2 * net/tipc/link.c: TIPC link code
4 * Copyright (c) 1996-2007, 2012-2016, 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.
42 #include "name_distr.h"
46 #include <linux/pkt_sched.h>
49 u32 sent_info; /* used in counting # sent packets */
50 u32 recv_info; /* used in counting # recv'd packets */
67 u32 link_congs; /* # port sends blocked by congestion */
70 u32 max_queue_sz; /* send queue size high water mark */
71 u32 accu_queue_sz; /* used for send queue size profiling */
72 u32 queue_sz_counts; /* used for send queue size profiling */
73 u32 msg_length_counts; /* used for message length profiling */
74 u32 msg_lengths_total; /* used for message length profiling */
75 u32 msg_length_profile[7]; /* used for msg. length profiling */
79 * struct tipc_link - TIPC link data structure
80 * @addr: network address of link's peer node
81 * @name: link name character string
82 * @media_addr: media address to use when sending messages over link
84 * @net: pointer to namespace struct
85 * @refcnt: reference counter for permanent references (owner node & timer)
86 * @peer_session: link session # being used by peer end of link
87 * @peer_bearer_id: bearer id used by link's peer endpoint
88 * @bearer_id: local bearer id used by link
89 * @tolerance: minimum link continuity loss needed to reset link [in ms]
90 * @keepalive_intv: link keepalive timer interval
91 * @abort_limit: # of unacknowledged continuity probes needed to reset link
92 * @state: current state of link FSM
93 * @peer_caps: bitmap describing capabilities of peer node
94 * @silent_intv_cnt: # of timer intervals without any reception from peer
95 * @proto_msg: template for control messages generated by link
96 * @pmsg: convenience pointer to "proto_msg" field
97 * @priority: current link priority
98 * @net_plane: current link network plane ('A' through 'H')
99 * @backlog_limit: backlog queue congestion thresholds (indexed by importance)
100 * @exp_msg_count: # of tunnelled messages expected during link changeover
101 * @reset_rcv_checkpt: seq # of last acknowledged message at time of link reset
102 * @mtu: current maximum packet size for this link
103 * @advertised_mtu: advertised own mtu when link is being established
104 * @transmitq: queue for sent, non-acked messages
105 * @backlogq: queue for messages waiting to be sent
106 * @snt_nxt: next sequence number to use for outbound messages
107 * @last_retransmitted: sequence number of most recently retransmitted message
108 * @stale_count: # of identical retransmit requests made by peer
109 * @ackers: # of peers that needs to ack each packet before it can be released
110 * @acked: # last packet acked by a certain peer. Used for broadcast.
111 * @rcv_nxt: next sequence number to expect for inbound messages
112 * @deferred_queue: deferred queue saved OOS b'cast message received from node
113 * @unacked_window: # of inbound messages rx'd without ack'ing back to peer
114 * @inputq: buffer queue for messages to be delivered upwards
115 * @namedq: buffer queue for name table messages to be delivered upwards
116 * @next_out: ptr to first unsent outbound message in queue
117 * @wakeupq: linked list of wakeup msgs waiting for link congestion to abate
118 * @long_msg_seq_no: next identifier to use for outbound fragmented messages
119 * @reasm_buf: head of partially reassembled inbound message fragments
120 * @bc_rcvr: marks that this is a broadcast receiver link
121 * @stats: collects statistics regarding link activity
125 char name[TIPC_MAX_LINK_NAME];
128 /* Management and link supervision data */
134 unsigned long keepalive_intv;
140 char if_name[TIPC_MAX_IF_NAME];
147 struct sk_buff *failover_reasm_skb;
149 /* Max packet negotiation */
154 struct sk_buff_head transmq;
155 struct sk_buff_head backlogq;
168 struct sk_buff_head deferdq;
169 struct sk_buff_head *inputq;
170 struct sk_buff_head *namedq;
172 /* Congestion handling */
173 struct sk_buff_head wakeupq;
175 /* Fragmentation/reassembly */
176 struct sk_buff *reasm_buf;
181 struct tipc_link *bc_rcvlink;
182 struct tipc_link *bc_sndlink;
187 struct tipc_stats stats;
191 * Error message prefixes
193 static const char *link_co_err = "Link tunneling error, ";
194 static const char *link_rst_msg = "Resetting link ";
196 /* Send states for broadcast NACKs
199 BC_NACK_SND_CONDITIONAL,
200 BC_NACK_SND_UNCONDITIONAL,
201 BC_NACK_SND_SUPPRESS,
205 * Interval between NACKs when packets arrive out of order
207 #define TIPC_NACK_INTV (TIPC_MIN_LINK_WIN * 2)
209 /* Wildcard value for link session numbers. When it is known that
210 * peer endpoint is down, any session number must be accepted.
