2 * net/tipc/link.c: TIPC link code
4 * Copyright (c) 1996-2007, 2012, 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
79 * struct tipc_link_name - deconstructed link name
80 * @addr_local: network address of node at this end
81 * @if_local: name of interface at this end
82 * @addr_peer: network address of node at far end
83 * @if_peer: name of interface at far end
85 struct tipc_link_name {
87 char if_local[TIPC_MAX_IF_NAME];
89 char if_peer[TIPC_MAX_IF_NAME];
92 static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
94 static void link_recv_proto_msg(struct tipc_link *l_ptr, struct sk_buff *buf);
95 static int link_recv_changeover_msg(struct tipc_link **l_ptr,
96 struct sk_buff **buf);
97 static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance);
98 static int link_send_sections_long(struct tipc_port *sender,
99 struct iovec const *msg_sect,
100 u32 num_sect, unsigned int total_len,
102 static void link_state_event(struct tipc_link *l_ptr, u32 event);
103 static void link_reset_statistics(struct tipc_link *l_ptr);
104 static void link_print(struct tipc_link *l_ptr, const char *str);
105 static void link_start(struct tipc_link *l_ptr);
106 static int link_send_long_buf(struct tipc_link *l_ptr, struct sk_buff *buf);
107 static void tipc_link_send_sync(struct tipc_link *l);
108 static void tipc_link_recv_sync(struct tipc_node *n, struct sk_buff *buf);
111 * Simple link routines
113 static unsigned int align(unsigned int i)
115 return (i + 3) & ~3u;
118 static void link_init_max_pkt(struct tipc_link *l_ptr)
122 max_pkt = (l_ptr->b_ptr->mtu & ~3);
123 if (max_pkt > MAX_MSG_SIZE)
124 max_pkt = MAX_MSG_SIZE;
126 l_ptr->max_pkt_target = max_pkt;
127 if (l_ptr->max_pkt_target < MAX_PKT_DEFAULT)
128 l_ptr->max_pkt = l_ptr->max_pkt_target;
130 l_ptr->max_pkt = MAX_PKT_DEFAULT;
132 l_ptr->max_pkt_probes = 0;
135 static u32 link_next_sent(struct tipc_link *l_ptr)
138 return buf_seqno(l_ptr->next_out);
139 return mod(l_ptr->next_out_no);
142 static u32 link_last_sent(struct tipc_link *l_ptr)
144 return mod(link_next_sent(l_ptr) - 1);
148 * Simple non-static link routines (i.e. referenced outside this file)
150 int tipc_link_is_up(struct tipc_link *l_ptr)
154 return link_working_working(l_ptr) || link_working_unknown(l_ptr);
157 int tipc_link_is_active(struct tipc_link *l_ptr)
159 return (l_ptr->owner->active_links[0] == l_ptr) ||
160 (l_ptr->owner->active_links[1] == l_ptr);
164 * link_name_validate - validate & (optionally) deconstruct tipc_link name
165 * @name: ptr to link name string
166 * @name_parts: ptr to area for link name components (or NULL if not needed)
168 * Returns 1 if link name is valid, otherwise 0.
170 static int link_name_validate(const char *name,
171 struct tipc_link_name *name_parts)
173 char name_copy[TIPC_MAX_LINK_NAME];
179 u32 z_local, c_local, n_local;
180 u32 z_peer, c_peer, n_peer;
184 /* copy link name & ensure length is OK */
185 name_copy[TIPC_MAX_LINK_NAME - 1] = 0;
186 /* need above in case non-Posix strncpy() doesn't pad with nulls */
187 strncpy(name_copy, name, TIPC_MAX_LINK_NAME);
188 if (name_copy[TIPC_MAX_LINK_NAME - 1] != 0)
191 /* ensure all component parts of link name are present */
192 addr_local = name_copy;
193 if_local = strchr(addr_local, ':');
194 if (if_local == NULL)
197 addr_peer = strchr(if_local, '-');
198 if (addr_peer == NULL)
201 if_local_len = addr_peer - if_local;
202 if_peer = strchr(addr_peer, ':');
206 if_peer_len = strlen(if_peer) + 1;
208 /* validate component parts of link name */
209 if ((sscanf(addr_local, "%u.%u.%u%c",
210 &z_local, &c_local, &n_local, &dummy) != 3) ||
211 (sscanf(addr_peer, "%u.%u.%u%c",
212 &z_peer, &c_peer, &n_peer, &dummy) != 3) ||
213 (z_local > 255) || (c_local > 4095) || (n_local > 4095) ||
214 (z_peer > 255) || (c_peer > 4095) || (n_peer > 4095) ||
215 (if_local_len <= 1) || (if_local_len > TIPC_MAX_IF_NAME) ||
216 (if_peer_len <= 1) || (if_peer_len > TIPC_MAX_IF_NAME))
219 /* return link name components, if necessary */
221 name_parts->addr_local = tipc_addr(z_local, c_local, n_local);
222 strcpy(name_parts->if_local, if_local);
223 name_parts->addr_peer = tipc_addr(z_peer, c_peer, n_peer);
224 strcpy(name_parts->if_peer, if_peer);
230 * link_timeout - handle expiration of link timer
231 * @l_ptr: pointer to link
233 * This routine must not grab "tipc_net_lock" to avoid a potential deadlock conflict
234 * with tipc_link_delete(). (There is no risk that the node will be deleted by
235 * another thread because tipc_link_delete() always cancels the link timer before
236 * tipc_node_delete() is called.)
238 static void link_timeout(struct tipc_link *l_ptr)
240 tipc_node_lock(l_ptr->owner);
242 /* update counters used in statistical profiling of send traffic */
243 l_ptr->stats.accu_queue_sz += l_ptr->out_queue_size;
244 l_ptr->stats.queue_sz_counts++;
246 if (l_ptr->first_out) {
247 struct tipc_msg *msg = buf_msg(l_ptr->first_out);
248 u32 length = msg_size(msg);
250 if ((msg_user(msg) == MSG_FRAGMENTER) &&
251 (msg_type(msg) == FIRST_FRAGMENT)) {
252 length = msg_size(msg_get_wrapped(msg));
255 l_ptr->stats.msg_lengths_total += length;
256 l_ptr->stats.msg_length_counts++;
258 l_ptr->stats.msg_length_profile[0]++;
259 else if (length <= 256)
260 l_ptr->stats.msg_length_profile[1]++;
261 else if (length <= 1024)
262 l_ptr->stats.msg_length_profile[2]++;
263 else if (length <= 4096)
264 l_ptr->stats.msg_length_profile[3]++;
265 else if (length <= 16384)
266 l_ptr->stats.msg_length_profile[4]++;
267 else if (length <= 32768)
268 l_ptr->stats.msg_length_profile[5]++;
270 l_ptr->stats.msg_length_profile[6]++;
274 /* do all other link processing performed on a periodic basis */
276 link_state_event(l_ptr, TIMEOUT_EVT);
279 tipc_link_push_queue(l_ptr);
281 tipc_node_unlock(l_ptr->owner);
284 static void link_set_timer(struct tipc_link *l_ptr, u32 time)
286 k_start_timer(&l_ptr->timer, time);
290 * tipc_link_create - create a new link
291 * @n_ptr: pointer to associated node
292 * @b_ptr: pointer to associated bearer
293 * @media_addr: media address to use when sending messages over link
295 * Returns pointer to link.
297 struct tipc_link *tipc_link_create(struct tipc_node *n_ptr,
298 struct tipc_bearer *b_ptr,
299 const struct tipc_media_addr *media_addr)
301 struct tipc_link *l_ptr;
302 struct tipc_msg *msg;
304 char addr_string[16];
305 u32 peer = n_ptr->addr;
307 if (n_ptr->link_cnt >= 2) {
308 tipc_addr_string_fill(addr_string, n_ptr->addr);
309 pr_err("Attempt to establish third link to %s\n", addr_string);
313 if (n_ptr->links[b_ptr->identity]) {
314 tipc_addr_string_fill(addr_string, n_ptr->addr);
315 pr_err("Attempt to establish second link on <%s> to %s\n",
316 b_ptr->name, addr_string);
320 l_ptr = kzalloc(sizeof(*l_ptr), GFP_ATOMIC);
322 pr_warn("Link creation failed, no memory\n");
327 if_name = strchr(b_ptr->name, ':') + 1;
328 sprintf(l_ptr->name, "%u.%u.%u:%s-%u.%u.%u:unknown",
329 tipc_zone(tipc_own_addr), tipc_cluster(tipc_own_addr),
330 tipc_node(tipc_own_addr),
332 tipc_zone(peer), tipc_cluster(peer), tipc_node(peer));
333 /* note: peer i/f name is updated by reset/activate message */
334 memcpy(&l_ptr->media_addr, media_addr, sizeof(*media_addr));
335 l_ptr->owner = n_ptr;
336 l_ptr->checkpoint = 1;
337 l_ptr->peer_session = INVALID_SESSION;
338 l_ptr->b_ptr = b_ptr;
339 link_set_supervision_props(l_ptr, b_ptr->tolerance);
340 l_ptr->state = RESET_UNKNOWN;
342 l_ptr->pmsg = (struct tipc_msg *)&l_ptr->proto_msg;
344 tipc_msg_init(msg, LINK_PROTOCOL, RESET_MSG, INT_H_SIZE, l_ptr->addr);
345 msg_set_size(msg, sizeof(l_ptr->proto_msg));
346 msg_set_session(msg, (tipc_random & 0xffff));
347 msg_set_bearer_id(msg, b_ptr->identity);
348 strcpy((char *)msg_data(msg), if_name);
350 l_ptr->priority = b_ptr->priority;
351 tipc_link_set_queue_limits(l_ptr, b_ptr->window);
353 link_init_max_pkt(l_ptr);
355 l_ptr->next_out_no = 1;
356 INIT_LIST_HEAD(&l_ptr->waiting_ports);
358 link_reset_statistics(l_ptr);
360 tipc_node_attach_link(n_ptr, l_ptr);
362 k_init_timer(&l_ptr->timer, (Handler)link_timeout, (unsigned long)l_ptr);
363 list_add_tail(&l_ptr->link_list, &b_ptr->links);
364 tipc_k_signal((Handler)link_start, (unsigned long)l_ptr);
370 * tipc_link_delete - delete a link
371 * @l_ptr: pointer to link
373 * Note: 'tipc_net_lock' is write_locked, bearer is locked.
374 * This routine must not grab the node lock until after link timer cancellation
375 * to avoid a potential deadlock situation.
