct tipc_port *temp_p_ptr;
347 int win = l_ptr->queue_limit[0] - l_ptr->out_queue_size;
353 if (!spin_trylock_bh(&tipc_port_list_lock))
355 if (link_congested(l_ptr))
357 list_for_each_entry_safe(p_ptr, temp_p_ptr, &l_ptr->waiting_ports,
361 list_del_init(&p_ptr->wait_list);
362 spin_lock_bh(p_ptr->lock);
363 p_ptr->congested = 0;
364 tipc_port_wakeup(p_ptr);
365 win -= p_ptr->waiting_pkts;
366 spin_unlock_bh(p_ptr->lock);
370 spin_unlock_bh(&tipc_port_list_lock);
374 * link_release_outqueue - purge link's outbound message queue
375 * @l_ptr: pointer to link
377 static void link_release_outqueue(struct tipc_link *l_ptr)
379 kfree_skb_list(l_ptr->first_out);
380 l_ptr->first_out = NULL;
381 l_ptr->out_queue_size = 0;
385 * tipc_link_reset_fragments - purge link's inbound message fragments queue
386 * @l_ptr: pointer to link
388 void tipc_link_reset_fragments(struct tipc_link *l_ptr)
390 kfree_skb(l_ptr->reasm_head);
391 l_ptr->reasm_head = NULL;
392 l_ptr->reasm_tail = NULL;
396 * tipc_link_purge_queues - purge all pkt queues associated with link
397 * @l_ptr: pointer to link
399 void tipc_link_purge_queues(struct tipc_link *l_ptr)
401 kfree_skb_list(l_ptr->oldest_deferred_in);
402 kfree_skb_list(l_ptr->first_out);
403 tipc_link_reset_fragments(l_ptr);
404 kfree_skb(l_ptr->proto_msg_queue);
405 l_ptr->proto_msg_queue = NULL;
408 void tipc_link_reset(struct tipc_link *l_ptr)
410 u32 prev_state = l_ptr->state;
411 u32 checkpoint = l_ptr->next_in_no;
412 int was_active_link = tipc_link_is_active(l_ptr);
414 msg_set_session(l_ptr->pmsg, ((msg_session(l_ptr->pmsg) + 1) & 0xffff));
416 /* Link is down, accept any session */
417 l_ptr->peer_session = INVALID_SESSION;
419 /* Prepare for max packet size negotiation */
420 link_init_max_pkt(l_ptr);
422 l_ptr->state = RESET_UNKNOWN;
424 if ((prev_state == RESET_UNKNOWN) || (prev_state == RESET_RESET))
427 tipc_node_link_down(l_ptr->owner, l_ptr);
428 tipc_bearer_remove_dest(l_ptr->b_ptr, l_ptr->addr);
430 if (was_active_link && tipc_node_active_links(l_ptr->owner)) {
431 l_ptr->reset_checkpoint = checkpoint;
432 l_ptr->exp_msg_count = START_CHANGEOVER;
435 /* Clean up all queues: */
436 link_release_outqueue(l_ptr);
437 kfree_skb(l_ptr->proto_msg_queue);
438 l_ptr->proto_msg_queue = NULL;
439 kfree_skb_list(l_ptr->oldest_deferred_in);
440 if (!list_empty(&l_ptr->waiting_ports))
441 tipc_link_wakeup_ports(l_ptr, 1);
443 l_ptr->retransm_queue_head = 0;
444 l_ptr->retransm_queue_size = 0;
445 l_ptr->last_out = NULL;
446 l_ptr->first_out = NULL;
447 l_ptr->next_out = NULL;
448 l_ptr->unacked_window = 0;
449 l_ptr->checkpoint = 1;
450 l_ptr->next_out_no = 1;
451 l_ptr->deferred_inqueue_sz = 0;
452 l_ptr->oldest_deferred_in = NULL;
453 l_ptr->newest_deferred_in = NULL;
454 l_ptr->fsm_msg_cnt = 0;
455 l_ptr->stale_count = 0;
456 link_reset_statistics(l_ptr);
459 void tipc_link_reset_list(unsigned int bearer_id)
461 struct tipc_link *l_ptr;
462 struct tipc_node *n_ptr;
464 list_for_each_entry(n_ptr, &tipc_node_list, list) {
465 spin_lock_bh(&n_ptr->lock);
466 l_ptr = n_ptr->links[bearer_id];
468 tipc_link_reset(l_ptr);
469 spin_unlock_bh(&n_ptr->lock);
473 static void link_activate(struct tipc_link *l_ptr)
475 l_ptr->next_in_no = l_ptr->stats.recv_info = 1;
476 tipc_node_link_up(l_ptr->owner, l_ptr);
477 tipc_bearer_add_dest(l_ptr->b_ptr, l_ptr->addr);
481 * link_state_event - link finite state machine
482 * @l_ptr: pointer to link
483 * @event: state machine event to process
485 static void link_state_event(struct tipc_link *l_ptr, unsigned int event)
487 struct tipc_link *other;
488 u32 cont_intv = l_ptr->continuity_interval;
490 if (l_ptr->flags & LINK_STOPPED)
493 if (!(l_ptr->flags & LINK_STARTED) && (event != STARTING_EVT))
494 return; /* Not yet. */
496 /* Check whether changeover is going on */
497 if (l_ptr->exp_msg_count) {
498 if (event == TIMEOUT_EVT)
499 link_set_timer(l_ptr, cont_intv);
503 switch (l_ptr->state) {
504 case WORKING_WORKING:
506 case TRAFFIC_MSG_EVT:
510 if (l_ptr->next_in_no != l_ptr->checkpoint) {
511 l_ptr->checkpoint = l_ptr->next_in_no;
512 if (tipc_bclink_acks_missing(l_ptr->owner)) {
513 tipc_link_proto_xmit(l_ptr, STATE_MSG,
515 l_ptr->fsm_msg_cnt++;
516 } else if (l_ptr->max_pkt < l_ptr->max_pkt_target) {
517 tipc_link_proto_xmit(l_ptr, STATE_MSG,
519 l_ptr->fsm_msg_cnt++;
521 link_set_timer(l_ptr, cont_intv);
524 l_ptr->state = WORKING_UNKNOWN;
525 l_ptr->fsm_msg_cnt = 0;
526 tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
527 l_ptr->fsm_msg_cnt++;
528 link_set_timer(l_ptr, cont_intv / 4);
531 pr_info("%s<%s>, requested by peer\n", link_rst_msg,
533 tipc_link_reset(l_ptr);
534 l_ptr->state = RESET_RESET;
535 l_ptr->fsm_msg_cnt = 0;
536 tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
538 l_ptr->fsm_msg_cnt++;
539 link_set_timer(l_ptr, cont_intv);
542 pr_err("%s%u in WW state\n", link_unk_evt, event);
545 case WORKING_UNKNOWN:
547 case TRAFFIC_MSG_EVT:
549 l_ptr->state = WORKING_WORKING;
550 l_ptr->fsm_msg_cnt = 0;
551 link_set_timer(l_ptr, cont_intv);
554 pr_info("%s<%s>, requested by peer while probing\n",
555 link_rst_msg, l_ptr->name);
556 tipc_link_reset(l_ptr);
557 l_ptr->state = RESET_RESET;
558 l_ptr->fsm_msg_cnt = 0;
559 tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
561 l_ptr->fsm_msg_cnt++;
562 link_set_timer(l_ptr, cont_intv);
565 if (l_ptr->next_in_no != l_ptr->checkpoint) {
566 l_ptr->state = WORKING_WORKING;
567 l_ptr->fsm_msg_cnt = 0;
568 l_ptr->checkpoint = l_ptr->next_in_no;
569 if (tipc_bclink_acks_missing(l_ptr->owner)) {
570 tipc_link_proto_xmit(l_ptr, STATE_MSG,
572 l_ptr->fsm_msg_cnt++;
574 link_set_timer(l_ptr, cont_intv);
575 } else if (l_ptr->fsm_msg_cnt < l_ptr->abort_limit) {
576 tipc_link_proto_xmit(l_ptr, STATE_MSG,
578 l_ptr->fsm_msg_cnt++;
579 link_set_timer(l_ptr, cont_intv / 4);
580 } else { /* Link has failed */
581 pr_warn("%s<%s>, peer not responding\n",
582 link_rst_msg, l_ptr->name);
583 tipc_link_reset(l_ptr);
584 l_ptr->state = RESET_UNKNOWN;
585 l_ptr->fsm_msg_cnt = 0;
586 tipc_link_proto_xmit(l_ptr, RESET_MSG,
588 l_ptr->fsm_msg_cnt++;
589 link_set_timer(l_ptr, cont_intv);
593 pr_err("%s%u in WU state\n", link_unk_evt, event);
598 case TRAFFIC_MSG_EVT:
601 other = l_ptr->owner->active_links[0];
602 if (other && link_working_unknown(other))
604 l_ptr->state = WORKING_WORKING;
605 l_ptr->fsm_msg_cnt = 0;
606 link_activate(l_ptr);
607 tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
608 l_ptr->fsm_msg_cnt++;
609 if (l_ptr->owner->working_links == 1)
610 tipc_link_sync_xmit(l_ptr);
611 link_set_timer(l_ptr, cont_intv);
614 l_ptr->state = RESET_RESET;
615 l_ptr->fsm_msg_cnt = 0;
616 tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
618 l_ptr->fsm_msg_cnt++;
619 link_set_timer(l_ptr, cont_intv);
622 l_ptr->flags |= LINK_STARTED;
625 tipc_link_proto_xmit(l_ptr, RESET_MSG, 0, 0, 0, 0, 0);
626 l_ptr->fsm_msg_cnt++;
627 link_set_timer(l_ptr, cont_intv);
630 pr_err("%s%u in RU state\n", link_unk_evt, event);
635 case TRAFFIC_MSG_EVT:
637 other = l_ptr->owner->active_links[0];
638 if (other && link_working_unknown(other))
640 l_ptr->state = WORKING_WORKING;
641 l_ptr->fsm_msg_cnt = 0;
642 link_activate(l_ptr);
643 tipc_link_proto_xmit(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
644 l_ptr->fsm_msg_cnt++;
645 if (l_ptr->owner->working_links == 1)
646 tipc_link_sync_xmit(l_ptr);
647 link_set_timer(l_ptr, cont_intv);
652 tipc_link_proto_xmit(l_ptr, ACTIVATE_MSG,
654 l_ptr->fsm_msg_cnt++;
655 link_set_timer(l_ptr, cont_intv);
658 pr_err("%s%u in RR state\n", link_unk_evt, event);
662 pr_err("Unknown link state %u/%u\n", l_ptr->state, event);
667 * link_bundle_buf(): Append contents of a buffer to
668 * the tail of an existing one.