212 #define ANY_SESSION 0x10000
217 LINK_ESTABLISHED = 0xe,
218 LINK_ESTABLISHING = 0xe << 4,
219 LINK_RESET = 0x1 << 8,
220 LINK_RESETTING = 0x2 << 12,
221 LINK_PEER_RESET = 0xd << 16,
222 LINK_FAILINGOVER = 0xf << 20,
223 LINK_SYNCHING = 0xc << 24
226 /* Link FSM state checking routines
228 static int link_is_up(struct tipc_link *l)
230 return l->state & (LINK_ESTABLISHED | LINK_SYNCHING);
233 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
234 struct sk_buff_head *xmitq);
235 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
236 u16 rcvgap, int tolerance, int priority,
237 struct sk_buff_head *xmitq);
238 static void link_print(struct tipc_link *l, const char *str);
239 static void tipc_link_build_nack_msg(struct tipc_link *l,
240 struct sk_buff_head *xmitq);
241 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
242 struct sk_buff_head *xmitq);
243 static bool tipc_link_release_pkts(struct tipc_link *l, u16 to);
246 * Simple non-static link routines (i.e. referenced outside this file)
248 bool tipc_link_is_up(struct tipc_link *l)
250 return link_is_up(l);
253 bool tipc_link_peer_is_down(struct tipc_link *l)
255 return l->state == LINK_PEER_RESET;
258 bool tipc_link_is_reset(struct tipc_link *l)
260 return l->state & (LINK_RESET | LINK_FAILINGOVER | LINK_ESTABLISHING);
263 bool tipc_link_is_establishing(struct tipc_link *l)
265 return l->state == LINK_ESTABLISHING;
268 bool tipc_link_is_synching(struct tipc_link *l)
270 return l->state == LINK_SYNCHING;
273 bool tipc_link_is_failingover(struct tipc_link *l)
275 return l->state == LINK_FAILINGOVER;
278 bool tipc_link_is_blocked(struct tipc_link *l)
280 return l->state & (LINK_RESETTING | LINK_PEER_RESET | LINK_FAILINGOVER);
283 static bool link_is_bc_sndlink(struct tipc_link *l)
285 return !l->bc_sndlink;
288 static bool link_is_bc_rcvlink(struct tipc_link *l)
290 return ((l->bc_rcvlink == l) && !link_is_bc_sndlink(l));
293 int tipc_link_is_active(struct tipc_link *l)
298 void tipc_link_set_active(struct tipc_link *l, bool active)
303 u32 tipc_link_id(struct tipc_link *l)
305 return l->peer_bearer_id << 16 | l->bearer_id;
308 int tipc_link_window(struct tipc_link *l)
313 int tipc_link_prio(struct tipc_link *l)
318 unsigned long tipc_link_tolerance(struct tipc_link *l)
323 struct sk_buff_head *tipc_link_inputq(struct tipc_link *l)
328 char tipc_link_plane(struct tipc_link *l)
333 void tipc_link_add_bc_peer(struct tipc_link *snd_l,
334 struct tipc_link *uc_l,
335 struct sk_buff_head *xmitq)
337 struct tipc_link *rcv_l = uc_l->bc_rcvlink;
340 rcv_l->acked = snd_l->snd_nxt - 1;
341 snd_l->state = LINK_ESTABLISHED;
342 tipc_link_build_bc_init_msg(uc_l, xmitq);
345 void tipc_link_remove_bc_peer(struct tipc_link *snd_l,
346 struct tipc_link *rcv_l,
347 struct sk_buff_head *xmitq)
349 u16 ack = snd_l->snd_nxt - 1;
352 tipc_link_bc_ack_rcv(rcv_l, ack, xmitq);
353 tipc_link_reset(rcv_l);
354 rcv_l->state = LINK_RESET;
355 if (!snd_l->ackers) {
356 tipc_link_reset(snd_l);
357 snd_l->state = LINK_RESET;
358 __skb_queue_purge(xmitq);
362 int tipc_link_bc_peers(struct tipc_link *l)
367 void tipc_link_set_mtu(struct tipc_link *l, int mtu)
372 int tipc_link_mtu(struct tipc_link *l)
377 u16 tipc_link_rcv_nxt(struct tipc_link *l)
382 u16 tipc_link_acked(struct tipc_link *l)
387 char *tipc_link_name(struct tipc_link *l)
393 * tipc_link_create - create a new link
394 * @n: pointer to associated node
395 * @if_name: associated interface name
396 * @bearer_id: id (index) of associated bearer
397 * @tolerance: link tolerance to be used by link
398 * @net_plane: network plane (A,B,c..) this link belongs to
399 * @mtu: mtu to be advertised by link
400 * @priority: priority to be used by link
401 * @window: send window to be used by link
402 * @session: session to be used by link
403 * @ownnode: identity of own node
404 * @peer: node id of peer node
405 * @peer_caps: bitmap describing peer node capabilities
406 * @bc_sndlink: the namespace global link used for broadcast sending
407 * @bc_rcvlink: the peer specific link used for broadcast reception
408 * @inputq: queue to put messages ready for delivery
409 * @namedq: queue to put binding table update messages ready for delivery
410 * @link: return value, pointer to put the created link
412 * Returns true if link was created, otherwise false
414 bool tipc_link_create(struct net *net, char *if_name, int bearer_id,
415 int tolerance, char net_plane, u32 mtu, int priority,
416 int window, u32 session, u32 ownnode, u32 peer,
418 struct tipc_link *bc_sndlink,