377 void tipc_link_delete(struct tipc_link *l_ptr)
380 pr_err("Attempt to delete non-existent link\n");
384 k_cancel_timer(&l_ptr->timer);
386 tipc_node_lock(l_ptr->owner);
387 tipc_link_reset(l_ptr);
388 tipc_node_detach_link(l_ptr->owner, l_ptr);
389 tipc_link_stop(l_ptr);
390 list_del_init(&l_ptr->link_list);
391 tipc_node_unlock(l_ptr->owner);
392 k_term_timer(&l_ptr->timer);
396 static void link_start(struct tipc_link *l_ptr)
398 tipc_node_lock(l_ptr->owner);
399 link_state_event(l_ptr, STARTING_EVT);
400 tipc_node_unlock(l_ptr->owner);
404 * link_schedule_port - schedule port for deferred sending
405 * @l_ptr: pointer to link
406 * @origport: reference to sending port
407 * @sz: amount of data to be sent
409 * Schedules port for renewed sending of messages after link congestion
412 static int link_schedule_port(struct tipc_link *l_ptr, u32 origport, u32 sz)
414 struct tipc_port *p_ptr;
416 spin_lock_bh(&tipc_port_list_lock);
417 p_ptr = tipc_port_lock(origport);
421 if (!list_empty(&p_ptr->wait_list))
423 p_ptr->congested = 1;
424 p_ptr->waiting_pkts = 1 + ((sz - 1) / l_ptr->max_pkt);
425 list_add_tail(&p_ptr->wait_list, &l_ptr->waiting_ports);
426 l_ptr->stats.link_congs++;
428 tipc_port_unlock(p_ptr);
430 spin_unlock_bh(&tipc_port_list_lock);
434 void tipc_link_wakeup_ports(struct tipc_link *l_ptr, int all)
436 struct tipc_port *p_ptr;
437 struct tipc_port *temp_p_ptr;
438 int win = l_ptr->queue_limit[0] - l_ptr->out_queue_size;
444 if (!spin_trylock_bh(&tipc_port_list_lock))
446 if (link_congested(l_ptr))
448 list_for_each_entry_safe(p_ptr, temp_p_ptr, &l_ptr->waiting_ports,
452 list_del_init(&p_ptr->wait_list);
453 spin_lock_bh(p_ptr->lock);
454 p_ptr->congested = 0;
455 p_ptr->wakeup(p_ptr);
456 win -= p_ptr->waiting_pkts;
457 spin_unlock_bh(p_ptr->lock);
461 spin_unlock_bh(&tipc_port_list_lock);
465 * link_release_outqueue - purge link's outbound message queue
466 * @l_ptr: pointer to link
468 static void link_release_outqueue(struct tipc_link *l_ptr)
470 struct sk_buff *buf = l_ptr->first_out;
471 struct sk_buff *next;
478 l_ptr->first_out = NULL;
479 l_ptr->out_queue_size = 0;
483 * tipc_link_reset_fragments - purge link's inbound message fragments queue
484 * @l_ptr: pointer to link
486 void tipc_link_reset_fragments(struct tipc_link *l_ptr)
488 struct sk_buff *buf = l_ptr->defragm_buf;
489 struct sk_buff *next;
496 l_ptr->defragm_buf = NULL;
500 * tipc_link_stop - purge all inbound and outbound messages associated with link
501 * @l_ptr: pointer to link
503 void tipc_link_stop(struct tipc_link *l_ptr)
506 struct sk_buff *next;
508 buf = l_ptr->oldest_deferred_in;
515 buf = l_ptr->first_out;
522 tipc_link_reset_fragments(l_ptr);
524 kfree_skb(l_ptr->proto_msg_queue);
525 l_ptr->proto_msg_queue = NULL;
528 void tipc_link_reset(struct tipc_link *l_ptr)
531 u32 prev_state = l_ptr->state;
532 u32 checkpoint = l_ptr->next_in_no;
533 int was_active_link = tipc_link_is_active(l_ptr);
535 msg_set_session(l_ptr->pmsg, ((msg_session(l_ptr->pmsg) + 1) & 0xffff));
537 /* Link is down, accept any session */
538 l_ptr->peer_session = INVALID_SESSION;
540 /* Prepare for max packet size negotiation */
541 link_init_max_pkt(l_ptr);
543 l_ptr->state = RESET_UNKNOWN;
545 if ((prev_state == RESET_UNKNOWN) || (prev_state == RESET_RESET))
548 tipc_node_link_down(l_ptr->owner, l_ptr);
549 tipc_bearer_remove_dest(l_ptr->b_ptr, l_ptr->addr);
551 if (was_active_link && tipc_node_active_links(l_ptr->owner) &&
552 l_ptr->owner->permit_changeover) {
553 l_ptr->reset_checkpoint = checkpoint;
554 l_ptr->exp_msg_count = START_CHANGEOVER;
557 /* Clean up all queues: */
558 link_release_outqueue(l_ptr);
559 kfree_skb(l_ptr->proto_msg_queue);
560 l_ptr->proto_msg_queue = NULL;
561 buf = l_ptr->oldest_deferred_in;
563 struct sk_buff *next = buf->next;
567 if (!list_empty(&l_ptr->waiting_ports))
568 tipc_link_wakeup_ports(l_ptr, 1);
570 l_ptr->retransm_queue_head = 0;
571 l_ptr->retransm_queue_size = 0;
572 l_ptr->last_out = NULL;
573 l_ptr->first_out = NULL;
574 l_ptr->next_out = NULL;
575 l_ptr->unacked_window = 0;
576 l_ptr->checkpoint = 1;
577 l_ptr->next_out_no = 1;
578 l_ptr->deferred_inqueue_sz = 0;
579 l_ptr->oldest_deferred_in = NULL;
580 l_ptr->newest_deferred_in = NULL;
581 l_ptr->fsm_msg_cnt = 0;
582 l_ptr->stale_count = 0;
583 link_reset_statistics(l_ptr);
587 static void link_activate(struct tipc_link *l_ptr)
589 l_ptr->next_in_no = l_ptr->stats.recv_info = 1;
590 tipc_node_link_up(l_ptr->owner, l_ptr);
591 tipc_bearer_add_dest(l_ptr->b_ptr, l_ptr->addr);
595 * link_state_event - link finite state machine
596 * @l_ptr: pointer to link
597 * @event: state machine event to process
599 static void link_state_event(struct tipc_link *l_ptr, unsigned int event)
601 struct tipc_link *other;
602 u32 cont_intv = l_ptr->continuity_interval;
604 if (!l_ptr->started && (event != STARTING_EVT))
605 return; /* Not yet. */
607 if (link_blocked(l_ptr)) {
608 if (event == TIMEOUT_EVT)
609 link_set_timer(l_ptr, cont_intv);
610 return; /* Changeover going on */
613 switch (l_ptr->state) {
614 case WORKING_WORKING:
616 case TRAFFIC_MSG_EVT:
620 if (l_ptr->next_in_no != l_ptr->checkpoint) {
621 l_ptr->checkpoint = l_ptr->next_in_no;
622 if (tipc_bclink_acks_missing(l_ptr->owner)) {
623 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
625 l_ptr->fsm_msg_cnt++;
626 } else if (l_ptr->max_pkt < l_ptr->max_pkt_target) {
627 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
629 l_ptr->fsm_msg_cnt++;
631 link_set_timer(l_ptr, cont_intv);
634 l_ptr->state = WORKING_UNKNOWN;
635 l_ptr->fsm_msg_cnt = 0;
636 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
637 l_ptr->fsm_msg_cnt++;
638 link_set_timer(l_ptr, cont_intv / 4);
641 pr_info("%s<%s>, requested by peer\n", link_rst_msg,
643 tipc_link_reset(l_ptr);
644 l_ptr->state = RESET_RESET;
645 l_ptr->fsm_msg_cnt = 0;
646 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 0, 0, 0, 0, 0);
647 l_ptr->fsm_msg_cnt++;
648 link_set_timer(l_ptr, cont_intv);
651 pr_err("%s%u in WW state\n", link_unk_evt, event);
654 case WORKING_UNKNOWN:
656 case TRAFFIC_MSG_EVT:
658 l_ptr->state = WORKING_WORKING;
659 l_ptr->fsm_msg_cnt = 0;
660 link_set_timer(l_ptr, cont_intv);
663 pr_info("%s<%s>, requested by peer while probing\n",
664 link_rst_msg, l_ptr->name);
665 tipc_link_reset(l_ptr);
666 l_ptr->state = RESET_RESET;
667 l_ptr->fsm_msg_cnt = 0;
668 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 0, 0, 0, 0, 0);
669 l_ptr->fsm_msg_cnt++;
670 link_set_timer(l_ptr, cont_intv);
673 if (l_ptr->next_in_no != l_ptr->checkpoint) {
674 l_ptr->state = WORKING_WORKING;
675 l_ptr->fsm_msg_cnt = 0;
676 l_ptr->checkpoint = l_ptr->next_in_no;
677 if (tipc_bclink_acks_missing(l_ptr->owner)) {
678 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
680 l_ptr->fsm_msg_cnt++;
682 link_set_timer(l_ptr, cont_intv);
683 } else if (l_ptr->fsm_msg_cnt < l_ptr->abort_limit) {
684 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
686 l_ptr->fsm_msg_cnt++;
687 link_set_timer(l_ptr, cont_intv / 4);
688 } else { /* Link has failed */
689 pr_warn("%s<%s>, peer not responding\n",
690 link_rst_msg, l_ptr->name);
691 tipc_link_reset(l_ptr);
692 l_ptr->state = RESET_UNKNOWN;
693 l_ptr->fsm_msg_cnt = 0;
694 tipc_link_send_proto_msg(l_ptr, RESET_MSG,
696 l_ptr->fsm_msg_cnt++;
697 link_set_timer(l_ptr, cont_intv);
701 pr_err("%s%u in WU state\n", link_unk_evt, event);
706 case TRAFFIC_MSG_EVT:
709 other = l_ptr->owner->active_links[0];
710 if (other && link_working_unknown(other))
712 l_ptr->state = WORKING_WORKING;
713 l_ptr->fsm_msg_cnt = 0;
714 link_activate(l_ptr);
715 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
716 l_ptr->fsm_msg_cnt++;
717 if (l_ptr->owner->working_links == 1)
718 tipc_link_send_sync(l_ptr);
719 link_set_timer(l_ptr, cont_intv);
722 l_ptr->state = RESET_RESET;
723 l_ptr->fsm_msg_cnt = 0;
724 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 1, 0, 0, 0, 0);
725 l_ptr->fsm_msg_cnt++;
726 link_set_timer(l_ptr, cont_intv);
732 tipc_link_send_proto_msg(l_ptr, RESET_MSG, 0, 0, 0, 0, 0);
733 l_ptr->fsm_msg_cnt++;
734 link_set_timer(l_ptr, cont_intv);
737 pr_err("%s%u in RU state\n", link_unk_evt, event);
742 case TRAFFIC_MSG_EVT:
744 other = l_ptr->owner->active_links[0];
745 if (other && link_working_unknown(other))
747 l_ptr->state = WORKING_WORKING;
748 l_ptr->fsm_msg_cnt = 0;
749 link_activate(l_ptr);
750 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
751 l_ptr->fsm_msg_cnt++;
752 if (l_ptr->owner->working_links == 1)
753 tipc_link_send_sync(l_ptr);
754 link_set_timer(l_ptr, cont_intv);
759 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 0, 0, 0, 0, 0);
760 l_ptr->fsm_msg_cnt++;
761 link_set_timer(l_ptr, cont_intv);
764 pr_err("%s%u in RR state\n", link_unk_evt, event);
768 pr_err("Unknown link state %u/%u\n", l_ptr->state, event);
773 * link_bundle_buf(): Append contents of a buffer to
774 * the tail of an existing one.