670 static int link_bundle_buf(struct tipc_link *l_ptr, struct sk_buff *bundler,
673 struct tipc_msg *bundler_msg = buf_msg(bundler);
674 struct tipc_msg *msg = buf_msg(buf);
675 u32 size = msg_size(msg);
676 u32 bundle_size = msg_size(bundler_msg);
677 u32 to_pos = align(bundle_size);
678 u32 pad = to_pos - bundle_size;
680 if (msg_user(bundler_msg) != MSG_BUNDLER)
682 if (msg_type(bundler_msg) != OPEN_MSG)
684 if (skb_tailroom(bundler) < (pad + size))
686 if (l_ptr->max_pkt < (to_pos + size))
689 skb_put(bundler, pad + size);
690 skb_copy_to_linear_data_offset(bundler, to_pos, buf->data, size);
691 msg_set_size(bundler_msg, to_pos + size);
692 msg_set_msgcnt(bundler_msg, msg_msgcnt(bundler_msg) + 1);
694 l_ptr->stats.sent_bundled++;
698 static void link_add_to_outqueue(struct tipc_link *l_ptr,
700 struct tipc_msg *msg)
702 u32 ack = mod(l_ptr->next_in_no - 1);
703 u32 seqno = mod(l_ptr->next_out_no++);
705 msg_set_word(msg, 2, ((ack << 16) | seqno));
706 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
708 if (l_ptr->first_out) {
709 l_ptr->last_out->next = buf;
710 l_ptr->last_out = buf;
712 l_ptr->first_out = l_ptr->last_out = buf;
714 l_ptr->out_queue_size++;
715 if (l_ptr->out_queue_size > l_ptr->stats.max_queue_sz)
716 l_ptr->stats.max_queue_sz = l_ptr->out_queue_size;
719 static void link_add_chain_to_outqueue(struct tipc_link *l_ptr,
720 struct sk_buff *buf_chain,
724 struct tipc_msg *msg;
726 if (!l_ptr->next_out)
727 l_ptr->next_out = buf_chain;
730 buf_chain = buf_chain->next;
733 msg_set_long_msgno(msg, long_msgno);
734 link_add_to_outqueue(l_ptr, buf, msg);
739 * tipc_link_xmit() is the 'full path' for messages, called from
740 * inside TIPC when the 'fast path' in tipc_send_xmit
741 * has failed, and from link_send()
743 int __tipc_link_xmit(struct tipc_link *l_ptr, struct sk_buff *buf)
745 struct tipc_msg *msg = buf_msg(buf);
746 u32 size = msg_size(msg);
747 u32 dsz = msg_data_sz(msg);
748 u32 queue_size = l_ptr->out_queue_size;
749 u32 imp = tipc_msg_tot_importance(msg);
750 u32 queue_limit = l_ptr->queue_limit[imp];
751 u32 max_packet = l_ptr->max_pkt;
753 /* Match msg importance against queue limits: */
754 if (unlikely(queue_size >= queue_limit)) {
755 if (imp <= TIPC_CRITICAL_IMPORTANCE) {
756 link_schedule_port(l_ptr, msg_origport(msg), size);
761 if (imp > CONN_MANAGER) {
762 pr_warn("%s<%s>, send queue full", link_rst_msg,
764 tipc_link_reset(l_ptr);
769 /* Fragmentation needed ? */
770 if (size > max_packet)
771 return tipc_link_frag_xmit(l_ptr, buf);
773 /* Packet can be queued or sent. */
774 if (likely(!link_congested(l_ptr))) {
775 link_add_to_outqueue(l_ptr, buf, msg);
777 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
778 l_ptr->unacked_window = 0;
781 /* Congestion: can message be bundled ? */
782 if ((msg_user(msg) != CHANGEOVER_PROTOCOL) &&
783 (msg_user(msg) != MSG_FRAGMENTER)) {
785 /* Try adding message to an existing bundle */
786 if (l_ptr->next_out &&
787 link_bundle_buf(l_ptr, l_ptr->last_out, buf))
790 /* Try creating a new bundle */
791 if (size <= max_packet * 2 / 3) {
792 struct sk_buff *bundler = tipc_buf_acquire(max_packet);
793 struct tipc_msg bundler_hdr;
796 tipc_msg_init(&bundler_hdr, MSG_BUNDLER, OPEN_MSG,
797 INT_H_SIZE, l_ptr->addr);
798 skb_copy_to_linear_data(bundler, &bundler_hdr,
800 skb_trim(bundler, INT_H_SIZE);
801 link_bundle_buf(l_ptr, bundler, buf);
804 l_ptr->stats.sent_bundles++;
808 if (!l_ptr->next_out)
809 l_ptr->next_out = buf;
810 link_add_to_outqueue(l_ptr, buf, msg);
815 * tipc_link_xmit(): same as __tipc_link_xmit(), but the link to use
816 * has not been selected yet, and the the owner node is not locked
817 * Called by TIPC internal users, e.g. the name distributor
819 int tipc_link_xmit(struct sk_buff *buf, u32 dest, u32 selector)
821 struct tipc_link *l_ptr;
822 struct tipc_node *n_ptr;
823 int res = -ELINKCONG;
825 read_lock_bh(&tipc_net_lock);
826 n_ptr = tipc_node_find(dest);
828 tipc_node_lock(n_ptr);
829 l_ptr = n_ptr->active_links[selector & 1];
831 res = __tipc_link_xmit(l_ptr, buf);
834 tipc_node_unlock(n_ptr);
838 read_unlock_bh(&tipc_net_lock);
843 * tipc_link_sync_xmit - synchronize broadcast link endpoints.
845 * Give a newly added peer node the sequence number where it should
846 * start receiving and acking broadcast packets.
848 * Called with node locked
850 static void tipc_link_sync_xmit(struct tipc_link *l)
853 struct tipc_msg *msg;
855 buf = tipc_buf_acquire(INT_H_SIZE);
860 tipc_msg_init(msg, BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE, l->addr);
861 msg_set_last_bcast(msg, l->owner->bclink.acked);
862 link_add_chain_to_outqueue(l, buf, 0);
863 tipc_link_push_queue(l);
867 * tipc_link_sync_rcv - synchronize broadcast link endpoints.
868 * Receive the sequence number where we should start receiving and
869 * acking broadcast packets from a newly added peer node, and open
870 * up for reception of such packets.
872 * Called with node locked
874 static void tipc_link_sync_rcv(struct tipc_node *n, struct sk_buff *buf)
876 struct tipc_msg *msg = buf_msg(buf);
878 n->bclink.last_sent = n->bclink.last_in = msg_last_bcast(msg);
879 n->bclink.recv_permitted = true;
884 * tipc_link_names_xmit - send name table entries to new neighbor
886 * Send routine for bulk delivery of name table messages when contact
887 * with a new neighbor occurs. No link congestion checking is performed
888 * because name table messages *must* be delivered. The messages must be
889 * small enough not to require fragmentation.
890 * Called without any locks held.
892 void tipc_link_names_xmit(struct list_head *message_list, u32 dest)
894 struct tipc_node *n_ptr;
895 struct tipc_link *l_ptr;
897 struct sk_buff *temp_buf;
899 if (list_empty(message_list))
902 read_lock_bh(&tipc_net_lock);
903 n_ptr = tipc_node_find(dest);
905 tipc_node_lock(n_ptr);
906 l_ptr = n_ptr->active_links[0];
908 /* convert circular list to linear list */
909 ((struct sk_buff *)message_list->prev)->next = NULL;
910 link_add_chain_to_outqueue(l_ptr,
911 (struct sk_buff *)message_list->next, 0);
912 tipc_link_push_queue(l_ptr);
913 INIT_LIST_HEAD(message_list);
915 tipc_node_unlock(n_ptr);
917 read_unlock_bh(&tipc_net_lock);
919 /* discard the messages if they couldn't be sent */
920 list_for_each_safe(buf, temp_buf, ((struct sk_buff *)message_list)) {
921 list_del((struct list_head *)buf);
927 * tipc_link_xmit_fast: Entry for data messages where the
928 * destination link is known and the header is complete,
929 * inclusive total message length. Very time critical.
930 * Link is locked. Returns user data length.
932 static int tipc_link_xmit_fast(struct tipc_link *l_ptr, struct sk_buff *buf,
935 struct tipc_msg *msg = buf_msg(buf);
936 int res = msg_data_sz(msg);
938 if (likely(!link_congested(l_ptr))) {
939 if (likely(msg_size(msg) <= l_ptr->max_pkt)) {
940 link_add_to_outqueue(l_ptr, buf, msg);
941 tipc_bearer_send(l_ptr->b_ptr, buf,
943 l_ptr->unacked_window = 0;
947 *used_max_pkt = l_ptr->max_pkt;
949 return __tipc_link_xmit(l_ptr, buf); /* All other cases */
953 * tipc_link_iovec_xmit_fast: Entry for messages where the
954 * destination processor is known and the header is complete,
955 * except for total message length.