419 struct tipc_link *bc_rcvlink,
420 struct sk_buff_head *inputq,
421 struct sk_buff_head *namedq,
422 struct tipc_link **link)
426 l = kzalloc(sizeof(*l), GFP_ATOMIC);
430 l->session = session;
432 /* Note: peer i/f name is completed by reset/activate message */
433 sprintf(l->name, "%u.%u.%u:%s-%u.%u.%u:unknown",
434 tipc_zone(ownnode), tipc_cluster(ownnode), tipc_node(ownnode),
435 if_name, tipc_zone(peer), tipc_cluster(peer), tipc_node(peer));
436 strcpy(l->if_name, if_name);
438 l->peer_caps = peer_caps;
440 l->peer_session = ANY_SESSION;
441 l->bearer_id = bearer_id;
442 l->tolerance = tolerance;
443 l->net_plane = net_plane;
444 l->advertised_mtu = mtu;
446 l->priority = priority;
447 tipc_link_set_queue_limits(l, window);
449 l->bc_sndlink = bc_sndlink;
450 l->bc_rcvlink = bc_rcvlink;
453 l->state = LINK_RESETTING;
454 __skb_queue_head_init(&l->transmq);
455 __skb_queue_head_init(&l->backlogq);
456 __skb_queue_head_init(&l->deferdq);
457 skb_queue_head_init(&l->wakeupq);
458 skb_queue_head_init(l->inputq);
463 * tipc_link_bc_create - create new link to be used for broadcast
464 * @n: pointer to associated node
465 * @mtu: mtu to be used
466 * @window: send window to be used
467 * @inputq: queue to put messages ready for delivery
468 * @namedq: queue to put binding table update messages ready for delivery
469 * @link: return value, pointer to put the created link
471 * Returns true if link was created, otherwise false
473 bool tipc_link_bc_create(struct net *net, u32 ownnode, u32 peer,
474 int mtu, int window, u16 peer_caps,
475 struct sk_buff_head *inputq,
476 struct sk_buff_head *namedq,
477 struct tipc_link *bc_sndlink,
478 struct tipc_link **link)
482 if (!tipc_link_create(net, "", MAX_BEARERS, 0, 'Z', mtu, 0, window,
483 0, ownnode, peer, peer_caps, bc_sndlink,
484 NULL, inputq, namedq, link))
488 strcpy(l->name, tipc_bclink_name);
490 l->state = LINK_RESET;
494 /* Broadcast send link is always up */
495 if (link_is_bc_sndlink(l))
496 l->state = LINK_ESTABLISHED;
502 * tipc_link_fsm_evt - link finite state machine
503 * @l: pointer to link
504 * @evt: state machine event to be processed
506 int tipc_link_fsm_evt(struct tipc_link *l, int evt)
513 case LINK_PEER_RESET_EVT:
514 l->state = LINK_PEER_RESET;
517 l->state = LINK_RESET;
519 case LINK_FAILURE_EVT:
520 case LINK_FAILOVER_BEGIN_EVT:
521 case LINK_ESTABLISH_EVT:
522 case LINK_FAILOVER_END_EVT:
523 case LINK_SYNCH_BEGIN_EVT:
524 case LINK_SYNCH_END_EVT:
531 case LINK_PEER_RESET_EVT:
532 l->state = LINK_ESTABLISHING;
534 case LINK_FAILOVER_BEGIN_EVT:
535 l->state = LINK_FAILINGOVER;
536 case LINK_FAILURE_EVT:
538 case LINK_ESTABLISH_EVT:
539 case LINK_FAILOVER_END_EVT:
541 case LINK_SYNCH_BEGIN_EVT:
542 case LINK_SYNCH_END_EVT:
547 case LINK_PEER_RESET:
550 l->state = LINK_ESTABLISHING;
552 case LINK_PEER_RESET_EVT:
553 case LINK_ESTABLISH_EVT:
554 case LINK_FAILURE_EVT:
556 case LINK_SYNCH_BEGIN_EVT:
557 case LINK_SYNCH_END_EVT:
558 case LINK_FAILOVER_BEGIN_EVT:
559 case LINK_FAILOVER_END_EVT:
564 case LINK_FAILINGOVER:
566 case LINK_FAILOVER_END_EVT:
567 l->state = LINK_RESET;
569 case LINK_PEER_RESET_EVT:
571 case LINK_ESTABLISH_EVT:
572 case LINK_FAILURE_EVT:
574 case LINK_FAILOVER_BEGIN_EVT:
575 case LINK_SYNCH_BEGIN_EVT:
576 case LINK_SYNCH_END_EVT:
581 case LINK_ESTABLISHING:
583 case LINK_ESTABLISH_EVT:
584 l->state = LINK_ESTABLISHED;
586 case LINK_FAILOVER_BEGIN_EVT:
587 l->state = LINK_FAILINGOVER;
590 l->state = LINK_RESET;
592 case LINK_FAILURE_EVT:
593 case LINK_PEER_RESET_EVT:
594 case LINK_SYNCH_BEGIN_EVT:
595 case LINK_FAILOVER_END_EVT:
597 case LINK_SYNCH_END_EVT:
602 case LINK_ESTABLISHED:
604 case LINK_PEER_RESET_EVT:
605 l->state = LINK_PEER_RESET;
606 rc |= TIPC_LINK_DOWN_EVT;
608 case LINK_FAILURE_EVT:
609 l->state = LINK_RESETTING;
610 rc |= TIPC_LINK_DOWN_EVT;
613 l->state = LINK_RESET;
615 case LINK_ESTABLISH_EVT:
616 case LINK_SYNCH_END_EVT:
618 case LINK_SYNCH_BEGIN_EVT:
619 l->state = LINK_SYNCHING;
621 case LINK_FAILOVER_BEGIN_EVT:
622 case LINK_FAILOVER_END_EVT:
629 case LINK_PEER_RESET_EVT:
630 l->state = LINK_PEER_RESET;
631 rc |= TIPC_LINK_DOWN_EVT;
633 case LINK_FAILURE_EVT:
634 l->state = LINK_RESETTING;
635 rc |= TIPC_LINK_DOWN_EVT;
638 l->state = LINK_RESET;
640 case LINK_ESTABLISH_EVT:
641 case LINK_SYNCH_BEGIN_EVT:
643 case LINK_SYNCH_END_EVT:
644 l->state = LINK_ESTABLISHED;
646 case LINK_FAILOVER_BEGIN_EVT:
647 case LINK_FAILOVER_END_EVT:
653 pr_err("Unknown FSM state %x in %s\n", l->state, l->name);
657 pr_err("Illegal FSM event %x in state %x on link %s\n",