776 static int link_bundle_buf(struct tipc_link *l_ptr, struct sk_buff *bundler,
779 struct tipc_msg *bundler_msg = buf_msg(bundler);
780 struct tipc_msg *msg = buf_msg(buf);
781 u32 size = msg_size(msg);
782 u32 bundle_size = msg_size(bundler_msg);
783 u32 to_pos = align(bundle_size);
784 u32 pad = to_pos - bundle_size;
786 if (msg_user(bundler_msg) != MSG_BUNDLER)
788 if (msg_type(bundler_msg) != OPEN_MSG)
790 if (skb_tailroom(bundler) < (pad + size))
792 if (l_ptr->max_pkt < (to_pos + size))
795 skb_put(bundler, pad + size);
796 skb_copy_to_linear_data_offset(bundler, to_pos, buf->data, size);
797 msg_set_size(bundler_msg, to_pos + size);
798 msg_set_msgcnt(bundler_msg, msg_msgcnt(bundler_msg) + 1);
800 l_ptr->stats.sent_bundled++;
804 static void link_add_to_outqueue(struct tipc_link *l_ptr,
806 struct tipc_msg *msg)
808 u32 ack = mod(l_ptr->next_in_no - 1);
809 u32 seqno = mod(l_ptr->next_out_no++);
811 msg_set_word(msg, 2, ((ack << 16) | seqno));
812 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
814 if (l_ptr->first_out) {
815 l_ptr->last_out->next = buf;
816 l_ptr->last_out = buf;
818 l_ptr->first_out = l_ptr->last_out = buf;
820 l_ptr->out_queue_size++;
821 if (l_ptr->out_queue_size > l_ptr->stats.max_queue_sz)
822 l_ptr->stats.max_queue_sz = l_ptr->out_queue_size;
825 static void link_add_chain_to_outqueue(struct tipc_link *l_ptr,
826 struct sk_buff *buf_chain,
830 struct tipc_msg *msg;
832 if (!l_ptr->next_out)
833 l_ptr->next_out = buf_chain;
836 buf_chain = buf_chain->next;
839 msg_set_long_msgno(msg, long_msgno);
840 link_add_to_outqueue(l_ptr, buf, msg);
845 * tipc_link_send_buf() is the 'full path' for messages, called from
846 * inside TIPC when the 'fast path' in tipc_send_buf
847 * has failed, and from link_send()
849 int tipc_link_send_buf(struct tipc_link *l_ptr, struct sk_buff *buf)
851 struct tipc_msg *msg = buf_msg(buf);
852 u32 size = msg_size(msg);
853 u32 dsz = msg_data_sz(msg);
854 u32 queue_size = l_ptr->out_queue_size;
855 u32 imp = tipc_msg_tot_importance(msg);
856 u32 queue_limit = l_ptr->queue_limit[imp];
857 u32 max_packet = l_ptr->max_pkt;
859 /* Match msg importance against queue limits: */
860 if (unlikely(queue_size >= queue_limit)) {
861 if (imp <= TIPC_CRITICAL_IMPORTANCE) {
862 link_schedule_port(l_ptr, msg_origport(msg), size);
867 if (imp > CONN_MANAGER) {
868 pr_warn("%s<%s>, send queue full", link_rst_msg,
870 tipc_link_reset(l_ptr);
875 /* Fragmentation needed ? */
876 if (size > max_packet)
877 return link_send_long_buf(l_ptr, buf);
879 /* Packet can be queued or sent. */
880 if (likely(!tipc_bearer_blocked(l_ptr->b_ptr) &&
881 !link_congested(l_ptr))) {
882 link_add_to_outqueue(l_ptr, buf, msg);
884 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
885 l_ptr->unacked_window = 0;
888 /* Congestion: can message be bundled ? */
889 if ((msg_user(msg) != CHANGEOVER_PROTOCOL) &&
890 (msg_user(msg) != MSG_FRAGMENTER)) {
892 /* Try adding message to an existing bundle */
893 if (l_ptr->next_out &&
894 link_bundle_buf(l_ptr, l_ptr->last_out, buf))
897 /* Try creating a new bundle */
898 if (size <= max_packet * 2 / 3) {
899 struct sk_buff *bundler = tipc_buf_acquire(max_packet);
900 struct tipc_msg bundler_hdr;
903 tipc_msg_init(&bundler_hdr, MSG_BUNDLER, OPEN_MSG,
904 INT_H_SIZE, l_ptr->addr);
905 skb_copy_to_linear_data(bundler, &bundler_hdr,
907 skb_trim(bundler, INT_H_SIZE);
908 link_bundle_buf(l_ptr, bundler, buf);
911 l_ptr->stats.sent_bundles++;
915 if (!l_ptr->next_out)
916 l_ptr->next_out = buf;
917 link_add_to_outqueue(l_ptr, buf, msg);
922 * tipc_link_send(): same as tipc_link_send_buf(), but the link to use has
923 * not been selected yet, and the the owner node is not locked
924 * Called by TIPC internal users, e.g. the name distributor
926 int tipc_link_send(struct sk_buff *buf, u32 dest, u32 selector)
928 struct tipc_link *l_ptr;
929 struct tipc_node *n_ptr;
930 int res = -ELINKCONG;
932 read_lock_bh(&tipc_net_lock);
933 n_ptr = tipc_node_find(dest);
935 tipc_node_lock(n_ptr);
936 l_ptr = n_ptr->active_links[selector & 1];
938 res = tipc_link_send_buf(l_ptr, buf);
941 tipc_node_unlock(n_ptr);
945 read_unlock_bh(&tipc_net_lock);
950 * tipc_link_send_sync - synchronize broadcast link endpoints.
952 * Give a newly added peer node the sequence number where it should
953 * start receiving and acking broadcast packets.
955 * Called with node locked
957 static void tipc_link_send_sync(struct tipc_link *l)
960 struct tipc_msg *msg;
962 buf = tipc_buf_acquire(INT_H_SIZE);
967 tipc_msg_init(msg, BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE, l->addr);
968 msg_set_last_bcast(msg, l->owner->bclink.acked);
969 link_add_chain_to_outqueue(l, buf, 0);
970 tipc_link_push_queue(l);
974 * tipc_link_recv_sync - synchronize broadcast link endpoints.
975 * Receive the sequence number where we should start receiving and
976 * acking broadcast packets from a newly added peer node, and open
977 * up for reception of such packets.
979 * Called with node locked
981 static void tipc_link_recv_sync(struct tipc_node *n, struct sk_buff *buf)
983 struct tipc_msg *msg = buf_msg(buf);
985 n->bclink.last_sent = n->bclink.last_in = msg_last_bcast(msg);
986 n->bclink.recv_permitted = true;
991 * tipc_link_send_names - send name table entries to new neighbor
993 * Send routine for bulk delivery of name table messages when contact
994 * with a new neighbor occurs. No link congestion checking is performed
995 * because name table messages *must* be delivered. The messages must be
996 * small enough not to require fragmentation.
997 * Called without any locks held.
999 void tipc_link_send_names(struct list_head *message_list, u32 dest)
1001 struct tipc_node *n_ptr;
1002 struct tipc_link *l_ptr;
1003 struct sk_buff *buf;
1004 struct sk_buff *temp_buf;
1006 if (list_empty(message_list))
1009 read_lock_bh(&tipc_net_lock);
1010 n_ptr = tipc_node_find(dest);
1012 tipc_node_lock(n_ptr);
1013 l_ptr = n_ptr->active_links[0];
1015 /* convert circular list to linear list */
1016 ((struct sk_buff *)message_list->prev)->next = NULL;
1017 link_add_chain_to_outqueue(l_ptr,
1018 (struct sk_buff *)message_list->next, 0);
1019 tipc_link_push_queue(l_ptr);
1020 INIT_LIST_HEAD(message_list);
1022 tipc_node_unlock(n_ptr);
1024 read_unlock_bh(&tipc_net_lock);
1026 /* discard the messages if they couldn't be sent */
1027 list_for_each_safe(buf, temp_buf, ((struct sk_buff *)message_list)) {
1028 list_del((struct list_head *)buf);
1034 * link_send_buf_fast: Entry for data messages where the
1035 * destination link is known and the header is complete,
1036 * inclusive total message length. Very time critical.
1037 * Link is locked. Returns user data length.
1039 static int link_send_buf_fast(struct tipc_link *l_ptr, struct sk_buff *buf,
1042 struct tipc_msg *msg = buf_msg(buf);
1043 int res = msg_data_sz(msg);
1045 if (likely(!link_congested(l_ptr))) {
1046 if (likely(msg_size(msg) <= l_ptr->max_pkt)) {
1047 if (likely(!tipc_bearer_blocked(l_ptr->b_ptr))) {
1048 link_add_to_outqueue(l_ptr, buf, msg);
1049 tipc_bearer_send(l_ptr->b_ptr, buf,
1050 &l_ptr->media_addr);
1051 l_ptr->unacked_window = 0;
1055 *used_max_pkt = l_ptr->max_pkt;
1057 return tipc_link_send_buf(l_ptr, buf); /* All other cases */
1061 * tipc_link_send_sections_fast: Entry for messages where the
1062 * destination processor is known and the header is complete,
1063 * except for total message length.
1064 * Returns user data length or errno.
1066 int tipc_link_send_sections_fast(struct tipc_port *sender,
1067 struct iovec const *msg_sect,
1068 const u32 num_sect, unsigned int total_len,
1071 struct tipc_msg *hdr = &sender->phdr;
1072 struct tipc_link *l_ptr;
1073 struct sk_buff *buf;
1074 struct tipc_node *node;
1076 u32 selector = msg_origport(hdr) & 1;
1080 * Try building message using port's max_pkt hint.
1081 * (Must not hold any locks while building message.)
1083 res = tipc_msg_build(hdr, msg_sect, num_sect, total_len,
1084 sender->max_pkt, &buf);
1085 /* Exit if build request was invalid */
1086 if (unlikely(res < 0))
1089 read_lock_bh(&tipc_net_lock);
1090 node = tipc_node_find(destaddr);
1092 tipc_node_lock(node);
1093 l_ptr = node->active_links[selector];
1094 if (likely(l_ptr)) {
1096 res = link_send_buf_fast(l_ptr, buf,
1099 tipc_node_unlock(node);
1100 read_unlock_bh(&tipc_net_lock);
1104 /* Exit if link (or bearer) is congested */
1105 if (link_congested(l_ptr) ||
1106 tipc_bearer_blocked(l_ptr->b_ptr)) {
1107 res = link_schedule_port(l_ptr,
1113 * Message size exceeds max_pkt hint; update hint,
1114 * then re-try fast path or fragment the message
1116 sender->max_pkt = l_ptr->max_pkt;
1117 tipc_node_unlock(node);
1118 read_unlock_bh(&tipc_net_lock);
1121 if ((msg_hdr_sz(hdr) + res) <= sender->max_pkt)
1124 return link_send_sections_long(sender, msg_sect,
1125 num_sect, total_len,
1128 tipc_node_unlock(node);
1130 read_unlock_bh(&tipc_net_lock);
1132 /* Couldn't find a link to the destination node */
1134 return tipc_reject_msg(buf, TIPC_ERR_NO_NODE);
1136 return tipc_port_reject_sections(sender, hdr, msg_sect, num_sect,
1137 total_len, TIPC_ERR_NO_NODE);
1142 * link_send_sections_long(): Entry for long messages where the
1143 * destination node is known and the header is complete,
1144 * inclusive total message length.
1145 * Link and bearer congestion status have been checked to be ok,
1146 * and are ignored if they change.
1148 * Note that fragments do not use the full link MTU so that they won't have
1149 * to undergo refragmentation if link changeover causes them to be sent
1150 * over another link with an additional tunnel header added as prefix.
1151 * (Refragmentation will still occur if the other link has a smaller MTU.)
1153 * Returns user data length or errno.