956 * Returns user data length or errno.
958 int tipc_link_iovec_xmit_fast(struct tipc_port *sender,
959 struct iovec const *msg_sect,
960 unsigned int len, u32 destaddr)
962 struct tipc_msg *hdr = &sender->phdr;
963 struct tipc_link *l_ptr;
965 struct tipc_node *node;
967 u32 selector = msg_origport(hdr) & 1;
971 * Try building message using port's max_pkt hint.
972 * (Must not hold any locks while building message.)
974 res = tipc_msg_build(hdr, msg_sect, len, sender->max_pkt, &buf);
975 /* Exit if build request was invalid */
976 if (unlikely(res < 0))
979 read_lock_bh(&tipc_net_lock);
980 node = tipc_node_find(destaddr);
982 tipc_node_lock(node);
983 l_ptr = node->active_links[selector];
986 res = tipc_link_xmit_fast(l_ptr, buf,
989 tipc_node_unlock(node);
990 read_unlock_bh(&tipc_net_lock);
994 /* Exit if link (or bearer) is congested */
995 if (link_congested(l_ptr)) {
996 res = link_schedule_port(l_ptr,
1002 * Message size exceeds max_pkt hint; update hint,
1003 * then re-try fast path or fragment the message
1005 sender->max_pkt = l_ptr->max_pkt;
1006 tipc_node_unlock(node);
1007 read_unlock_bh(&tipc_net_lock);
1010 if ((msg_hdr_sz(hdr) + res) <= sender->max_pkt)
1013 return tipc_link_iovec_long_xmit(sender, msg_sect,
1016 tipc_node_unlock(node);
1018 read_unlock_bh(&tipc_net_lock);
1020 /* Couldn't find a link to the destination node */
1022 tipc_port_iovec_reject(sender, hdr, msg_sect, len, TIPC_ERR_NO_NODE);
1023 return -ENETUNREACH;
1027 * tipc_link_iovec_long_xmit(): Entry for long messages where the
1028 * destination node is known and the header is complete,
1029 * inclusive total message length.
1030 * Link and bearer congestion status have been checked to be ok,
1031 * and are ignored if they change.
1033 * Note that fragments do not use the full link MTU so that they won't have
1034 * to undergo refragmentation if link changeover causes them to be sent
1035 * over another link with an additional tunnel header added as prefix.
1036 * (Refragmentation will still occur if the other link has a smaller MTU.)
1038 * Returns user data length or errno.
1040 static int tipc_link_iovec_long_xmit(struct tipc_port *sender,
1041 struct iovec const *msg_sect,
1042 unsigned int len, u32 destaddr)
1044 struct tipc_link *l_ptr;
1045 struct tipc_node *node;
1046 struct tipc_msg *hdr = &sender->phdr;
1048 u32 max_pkt, fragm_sz, rest;
1049 struct tipc_msg fragm_hdr;
1050 struct sk_buff *buf, *buf_chain, *prev;
1051 u32 fragm_crs, fragm_rest, hsz, sect_rest;
1052 const unchar __user *sect_crs;
1059 max_pkt = sender->max_pkt - INT_H_SIZE;
1060 /* leave room for tunnel header in case of link changeover */
1061 fragm_sz = max_pkt - INT_H_SIZE;
1062 /* leave room for fragmentation header in each fragment */
1070 /* Prepare reusable fragment header */
1071 tipc_msg_init(&fragm_hdr, MSG_FRAGMENTER, FIRST_FRAGMENT,
1072 INT_H_SIZE, msg_destnode(hdr));
1073 msg_set_size(&fragm_hdr, max_pkt);
1074 msg_set_fragm_no(&fragm_hdr, 1);
1076 /* Prepare header of first fragment */
1077 buf_chain = buf = tipc_buf_acquire(max_pkt);
1081 skb_copy_to_linear_data(buf, &fragm_hdr, INT_H_SIZE);
1082 hsz = msg_hdr_sz(hdr);
1083 skb_copy_to_linear_data_offset(buf, INT_H_SIZE, hdr, hsz);
1085 /* Chop up message */
1086 fragm_crs = INT_H_SIZE + hsz;
1087 fragm_rest = fragm_sz - hsz;
1089 do { /* For all sections */
1093 sect_rest = msg_sect[++curr_sect].iov_len;
1094 sect_crs = msg_sect[curr_sect].iov_base;
1097 if (sect_rest < fragm_rest)
1102 if (copy_from_user(buf->data + fragm_crs, sect_crs, sz)) {
1105 kfree_skb_list(buf_chain);
1114 if (!fragm_rest && rest) {
1116 /* Initiate new fragment: */
1117 if (rest <= fragm_sz) {
1119 msg_set_type(&fragm_hdr, LAST_FRAGMENT);
1121 msg_set_type(&fragm_hdr, FRAGMENT);
1123 msg_set_size(&fragm_hdr, fragm_sz + INT_H_SIZE);
1124 msg_set_fragm_no(&fragm_hdr, ++fragm_no);
1126 buf = tipc_buf_acquire(fragm_sz + INT_H_SIZE);
1134 skb_copy_to_linear_data(buf, &fragm_hdr, INT_H_SIZE);
1135 fragm_crs = INT_H_SIZE;
1136 fragm_rest = fragm_sz;
1141 * Now we have a buffer chain. Select a link and check
1142 * that packet size is still OK
1144 node = tipc_node_find(destaddr);
1146 tipc_node_lock(node);
1147 l_ptr = node->active_links[sender->ref & 1];
1149 tipc_node_unlock(node);
1152 if (l_ptr->max_pkt < max_pkt) {
1153 sender->max_pkt = l_ptr->max_pkt;
1154 tipc_node_unlock(node);
1155 kfree_skb_list(buf_chain);
1160 kfree_skb_list(buf_chain);
1161 tipc_port_iovec_reject(sender, hdr, msg_sect, len,
1163 return -ENETUNREACH;
1166 /* Append chain of fragments to send queue & send them */
1167 l_ptr->long_msg_seq_no++;
1168 link_add_chain_to_outqueue(l_ptr, buf_chain, l_ptr->long_msg_seq_no);
1169 l_ptr->stats.sent_fragments += fragm_no;
1170 l_ptr->stats.sent_fragmented++;
1171 tipc_link_push_queue(l_ptr);
1172 tipc_node_unlock(node);
1177 * tipc_link_push_packet: Push one unsent packet to the media
1179 static u32 tipc_link_push_packet(struct tipc_link *l_ptr)
1181 struct sk_buff *buf = l_ptr->first_out;
1182 u32 r_q_size = l_ptr->retransm_queue_size;
1183 u32 r_q_head = l_ptr->retransm_queue_head;
1185 /* Step to position where retransmission failed, if any, */
1186 /* consider that buffers may have been released in meantime */
1187 if (r_q_size && buf) {
1188 u32 last = lesser(mod(r_q_head + r_q_size),
1189 link_last_sent(l_ptr));
1190 u32 first = buf_seqno(buf);
1192 while (buf && less(first, r_q_head)) {
1193 first = mod(first + 1);
1196 l_ptr->retransm_queue_head = r_q_head = first;
1197 l_ptr->retransm_queue_size = r_q_size = mod(last - first);
1200 /* Continue retransmission now, if there is anything: */
1201 if (r_q_size && buf) {
1202 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
1203 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
1204 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1205 l_ptr->retransm_queue_head = mod(++r_q_head);
1206 l_ptr->retransm_queue_size = --r_q_size;
1207 l_ptr->stats.retransmitted++;
1211 /* Send deferred protocol message, if any: */
1212 buf = l_ptr->proto_msg_queue;
1214 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
1215 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
1216 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1217 l_ptr->unacked_window = 0;
1219 l_ptr->proto_msg_queue = NULL;
1223 /* Send one deferred data message, if send window not full: */
1224 buf = l_ptr->next_out;
1226 struct tipc_msg *msg = buf_msg(buf);
1227 u32 next = msg_seqno(msg);
1228 u32 first = buf_seqno(l_ptr->first_out);
1230 if (mod(next - first) < l_ptr->queue_limit[0]) {
1231 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1232 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1233 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1234 if (msg_user(msg) == MSG_BUNDLER)
1235 msg_set_type(msg, CLOSED_MSG);
1236 l_ptr->next_out = buf->next;
1244 * push_queue(): push out the unsent messages of a link where
1245 * congestion has abated. Node is locked
1247 void tipc_link_push_queue(struct tipc_link *l_ptr)
1252 res = tipc_link_push_packet(l_ptr);
1256 static void link_reset_all(unsigned long addr)
1258 struct tipc_node *n_ptr;
1259 char addr_string[16];
1262 read_lock_bh(&tipc_net_lock);
1263 n_ptr = tipc_node_find((u32)addr);
1265 read_unlock_bh(&tipc_net_lock);
1266 return; /* node no longer exists */
1269 tipc_node_lock(n_ptr);
1271 pr_warn("Resetting all links to %s\n",
1272 tipc_addr_string_fill(addr_string, n_ptr->addr));
1274 for (i = 0; i < MAX_BEARERS; i++) {
1275 if (n_ptr->links[i]) {
1276 link_print(n_ptr->links[i], "Resetting link\n");
1277 tipc_link_reset(n_ptr->links[i]);
1281 tipc_node_unlock(n_ptr);
1282 read_unlock_bh(&tipc_net_lock);
1285 static void link_retransmit_failure(struct tipc_link *l_ptr,
1286 struct sk_buff *buf)
1288 struct tipc_msg *msg = buf_msg(buf);
1290 pr_warn("Retransmission failure on link <%s>\n", l_ptr->name);
1293 /* Handle failure on standard link */
1294 link_print(l_ptr, "Resetting link\n");
1295 tipc_link_reset(l_ptr);
1298 /* Handle failure on broadcast link */
1299 struct tipc_node *n_ptr;
1300 char addr_string[16];
1302 pr_info("Msg seq number: %u, ", msg_seqno(msg));
1303 pr_cont("Outstanding acks: %lu\n",
1304 (unsigned long) TIPC_SKB_CB(buf)->handle);
1306 n_ptr = tipc_bclink_retransmit_to();
1307 tipc_node_lock(n_ptr);
1309 tipc_addr_string_fill(addr_string, n_ptr->addr);
1310 pr_info("Broadcast link info for %s\n", addr_string);
1311 pr_info("Reception permitted: %d, Acked: %u\n",
1312 n_ptr->bclink.recv_permitted,
1313 n_ptr->bclink.acked);
1314 pr_info("Last in: %u, Oos state: %u, Last sent: %u\n",
1315 n_ptr->bclink.last_in,
1316 n_ptr->bclink.oos_state,
1317 n_ptr->bclink.last_sent);
1319 tipc_k_signal((Handler)link_reset_all, (unsigned long)n_ptr->addr);
1321 tipc_node_unlock(n_ptr);
1323 l_ptr->stale_count = 0;
1327 void tipc_link_retransmit(struct tipc_link *l_ptr, struct sk_buff *buf,
1330 struct tipc_msg *msg;
1337 /* Detect repeated retransmit failures */
1338 if (l_ptr->last_retransmitted == msg_seqno(msg)) {
1339 if (++l_ptr->stale_count > 100) {
1340 link_retransmit_failure(l_ptr, buf);
1344 l_ptr->last_retransmitted = msg_seqno(msg);
1345 l_ptr->stale_count = 1;
1348 while (retransmits && (buf != l_ptr->next_out) && buf) {
1350 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1351 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1352 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1355 l_ptr->stats.retransmitted++;
1358 l_ptr->retransm_queue_head = l_ptr->retransm_queue_size = 0;
1362 * link_insert_deferred_queue - insert deferred messages back into receive chain
1364 static struct sk_buff *link_insert_deferred_queue(struct tipc_link *l_ptr,
1365 struct sk_buff *buf)
1369 if (l_ptr->oldest_deferred_in == NULL)
1372 seq_no = buf_seqno(l_ptr->oldest_deferred_in);
1373 if (seq_no == mod(l_ptr->next_in_no)) {
1374 l_ptr->newest_deferred_in->next = buf;
1375 buf = l_ptr->oldest_deferred_in;
1376 l_ptr->oldest_deferred_in = NULL;
1377 l_ptr->deferred_inqueue_sz = 0;
1383 * link_recv_buf_validate - validate basic format of received message
1385 * This routine ensures a TIPC message has an acceptable header, and at least
1386 * as much data as the header indicates it should. The routine also ensures
1387 * that the entire message header is stored in the main fragment of the message
1388 * buffer, to simplify future access to message header fields.