658 evt, l->state, l->name);
662 /* link_profile_stats - update statistical profiling of traffic
664 static void link_profile_stats(struct tipc_link *l)
667 struct tipc_msg *msg;
670 /* Update counters used in statistical profiling of send traffic */
671 l->stats.accu_queue_sz += skb_queue_len(&l->transmq);
672 l->stats.queue_sz_counts++;
674 skb = skb_peek(&l->transmq);
678 length = msg_size(msg);
680 if (msg_user(msg) == MSG_FRAGMENTER) {
681 if (msg_type(msg) != FIRST_FRAGMENT)
683 length = msg_size(msg_get_wrapped(msg));
685 l->stats.msg_lengths_total += length;
686 l->stats.msg_length_counts++;
688 l->stats.msg_length_profile[0]++;
689 else if (length <= 256)
690 l->stats.msg_length_profile[1]++;
691 else if (length <= 1024)
692 l->stats.msg_length_profile[2]++;
693 else if (length <= 4096)
694 l->stats.msg_length_profile[3]++;
695 else if (length <= 16384)
696 l->stats.msg_length_profile[4]++;
697 else if (length <= 32768)
698 l->stats.msg_length_profile[5]++;
700 l->stats.msg_length_profile[6]++;
703 /* tipc_link_timeout - perform periodic task as instructed from node timeout
705 int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq)
711 u16 bc_snt = l->bc_sndlink->snd_nxt - 1;
712 u16 bc_acked = l->bc_rcvlink->acked;
714 link_profile_stats(l);
717 case LINK_ESTABLISHED:
719 if (l->silent_intv_cnt > l->abort_limit)
720 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
722 state = bc_acked != bc_snt;
723 probe = l->silent_intv_cnt;
724 l->silent_intv_cnt++;
727 setup = l->rst_cnt++ <= 4;
728 setup |= !(l->rst_cnt % 16);
731 case LINK_ESTABLISHING:
735 case LINK_PEER_RESET:
737 case LINK_FAILINGOVER:
743 if (state || probe || setup)
744 tipc_link_build_proto_msg(l, mtyp, probe, 0, 0, 0, xmitq);
750 * link_schedule_user - schedule a message sender for wakeup after congestion
751 * @link: congested link
752 * @list: message that was attempted sent
753 * Create pseudo msg to send back to user when congestion abates
754 * Does not consume buffer list
756 static int link_schedule_user(struct tipc_link *link, struct sk_buff_head *list)
758 struct tipc_msg *msg = buf_msg(skb_peek(list));
759 int imp = msg_importance(msg);
760 u32 oport = msg_origport(msg);
761 u32 addr = tipc_own_addr(link->net);
764 /* This really cannot happen... */
765 if (unlikely(imp > TIPC_CRITICAL_IMPORTANCE)) {
766 pr_warn("%s<%s>, send queue full", link_rst_msg, link->name);
769 /* Non-blocking sender: */
770 if (TIPC_SKB_CB(skb_peek(list))->wakeup_pending)
773 /* Create and schedule wakeup pseudo message */
774 skb = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
775 addr, addr, oport, 0, 0);
778 TIPC_SKB_CB(skb)->chain_sz = skb_queue_len(list);
779 TIPC_SKB_CB(skb)->chain_imp = imp;
780 skb_queue_tail(&link->wakeupq, skb);
781 link->stats.link_congs++;
786 * link_prepare_wakeup - prepare users for wakeup after congestion
787 * @link: congested link
788 * Move a number of waiting users, as permitted by available space in
789 * the send queue, from link wait queue to node wait queue for wakeup
791 void link_prepare_wakeup(struct tipc_link *l)
793 int pnd[TIPC_SYSTEM_IMPORTANCE + 1] = {0,};
795 struct sk_buff *skb, *tmp;
797 skb_queue_walk_safe(&l->wakeupq, skb, tmp) {
798 imp = TIPC_SKB_CB(skb)->chain_imp;
799 lim = l->window + l->backlog[imp].limit;
800 pnd[imp] += TIPC_SKB_CB(skb)->chain_sz;
801 if ((pnd[imp] + l->backlog[imp].len) >= lim)
803 skb_unlink(skb, &l->wakeupq);
804 skb_queue_tail(l->inputq, skb);
808 void tipc_link_reset(struct tipc_link *l)
810 l->peer_session = ANY_SESSION;
812 l->mtu = l->advertised_mtu;
813 __skb_queue_purge(&l->transmq);
814 __skb_queue_purge(&l->deferdq);
815 skb_queue_splice_init(&l->wakeupq, l->inputq);
816 __skb_queue_purge(&l->backlogq);
817 l->backlog[TIPC_LOW_IMPORTANCE].len = 0;
818 l->backlog[TIPC_MEDIUM_IMPORTANCE].len = 0;
819 l->backlog[TIPC_HIGH_IMPORTANCE].len = 0;
820 l->backlog[TIPC_CRITICAL_IMPORTANCE].len = 0;
821 l->backlog[TIPC_SYSTEM_IMPORTANCE].len = 0;
822 kfree_skb(l->reasm_buf);
823 kfree_skb(l->failover_reasm_skb);
825 l->failover_reasm_skb = NULL;
830 l->silent_intv_cnt = 0;
832 l->stats.recv_info = 0;
834 l->bc_peer_is_up = false;
835 tipc_link_reset_stats(l);
839 * tipc_link_xmit(): enqueue buffer list according to queue situation
841 * @list: chain of buffers containing message
842 * @xmitq: returned list of packets to be sent by caller
844 * Consumes the buffer chain, except when returning -ELINKCONG,
845 * since the caller then may want to make more send attempts.