1155 static int link_send_sections_long(struct tipc_port *sender,
1156 struct iovec const *msg_sect,
1157 u32 num_sect, unsigned int total_len,
1160 struct tipc_link *l_ptr;
1161 struct tipc_node *node;
1162 struct tipc_msg *hdr = &sender->phdr;
1163 u32 dsz = total_len;
1164 u32 max_pkt, fragm_sz, rest;
1165 struct tipc_msg fragm_hdr;
1166 struct sk_buff *buf, *buf_chain, *prev;
1167 u32 fragm_crs, fragm_rest, hsz, sect_rest;
1168 const unchar *sect_crs;
1174 max_pkt = sender->max_pkt - INT_H_SIZE;
1175 /* leave room for tunnel header in case of link changeover */
1176 fragm_sz = max_pkt - INT_H_SIZE;
1177 /* leave room for fragmentation header in each fragment */
1185 /* Prepare reusable fragment header */
1186 tipc_msg_init(&fragm_hdr, MSG_FRAGMENTER, FIRST_FRAGMENT,
1187 INT_H_SIZE, msg_destnode(hdr));
1188 msg_set_size(&fragm_hdr, max_pkt);
1189 msg_set_fragm_no(&fragm_hdr, 1);
1191 /* Prepare header of first fragment */
1192 buf_chain = buf = tipc_buf_acquire(max_pkt);
1196 skb_copy_to_linear_data(buf, &fragm_hdr, INT_H_SIZE);
1197 hsz = msg_hdr_sz(hdr);
1198 skb_copy_to_linear_data_offset(buf, INT_H_SIZE, hdr, hsz);
1200 /* Chop up message */
1201 fragm_crs = INT_H_SIZE + hsz;
1202 fragm_rest = fragm_sz - hsz;
1204 do { /* For all sections */
1208 sect_rest = msg_sect[++curr_sect].iov_len;
1209 sect_crs = (const unchar *)msg_sect[curr_sect].iov_base;
1212 if (sect_rest < fragm_rest)
1217 if (copy_from_user(buf->data + fragm_crs, sect_crs, sz)) {
1219 for (; buf_chain; buf_chain = buf) {
1220 buf = buf_chain->next;
1221 kfree_skb(buf_chain);
1231 if (!fragm_rest && rest) {
1233 /* Initiate new fragment: */
1234 if (rest <= fragm_sz) {
1236 msg_set_type(&fragm_hdr, LAST_FRAGMENT);
1238 msg_set_type(&fragm_hdr, FRAGMENT);
1240 msg_set_size(&fragm_hdr, fragm_sz + INT_H_SIZE);
1241 msg_set_fragm_no(&fragm_hdr, ++fragm_no);
1243 buf = tipc_buf_acquire(fragm_sz + INT_H_SIZE);
1249 skb_copy_to_linear_data(buf, &fragm_hdr, INT_H_SIZE);
1250 fragm_crs = INT_H_SIZE;
1251 fragm_rest = fragm_sz;
1256 * Now we have a buffer chain. Select a link and check
1257 * that packet size is still OK
1259 node = tipc_node_find(destaddr);
1261 tipc_node_lock(node);
1262 l_ptr = node->active_links[sender->ref & 1];
1264 tipc_node_unlock(node);
1267 if (l_ptr->max_pkt < max_pkt) {
1268 sender->max_pkt = l_ptr->max_pkt;
1269 tipc_node_unlock(node);
1270 for (; buf_chain; buf_chain = buf) {
1271 buf = buf_chain->next;
1272 kfree_skb(buf_chain);
1278 for (; buf_chain; buf_chain = buf) {
1279 buf = buf_chain->next;
1280 kfree_skb(buf_chain);
1282 return tipc_port_reject_sections(sender, hdr, msg_sect, num_sect,
1283 total_len, TIPC_ERR_NO_NODE);
1286 /* Append chain of fragments to send queue & send them */
1287 l_ptr->long_msg_seq_no++;
1288 link_add_chain_to_outqueue(l_ptr, buf_chain, l_ptr->long_msg_seq_no);
1289 l_ptr->stats.sent_fragments += fragm_no;
1290 l_ptr->stats.sent_fragmented++;
1291 tipc_link_push_queue(l_ptr);
1292 tipc_node_unlock(node);
1297 * tipc_link_push_packet: Push one unsent packet to the media
1299 u32 tipc_link_push_packet(struct tipc_link *l_ptr)
1301 struct sk_buff *buf = l_ptr->first_out;
1302 u32 r_q_size = l_ptr->retransm_queue_size;
1303 u32 r_q_head = l_ptr->retransm_queue_head;
1305 /* Step to position where retransmission failed, if any, */
1306 /* consider that buffers may have been released in meantime */
1307 if (r_q_size && buf) {
1308 u32 last = lesser(mod(r_q_head + r_q_size),
1309 link_last_sent(l_ptr));
1310 u32 first = buf_seqno(buf);
1312 while (buf && less(first, r_q_head)) {
1313 first = mod(first + 1);
1316 l_ptr->retransm_queue_head = r_q_head = first;
1317 l_ptr->retransm_queue_size = r_q_size = mod(last - first);
1320 /* Continue retransmission now, if there is anything: */
1321 if (r_q_size && buf) {
1322 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
1323 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
1324 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1325 l_ptr->retransm_queue_head = mod(++r_q_head);
1326 l_ptr->retransm_queue_size = --r_q_size;
1327 l_ptr->stats.retransmitted++;
1331 /* Send deferred protocol message, if any: */
1332 buf = l_ptr->proto_msg_queue;
1334 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
1335 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
1336 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1337 l_ptr->unacked_window = 0;
1339 l_ptr->proto_msg_queue = NULL;
1343 /* Send one deferred data message, if send window not full: */
1344 buf = l_ptr->next_out;
1346 struct tipc_msg *msg = buf_msg(buf);
1347 u32 next = msg_seqno(msg);
1348 u32 first = buf_seqno(l_ptr->first_out);
1350 if (mod(next - first) < l_ptr->queue_limit[0]) {
1351 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1352 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1353 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1354 if (msg_user(msg) == MSG_BUNDLER)
1355 msg_set_type(msg, CLOSED_MSG);
1356 l_ptr->next_out = buf->next;
1364 * push_queue(): push out the unsent messages of a link where
1365 * congestion has abated. Node is locked
1367 void tipc_link_push_queue(struct tipc_link *l_ptr)
1371 if (tipc_bearer_blocked(l_ptr->b_ptr))
1375 res = tipc_link_push_packet(l_ptr);
1379 static void link_reset_all(unsigned long addr)
1381 struct tipc_node *n_ptr;
1382 char addr_string[16];
1385 read_lock_bh(&tipc_net_lock);
1386 n_ptr = tipc_node_find((u32)addr);
1388 read_unlock_bh(&tipc_net_lock);
1389 return; /* node no longer exists */
1392 tipc_node_lock(n_ptr);
1394 pr_warn("Resetting all links to %s\n",
1395 tipc_addr_string_fill(addr_string, n_ptr->addr));
1397 for (i = 0; i < MAX_BEARERS; i++) {
1398 if (n_ptr->links[i]) {
1399 link_print(n_ptr->links[i], "Resetting link\n");
1400 tipc_link_reset(n_ptr->links[i]);
1404 tipc_node_unlock(n_ptr);
1405 read_unlock_bh(&tipc_net_lock);
1408 static void link_retransmit_failure(struct tipc_link *l_ptr,
1409 struct sk_buff *buf)
1411 struct tipc_msg *msg = buf_msg(buf);
1413 pr_warn("Retransmission failure on link <%s>\n", l_ptr->name);
1416 /* Handle failure on standard link */
1417 link_print(l_ptr, "Resetting link\n");
1418 tipc_link_reset(l_ptr);
1421 /* Handle failure on broadcast link */
1422 struct tipc_node *n_ptr;
1423 char addr_string[16];
1425 pr_info("Msg seq number: %u, ", msg_seqno(msg));
1426 pr_cont("Outstanding acks: %lu\n",
1427 (unsigned long) TIPC_SKB_CB(buf)->handle);
1429 n_ptr = tipc_bclink_retransmit_to();
1430 tipc_node_lock(n_ptr);
1432 tipc_addr_string_fill(addr_string, n_ptr->addr);
1433 pr_info("Broadcast link info for %s\n", addr_string);
1434 pr_info("Reception permitted: %d, Acked: %u\n",
1435 n_ptr->bclink.recv_permitted,
1436 n_ptr->bclink.acked);
1437 pr_info("Last in: %u, Oos state: %u, Last sent: %u\n",
1438 n_ptr->bclink.last_in,
1439 n_ptr->bclink.oos_state,
1440 n_ptr->bclink.last_sent);
1442 tipc_k_signal((Handler)link_reset_all, (unsigned long)n_ptr->addr);
1444 tipc_node_unlock(n_ptr);
1446 l_ptr->stale_count = 0;
1450 void tipc_link_retransmit(struct tipc_link *l_ptr, struct sk_buff *buf,
1453 struct tipc_msg *msg;
1460 if (tipc_bearer_blocked(l_ptr->b_ptr)) {
1461 if (l_ptr->retransm_queue_size == 0) {
1462 l_ptr->retransm_queue_head = msg_seqno(msg);
1463 l_ptr->retransm_queue_size = retransmits;
1465 pr_err("Unexpected retransmit on link %s (qsize=%d)\n",
1466 l_ptr->name, l_ptr->retransm_queue_size);
1470 /* Detect repeated retransmit failures on unblocked bearer */
1471 if (l_ptr->last_retransmitted == msg_seqno(msg)) {
1472 if (++l_ptr->stale_count > 100) {
1473 link_retransmit_failure(l_ptr, buf);
1477 l_ptr->last_retransmitted = msg_seqno(msg);
1478 l_ptr->stale_count = 1;
1482 while (retransmits && (buf != l_ptr->next_out) && buf) {
1484 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1485 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1486 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1489 l_ptr->stats.retransmitted++;
1492 l_ptr->retransm_queue_head = l_ptr->retransm_queue_size = 0;
1496 * link_insert_deferred_queue - insert deferred messages back into receive chain
1498 static struct sk_buff *link_insert_deferred_queue(struct tipc_link *l_ptr,
1499 struct sk_buff *buf)
1503 if (l_ptr->oldest_deferred_in == NULL)
1506 seq_no = buf_seqno(l_ptr->oldest_deferred_in);
1507 if (seq_no == mod(l_ptr->next_in_no)) {
1508 l_ptr->newest_deferred_in->next = buf;
1509 buf = l_ptr->oldest_deferred_in;
1510 l_ptr->oldest_deferred_in = NULL;
1511 l_ptr->deferred_inqueue_sz = 0;
1517 * link_recv_buf_validate - validate basic format of received message
1519 * This routine ensures a TIPC message has an acceptable header, and at least
1520 * as much data as the header indicates it should. The routine also ensures
1521 * that the entire message header is stored in the main fragment of the message
1522 * buffer, to simplify future access to message header fields.
1524 * Note: Having extra info present in the message header or data areas is OK.
1525 * TIPC will ignore the excess, under the assumption that it is optional info
1526 * introduced by a later release of the protocol.