1390 * Note: Having extra info present in the message header or data areas is OK.
1391 * TIPC will ignore the excess, under the assumption that it is optional info
1392 * introduced by a later release of the protocol.
1394 static int link_recv_buf_validate(struct sk_buff *buf)
1396 static u32 min_data_hdr_size[8] = {
1397 SHORT_H_SIZE, MCAST_H_SIZE, NAMED_H_SIZE, BASIC_H_SIZE,
1398 MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE
1401 struct tipc_msg *msg;
1407 /* If this packet comes from the defer queue, the skb has already
1410 if (unlikely(TIPC_SKB_CB(buf)->deferred))
1413 if (unlikely(buf->len < MIN_H_SIZE))
1416 msg = skb_header_pointer(buf, 0, sizeof(tipc_hdr), tipc_hdr);
1420 if (unlikely(msg_version(msg) != TIPC_VERSION))
1423 size = msg_size(msg);
1424 hdr_size = msg_hdr_sz(msg);
1425 min_hdr_size = msg_isdata(msg) ?
1426 min_data_hdr_size[msg_type(msg)] : INT_H_SIZE;
1428 if (unlikely((hdr_size < min_hdr_size) ||
1429 (size < hdr_size) ||
1430 (buf->len < size) ||
1431 (size - hdr_size > TIPC_MAX_USER_MSG_SIZE)))
1434 return pskb_may_pull(buf, hdr_size);
1438 * tipc_rcv - process TIPC packets/messages arriving from off-node
1439 * @head: pointer to message buffer chain
1440 * @tb_ptr: pointer to bearer message arrived on
1442 * Invoked with no locks held. Bearer pointer must point to a valid bearer
1443 * structure (i.e. cannot be NULL), but bearer can be inactive.
1445 void tipc_rcv(struct sk_buff *head, struct tipc_bearer *b_ptr)
1447 read_lock_bh(&tipc_net_lock);
1449 struct tipc_node *n_ptr;
1450 struct tipc_link *l_ptr;
1451 struct sk_buff *crs;
1452 struct sk_buff *buf = head;
1453 struct tipc_msg *msg;
1461 /* Ensure message is well-formed */
1462 if (unlikely(!link_recv_buf_validate(buf)))
1465 /* Ensure message data is a single contiguous unit */
1466 if (unlikely(skb_linearize(buf)))
1469 /* Handle arrival of a non-unicast link message */
1472 if (unlikely(msg_non_seq(msg))) {
1473 if (msg_user(msg) == LINK_CONFIG)
1474 tipc_disc_rcv(buf, b_ptr);
1476 tipc_bclink_rcv(buf);
1480 /* Discard unicast link messages destined for another node */
1481 if (unlikely(!msg_short(msg) &&
1482 (msg_destnode(msg) != tipc_own_addr)))
1485 /* Locate neighboring node that sent message */
1486 n_ptr = tipc_node_find(msg_prevnode(msg));
1487 if (unlikely(!n_ptr))
1489 tipc_node_lock(n_ptr);
1491 /* Locate unicast link endpoint that should handle message */
1492 l_ptr = n_ptr->links[b_ptr->identity];
1493 if (unlikely(!l_ptr))
1494 goto unlock_discard;
1496 /* Verify that communication with node is currently allowed */
1497 if ((n_ptr->block_setup & WAIT_PEER_DOWN) &&
1498 msg_user(msg) == LINK_PROTOCOL &&
1499 (msg_type(msg) == RESET_MSG ||
1500 msg_type(msg) == ACTIVATE_MSG) &&
1501 !msg_redundant_link(msg))
1502 n_ptr->block_setup &= ~WAIT_PEER_DOWN;
1504 if (n_ptr->block_setup)
1505 goto unlock_discard;
1507 /* Validate message sequence number info */
1508 seq_no = msg_seqno(msg);
1509 ackd = msg_ack(msg);
1511 /* Release acked messages */
1512 if (n_ptr->bclink.recv_permitted)
1513 tipc_bclink_acknowledge(n_ptr, msg_bcast_ack(msg));
1515 crs = l_ptr->first_out;
1516 while ((crs != l_ptr->next_out) &&
1517 less_eq(buf_seqno(crs), ackd)) {
1518 struct sk_buff *next = crs->next;
1524 l_ptr->first_out = crs;
1525 l_ptr->out_queue_size -= released;
1528 /* Try sending any messages link endpoint has pending */
1529 if (unlikely(l_ptr->next_out))
1530 tipc_link_push_queue(l_ptr);
1532 if (unlikely(!list_empty(&l_ptr->waiting_ports)))
1533 tipc_link_wakeup_ports(l_ptr, 0);
1535 if (unlikely(++l_ptr->unacked_window >= TIPC_MIN_LINK_WIN)) {
1536 l_ptr->stats.sent_acks++;
1537 tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1540 /* Process the incoming packet */
1541 if (unlikely(!link_working_working(l_ptr))) {
1542 if (msg_user(msg) == LINK_PROTOCOL) {
1543 tipc_link_proto_rcv(l_ptr, buf);
1544 head = link_insert_deferred_queue(l_ptr, head);
1545 tipc_node_unlock(n_ptr);
1549 /* Traffic message. Conditionally activate link */
1550 link_state_event(l_ptr, TRAFFIC_MSG_EVT);
1552 if (link_working_working(l_ptr)) {
1553 /* Re-insert buffer in front of queue */
1556 tipc_node_unlock(n_ptr);
1559 goto unlock_discard;
1562 /* Link is now in state WORKING_WORKING */
1563 if (unlikely(seq_no != mod(l_ptr->next_in_no))) {
1564 link_handle_out_of_seq_msg(l_ptr, buf);
1565 head = link_insert_deferred_queue(l_ptr, head);
1566 tipc_node_unlock(n_ptr);
1569 l_ptr->next_in_no++;
1570 if (unlikely(l_ptr->oldest_deferred_in))
1571 head = link_insert_deferred_queue(l_ptr, head);
1573 /* Deliver packet/message to correct user: */
1574 if (unlikely(msg_user(msg) == CHANGEOVER_PROTOCOL)) {
1575 if (!tipc_link_tunnel_rcv(n_ptr, &buf)) {
1576 tipc_node_unlock(n_ptr);
1580 } else if (msg_user(msg) == MSG_FRAGMENTER) {
1583 l_ptr->stats.recv_fragments++;
1584 rc = tipc_link_frag_rcv(&l_ptr->reasm_head,
1587 if (rc == LINK_REASM_COMPLETE) {
1588 l_ptr->stats.recv_fragmented++;
1591 if (rc == LINK_REASM_ERROR)
1592 tipc_link_reset(l_ptr);
1593 tipc_node_unlock(n_ptr);
1598 switch (msg_user(msg)) {
1599 case TIPC_LOW_IMPORTANCE:
1600 case TIPC_MEDIUM_IMPORTANCE:
1601 case TIPC_HIGH_IMPORTANCE:
1602 case TIPC_CRITICAL_IMPORTANCE:
1603 tipc_node_unlock(n_ptr);
1607 l_ptr->stats.recv_bundles++;
1608 l_ptr->stats.recv_bundled += msg_msgcnt(msg);
1609 tipc_node_unlock(n_ptr);
1610 tipc_link_bundle_rcv(buf);
1612 case NAME_DISTRIBUTOR:
1613 n_ptr->bclink.recv_permitted = true;
1614 tipc_node_unlock(n_ptr);
1615 tipc_named_rcv(buf);
1618 tipc_node_unlock(n_ptr);
1619 tipc_port_proto_rcv(buf);
1621 case BCAST_PROTOCOL:
1622 tipc_link_sync_rcv(n_ptr, buf);
1628 tipc_node_unlock(n_ptr);
1631 tipc_node_unlock(n_ptr);
1635 read_unlock_bh(&tipc_net_lock);
1639 * tipc_link_defer_pkt - Add out-of-sequence message to deferred reception queue
1641 * Returns increase in queue length (i.e. 0 or 1)
1643 u32 tipc_link_defer_pkt(struct sk_buff **head, struct sk_buff **tail,
1644 struct sk_buff *buf)
1646 struct sk_buff *queue_buf;
1647 struct sk_buff **prev;
1648 u32 seq_no = buf_seqno(buf);
1653 if (*head == NULL) {
1654 *head = *tail = buf;
1659 if (less(buf_seqno(*tail), seq_no)) {
1660 (*tail)->next = buf;
1665 /* Locate insertion point in queue, then insert; discard if duplicate */
1669 u32 curr_seqno = buf_seqno(queue_buf);
1671 if (seq_no == curr_seqno) {
1676 if (less(seq_no, curr_seqno))
1679 prev = &queue_buf->next;
1680 queue_buf = queue_buf->next;
1683 buf->next = queue_buf;
1689 * link_handle_out_of_seq_msg - handle arrival of out-of-sequence packet
1691 static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
1692 struct sk_buff *buf)
1694 u32 seq_no = buf_seqno(buf);
1696 if (likely(msg_user(buf_msg(buf)) == LINK_PROTOCOL)) {
1697 tipc_link_proto_rcv(l_ptr, buf);
1701 /* Record OOS packet arrival (force mismatch on next timeout) */
1702 l_ptr->checkpoint--;
1705 * Discard packet if a duplicate; otherwise add it to deferred queue
1706 * and notify peer of gap as per protocol specification