846 * Returns 0 if success, or errno: -ELINKCONG, -EMSGSIZE or -ENOBUFS
847 * Messages at TIPC_SYSTEM_IMPORTANCE are always accepted
849 int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list,
850 struct sk_buff_head *xmitq)
852 struct tipc_msg *hdr = buf_msg(skb_peek(list));
853 unsigned int maxwin = l->window;
854 unsigned int i, imp = msg_importance(hdr);
855 unsigned int mtu = l->mtu;
856 u16 ack = l->rcv_nxt - 1;
857 u16 seqno = l->snd_nxt;
858 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
859 struct sk_buff_head *transmq = &l->transmq;
860 struct sk_buff_head *backlogq = &l->backlogq;
861 struct sk_buff *skb, *_skb, *bskb;
863 /* Match msg importance against this and all higher backlog limits: */
864 for (i = imp; i <= TIPC_SYSTEM_IMPORTANCE; i++) {
865 if (unlikely(l->backlog[i].len >= l->backlog[i].limit))
866 return link_schedule_user(l, list);
868 if (unlikely(msg_size(hdr) > mtu)) {
869 skb_queue_purge(list);
873 /* Prepare each packet for sending, and add to relevant queue: */
874 while (skb_queue_len(list)) {
875 skb = skb_peek(list);
877 msg_set_seqno(hdr, seqno);
878 msg_set_ack(hdr, ack);
879 msg_set_bcast_ack(hdr, bc_ack);
881 if (likely(skb_queue_len(transmq) < maxwin)) {
882 _skb = skb_clone(skb, GFP_ATOMIC);
884 skb_queue_purge(list);
888 __skb_queue_tail(transmq, skb);
889 __skb_queue_tail(xmitq, _skb);
890 TIPC_SKB_CB(skb)->ackers = l->ackers;
895 if (tipc_msg_bundle(skb_peek_tail(backlogq), hdr, mtu)) {
896 kfree_skb(__skb_dequeue(list));
897 l->stats.sent_bundled++;
900 if (tipc_msg_make_bundle(&bskb, hdr, mtu, l->addr)) {
901 kfree_skb(__skb_dequeue(list));
902 __skb_queue_tail(backlogq, bskb);
903 l->backlog[msg_importance(buf_msg(bskb))].len++;
904 l->stats.sent_bundled++;
905 l->stats.sent_bundles++;
908 l->backlog[imp].len += skb_queue_len(list);
909 skb_queue_splice_tail_init(list, backlogq);
915 void tipc_link_advance_backlog(struct tipc_link *l, struct sk_buff_head *xmitq)
917 struct sk_buff *skb, *_skb;
918 struct tipc_msg *hdr;
919 u16 seqno = l->snd_nxt;
920 u16 ack = l->rcv_nxt - 1;
921 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
923 while (skb_queue_len(&l->transmq) < l->window) {
924 skb = skb_peek(&l->backlogq);
927 _skb = skb_clone(skb, GFP_ATOMIC);
930 __skb_dequeue(&l->backlogq);
932 l->backlog[msg_importance(hdr)].len--;
933 __skb_queue_tail(&l->transmq, skb);
934 __skb_queue_tail(xmitq, _skb);
935 TIPC_SKB_CB(skb)->ackers = l->ackers;
936 msg_set_seqno(hdr, seqno);
937 msg_set_ack(hdr, ack);
938 msg_set_bcast_ack(hdr, bc_ack);
945 static void link_retransmit_failure(struct tipc_link *l, struct sk_buff *skb)
947 struct tipc_msg *hdr = buf_msg(skb);
949 pr_warn("Retransmission failure on link <%s>\n", l->name);
950 link_print(l, "Resetting link ");
951 pr_info("Failed msg: usr %u, typ %u, len %u, err %u\n",
952 msg_user(hdr), msg_type(hdr), msg_size(hdr), msg_errcode(hdr));
953 pr_info("sqno %u, prev: %x, src: %x\n",
954 msg_seqno(hdr), msg_prevnode(hdr), msg_orignode(hdr));
957 int tipc_link_retrans(struct tipc_link *l, u16 from, u16 to,
958 struct sk_buff_head *xmitq)
960 struct sk_buff *_skb, *skb = skb_peek(&l->transmq);
961 struct tipc_msg *hdr;
962 u16 ack = l->rcv_nxt - 1;
963 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
968 /* Detect repeated retransmit failures on same packet */
969 if (likely(l->last_retransm != buf_seqno(skb))) {
970 l->last_retransm = buf_seqno(skb);
972 } else if (++l->stale_count > 100) {
973 link_retransmit_failure(l, skb);
974 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
977 /* Move forward to where retransmission should start */
978 skb_queue_walk(&l->transmq, skb) {
979 if (!less(buf_seqno(skb), from))
983 skb_queue_walk_from(&l->transmq, skb) {
984 if (more(buf_seqno(skb), to))
987 _skb = __pskb_copy(skb, MIN_H_SIZE, GFP_ATOMIC);
991 msg_set_ack(hdr, ack);
992 msg_set_bcast_ack(hdr, bc_ack);
993 _skb->priority = TC_PRIO_CONTROL;
994 __skb_queue_tail(xmitq, _skb);
995 l->stats.retransmitted++;
1000 /* tipc_data_input - deliver data and name distr msgs to upper layer
1002 * Consumes buffer if message is of right type
1003 * Node lock must be held
1005 static bool tipc_data_input(struct tipc_link *l, struct sk_buff *skb,
1006 struct sk_buff_head *inputq)
1008 switch (msg_user(buf_msg(skb))) {
1009 case TIPC_LOW_IMPORTANCE:
1010 case TIPC_MEDIUM_IMPORTANCE:
1011 case TIPC_HIGH_IMPORTANCE:
1012 case TIPC_CRITICAL_IMPORTANCE:
1014 skb_queue_tail(inputq, skb);
1016 case NAME_DISTRIBUTOR:
1017 l->bc_rcvlink->state = LINK_ESTABLISHED;
1018 skb_queue_tail(l->namedq, skb);
1021 case TUNNEL_PROTOCOL:
1022 case MSG_FRAGMENTER:
1023 case BCAST_PROTOCOL:
1026 pr_warn("Dropping received illegal msg type\n");
1032 /* tipc_link_input - process packet that has passed link protocol check
1036 static int tipc_link_input(struct tipc_link *l, struct sk_buff *skb,
1037 struct sk_buff_head *inputq)
1039 struct tipc_msg *hdr = buf_msg(skb);
1040 struct sk_buff **reasm_skb = &l->reasm_buf;
1041 struct sk_buff *iskb;
1042 struct sk_buff_head tmpq;
1043 int usr = msg_user(hdr);
1048 if (unlikely(usr == TUNNEL_PROTOCOL)) {
1049 if (msg_type(hdr) == SYNCH_MSG) {
1050 __skb_queue_purge(&l->deferdq);
1053 if (!tipc_msg_extract(skb, &iskb, &ipos))
1058 if (less(msg_seqno(hdr), l->drop_point))
1060 if (tipc_data_input(l, skb, inputq))
1062 usr = msg_user(hdr);
1063 reasm_skb = &l->failover_reasm_skb;
1066 if (usr == MSG_BUNDLER) {
1067 skb_queue_head_init(&tmpq);
1068 l->stats.recv_bundles++;
1069 l->stats.recv_bundled += msg_msgcnt(hdr);
1070 while (tipc_msg_extract(skb, &iskb, &pos))
1071 tipc_data_input(l, iskb, &tmpq);
1072 tipc_skb_queue_splice_tail(&tmpq, inputq);
1074 } else if (usr == MSG_FRAGMENTER) {
1075 l->stats.recv_fragments++;
1076 if (tipc_buf_append(reasm_skb, &skb)) {
1077 l->stats.recv_fragmented++;
1078 tipc_data_input(l, skb, inputq);
1079 } else if (!*reasm_skb && !link_is_bc_rcvlink(l)) {
1080 pr_warn_ratelimited("Unable to build fragment list\n");
1081 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1084 } else if (usr == BCAST_PROTOCOL) {
1085 tipc_bcast_lock(l->net);
1086 tipc_link_bc_init_rcv(l->bc_rcvlink, hdr);
1087 tipc_bcast_unlock(l->net);
1094 static bool tipc_link_release_pkts(struct tipc_link *l, u16 acked)
1096 bool released = false;
1097 struct sk_buff *skb, *tmp;
1099 skb_queue_walk_safe(&l->transmq, skb, tmp) {
1100 if (more(buf_seqno(skb), acked))
1102 __skb_unlink(skb, &l->transmq);
1109 /* tipc_link_build_state_msg: prepare link state message for transmission
1111 * Note that sending of broadcast ack is coordinated among nodes, to reduce
1112 * risk of ack storms towards the sender
1114 int tipc_link_build_state_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1119 /* Broadcast ACK must be sent via a unicast link => defer to caller */
1120 if (link_is_bc_rcvlink(l)) {
1121 if (((l->rcv_nxt ^ tipc_own_addr(l->net)) & 0xf) != 0xf)
1124 return TIPC_LINK_SND_BC_ACK;
1129 l->stats.sent_acks++;
1130 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, xmitq);
1134 /* tipc_link_build_reset_msg: prepare link RESET or ACTIVATE message
1136 void tipc_link_build_reset_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1138 int mtyp = RESET_MSG;
1139 struct sk_buff *skb;
1141 if (l->state == LINK_ESTABLISHING)
1142 mtyp = ACTIVATE_MSG;
1144 tipc_link_build_proto_msg(l, mtyp, 0, 0, 0, 0, xmitq);
1146 /* Inform peer that this endpoint is going down if applicable */
1147 skb = skb_peek_tail(xmitq);
1148 if (skb && (l->state == LINK_RESET))
1149 msg_set_peer_stopping(buf_msg(skb), 1);
1152 /* tipc_link_build_nack_msg: prepare link nack message for transmission
1154 static void tipc_link_build_nack_msg(struct tipc_link *l,
1155 struct sk_buff_head *xmitq)
1157 u32 def_cnt = ++l->stats.deferred_recv;
1159 if (link_is_bc_rcvlink(l))
1162 if ((skb_queue_len(&l->deferdq) == 1) || !(def_cnt % TIPC_NACK_INTV))
1163 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, xmitq);
1166 /* tipc_link_rcv - process TIPC packets/messages arriving from off-node
1167 * @l: the link that should handle the message
1169 * @xmitq: queue to place packets to be sent after this call
1171 int tipc_link_rcv(struct tipc_link *l, struct sk_buff *skb,
1172 struct sk_buff_head *xmitq)
1174 struct sk_buff_head *defq = &l->deferdq;
1175 struct tipc_msg *hdr;
1176 u16 seqno, rcv_nxt, win_lim;
1181 seqno = msg_seqno(hdr);
1182 rcv_nxt = l->rcv_nxt;
1183 win_lim = rcv_nxt + TIPC_MAX_LINK_WIN;
1185 /* Verify and update link state */
1186 if (unlikely(msg_user(hdr) == LINK_PROTOCOL))
1187 return tipc_link_proto_rcv(l, skb, xmitq);
1189 if (unlikely(!link_is_up(l))) {
1190 if (l->state == LINK_ESTABLISHING)
1191 rc = TIPC_LINK_UP_EVT;
1195 /* Don't send probe at next timeout expiration */
1196 l->silent_intv_cnt = 0;
1198 /* Drop if outside receive window */
1199 if (unlikely(less(seqno, rcv_nxt) || more(seqno, win_lim))) {
1200 l->stats.duplicates++;
1204 /* Forward queues and wake up waiting users */