1528 static int link_recv_buf_validate(struct sk_buff *buf)
1530 static u32 min_data_hdr_size[8] = {
1531 SHORT_H_SIZE, MCAST_H_SIZE, NAMED_H_SIZE, BASIC_H_SIZE,
1532 MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE
1535 struct tipc_msg *msg;
1541 if (unlikely(buf->len < MIN_H_SIZE))
1544 msg = skb_header_pointer(buf, 0, sizeof(tipc_hdr), tipc_hdr);
1548 if (unlikely(msg_version(msg) != TIPC_VERSION))
1551 size = msg_size(msg);
1552 hdr_size = msg_hdr_sz(msg);
1553 min_hdr_size = msg_isdata(msg) ?
1554 min_data_hdr_size[msg_type(msg)] : INT_H_SIZE;
1556 if (unlikely((hdr_size < min_hdr_size) ||
1557 (size < hdr_size) ||
1558 (buf->len < size) ||
1559 (size - hdr_size > TIPC_MAX_USER_MSG_SIZE)))
1562 return pskb_may_pull(buf, hdr_size);
1566 * tipc_recv_msg - process TIPC messages arriving from off-node
1567 * @head: pointer to message buffer chain
1568 * @tb_ptr: pointer to bearer message arrived on
1570 * Invoked with no locks held. Bearer pointer must point to a valid bearer
1571 * structure (i.e. cannot be NULL), but bearer can be inactive.
1573 void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *b_ptr)
1575 read_lock_bh(&tipc_net_lock);
1577 struct tipc_node *n_ptr;
1578 struct tipc_link *l_ptr;
1579 struct sk_buff *crs;
1580 struct sk_buff *buf = head;
1581 struct tipc_msg *msg;
1589 /* Ensure bearer is still enabled */
1590 if (unlikely(!b_ptr->active))
1593 /* Ensure message is well-formed */
1594 if (unlikely(!link_recv_buf_validate(buf)))
1597 /* Ensure message data is a single contiguous unit */
1598 if (unlikely(skb_linearize(buf)))
1601 /* Handle arrival of a non-unicast link message */
1604 if (unlikely(msg_non_seq(msg))) {
1605 if (msg_user(msg) == LINK_CONFIG)
1606 tipc_disc_recv_msg(buf, b_ptr);
1608 tipc_bclink_recv_pkt(buf);
1612 /* Discard unicast link messages destined for another node */
1613 if (unlikely(!msg_short(msg) &&
1614 (msg_destnode(msg) != tipc_own_addr)))
1617 /* Locate neighboring node that sent message */
1618 n_ptr = tipc_node_find(msg_prevnode(msg));
1619 if (unlikely(!n_ptr))
1621 tipc_node_lock(n_ptr);
1623 /* Locate unicast link endpoint that should handle message */
1624 l_ptr = n_ptr->links[b_ptr->identity];
1625 if (unlikely(!l_ptr)) {
1626 tipc_node_unlock(n_ptr);
1630 /* Verify that communication with node is currently allowed */
1631 if ((n_ptr->block_setup & WAIT_PEER_DOWN) &&
1632 msg_user(msg) == LINK_PROTOCOL &&
1633 (msg_type(msg) == RESET_MSG ||
1634 msg_type(msg) == ACTIVATE_MSG) &&
1635 !msg_redundant_link(msg))
1636 n_ptr->block_setup &= ~WAIT_PEER_DOWN;
1638 if (n_ptr->block_setup) {
1639 tipc_node_unlock(n_ptr);
1643 /* Validate message sequence number info */
1644 seq_no = msg_seqno(msg);
1645 ackd = msg_ack(msg);
1647 /* Release acked messages */
1648 if (n_ptr->bclink.recv_permitted)
1649 tipc_bclink_acknowledge(n_ptr, msg_bcast_ack(msg));
1651 crs = l_ptr->first_out;
1652 while ((crs != l_ptr->next_out) &&
1653 less_eq(buf_seqno(crs), ackd)) {
1654 struct sk_buff *next = crs->next;
1661 l_ptr->first_out = crs;
1662 l_ptr->out_queue_size -= released;
1665 /* Try sending any messages link endpoint has pending */
1666 if (unlikely(l_ptr->next_out))
1667 tipc_link_push_queue(l_ptr);
1668 if (unlikely(!list_empty(&l_ptr->waiting_ports)))
1669 tipc_link_wakeup_ports(l_ptr, 0);
1670 if (unlikely(++l_ptr->unacked_window >= TIPC_MIN_LINK_WIN)) {
1671 l_ptr->stats.sent_acks++;
1672 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1675 /* Now (finally!) process the incoming message */
1677 if (likely(link_working_working(l_ptr))) {
1678 if (likely(seq_no == mod(l_ptr->next_in_no))) {
1679 l_ptr->next_in_no++;
1680 if (unlikely(l_ptr->oldest_deferred_in))
1681 head = link_insert_deferred_queue(l_ptr,
1684 if (likely(msg_isdata(msg))) {
1685 tipc_node_unlock(n_ptr);
1686 tipc_port_recv_msg(buf);
1689 switch (msg_user(msg)) {
1692 l_ptr->stats.recv_bundles++;
1693 l_ptr->stats.recv_bundled +=
1695 tipc_node_unlock(n_ptr);
1696 tipc_link_recv_bundle(buf);
1698 case NAME_DISTRIBUTOR:
1699 n_ptr->bclink.recv_permitted = true;
1700 tipc_node_unlock(n_ptr);
1701 tipc_named_recv(buf);
1703 case BCAST_PROTOCOL:
1704 tipc_link_recv_sync(n_ptr, buf);
1705 tipc_node_unlock(n_ptr);
1708 tipc_node_unlock(n_ptr);
1709 tipc_port_recv_proto_msg(buf);
1711 case MSG_FRAGMENTER:
1712 l_ptr->stats.recv_fragments++;
1713 ret = tipc_link_recv_fragment(
1714 &l_ptr->defragm_buf,
1717 l_ptr->stats.recv_fragmented++;
1721 l_ptr->next_in_no--;
1723 case CHANGEOVER_PROTOCOL:
1724 type = msg_type(msg);
1725 if (link_recv_changeover_msg(&l_ptr,
1728 seq_no = msg_seqno(msg);
1729 if (type == ORIGINAL_MSG)
1731 goto protocol_check;
1739 tipc_node_unlock(n_ptr);
1740 tipc_net_route_msg(buf);
1743 link_handle_out_of_seq_msg(l_ptr, buf);
1744 head = link_insert_deferred_queue(l_ptr, head);
1745 tipc_node_unlock(n_ptr);
1749 /* Link is not in state WORKING_WORKING */
1750 if (msg_user(msg) == LINK_PROTOCOL) {
1751 link_recv_proto_msg(l_ptr, buf);
1752 head = link_insert_deferred_queue(l_ptr, head);
1753 tipc_node_unlock(n_ptr);
1757 /* Traffic message. Conditionally activate link */
1758 link_state_event(l_ptr, TRAFFIC_MSG_EVT);
1760 if (link_working_working(l_ptr)) {
1761 /* Re-insert buffer in front of queue */
1764 tipc_node_unlock(n_ptr);
1767 tipc_node_unlock(n_ptr);
1771 read_unlock_bh(&tipc_net_lock);
1775 * tipc_link_defer_pkt - Add out-of-sequence message to deferred reception queue
1777 * Returns increase in queue length (i.e. 0 or 1)
1779 u32 tipc_link_defer_pkt(struct sk_buff **head, struct sk_buff **tail,
1780 struct sk_buff *buf)
1782 struct sk_buff *queue_buf;
1783 struct sk_buff **prev;
1784 u32 seq_no = buf_seqno(buf);
1789 if (*head == NULL) {
1790 *head = *tail = buf;
1795 if (less(buf_seqno(*tail), seq_no)) {
1796 (*tail)->next = buf;
1801 /* Locate insertion point in queue, then insert; discard if duplicate */
1805 u32 curr_seqno = buf_seqno(queue_buf);
1807 if (seq_no == curr_seqno) {
1812 if (less(seq_no, curr_seqno))
1815 prev = &queue_buf->next;
1816 queue_buf = queue_buf->next;
1819 buf->next = queue_buf;
1825 * link_handle_out_of_seq_msg - handle arrival of out-of-sequence packet
1827 static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
1828 struct sk_buff *buf)
1830 u32 seq_no = buf_seqno(buf);
1832 if (likely(msg_user(buf_msg(buf)) == LINK_PROTOCOL)) {
1833 link_recv_proto_msg(l_ptr, buf);
1837 /* Record OOS packet arrival (force mismatch on next timeout) */
1838 l_ptr->checkpoint--;
1841 * Discard packet if a duplicate; otherwise add it to deferred queue
1842 * and notify peer of gap as per protocol specification
1844 if (less(seq_no, mod(l_ptr->next_in_no))) {
1845 l_ptr->stats.duplicates++;
1850 if (tipc_link_defer_pkt(&l_ptr->oldest_deferred_in,
1851 &l_ptr->newest_deferred_in, buf)) {
1852 l_ptr->deferred_inqueue_sz++;
1853 l_ptr->stats.deferred_recv++;
1854 if ((l_ptr->deferred_inqueue_sz % 16) == 1)
1855 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1857 l_ptr->stats.duplicates++;
1861 * Send protocol message to the other endpoint.