1708 if (less(seq_no, mod(l_ptr->next_in_no))) {
1709 l_ptr->stats.duplicates++;
1714 if (tipc_link_defer_pkt(&l_ptr->oldest_deferred_in,
1715 &l_ptr->newest_deferred_in, buf)) {
1716 l_ptr->deferred_inqueue_sz++;
1717 l_ptr->stats.deferred_recv++;
1718 TIPC_SKB_CB(buf)->deferred = true;
1719 if ((l_ptr->deferred_inqueue_sz % 16) == 1)
1720 tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1722 l_ptr->stats.duplicates++;
1726 * Send protocol message to the other endpoint.
1728 void tipc_link_proto_xmit(struct tipc_link *l_ptr, u32 msg_typ, int probe_msg,
1729 u32 gap, u32 tolerance, u32 priority, u32 ack_mtu)
1731 struct sk_buff *buf = NULL;
1732 struct tipc_msg *msg = l_ptr->pmsg;
1733 u32 msg_size = sizeof(l_ptr->proto_msg);
1736 /* Discard any previous message that was deferred due to congestion */
1737 if (l_ptr->proto_msg_queue) {
1738 kfree_skb(l_ptr->proto_msg_queue);
1739 l_ptr->proto_msg_queue = NULL;
1742 /* Don't send protocol message during link changeover */
1743 if (l_ptr->exp_msg_count)
1746 /* Abort non-RESET send if communication with node is prohibited */
1747 if ((l_ptr->owner->block_setup) && (msg_typ != RESET_MSG))
1750 /* Create protocol message with "out-of-sequence" sequence number */
1751 msg_set_type(msg, msg_typ);
1752 msg_set_net_plane(msg, l_ptr->b_ptr->net_plane);
1753 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1754 msg_set_last_bcast(msg, tipc_bclink_get_last_sent());
1756 if (msg_typ == STATE_MSG) {
1757 u32 next_sent = mod(l_ptr->next_out_no);
1759 if (!tipc_link_is_up(l_ptr))
1761 if (l_ptr->next_out)
1762 next_sent = buf_seqno(l_ptr->next_out);
1763 msg_set_next_sent(msg, next_sent);
1764 if (l_ptr->oldest_deferred_in) {
1765 u32 rec = buf_seqno(l_ptr->oldest_deferred_in);
1766 gap = mod(rec - mod(l_ptr->next_in_no));
1768 msg_set_seq_gap(msg, gap);
1770 l_ptr->stats.sent_nacks++;
1771 msg_set_link_tolerance(msg, tolerance);
1772 msg_set_linkprio(msg, priority);
1773 msg_set_max_pkt(msg, ack_mtu);
1774 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1775 msg_set_probe(msg, probe_msg != 0);
1777 u32 mtu = l_ptr->max_pkt;
1779 if ((mtu < l_ptr->max_pkt_target) &&
1780 link_working_working(l_ptr) &&
1781 l_ptr->fsm_msg_cnt) {
1782 msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1783 if (l_ptr->max_pkt_probes == 10) {
1784 l_ptr->max_pkt_target = (msg_size - 4);
1785 l_ptr->max_pkt_probes = 0;
1786 msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1788 l_ptr->max_pkt_probes++;
1791 l_ptr->stats.sent_probes++;
1793 l_ptr->stats.sent_states++;
1794 } else { /* RESET_MSG or ACTIVATE_MSG */
1795 msg_set_ack(msg, mod(l_ptr->reset_checkpoint - 1));
1796 msg_set_seq_gap(msg, 0);
1797 msg_set_next_sent(msg, 1);
1798 msg_set_probe(msg, 0);
1799 msg_set_link_tolerance(msg, l_ptr->tolerance);
1800 msg_set_linkprio(msg, l_ptr->priority);
1801 msg_set_max_pkt(msg, l_ptr->max_pkt_target);
1804 r_flag = (l_ptr->owner->working_links > tipc_link_is_up(l_ptr));
1805 msg_set_redundant_link(msg, r_flag);
1806 msg_set_linkprio(msg, l_ptr->priority);
1807 msg_set_size(msg, msg_size);
1809 msg_set_seqno(msg, mod(l_ptr->next_out_no + (0xffff/2)));
1811 buf = tipc_buf_acquire(msg_size);
1815 skb_copy_to_linear_data(buf, msg, sizeof(l_ptr->proto_msg));
1816 buf->priority = TC_PRIO_CONTROL;
1818 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1819 l_ptr->unacked_window = 0;
1824 * Receive protocol message :
1825 * Note that network plane id propagates through the network, and may
1826 * change at any time. The node with lowest address rules
1828 static void tipc_link_proto_rcv(struct tipc_link *l_ptr, struct sk_buff *buf)
1834 struct tipc_msg *msg = buf_msg(buf);
1836 /* Discard protocol message during link changeover */
1837 if (l_ptr->exp_msg_count)
1840 /* record unnumbered packet arrival (force mismatch on next timeout) */
1841 l_ptr->checkpoint--;
1843 if (l_ptr->b_ptr->net_plane != msg_net_plane(msg))
1844 if (tipc_own_addr > msg_prevnode(msg))
1845 l_ptr->b_ptr->net_plane = msg_net_plane(msg);
1847 switch (msg_type(msg)) {
1850 if (!link_working_unknown(l_ptr) &&
1851 (l_ptr->peer_session != INVALID_SESSION)) {
1852 if (less_eq(msg_session(msg), l_ptr->peer_session))
1853 break; /* duplicate or old reset: ignore */
1856 if (!msg_redundant_link(msg) && (link_working_working(l_ptr) ||
1857 link_working_unknown(l_ptr))) {
1859 * peer has lost contact -- don't allow peer's links
1860 * to reactivate before we recognize loss & clean up
1862 l_ptr->owner->block_setup = WAIT_NODE_DOWN;
1865 link_state_event(l_ptr, RESET_MSG);
1869 /* Update link settings according other endpoint's values */
1870 strcpy((strrchr(l_ptr->name, ':') + 1), (char *)msg_data(msg));
1872 msg_tol = msg_link_tolerance(msg);
1873 if (msg_tol > l_ptr->tolerance)
1874 link_set_supervision_props(l_ptr, msg_tol);
1876 if (msg_linkprio(msg) > l_ptr->priority)
1877 l_ptr->priority = msg_linkprio(msg);
1879 max_pkt_info = msg_max_pkt(msg);
1881 if (max_pkt_info < l_ptr->max_pkt_target)
1882 l_ptr->max_pkt_target = max_pkt_info;
1883 if (l_ptr->max_pkt > l_ptr->max_pkt_target)
1884 l_ptr->max_pkt = l_ptr->max_pkt_target;
1886 l_ptr->max_pkt = l_ptr->max_pkt_target;
1889 /* Synchronize broadcast link info, if not done previously */
1890 if (!tipc_node_is_up(l_ptr->owner)) {
1891 l_ptr->owner->bclink.last_sent =
1892 l_ptr->owner->bclink.last_in =
1893 msg_last_bcast(msg);
1894 l_ptr->owner->bclink.oos_state = 0;
1897 l_ptr->peer_session = msg_session(msg);
1898 l_ptr->peer_bearer_id = msg_bearer_id(msg);
1900 if (msg_type(msg) == ACTIVATE_MSG)
1901 link_state_event(l_ptr, ACTIVATE_MSG);
1905 msg_tol = msg_link_tolerance(msg);
1907 link_set_supervision_props(l_ptr, msg_tol);
1909 if (msg_linkprio(msg) &&
1910 (msg_linkprio(msg) != l_ptr->priority)) {
1911 pr_warn("%s<%s>, priority change %u->%u\n",
1912 link_rst_msg, l_ptr->name, l_ptr->priority,
1914 l_ptr->priority = msg_linkprio(msg);
1915 tipc_link_reset(l_ptr); /* Enforce change to take effect */
1918 link_state_event(l_ptr, TRAFFIC_MSG_EVT);
1919 l_ptr->stats.recv_states++;
1920 if (link_reset_unknown(l_ptr))
1923 if (less_eq(mod(l_ptr->next_in_no), msg_next_sent(msg))) {
1924 rec_gap = mod(msg_next_sent(msg) -
1925 mod(l_ptr->next_in_no));
1928 max_pkt_ack = msg_max_pkt(msg);
1929 if (max_pkt_ack > l_ptr->max_pkt) {
1930 l_ptr->max_pkt = max_pkt_ack;
1931 l_ptr->max_pkt_probes = 0;
1935 if (msg_probe(msg)) {
1936 l_ptr->stats.recv_probes++;
1937 if (msg_size(msg) > sizeof(l_ptr->proto_msg))
1938 max_pkt_ack = msg_size(msg);
1941 /* Protocol message before retransmits, reduce loss risk */
1942 if (l_ptr->owner->bclink.recv_permitted)
1943 tipc_bclink_update_link_state(l_ptr->owner,
1944 msg_last_bcast(msg));
1946 if (rec_gap || (msg_probe(msg))) {
1947 tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, rec_gap, 0,
1950 if (msg_seq_gap(msg)) {
1951 l_ptr->stats.recv_nacks++;
1952 tipc_link_retransmit(l_ptr, l_ptr->first_out,
1962 /* tipc_link_tunnel_xmit(): Tunnel one packet via a link belonging to
1963 * a different bearer. Owner node is locked.