1205 if (likely(tipc_link_release_pkts(l, msg_ack(hdr)))) {
1206 tipc_link_advance_backlog(l, xmitq);
1207 if (unlikely(!skb_queue_empty(&l->wakeupq)))
1208 link_prepare_wakeup(l);
1211 /* Defer delivery if sequence gap */
1212 if (unlikely(seqno != rcv_nxt)) {
1213 __tipc_skb_queue_sorted(defq, seqno, skb);
1214 tipc_link_build_nack_msg(l, xmitq);
1218 /* Deliver packet */
1220 l->stats.recv_info++;
1221 if (!tipc_data_input(l, skb, l->inputq))
1222 rc |= tipc_link_input(l, skb, l->inputq);
1223 if (unlikely(++l->rcv_unacked >= TIPC_MIN_LINK_WIN))
1224 rc |= tipc_link_build_state_msg(l, xmitq);
1225 if (unlikely(rc & ~TIPC_LINK_SND_BC_ACK))
1227 } while ((skb = __skb_dequeue(defq)));
1235 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
1236 u16 rcvgap, int tolerance, int priority,
1237 struct sk_buff_head *xmitq)
1239 struct sk_buff *skb;
1240 struct tipc_msg *hdr;
1241 struct sk_buff_head *dfq = &l->deferdq;
1242 bool node_up = link_is_up(l->bc_rcvlink);
1244 /* Don't send protocol message during reset or link failover */
1245 if (tipc_link_is_blocked(l))
1248 if (!tipc_link_is_up(l) && (mtyp == STATE_MSG))
1251 if (!skb_queue_empty(dfq))
1252 rcvgap = buf_seqno(skb_peek(dfq)) - l->rcv_nxt;
1254 skb = tipc_msg_create(LINK_PROTOCOL, mtyp, INT_H_SIZE,
1255 TIPC_MAX_IF_NAME, l->addr,
1256 tipc_own_addr(l->net), 0, 0, 0);
1261 msg_set_session(hdr, l->session);
1262 msg_set_bearer_id(hdr, l->bearer_id);
1263 msg_set_net_plane(hdr, l->net_plane);
1264 msg_set_next_sent(hdr, l->snd_nxt);
1265 msg_set_ack(hdr, l->rcv_nxt - 1);
1266 msg_set_bcast_ack(hdr, l->bc_rcvlink->rcv_nxt - 1);
1267 msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1268 msg_set_link_tolerance(hdr, tolerance);
1269 msg_set_linkprio(hdr, priority);
1270 msg_set_redundant_link(hdr, node_up);
1271 msg_set_seq_gap(hdr, 0);
1272 msg_set_seqno(hdr, l->snd_nxt + U16_MAX / 2);
1274 if (mtyp == STATE_MSG) {
1275 msg_set_seq_gap(hdr, rcvgap);
1276 msg_set_size(hdr, INT_H_SIZE);
1277 msg_set_probe(hdr, probe);
1278 l->stats.sent_states++;
1281 /* RESET_MSG or ACTIVATE_MSG */
1282 msg_set_max_pkt(hdr, l->advertised_mtu);
1283 strcpy(msg_data(hdr), l->if_name);
1286 l->stats.sent_probes++;
1288 l->stats.sent_nacks++;
1289 skb->priority = TC_PRIO_CONTROL;
1290 __skb_queue_tail(xmitq, skb);
1293 /* tipc_link_tnl_prepare(): prepare and return a list of tunnel packets
1294 * with contents of the link's transmit and backlog queues.
1296 void tipc_link_tnl_prepare(struct tipc_link *l, struct tipc_link *tnl,
1297 int mtyp, struct sk_buff_head *xmitq)
1299 struct sk_buff *skb, *tnlskb;
1300 struct tipc_msg *hdr, tnlhdr;
1301 struct sk_buff_head *queue = &l->transmq;
1302 struct sk_buff_head tmpxq, tnlq;
1303 u16 pktlen, pktcnt, seqno = l->snd_nxt;
1308 skb_queue_head_init(&tnlq);
1309 skb_queue_head_init(&tmpxq);
1311 /* At least one packet required for safe algorithm => add dummy */
1312 skb = tipc_msg_create(TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1313 BASIC_H_SIZE, 0, l->addr, tipc_own_addr(l->net),
1314 0, 0, TIPC_ERR_NO_PORT);
1316 pr_warn("%sunable to create tunnel packet\n", link_co_err);
1319 skb_queue_tail(&tnlq, skb);
1320 tipc_link_xmit(l, &tnlq, &tmpxq);
1321 __skb_queue_purge(&tmpxq);
1323 /* Initialize reusable tunnel packet header */
1324 tipc_msg_init(tipc_own_addr(l->net), &tnlhdr, TUNNEL_PROTOCOL,
1325 mtyp, INT_H_SIZE, l->addr);
1326 pktcnt = skb_queue_len(&l->transmq) + skb_queue_len(&l->backlogq);
1327 msg_set_msgcnt(&tnlhdr, pktcnt);
1328 msg_set_bearer_id(&tnlhdr, l->peer_bearer_id);
1330 /* Wrap each packet into a tunnel packet */
1331 skb_queue_walk(queue, skb) {
1333 if (queue == &l->backlogq)
1334 msg_set_seqno(hdr, seqno++);
1335 pktlen = msg_size(hdr);
1336 msg_set_size(&tnlhdr, pktlen + INT_H_SIZE);
1337 tnlskb = tipc_buf_acquire(pktlen + INT_H_SIZE);
1339 pr_warn("%sunable to send packet\n", link_co_err);
1342 skb_copy_to_linear_data(tnlskb, &tnlhdr, INT_H_SIZE);
1343 skb_copy_to_linear_data_offset(tnlskb, INT_H_SIZE, hdr, pktlen);
1344 __skb_queue_tail(&tnlq, tnlskb);
1346 if (queue != &l->backlogq) {
1347 queue = &l->backlogq;
1351 tipc_link_xmit(tnl, &tnlq, xmitq);
1353 if (mtyp == FAILOVER_MSG) {
1354 tnl->drop_point = l->rcv_nxt;
1355 tnl->failover_reasm_skb = l->reasm_buf;
1356 l->reasm_buf = NULL;
1360 /* tipc_link_proto_rcv(): receive link level protocol message :
1361 * Note that network plane id propagates through the network, and may
1362 * change at any time. The node with lowest numerical id determines