1863 void tipc_link_send_proto_msg(struct tipc_link *l_ptr, u32 msg_typ,
1864 int probe_msg, u32 gap, u32 tolerance,
1865 u32 priority, u32 ack_mtu)
1867 struct sk_buff *buf = NULL;
1868 struct tipc_msg *msg = l_ptr->pmsg;
1869 u32 msg_size = sizeof(l_ptr->proto_msg);
1872 /* Discard any previous message that was deferred due to congestion */
1873 if (l_ptr->proto_msg_queue) {
1874 kfree_skb(l_ptr->proto_msg_queue);
1875 l_ptr->proto_msg_queue = NULL;
1878 if (link_blocked(l_ptr))
1881 /* Abort non-RESET send if communication with node is prohibited */
1882 if ((l_ptr->owner->block_setup) && (msg_typ != RESET_MSG))
1885 /* Create protocol message with "out-of-sequence" sequence number */
1886 msg_set_type(msg, msg_typ);
1887 msg_set_net_plane(msg, l_ptr->b_ptr->net_plane);
1888 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1889 msg_set_last_bcast(msg, tipc_bclink_get_last_sent());
1891 if (msg_typ == STATE_MSG) {
1892 u32 next_sent = mod(l_ptr->next_out_no);
1894 if (!tipc_link_is_up(l_ptr))
1896 if (l_ptr->next_out)
1897 next_sent = buf_seqno(l_ptr->next_out);
1898 msg_set_next_sent(msg, next_sent);
1899 if (l_ptr->oldest_deferred_in) {
1900 u32 rec = buf_seqno(l_ptr->oldest_deferred_in);
1901 gap = mod(rec - mod(l_ptr->next_in_no));
1903 msg_set_seq_gap(msg, gap);
1905 l_ptr->stats.sent_nacks++;
1906 msg_set_link_tolerance(msg, tolerance);
1907 msg_set_linkprio(msg, priority);
1908 msg_set_max_pkt(msg, ack_mtu);
1909 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1910 msg_set_probe(msg, probe_msg != 0);
1912 u32 mtu = l_ptr->max_pkt;
1914 if ((mtu < l_ptr->max_pkt_target) &&
1915 link_working_working(l_ptr) &&
1916 l_ptr->fsm_msg_cnt) {
1917 msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1918 if (l_ptr->max_pkt_probes == 10) {
1919 l_ptr->max_pkt_target = (msg_size - 4);
1920 l_ptr->max_pkt_probes = 0;
1921 msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1923 l_ptr->max_pkt_probes++;
1926 l_ptr->stats.sent_probes++;
1928 l_ptr->stats.sent_states++;
1929 } else { /* RESET_MSG or ACTIVATE_MSG */
1930 msg_set_ack(msg, mod(l_ptr->reset_checkpoint - 1));
1931 msg_set_seq_gap(msg, 0);
1932 msg_set_next_sent(msg, 1);
1933 msg_set_probe(msg, 0);
1934 msg_set_link_tolerance(msg, l_ptr->tolerance);
1935 msg_set_linkprio(msg, l_ptr->priority);
1936 msg_set_max_pkt(msg, l_ptr->max_pkt_target);
1939 r_flag = (l_ptr->owner->working_links > tipc_link_is_up(l_ptr));
1940 msg_set_redundant_link(msg, r_flag);
1941 msg_set_linkprio(msg, l_ptr->priority);
1942 msg_set_size(msg, msg_size);
1944 msg_set_seqno(msg, mod(l_ptr->next_out_no + (0xffff/2)));
1946 buf = tipc_buf_acquire(msg_size);
1950 skb_copy_to_linear_data(buf, msg, sizeof(l_ptr->proto_msg));
1951 buf->priority = TC_PRIO_CONTROL;
1953 /* Defer message if bearer is already blocked */
1954 if (tipc_bearer_blocked(l_ptr->b_ptr)) {
1955 l_ptr->proto_msg_queue = buf;
1959 tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr);
1960 l_ptr->unacked_window = 0;
1965 * Receive protocol message :
1966 * Note that network plane id propagates through the network, and may
1967 * change at any time. The node with lowest address rules
1969 static void link_recv_proto_msg(struct tipc_link *l_ptr, struct sk_buff *buf)
1975 struct tipc_msg *msg = buf_msg(buf);
1977 if (link_blocked(l_ptr))
1980 /* record unnumbered packet arrival (force mismatch on next timeout) */
1981 l_ptr->checkpoint--;
1983 if (l_ptr->b_ptr->net_plane != msg_net_plane(msg))
1984 if (tipc_own_addr > msg_prevnode(msg))
1985 l_ptr->b_ptr->net_plane = msg_net_plane(msg);
1987 l_ptr->owner->permit_changeover = msg_redundant_link(msg);
1989 switch (msg_type(msg)) {
1992 if (!link_working_unknown(l_ptr) &&
1993 (l_ptr->peer_session != INVALID_SESSION)) {
1994 if (less_eq(msg_session(msg), l_ptr->peer_session))
1995 break; /* duplicate or old reset: ignore */
1998 if (!msg_redundant_link(msg) && (link_working_working(l_ptr) ||
1999 link_working_unknown(l_ptr))) {
2001 * peer has lost contact -- don't allow peer's links
2002 * to reactivate before we recognize loss & clean up
2004 l_ptr->owner->block_setup = WAIT_NODE_DOWN;
2007 link_state_event(l_ptr, RESET_MSG);
2011 /* Update link settings according other endpoint's values */
2012 strcpy((strrchr(l_ptr->name, ':') + 1), (char *)msg_data(msg));
2014 msg_tol = msg_link_tolerance(msg);
2015 if (msg_tol > l_ptr->tolerance)
2016 link_set_supervision_props(l_ptr, msg_tol);
2018 if (msg_linkprio(msg) > l_ptr->priority)
2019 l_ptr->priority = msg_linkprio(msg);
2021 max_pkt_info = msg_max_pkt(msg);
2023 if (max_pkt_info < l_ptr->max_pkt_target)
2024 l_ptr->max_pkt_target = max_pkt_info;
2025 if (l_ptr->max_pkt > l_ptr->max_pkt_target)
2026 l_ptr->max_pkt = l_ptr->max_pkt_target;
2028 l_ptr->max_pkt = l_ptr->max_pkt_target;
2031 /* Synchronize broadcast link info, if not done previously */
2032 if (!tipc_node_is_up(l_ptr->owner)) {
2033 l_ptr->owner->bclink.last_sent =
2034 l_ptr->owner->bclink.last_in =
2035 msg_last_bcast(msg);
2036 l_ptr->owner->bclink.oos_state = 0;
2039 l_ptr->peer_session = msg_session(msg);
2040 l_ptr->peer_bearer_id = msg_bearer_id(msg);
2042 if (msg_type(msg) == ACTIVATE_MSG)
2043 link_state_event(l_ptr, ACTIVATE_MSG);
2047 msg_tol = msg_link_tolerance(msg);
2049 link_set_supervision_props(l_ptr, msg_tol);
2051 if (msg_linkprio(msg) &&
2052 (msg_linkprio(msg) != l_ptr->priority)) {
2053 pr_warn("%s<%s>, priority change %u->%u\n",
2054 link_rst_msg, l_ptr->name, l_ptr->priority,
2056 l_ptr->priority = msg_linkprio(msg);
2057 tipc_link_reset(l_ptr); /* Enforce change to take effect */
2060 link_state_event(l_ptr, TRAFFIC_MSG_EVT);
2061 l_ptr->stats.recv_states++;
2062 if (link_reset_unknown(l_ptr))
2065 if (less_eq(mod(l_ptr->next_in_no), msg_next_sent(msg))) {
2066 rec_gap = mod(msg_next_sent(msg) -
2067 mod(l_ptr->next_in_no));
2070 max_pkt_ack = msg_max_pkt(msg);
2071 if (max_pkt_ack > l_ptr->max_pkt) {
2072 l_ptr->max_pkt = max_pkt_ack;
2073 l_ptr->max_pkt_probes = 0;
2077 if (msg_probe(msg)) {
2078 l_ptr->stats.recv_probes++;
2079 if (msg_size(msg) > sizeof(l_ptr->proto_msg))
2080 max_pkt_ack = msg_size(msg);
2083 /* Protocol message before retransmits, reduce loss risk */
2084 if (l_ptr->owner->bclink.recv_permitted)
2085 tipc_bclink_update_link_state(l_ptr->owner,
2086 msg_last_bcast(msg));
2088 if (rec_gap || (msg_probe(msg))) {
2089 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
2090 0, rec_gap, 0, 0, max_pkt_ack);
2092 if (msg_seq_gap(msg)) {
2093 l_ptr->stats.recv_nacks++;
2094 tipc_link_retransmit(l_ptr, l_ptr->first_out,
2105 * tipc_link_tunnel(): Send one message via a link belonging to
2106 * another bearer. Owner node is locked.
2108 static void tipc_link_tunnel(struct tipc_link *l_ptr,
2109 struct tipc_msg *tunnel_hdr, struct tipc_msg *msg,
2112 struct tipc_link *tunnel;
2113 struct sk_buff *buf;
2114 u32 length = msg_size(msg);
2116 tunnel = l_ptr->owner->active_links[selector & 1];
2117 if (!tipc_link_is_up(tunnel)) {
2118 pr_warn("%stunnel link no longer available\n", link_co_err);
2121 msg_set_size(tunnel_hdr, length + INT_H_SIZE);
2122 buf = tipc_buf_acquire(length + INT_H_SIZE);
2124 pr_warn("%sunable to send tunnel msg\n", link_co_err);
2127 skb_copy_to_linear_data(buf, tunnel_hdr, INT_H_SIZE);
2128 skb_copy_to_linear_data_offset(buf, INT_H_SIZE, msg, length);
2129 tipc_link_send_buf(tunnel, buf);
2135 * changeover(): Send whole message queue via the remaining link
2136 * Owner node is locked.
2138 void tipc_link_changeover(struct tipc_link *l_ptr)
2140 u32 msgcount = l_ptr->out_queue_size;
2141 struct sk_buff *crs = l_ptr->first_out;
2142 struct tipc_link *tunnel = l_ptr->owner->active_links[0];
2143 struct tipc_msg tunnel_hdr;
2149 if (!l_ptr->owner->permit_changeover) {
2150 pr_warn("%speer did not permit changeover\n", link_co_err);
2154 tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
2155 ORIGINAL_MSG, INT_H_SIZE, l_ptr->addr);
2156 msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
2157 msg_set_msgcnt(&tunnel_hdr, msgcount);
2159 if (!l_ptr->first_out) {
2160 struct sk_buff *buf;
2162 buf = tipc_buf_acquire(INT_H_SIZE);
2164 skb_copy_to_linear_data(buf, &tunnel_hdr, INT_H_SIZE);
2165 msg_set_size(&tunnel_hdr, INT_H_SIZE);
2166 tipc_link_send_buf(tunnel, buf);
2168 pr_warn("%sunable to send changeover msg\n",
2174 split_bundles = (l_ptr->owner->active_links[0] !=
2175 l_ptr->owner->active_links[1]);
2178 struct tipc_msg *msg = buf_msg(crs);
2180 if ((msg_user(msg) == MSG_BUNDLER) && split_bundles) {
2181 struct tipc_msg *m = msg_get_wrapped(msg);
2182 unchar *pos = (unchar *)m;
2184 msgcount = msg_msgcnt(msg);
2185 while (msgcount--) {
2186 msg_set_seqno(m, msg_seqno(msg));
2187 tipc_link_tunnel(l_ptr, &tunnel_hdr, m,
2188 msg_link_selector(m));
2189 pos += align(msg_size(m));
2190 m = (struct tipc_msg *)pos;
2193 tipc_link_tunnel(l_ptr, &tunnel_hdr, msg,
2194 msg_link_selector(msg));
2200 void tipc_link_send_duplicate(struct tipc_link *l_ptr, struct tipc_link *tunnel)
2202 struct sk_buff *iter;
2203 struct tipc_msg tunnel_hdr;
2205 tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
2206 DUPLICATE_MSG, INT_H_SIZE, l_ptr->addr);
2207 msg_set_msgcnt(&tunnel_hdr, l_ptr->out_queue_size);
2208 msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
2209 iter = l_ptr->first_out;
2211 struct sk_buff *outbuf;
2212 struct tipc_msg *msg = buf_msg(iter);
2213 u32 length = msg_size(msg);
2215 if (msg_user(msg) == MSG_BUNDLER)
2216 msg_set_type(msg, CLOSED_MSG);
2217 msg_set_ack(msg, mod(l_ptr->next_in_no - 1)); /* Update */
2218 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
2219 msg_set_size(&tunnel_hdr, length + INT_H_SIZE);
2220 outbuf = tipc_buf_acquire(length + INT_H_SIZE);
2221 if (outbuf == NULL) {
2222 pr_warn("%sunable to send duplicate msg\n",
2226 skb_copy_to_linear_data(outbuf, &tunnel_hdr, INT_H_SIZE);
2227 skb_copy_to_linear_data_offset(outbuf, INT_H_SIZE, iter->data,
2229 tipc_link_send_buf(tunnel, outbuf);
2230 if (!tipc_link_is_up(l_ptr))
2237 * buf_extract - extracts embedded TIPC message from another message
2238 * @skb: encapsulating message buffer
2239 * @from_pos: offset to extract from
2241 * Returns a new message buffer containing an embedded message. The
2242 * encapsulating message itself is left unchanged.
2244 static struct sk_buff *buf_extract(struct sk_buff *skb, u32 from_pos)
2246 struct tipc_msg *msg = (struct tipc_msg *)(skb->data + from_pos);
2247 u32 size = msg_size(msg);
2250 eb = tipc_buf_acquire(size);
2252 skb_copy_to_linear_data(eb, msg, size);
2257 * link_recv_changeover_msg(): Receive tunneled packet sent
2258 * via other link. Node is locked. Return extracted buffer.