1965 static void tipc_link_tunnel_xmit(struct tipc_link *l_ptr,
1966 struct tipc_msg *tunnel_hdr,
1967 struct tipc_msg *msg,
1970 struct tipc_link *tunnel;
1971 struct sk_buff *buf;
1972 u32 length = msg_size(msg);
1974 tunnel = l_ptr->owner->active_links[selector & 1];
1975 if (!tipc_link_is_up(tunnel)) {
1976 pr_warn("%stunnel link no longer available\n", link_co_err);
1979 msg_set_size(tunnel_hdr, length + INT_H_SIZE);
1980 buf = tipc_buf_acquire(length + INT_H_SIZE);
1982 pr_warn("%sunable to send tunnel msg\n", link_co_err);
1985 skb_copy_to_linear_data(buf, tunnel_hdr, INT_H_SIZE);
1986 skb_copy_to_linear_data_offset(buf, INT_H_SIZE, msg, length);
1987 __tipc_link_xmit(tunnel, buf);
1991 /* tipc_link_failover_send_queue(): A link has gone down, but a second
1992 * link is still active. We can do failover. Tunnel the failing link's
1993 * whole send queue via the remaining link. This way, we don't lose
1994 * any packets, and sequence order is preserved for subsequent traffic
1995 * sent over the remaining link. Owner node is locked.
1997 void tipc_link_failover_send_queue(struct tipc_link *l_ptr)
1999 u32 msgcount = l_ptr->out_queue_size;
2000 struct sk_buff *crs = l_ptr->first_out;
2001 struct tipc_link *tunnel = l_ptr->owner->active_links[0];
2002 struct tipc_msg tunnel_hdr;
2008 tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
2009 ORIGINAL_MSG, INT_H_SIZE, l_ptr->addr);
2010 msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
2011 msg_set_msgcnt(&tunnel_hdr, msgcount);
2013 if (!l_ptr->first_out) {
2014 struct sk_buff *buf;
2016 buf = tipc_buf_acquire(INT_H_SIZE);
2018 skb_copy_to_linear_data(buf, &tunnel_hdr, INT_H_SIZE);
2019 msg_set_size(&tunnel_hdr, INT_H_SIZE);
2020 __tipc_link_xmit(tunnel, buf);
2022 pr_warn("%sunable to send changeover msg\n",
2028 split_bundles = (l_ptr->owner->active_links[0] !=
2029 l_ptr->owner->active_links[1]);
2032 struct tipc_msg *msg = buf_msg(crs);
2034 if ((msg_user(msg) == MSG_BUNDLER) && split_bundles) {
2035 struct tipc_msg *m = msg_get_wrapped(msg);
2036 unchar *pos = (unchar *)m;
2038 msgcount = msg_msgcnt(msg);
2039 while (msgcount--) {
2040 msg_set_seqno(m, msg_seqno(msg));
2041 tipc_link_tunnel_xmit(l_ptr, &tunnel_hdr, m,
2042 msg_link_selector(m));
2043 pos += align(msg_size(m));
2044 m = (struct tipc_msg *)pos;
2047 tipc_link_tunnel_xmit(l_ptr, &tunnel_hdr, msg,
2048 msg_link_selector(msg));
2054 /* tipc_link_dup_queue_xmit(): A second link has become active. Tunnel a
2055 * duplicate of the first link's send queue via the new link. This way, we
2056 * are guaranteed that currently queued packets from a socket are delivered
2057 * before future traffic from the same socket, even if this is using the
2058 * new link. The last arriving copy of each duplicate packet is dropped at
2059 * the receiving end by the regular protocol check, so packet cardinality
2060 * and sequence order is preserved per sender/receiver socket pair.
2061 * Owner node is locked.
2063 void tipc_link_dup_queue_xmit(struct tipc_link *l_ptr,
2064 struct tipc_link *tunnel)
2066 struct sk_buff *iter;
2067 struct tipc_msg tunnel_hdr;
2069 tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
2070 DUPLICATE_MSG, INT_H_SIZE, l_ptr->addr);
2071 msg_set_msgcnt(&tunnel_hdr, l_ptr->out_queue_size);
2072 msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
2073 iter = l_ptr->first_out;
2075 struct sk_buff *outbuf;
2076 struct tipc_msg *msg = buf_msg(iter);
2077 u32 length = msg_size(msg);
2079 if (msg_user(msg) == MSG_BUNDLER)
2080 msg_set_type(msg, CLOSED_MSG);
2081 msg_set_ack(msg, mod(l_ptr->next_in_no - 1)); /* Update */
2082 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
2083 msg_set_size(&tunnel_hdr, length + INT_H_SIZE);
2084 outbuf = tipc_buf_acquire(length + INT_H_SIZE);
2085 if (outbuf == NULL) {
2086 pr_warn("%sunable to send duplicate msg\n",
2090 skb_copy_to_linear_data(outbuf, &tunnel_hdr, INT_H_SIZE);
2091 skb_copy_to_linear_data_offset(outbuf, INT_H_SIZE, iter->data,
2093 __tipc_link_xmit(tunnel, outbuf);
2094 if (!tipc_link_is_up(l_ptr))
2101 * buf_extract - extracts embedded TIPC message from another message
2102 * @skb: encapsulating message buffer
2103 * @from_pos: offset to extract from
2105 * Returns a new message buffer containing an embedded message. The
2106 * encapsulating message itself is left unchanged.
2108 static struct sk_buff *buf_extract(struct sk_buff *skb, u32 from_pos)
2110 struct tipc_msg *msg = (struct tipc_msg *)(skb->data + from_pos);
2111 u32 size = msg_size(msg);
2114 eb = tipc_buf_acquire(size);
2116 skb_copy_to_linear_data(eb, msg, size);
2122 /* tipc_link_dup_rcv(): Receive a tunnelled DUPLICATE_MSG packet.
2123 * Owner node is locked.
2125 static void tipc_link_dup_rcv(struct tipc_link *l_ptr,
2126 struct sk_buff *t_buf)
2128 struct sk_buff *buf;
2130 if (!tipc_link_is_up(l_ptr))
2133 buf = buf_extract(t_buf, INT_H_SIZE);
2135 pr_warn("%sfailed to extract inner dup pkt\n", link_co_err);
2139 /* Add buffer to deferred queue, if applicable: */
2140 link_handle_out_of_seq_msg(l_ptr, buf);
2143 /* tipc_link_failover_rcv(): Receive a tunnelled ORIGINAL_MSG packet
2144 * Owner node is locked.
2146 static struct sk_buff *tipc_link_failover_rcv(struct tipc_link *l_ptr,
2147 struct sk_buff *t_buf)
2149 struct tipc_msg *t_msg = buf_msg(t_buf);
2150 struct sk_buff *buf = NULL;
2151 struct tipc_msg *msg;
2153 if (tipc_link_is_up(l_ptr))
2154 tipc_link_reset(l_ptr);
2156 /* First failover packet? */
2157 if (l_ptr->exp_msg_count == START_CHANGEOVER)
2158 l_ptr->exp_msg_count = msg_msgcnt(t_msg);
2160 /* Should there be an inner packet? */
2161 if (l_ptr->exp_msg_count) {
2162 l_ptr->exp_msg_count--;
2163 buf = buf_extract(t_buf, INT_H_SIZE);
2165 pr_warn("%sno inner failover pkt\n", link_co_err);
2170 if (less(msg_seqno(msg), l_ptr->reset_checkpoint)) {
2175 if (msg_user(msg) == MSG_FRAGMENTER) {
2176 l_ptr->stats.recv_fragments++;
2177 tipc_link_frag_rcv(&l_ptr->reasm_head,
2183 if ((l_ptr->exp_msg_count == 0) && (l_ptr->flags & LINK_STOPPED)) {
2184 tipc_node_detach_link(l_ptr->owner, l_ptr);
2190 /* tipc_link_tunnel_rcv(): Receive a tunnelled packet, sent
2191 * via other link as result of a failover (ORIGINAL_MSG) or
2192 * a new active link (DUPLICATE_MSG). Failover packets are
2193 * returned to the active link for delivery upwards.