2260 static int link_recv_changeover_msg(struct tipc_link **l_ptr,
2261 struct sk_buff **buf)
2263 struct sk_buff *tunnel_buf = *buf;
2264 struct tipc_link *dest_link;
2265 struct tipc_msg *msg;
2266 struct tipc_msg *tunnel_msg = buf_msg(tunnel_buf);
2267 u32 msg_typ = msg_type(tunnel_msg);
2268 u32 msg_count = msg_msgcnt(tunnel_msg);
2269 u32 bearer_id = msg_bearer_id(tunnel_msg);
2271 if (bearer_id >= MAX_BEARERS)
2273 dest_link = (*l_ptr)->owner->links[bearer_id];
2276 if (dest_link == *l_ptr) {
2277 pr_err("Unexpected changeover message on link <%s>\n",
2282 msg = msg_get_wrapped(tunnel_msg);
2284 if (msg_typ == DUPLICATE_MSG) {
2285 if (less(msg_seqno(msg), mod(dest_link->next_in_no)))
2287 *buf = buf_extract(tunnel_buf, INT_H_SIZE);
2289 pr_warn("%sduplicate msg dropped\n", link_co_err);
2292 kfree_skb(tunnel_buf);
2296 /* First original message ?: */
2297 if (tipc_link_is_up(dest_link)) {
2298 pr_info("%s<%s>, changeover initiated by peer\n", link_rst_msg,
2300 tipc_link_reset(dest_link);
2301 dest_link->exp_msg_count = msg_count;
2304 } else if (dest_link->exp_msg_count == START_CHANGEOVER) {
2305 dest_link->exp_msg_count = msg_count;
2310 /* Receive original message */
2311 if (dest_link->exp_msg_count == 0) {
2312 pr_warn("%sgot too many tunnelled messages\n", link_co_err);
2315 dest_link->exp_msg_count--;
2316 if (less(msg_seqno(msg), dest_link->reset_checkpoint)) {
2319 *buf = buf_extract(tunnel_buf, INT_H_SIZE);
2321 kfree_skb(tunnel_buf);
2324 pr_warn("%soriginal msg dropped\n", link_co_err);
2329 kfree_skb(tunnel_buf);
2334 * Bundler functionality:
2336 void tipc_link_recv_bundle(struct sk_buff *buf)
2338 u32 msgcount = msg_msgcnt(buf_msg(buf));
2339 u32 pos = INT_H_SIZE;
2340 struct sk_buff *obuf;
2342 while (msgcount--) {
2343 obuf = buf_extract(buf, pos);
2345 pr_warn("Link unable to unbundle message(s)\n");
2348 pos += align(msg_size(buf_msg(obuf)));
2349 tipc_net_route_msg(obuf);
2355 * Fragmentation/defragmentation:
2359 * link_send_long_buf: Entry for buffers needing fragmentation.
2360 * The buffer is complete, inclusive total message length.
2361 * Returns user data length.
2363 static int link_send_long_buf(struct tipc_link *l_ptr, struct sk_buff *buf)
2365 struct sk_buff *buf_chain = NULL;
2366 struct sk_buff *buf_chain_tail = (struct sk_buff *)&buf_chain;
2367 struct tipc_msg *inmsg = buf_msg(buf);
2368 struct tipc_msg fragm_hdr;
2369 u32 insize = msg_size(inmsg);
2370 u32 dsz = msg_data_sz(inmsg);
2371 unchar *crs = buf->data;
2373 u32 pack_sz = l_ptr->max_pkt;
2374 u32 fragm_sz = pack_sz - INT_H_SIZE;
2378 if (msg_short(inmsg))
2379 destaddr = l_ptr->addr;
2381 destaddr = msg_destnode(inmsg);
2383 /* Prepare reusable fragment header: */
2384 tipc_msg_init(&fragm_hdr, MSG_FRAGMENTER, FIRST_FRAGMENT,
2385 INT_H_SIZE, destaddr);
2387 /* Chop up message: */
2389 struct sk_buff *fragm;
2391 if (rest <= fragm_sz) {
2393 msg_set_type(&fragm_hdr, LAST_FRAGMENT);
2395 fragm = tipc_buf_acquire(fragm_sz + INT_H_SIZE);
2396 if (fragm == NULL) {
2400 buf_chain = buf_chain->next;
2405 msg_set_size(&fragm_hdr, fragm_sz + INT_H_SIZE);
2407 msg_set_fragm_no(&fragm_hdr, fragm_no);
2408 skb_copy_to_linear_data(fragm, &fragm_hdr, INT_H_SIZE);
2409 skb_copy_to_linear_data_offset(fragm, INT_H_SIZE, crs,
2411 buf_chain_tail->next = fragm;
2412 buf_chain_tail = fragm;
2416 msg_set_type(&fragm_hdr, FRAGMENT);
2420 /* Append chain of fragments to send queue & send them */
2421 l_ptr->long_msg_seq_no++;
2422 link_add_chain_to_outqueue(l_ptr, buf_chain, l_ptr->long_msg_seq_no);
2423 l_ptr->stats.sent_fragments += fragm_no;
2424 l_ptr->stats.sent_fragmented++;
2425 tipc_link_push_queue(l_ptr);
2431 * A pending message being re-assembled must store certain values
2432 * to handle subsequent fragments correctly. The following functions
2433 * help storing these values in unused, available fields in the
2434 * pending message. This makes dynamic memory allocation unnecessary.
2436 static void set_long_msg_seqno(struct sk_buff *buf, u32 seqno)
2438 msg_set_seqno(buf_msg(buf), seqno);
2441 static u32 get_fragm_size(struct sk_buff *buf)
2443 return msg_ack(buf_msg(buf));
2446 static void set_fragm_size(struct sk_buff *buf, u32 sz)
2448 msg_set_ack(buf_msg(buf), sz);
2451 static u32 get_expected_frags(struct sk_buff *buf)
2453 return msg_bcast_ack(buf_msg(buf));
2456 static void set_expected_frags(struct sk_buff *buf, u32 exp)
2458 msg_set_bcast_ack(buf_msg(buf), exp);
2462 * tipc_link_recv_fragment(): Called with node lock on. Returns
2463 * the reassembled buffer if message is complete.
2465 int tipc_link_recv_fragment(struct sk_buff **pending, struct sk_buff **fb,
2466 struct tipc_msg **m)
2468 struct sk_buff *prev = NULL;
2469 struct sk_buff *fbuf = *fb;
2470 struct tipc_msg *fragm = buf_msg(fbuf);
2471 struct sk_buff *pbuf = *pending;
2472 u32 long_msg_seq_no = msg_long_msgno(fragm);
2476 /* Is there an incomplete message waiting for this fragment? */
2477 while (pbuf && ((buf_seqno(pbuf) != long_msg_seq_no) ||
2478 (msg_orignode(fragm) != msg_orignode(buf_msg(pbuf))))) {
2483 if (!pbuf && (msg_type(fragm) == FIRST_FRAGMENT)) {
2484 struct tipc_msg *imsg = (struct tipc_msg *)msg_data(fragm);
2485 u32 msg_sz = msg_size(imsg);
2486 u32 fragm_sz = msg_data_sz(fragm);
2488 u32 max = TIPC_MAX_USER_MSG_SIZE + NAMED_H_SIZE;
2490 if (msg_type(imsg) == TIPC_MCAST_MSG)
2491 max = TIPC_MAX_USER_MSG_SIZE + MCAST_H_SIZE;
2492 if (fragm_sz == 0 || msg_size(imsg) > max) {
2496 exp_fragm_cnt = msg_sz / fragm_sz + !!(msg_sz % fragm_sz);
2497 pbuf = tipc_buf_acquire(msg_size(imsg));
2499 pbuf->next = *pending;
2501 skb_copy_to_linear_data(pbuf, imsg,
2502 msg_data_sz(fragm));
2503 /* Prepare buffer for subsequent fragments. */
2504 set_long_msg_seqno(pbuf, long_msg_seq_no);
2505 set_fragm_size(pbuf, fragm_sz);
2506 set_expected_frags(pbuf, exp_fragm_cnt - 1);
2508 pr_debug("Link unable to reassemble fragmented message\n");
2514 } else if (pbuf && (msg_type(fragm) != FIRST_FRAGMENT)) {
2515 u32 dsz = msg_data_sz(fragm);
2516 u32 fsz = get_fragm_size(pbuf);
2517 u32 crs = ((msg_fragm_no(fragm) - 1) * fsz);
2518 u32 exp_frags = get_expected_frags(pbuf) - 1;
2519 skb_copy_to_linear_data_offset(pbuf, crs,
2520 msg_data(fragm), dsz);
2523 /* Is message complete? */
2524 if (exp_frags == 0) {
2526 prev->next = pbuf->next;
2528 *pending = pbuf->next;
2529 msg_reset_reroute_cnt(buf_msg(pbuf));
2534 set_expected_frags(pbuf, exp_frags);
2541 static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance)
2543 if ((tolerance < TIPC_MIN_LINK_TOL) || (tolerance > TIPC_MAX_LINK_TOL))
2546 l_ptr->tolerance = tolerance;
2547 l_ptr->continuity_interval =
2548 ((tolerance / 4) > 500) ? 500 : tolerance / 4;
2549 l_ptr->abort_limit = tolerance / (l_ptr->continuity_interval / 4);
2552 void tipc_link_set_queue_limits(struct tipc_link *l_ptr, u32 window)
2554 /* Data messages from this node, inclusive FIRST_FRAGM */
2555 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE] = window;
2556 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE] = (window / 3) * 4;
2557 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE] = (window / 3) * 5;
2558 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE] = (window / 3) * 6;
2559 /* Transiting data messages,inclusive FIRST_FRAGM */
2560 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE + 4] = 300;
2561 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE + 4] = 600;
2562 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE + 4] = 900;
2563 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE + 4] = 1200;
2564 l_ptr->queue_limit[CONN_MANAGER] = 1200;
2565 l_ptr->queue_limit[CHANGEOVER_PROTOCOL] = 2500;
2566 l_ptr->queue_limit[NAME_DISTRIBUTOR] = 3000;
2567 /* FRAGMENT and LAST_FRAGMENT packets */
2568 l_ptr->queue_limit[MSG_FRAGMENTER] = 4000;
2572 * link_find_link - locate link by name
2573 * @name: ptr to link name string
2574 * @node: ptr to area to be filled with ptr to associated node
2576 * Caller must hold 'tipc_net_lock' to ensure node and bearer are not deleted;
2577 * this also prevents link deletion.
2579 * Returns pointer to link (or 0 if invalid link name).
2581 static struct tipc_link *link_find_link(const char *name,
2582 struct tipc_node **node)
2584 struct tipc_link_name link_name_parts;
2585 struct tipc_bearer *b_ptr;
2586 struct tipc_link *l_ptr;
2588 if (!link_name_validate(name, &link_name_parts))
2591 b_ptr = tipc_bearer_find_interface(link_name_parts.if_local);
2595 *node = tipc_node_find(link_name_parts.addr_peer);
2599 l_ptr = (*node)->links[b_ptr->identity];
2600 if (!l_ptr || strcmp(l_ptr->name, name))
2607 * link_value_is_valid -- validate proposed link tolerance/priority/window
2609 * @cmd: value type (TIPC_CMD_SET_LINK_*)
2610 * @new_value: the new value
2612 * Returns 1 if value is within range, 0 if not.
2614 static int link_value_is_valid(u16 cmd, u32 new_value)
2617 case TIPC_CMD_SET_LINK_TOL:
2618 return (new_value >= TIPC_MIN_LINK_TOL) &&
2619 (new_value <= TIPC_MAX_LINK_TOL);
2620 case TIPC_CMD_SET_LINK_PRI:
2621 return (new_value <= TIPC_MAX_LINK_PRI);
2622 case TIPC_CMD_SET_LINK_WINDOW:
2623 return (new_value >= TIPC_MIN_LINK_WIN) &&
2624 (new_value <= TIPC_MAX_LINK_WIN);
2630 * link_cmd_set_value - change priority/tolerance/window for link/bearer/media
2631 * @name: ptr to link, bearer, or media name
2632 * @new_value: new value of link, bearer, or media setting
2633 * @cmd: which link, bearer, or media attribute to set (TIPC_CMD_SET_LINK_*)
2635 * Caller must hold 'tipc_net_lock' to ensure link/bearer/media is not deleted.
2637 * Returns 0 if value updated and negative value on error.
2639 static int link_cmd_set_value(const char *name, u32 new_value, u16 cmd)
2641 struct tipc_node *node;
2642 struct tipc_link *l_ptr;
2643 struct tipc_bearer *b_ptr;
2644 struct tipc_media *m_ptr;
2646 l_ptr = link_find_link(name, &node);
2649 * acquire node lock for tipc_link_send_proto_msg().
2650 * see "TIPC locking policy" in net.c.
2652 tipc_node_lock(node);
2654 case TIPC_CMD_SET_LINK_TOL:
2655 link_set_supervision_props(l_ptr, new_value);
2656 tipc_link_send_proto_msg(l_ptr,
2657 STATE_MSG, 0, 0, new_value, 0, 0);
2659 case TIPC_CMD_SET_LINK_PRI:
2660 l_ptr->priority = new_value;
2661 tipc_link_send_proto_msg(l_ptr,
2662 STATE_MSG, 0, 0, 0, new_value, 0);
2664 case TIPC_CMD_SET_LINK_WINDOW:
2665 tipc_link_set_queue_limits(l_ptr, new_value);
2668 tipc_node_unlock(node);
2672 b_ptr = tipc_bearer_find(name);
2675 case TIPC_CMD_SET_LINK_TOL:
2676 b_ptr->tolerance = new_value;
2678 case TIPC_CMD_SET_LINK_PRI:
2679 b_ptr->priority = new_value;
2681 case TIPC_CMD_SET_LINK_WINDOW:
2682 b_ptr->window = new_value;
2688 m_ptr = tipc_media_find(name);
2692 case TIPC_CMD_SET_LINK_TOL:
2693 m_ptr->tolerance = new_value;
2695 case TIPC_CMD_SET_LINK_PRI:
2696 m_ptr->priority = new_value;
2698 case TIPC_CMD_SET_LINK_WINDOW:
2699 m_ptr->window = new_value;
2705 struct sk_buff *tipc_link_cmd_config(const void *req_tlv_area, int req_tlv_space,
2708 struct tipc_link_config *args;
2712 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_CONFIG))
2713 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2715 args = (struct tipc_link_config *)TLV_DATA(req_tlv_area);
2716 new_value = ntohl(args->value);
2718 if (!link_value_is_valid(cmd, new_value))
2719 return tipc_cfg_reply_error_string(
2720 "cannot change, value invalid");
2722 if (!strcmp(args->name, tipc_bclink_name)) {
2723 if ((cmd == TIPC_CMD_SET_LINK_WINDOW) &&
2724 (tipc_bclink_set_queue_limits(new_value) == 0))
2725 return tipc_cfg_reply_none();
2726 return tipc_cfg_reply_error_string(TIPC_CFG_NOT_SUPPORTED
2727 " (cannot change setting on broadcast link)");
2730 read_lock_bh(&tipc_net_lock);
2731 res = link_cmd_set_value(args->name, new_value, cmd);
2732 read_unlock_bh(&tipc_net_lock);
2734 return tipc_cfg_reply_error_string("cannot change link setting");
2736 return tipc_cfg_reply_none();
2740 * link_reset_statistics - reset link statistics
2741 * @l_ptr: pointer to link
2743 static void link_reset_statistics(struct tipc_link *l_ptr)
2745 memset(&l_ptr->stats, 0, sizeof(l_ptr->stats));
2746 l_ptr->stats.sent_info = l_ptr->next_out_no;
2747 l_ptr->stats.recv_info = l_ptr->next_in_no;
2750 struct sk_buff *tipc_link_cmd_reset_stats(const void *req_tlv_area, int req_tlv_space)
2753 struct tipc_link *l_ptr;
2754 struct tipc_node *node;
2756 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2757 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2759 link_name = (char *)TLV_DATA(req_tlv_area);
2760 if (!strcmp(link_name, tipc_bclink_name)) {
2761 if (tipc_bclink_reset_stats())
2762 return tipc_cfg_reply_error_string("link not found");
2763 return tipc_cfg_reply_none();
2766 read_lock_bh(&tipc_net_lock);
2767 l_ptr = link_find_link(link_name, &node);
2769 read_unlock_bh(&tipc_net_lock);
2770 return tipc_cfg_reply_error_string("link not found");
2773 tipc_node_lock(node);
2774 link_reset_statistics(l_ptr);
2775 tipc_node_unlock(node);
2776 read_unlock_bh(&tipc_net_lock);
2777 return tipc_cfg_reply_none();
2781 * percent - convert count to a percentage of total (rounding up or down)
2783 static u32 percent(u32 count, u32 total)
2785 return (count * 100 + (total / 2)) / total;
2789 * tipc_link_stats - print link statistics
2791 * @buf: print buffer area
2792 * @buf_size: size of print buffer area
2794 * Returns length of print buffer data string (or 0 if error)
2796 static int tipc_link_stats(const char *name, char *buf, const u32 buf_size)
2798 struct tipc_link *l;
2799 struct tipc_stats *s;
2800 struct tipc_node *node;
2802 u32 profile_total = 0;
2805 if (!strcmp(name, tipc_bclink_name))
2806 return tipc_bclink_stats(buf, buf_size);
2808 read_lock_bh(&tipc_net_lock);
2809 l = link_find_link(name, &node);
2811 read_unlock_bh(&tipc_net_lock);
2814 tipc_node_lock(node);
2817 if (tipc_link_is_active(l))
2819 else if (tipc_link_is_up(l))
2824 ret = tipc_snprintf(buf, buf_size, "Link <%s>\n"
2825 " %s MTU:%u Priority:%u Tolerance:%u ms"
2826 " Window:%u packets\n",
2827 l->name, status, l->max_pkt, l->priority,
2828 l->tolerance, l->queue_limit[0]);
2830 ret += tipc_snprintf(buf + ret, buf_size - ret,
2831 " RX packets:%u fragments:%u/%u bundles:%u/%u\n",
2832 l->next_in_no - s->recv_info, s->recv_fragments,
2833 s->recv_fragmented, s->recv_bundles,
2836 ret += tipc_snprintf(buf + ret, buf_size - ret,
2837 " TX packets:%u fragments:%u/%u bundles:%u/%u\n",
2838 l->next_out_no - s->sent_info, s->sent_fragments,
2839 s->sent_fragmented, s->sent_bundles,
2842 profile_total = s->msg_length_counts;
2846 ret += tipc_snprintf(buf + ret, buf_size - ret,
2847 " TX profile sample:%u packets average:%u octets\n"
2848 " 0-64:%u%% -256:%u%% -1024:%u%% -4096:%u%% "
2849 "-16384:%u%% -32768:%u%% -66000:%u%%\n",
2850 s->msg_length_counts,
2851 s->msg_lengths_total / profile_total,
2852 percent(s->msg_length_profile[0], profile_total),
2853 percent(s->msg_length_profile[1], profile_total),
2854 percent(s->msg_length_profile[2], profile_total),
2855 percent(s->msg_length_profile[3], profile_total),
2856 percent(s->msg_length_profile[4], profile_total),
2857 percent(s->msg_length_profile[5], profile_total),
2858 percent(s->msg_length_profile[6], profile_total));
2860 ret += tipc_snprintf(buf + ret, buf_size - ret,
2861 " RX states:%u probes:%u naks:%u defs:%u"
2862 " dups:%u\n", s->recv_states, s->recv_probes,
2863 s->recv_nacks, s->deferred_recv, s->duplicates);
2865 ret += tipc_snprintf(buf + ret, buf_size - ret,
2866 " TX states:%u probes:%u naks:%u acks:%u"
2867 " dups:%u\n", s->sent_states, s->sent_probes,
2868 s->sent_nacks, s->sent_acks, s->retransmitted);
2870 ret += tipc_snprintf(buf + ret, buf_size - ret,
2871 " Congestion link:%u Send queue"
2872 " max:%u avg:%u\n", s->link_congs,
2873 s->max_queue_sz, s->queue_sz_counts ?
2874 (s->accu_queue_sz / s->queue_sz_counts) : 0);
2876 tipc_node_unlock(node);
2877 read_unlock_bh(&tipc_net_lock);
2881 struct sk_buff *tipc_link_cmd_show_stats(const void *req_tlv_area, int req_tlv_space)
2883 struct sk_buff *buf;
2884 struct tlv_desc *rep_tlv;
2889 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2890 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2892 buf = tipc_cfg_reply_alloc(TLV_SPACE(ULTRA_STRING_MAX_LEN));
2896 rep_tlv = (struct tlv_desc *)buf->data;
2897 pb = TLV_DATA(rep_tlv);
2898 pb_len = ULTRA_STRING_MAX_LEN;
2899 str_len = tipc_link_stats((char *)TLV_DATA(req_tlv_area),
2903 return tipc_cfg_reply_error_string("link not found");
2905 str_len += 1; /* for "\0" */
2906 skb_put(buf, TLV_SPACE(str_len));
2907 TLV_SET(rep_tlv, TIPC_TLV_ULTRA_STRING, NULL, str_len);
2913 * tipc_link_get_max_pkt - get maximum packet size to use when sending to destination
2914 * @dest: network address of destination node
2915 * @selector: used to select from set of active links
2917 * If no active link can be found, uses default maximum packet size.
2919 u32 tipc_link_get_max_pkt(u32 dest, u32 selector)
2921 struct tipc_node *n_ptr;
2922 struct tipc_link *l_ptr;
2923 u32 res = MAX_PKT_DEFAULT;
2925 if (dest == tipc_own_addr)
2926 return MAX_MSG_SIZE;
2928 read_lock_bh(&tipc_net_lock);
2929 n_ptr = tipc_node_find(dest);
2931 tipc_node_lock(n_ptr);
2932 l_ptr = n_ptr->active_links[selector & 1];
2934 res = l_ptr->max_pkt;
2935 tipc_node_unlock(n_ptr);
2937 read_unlock_bh(&tipc_net_lock);
2941 static void link_print(struct tipc_link *l_ptr, const char *str)
2943 pr_info("%s Link %x<%s>:", str, l_ptr->addr, l_ptr->b_ptr->name);
2945 if (link_working_unknown(l_ptr))
2947 else if (link_reset_reset(l_ptr))
2949 else if (link_reset_unknown(l_ptr))
2951 else if (link_working_working(l_ptr))