2194 * Owner node is locked.
2196 static int tipc_link_tunnel_rcv(struct tipc_node *n_ptr,
2197 struct sk_buff **buf)
2199 struct sk_buff *t_buf = *buf;
2200 struct tipc_link *l_ptr;
2201 struct tipc_msg *t_msg = buf_msg(t_buf);
2202 u32 bearer_id = msg_bearer_id(t_msg);
2206 if (bearer_id >= MAX_BEARERS)
2209 l_ptr = n_ptr->links[bearer_id];
2213 if (msg_type(t_msg) == DUPLICATE_MSG)
2214 tipc_link_dup_rcv(l_ptr, t_buf);
2215 else if (msg_type(t_msg) == ORIGINAL_MSG)
2216 *buf = tipc_link_failover_rcv(l_ptr, t_buf);
2218 pr_warn("%sunknown tunnel pkt received\n", link_co_err);
2221 return *buf != NULL;
2225 * Bundler functionality:
2227 void tipc_link_bundle_rcv(struct sk_buff *buf)
2229 u32 msgcount = msg_msgcnt(buf_msg(buf));
2230 u32 pos = INT_H_SIZE;
2231 struct sk_buff *obuf;
2233 while (msgcount--) {
2234 obuf = buf_extract(buf, pos);
2236 pr_warn("Link unable to unbundle message(s)\n");
2239 pos += align(msg_size(buf_msg(obuf)));
2240 tipc_net_route_msg(obuf);
2246 * Fragmentation/defragmentation:
2250 * tipc_link_frag_xmit: Entry for buffers needing fragmentation.
2251 * The buffer is complete, inclusive total message length.
2252 * Returns user data length.
2254 static int tipc_link_frag_xmit(struct tipc_link *l_ptr, struct sk_buff *buf)
2256 struct sk_buff *buf_chain = NULL;
2257 struct sk_buff *buf_chain_tail = (struct sk_buff *)&buf_chain;
2258 struct tipc_msg *inmsg = buf_msg(buf);
2259 struct tipc_msg fragm_hdr;
2260 u32 insize = msg_size(inmsg);
2261 u32 dsz = msg_data_sz(inmsg);
2262 unchar *crs = buf->data;
2264 u32 pack_sz = l_ptr->max_pkt;
2265 u32 fragm_sz = pack_sz - INT_H_SIZE;
2269 if (msg_short(inmsg))
2270 destaddr = l_ptr->addr;
2272 destaddr = msg_destnode(inmsg);
2274 /* Prepare reusable fragment header: */
2275 tipc_msg_init(&fragm_hdr, MSG_FRAGMENTER, FIRST_FRAGMENT,
2276 INT_H_SIZE, destaddr);
2278 /* Chop up message: */
2280 struct sk_buff *fragm;
2282 if (rest <= fragm_sz) {
2284 msg_set_type(&fragm_hdr, LAST_FRAGMENT);
2286 fragm = tipc_buf_acquire(fragm_sz + INT_H_SIZE);
2287 if (fragm == NULL) {
2289 kfree_skb_list(buf_chain);
2292 msg_set_size(&fragm_hdr, fragm_sz + INT_H_SIZE);
2294 msg_set_fragm_no(&fragm_hdr, fragm_no);
2295 skb_copy_to_linear_data(fragm, &fragm_hdr, INT_H_SIZE);
2296 skb_copy_to_linear_data_offset(fragm, INT_H_SIZE, crs,
2298 buf_chain_tail->next = fragm;
2299 buf_chain_tail = fragm;
2303 msg_set_type(&fragm_hdr, FRAGMENT);
2307 /* Append chain of fragments to send queue & send them */
2308 l_ptr->long_msg_seq_no++;
2309 link_add_chain_to_outqueue(l_ptr, buf_chain, l_ptr->long_msg_seq_no);
2310 l_ptr->stats.sent_fragments += fragm_no;
2311 l_ptr->stats.sent_fragmented++;
2312 tipc_link_push_queue(l_ptr);
2317 /* tipc_link_frag_rcv(): Called with node lock on. Returns
2318 * the reassembled buffer if message is complete.
2320 int tipc_link_frag_rcv(struct sk_buff **head, struct sk_buff **tail,
2321 struct sk_buff **fbuf)
2323 struct sk_buff *frag = *fbuf;
2324 struct tipc_msg *msg = buf_msg(frag);
2325 u32 fragid = msg_type(msg);
2329 skb_pull(frag, msg_hdr_sz(msg));
2330 if (fragid == FIRST_FRAGMENT) {
2331 if (*head || skb_unclone(frag, GFP_ATOMIC))
2334 skb_frag_list_init(*head);
2338 skb_try_coalesce(*head, frag, &headstolen, &delta)) {
2339 kfree_skb_partial(frag, headstolen);
2343 if (!skb_has_frag_list(*head))
2344 skb_shinfo(*head)->frag_list = frag;
2346 (*tail)->next = frag;
2348 (*head)->truesize += frag->truesize;
2350 if (fragid == LAST_FRAGMENT) {
2352 *tail = *head = NULL;
2353 return LINK_REASM_COMPLETE;
2358 pr_warn_ratelimited("Link unable to reassemble fragmented message\n");
2361 return LINK_REASM_ERROR;
2364 static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance)
2366 if ((tolerance < TIPC_MIN_LINK_TOL) || (tolerance > TIPC_MAX_LINK_TOL))
2369 l_ptr->tolerance = tolerance;
2370 l_ptr->continuity_interval =
2371 ((tolerance / 4) > 500) ? 500 : tolerance / 4;
2372 l_ptr->abort_limit = tolerance / (l_ptr->continuity_interval / 4);
2375 void tipc_link_set_queue_limits(struct tipc_link *l_ptr, u32 window)
2377 /* Data messages from this node, inclusive FIRST_FRAGM */
2378 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE] = window;
2379 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE] = (window / 3) * 4;
2380 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE] = (window / 3) * 5;
2381 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE] = (window / 3) * 6;
2382 /* Transiting data messages,inclusive FIRST_FRAGM */
2383 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE + 4] = 300;
2384 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE + 4] = 600;
2385 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE + 4] = 900;
2386 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE + 4] = 1200;
2387 l_ptr->queue_limit[CONN_MANAGER] = 1200;
2388 l_ptr->queue_limit[CHANGEOVER_PROTOCOL] = 2500;
2389 l_ptr->queue_limit[NAME_DISTRIBUTOR] = 3000;
2390 /* FRAGMENT and LAST_FRAGMENT packets */
2391 l_ptr->queue_limit[MSG_FRAGMENTER] = 4000;
2394 /* tipc_link_find_owner - locate owner node of link by link's name
2395 * @name: pointer to link name string
2396 * @bearer_id: pointer to index in 'node->links' array where the link was found.
2397 * Caller must hold 'tipc_net_lock' to ensure node and bearer are not deleted;
2398 * this also prevents link deletion.
2400 * Returns pointer to node owning the link, or 0 if no matching link is found.
2402 static struct tipc_node *tipc_link_find_owner(const char *link_name,
2403 unsigned int *bearer_id)
2405 struct tipc_link *l_ptr;
2406 struct tipc_node *n_ptr;
2407 struct tipc_node *tmp_n_ptr;
2408 struct tipc_node *found_node = 0;
2413 list_for_each_entry_safe(n_ptr, tmp_n_ptr, &tipc_node_list, list) {
2414 tipc_node_lock(n_ptr);
2415 for (i = 0; i < MAX_BEARERS; i++) {
2416 l_ptr = n_ptr->links[i];
2417 if (l_ptr && !strcmp(l_ptr->name, link_name)) {
2423 tipc_node_unlock(n_ptr);
2431 * link_value_is_valid -- validate proposed link tolerance/priority/window
2433 * @cmd: value type (TIPC_CMD_SET_LINK_*)
2434 * @new_value: the new value
2436 * Returns 1 if value is within range, 0 if not.
2438 static int link_value_is_valid(u16 cmd, u32 new_value)
2441 case TIPC_CMD_SET_LINK_TOL:
2442 return (new_value >= TIPC_MIN_LINK_TOL) &&
2443 (new_value <= TIPC_MAX_LINK_TOL);
2444 case TIPC_CMD_SET_LINK_PRI:
2445 return (new_value <= TIPC_MAX_LINK_PRI);
2446 case TIPC_CMD_SET_LINK_WINDOW:
2447 return (new_value >= TIPC_MIN_LINK_WIN) &&
2448 (new_value <= TIPC_MAX_LINK_WIN);
2454 * link_cmd_set_value - change priority/tolerance/window for link/bearer/media
2455 * @name: ptr to link, bearer, or media name
2456 * @new_value: new value of link, bearer, or media setting
2457 * @cmd: which link, bearer, or media attribute to set (TIPC_CMD_SET_LINK_*)
2459 * Caller must hold 'tipc_net_lock' to ensure link/bearer/media is not deleted.
2461 * Returns 0 if value updated and negative value on error.
2463 static int link_cmd_set_value(const char *name, u32 new_value, u16 cmd)
2465 struct tipc_node *node;
2466 struct tipc_link *l_ptr;
2467 struct tipc_bearer *b_ptr;
2468 struct tipc_media *m_ptr;
2472 node = tipc_link_find_owner(name, &bearer_id);
2474 tipc_node_lock(node);
2475 l_ptr = node->links[bearer_id];
2479 case TIPC_CMD_SET_LINK_TOL:
2480 link_set_supervision_props(l_ptr, new_value);
2481 tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0,
2484 case TIPC_CMD_SET_LINK_PRI:
2485 l_ptr->priority = new_value;
2486 tipc_link_proto_xmit(l_ptr, STATE_MSG, 0, 0,
2489 case TIPC_CMD_SET_LINK_WINDOW:
2490 tipc_link_set_queue_limits(l_ptr, new_value);
2497 tipc_node_unlock(node);
2501 b_ptr = tipc_bearer_find(name);
2504 case TIPC_CMD_SET_LINK_TOL:
2505 b_ptr->tolerance = new_value;
2507 case TIPC_CMD_SET_LINK_PRI:
2508 b_ptr->priority = new_value;
2510 case TIPC_CMD_SET_LINK_WINDOW:
2511 b_ptr->window = new_value;
2520 m_ptr = tipc_media_find(name);
2524 case TIPC_CMD_SET_LINK_TOL:
2525 m_ptr->tolerance = new_value;
2527 case TIPC_CMD_SET_LINK_PRI:
2528 m_ptr->priority = new_value;
2530 case TIPC_CMD_SET_LINK_WINDOW:
2531 m_ptr->window = new_value;
2540 struct sk_buff *tipc_link_cmd_config(const void *req_tlv_area, int req_tlv_space,
2543 struct tipc_link_config *args;
2547 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_CONFIG))
2548 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2550 args = (struct tipc_link_config *)TLV_DATA(req_tlv_area);
2551 new_value = ntohl(args->value);
2553 if (!link_value_is_valid(cmd, new_value))
2554 return tipc_cfg_reply_error_string(
2555 "cannot change, value invalid");
2557 if (!strcmp(args->name, tipc_bclink_name)) {
2558 if ((cmd == TIPC_CMD_SET_LINK_WINDOW) &&
2559 (tipc_bclink_set_queue_limits(new_value) == 0))
2560 return tipc_cfg_reply_none();
2561 return tipc_cfg_reply_error_string(TIPC_CFG_NOT_SUPPORTED
2562 " (cannot change setting on broadcast link)");
2565 read_lock_bh(&tipc_net_lock);
2566 res = link_cmd_set_value(args->name, new_value, cmd);
2567 read_unlock_bh(&tipc_net_lock);
2569 return tipc_cfg_reply_error_string("cannot change link setting");
2571 return tipc_cfg_reply_none();
2575 * link_reset_statistics - reset link statistics
2576 * @l_ptr: pointer to link
2578 static void link_reset_statistics(struct tipc_link *l_ptr)
2580 memset(&l_ptr->stats, 0, sizeof(l_ptr->stats));
2581 l_ptr->stats.sent_info = l_ptr->next_out_no;
2582 l_ptr->stats.recv_info = l_ptr->next_in_no;
2585 struct sk_buff *tipc_link_cmd_reset_stats(const void *req_tlv_area, int req_tlv_space)
2588 struct tipc_link *l_ptr;
2589 struct tipc_node *node;
2590 unsigned int bearer_id;
2592 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2593 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2595 link_name = (char *)TLV_DATA(req_tlv_area);
2596 if (!strcmp(link_name, tipc_bclink_name)) {
2597 if (tipc_bclink_reset_stats())
2598 return tipc_cfg_reply_error_string("link not found");
2599 return tipc_cfg_reply_none();
2601 read_lock_bh(&tipc_net_lock);
2602 node = tipc_link_find_owner(link_name, &bearer_id);
2604 read_unlock_bh(&tipc_net_lock);
2605 return tipc_cfg_reply_error_string("link not found");
2607 tipc_node_lock(node);
2608 l_ptr = node->links[bearer_id];
2610 tipc_node_unlock(node);
2611 read_unlock_bh(&tipc_net_lock);
2612 return tipc_cfg_reply_error_string("link not found");
2614 link_reset_statistics(l_ptr);
2615 tipc_node_unlock(node);
2616 read_unlock_bh(&tipc_net_lock);
2617 return tipc_cfg_reply_none();
2621 * percent - convert count to a percentage of total (rounding up or down)
2623 static u32 percent(u32 count, u32 total)
2625 return (count * 100 + (total / 2)) / total;
2629 * tipc_link_stats - print link statistics
2631 * @buf: print buffer area
2632 * @buf_size: size of print buffer area
2634 * Returns length of print buffer data string (or 0 if error)
2636 static int tipc_link_stats(const char *name, char *buf, const u32 buf_size)
2638 struct tipc_link *l;
2639 struct tipc_stats *s;
2640 struct tipc_node *node;
2642 u32 profile_total = 0;
2643 unsigned int bearer_id;
2646 if (!strcmp(name, tipc_bclink_name))
2647 return tipc_bclink_stats(buf, buf_size);
2649 read_lock_bh(&tipc_net_lock);
2650 node = tipc_link_find_owner(name, &bearer_id);
2652 read_unlock_bh(&tipc_net_lock);
2655 tipc_node_lock(node);
2657 l = node->links[bearer_id];
2659 tipc_node_unlock(node);
2660 read_unlock_bh(&tipc_net_lock);
2666 if (tipc_link_is_active(l))
2668 else if (tipc_link_is_up(l))
2673 ret = tipc_snprintf(buf, buf_size, "Link <%s>\n"
2674 " %s MTU:%u Priority:%u Tolerance:%u ms"
2675 " Window:%u packets\n",
2676 l->name, status, l->max_pkt, l->priority,
2677 l->tolerance, l->queue_limit[0]);
2679 ret += tipc_snprintf(buf + ret, buf_size - ret,
2680 " RX packets:%u fragments:%u/%u bundles:%u/%u\n",
2681 l->next_in_no - s->recv_info, s->recv_fragments,
2682 s->recv_fragmented, s->recv_bundles,
2685 ret += tipc_snprintf(buf + ret, buf_size - ret,
2686 " TX packets:%u fragments:%u/%u bundles:%u/%u\n",
2687 l->next_out_no - s->sent_info, s->sent_fragments,
2688 s->sent_fragmented, s->sent_bundles,
2691 profile_total = s->msg_length_counts;
2695 ret += tipc_snprintf(buf + ret, buf_size - ret,
2696 " TX profile sample:%u packets average:%u octets\n"
2697 " 0-64:%u%% -256:%u%% -1024:%u%% -4096:%u%% "
2698 "-16384:%u%% -32768:%u%% -66000:%u%%\n",
2699 s->msg_length_counts,
2700 s->msg_lengths_total / profile_total,
2701 percent(s->msg_length_profile[0], profile_total),
2702 percent(s->msg_length_profile[1], profile_total),
2703 percent(s->msg_length_profile[2], profile_total),
2704 percent(s->msg_length_profile[3], profile_total),
2705 percent(s->msg_length_profile[4], profile_total),
2706 percent(s->msg_length_profile[5], profile_total),
2707 percent(s->msg_length_profile[6], profile_total));
2709 ret += tipc_snprintf(buf + ret, buf_size - ret,
2710 " RX states:%u probes:%u naks:%u defs:%u"
2711 " dups:%u\n", s->recv_states, s->recv_probes,
2712 s->recv_nacks, s->deferred_recv, s->duplicates);
2714 ret += tipc_snprintf(buf + ret, buf_size - ret,
2715 " TX states:%u probes:%u naks:%u acks:%u"
2716 " dups:%u\n", s->sent_states, s->sent_probes,
2717 s->sent_nacks, s->sent_acks, s->retransmitted);
2719 ret += tipc_snprintf(buf + ret, buf_size - ret,
2720 " Congestion link:%u Send queue"
2721 " max:%u avg:%u\n", s->link_congs,
2722 s->max_queue_sz, s->queue_sz_counts ?
2723 (s->accu_queue_sz / s->queue_sz_counts) : 0);
2725 tipc_node_unlock(node);
2726 read_unlock_bh(&tipc_net_lock);
2730 struct sk_buff *tipc_link_cmd_show_stats(const void *req_tlv_area, int req_tlv_space)
2732 struct sk_buff *buf;
2733 struct tlv_desc *rep_tlv;
2738 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2739 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2741 buf = tipc_cfg_reply_alloc(TLV_SPACE(ULTRA_STRING_MAX_LEN));
2745 rep_tlv = (struct tlv_desc *)buf->data;
2746 pb = TLV_DATA(rep_tlv);
2747 pb_len = ULTRA_STRING_MAX_LEN;
2748 str_len = tipc_link_stats((char *)TLV_DATA(req_tlv_area),
2752 return tipc_cfg_reply_error_string("link not found");
2754 str_len += 1; /* for "\0" */
2755 skb_put(buf, TLV_SPACE(str_len));
2756 TLV_SET(rep_tlv, TIPC_TLV_ULTRA_STRING, NULL, str_len);
2762 * tipc_link_get_max_pkt - get maximum packet size to use when sending to destination
2763 * @dest: network address of destination node
2764 * @selector: used to select from set of active links
2766 * If no active link can be found, uses default maximum packet size.
2768 u32 tipc_link_get_max_pkt(u32 dest, u32 selector)
2770 struct tipc_node *n_ptr;
2771 struct tipc_link *l_ptr;
2772 u32 res = MAX_PKT_DEFAULT;
2774 if (dest == tipc_own_addr)
2775 return MAX_MSG_SIZE;
2777 read_lock_bh(&tipc_net_lock);
2778 n_ptr = tipc_node_find(dest);
2780 tipc_node_lock(n_ptr);
2781 l_ptr = n_ptr->active_links[selector & 1];
2783 res = l_ptr->max_pkt;
2784 tipc_node_unlock(n_ptr);
2786 read_unlock_bh(&tipc_net_lock);
2790 static void link_print(struct tipc_link *l_ptr, const char *str)
2792 pr_info("%s Link %x<%s>:", str, l_ptr->addr, l_ptr->b_ptr->name);
2794 if (link_working_unknown(l_ptr))
2796 else if (link_reset_reset(l_ptr))
2798 else if (link_reset_unknown(l_ptr))
2800 else if (link_working_working(l_ptr))