1365 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
1366 struct sk_buff_head *xmitq)
1368 struct tipc_msg *hdr = buf_msg(skb);
1370 u16 ack = msg_ack(hdr);
1371 u16 gap = msg_seq_gap(hdr);
1372 u16 peers_snd_nxt = msg_next_sent(hdr);
1373 u16 peers_tol = msg_link_tolerance(hdr);
1374 u16 peers_prio = msg_linkprio(hdr);
1375 u16 rcv_nxt = l->rcv_nxt;
1376 int mtyp = msg_type(hdr);
1380 if (tipc_link_is_blocked(l) || !xmitq)
1383 if (tipc_own_addr(l->net) > msg_prevnode(hdr))
1384 l->net_plane = msg_net_plane(hdr);
1389 /* Ignore duplicate RESET with old session number */
1390 if ((less_eq(msg_session(hdr), l->peer_session)) &&
1391 (l->peer_session != ANY_SESSION))
1399 /* Complete own link name with peer's interface name */
1400 if_name = strrchr(l->name, ':') + 1;
1401 if (sizeof(l->name) - (if_name - l->name) <= TIPC_MAX_IF_NAME)
1403 if (msg_data_sz(hdr) < TIPC_MAX_IF_NAME)
1405 strncpy(if_name, msg_data(hdr), TIPC_MAX_IF_NAME);
1407 /* Update own tolerance if peer indicates a non-zero value */
1408 if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL))
1409 l->tolerance = peers_tol;
1411 /* Update own priority if peer's priority is higher */
1412 if (in_range(peers_prio, l->priority + 1, TIPC_MAX_LINK_PRI))
1413 l->priority = peers_prio;
1415 /* ACTIVATE_MSG serves as PEER_RESET if link is already down */
1416 if (msg_peer_stopping(hdr))
1417 rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1418 else if ((mtyp == RESET_MSG) || !link_is_up(l))
1419 rc = tipc_link_fsm_evt(l, LINK_PEER_RESET_EVT);
1421 /* ACTIVATE_MSG takes up link if it was already locally reset */
1422 if ((mtyp == ACTIVATE_MSG) && (l->state == LINK_ESTABLISHING))
1423 rc = TIPC_LINK_UP_EVT;
1425 l->peer_session = msg_session(hdr);
1426 l->peer_bearer_id = msg_bearer_id(hdr);
1427 if (l->mtu > msg_max_pkt(hdr))
1428 l->mtu = msg_max_pkt(hdr);
1433 /* Update own tolerance if peer indicates a non-zero value */
1434 if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL))
1435 l->tolerance = peers_tol;
1437 if (peers_prio && in_range(peers_prio, TIPC_MIN_LINK_PRI,
1438 TIPC_MAX_LINK_PRI)) {
1439 l->priority = peers_prio;
1440 rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1443 l->silent_intv_cnt = 0;
1444 l->stats.recv_states++;
1446 l->stats.recv_probes++;
1448 if (!link_is_up(l)) {
1449 if (l->state == LINK_ESTABLISHING)
1450 rc = TIPC_LINK_UP_EVT;
1454 /* Send NACK if peer has sent pkts we haven't received yet */
1455 if (more(peers_snd_nxt, rcv_nxt) && !tipc_link_is_synching(l))
1456 rcvgap = peers_snd_nxt - l->rcv_nxt;
1457 if (rcvgap || (msg_probe(hdr)))
1458 tipc_link_build_proto_msg(l, STATE_MSG, 0, rcvgap,
1460 tipc_link_release_pkts(l, ack);
1462 /* If NACK, retransmit will now start at right position */
1464 rc = tipc_link_retrans(l, ack + 1, ack + gap, xmitq);
1465 l->stats.recv_nacks++;
1468 tipc_link_advance_backlog(l, xmitq);
1469 if (unlikely(!skb_queue_empty(&l->wakeupq)))
1470 link_prepare_wakeup(l);
1477 /* tipc_link_build_bc_proto_msg() - create broadcast protocol message
1479 static bool tipc_link_build_bc_proto_msg(struct tipc_link *l, bool bcast,
1481 struct sk_buff_head *xmitq)
1483 struct sk_buff *skb;
1484 struct tipc_msg *hdr;
1485 struct sk_buff *dfrd_skb = skb_peek(&l->deferdq);
1486 u16 ack = l->rcv_nxt - 1;
1487 u16 gap_to = peers_snd_nxt - 1;
1489 skb = tipc_msg_create(BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE,
1490 0, l->addr, tipc_own_addr(l->net), 0, 0, 0);
1494 msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1495 msg_set_bcast_ack(hdr, ack);
1496 msg_set_bcgap_after(hdr, ack);
1498 gap_to = buf_seqno(dfrd_skb) - 1;
1499 msg_set_bcgap_to(hdr, gap_to);
1500 msg_set_non_seq(hdr, bcast);
1501 __skb_queue_tail(xmitq, skb);
1505 /* tipc_link_build_bc_init_msg() - synchronize broadcast link endpoints.
1507 * Give a newly added peer node the sequence number where it should
1508 * start receiving and acking broadcast packets.
1510 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
1511 struct sk_buff_head *xmitq)
1513 struct sk_buff_head list;
1515 __skb_queue_head_init(&list);
1516 if (!tipc_link_build_bc_proto_msg(l->bc_rcvlink, false, 0, &list))
1518 tipc_link_xmit(l, &list, xmitq);
1521 /* tipc_link_bc_init_rcv - receive initial broadcast synch data from peer
1523 void tipc_link_bc_init_rcv(struct tipc_link *l, struct tipc_msg *hdr)
1525 int mtyp = msg_type(hdr);
1526 u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
1531 if (msg_user(hdr) == BCAST_PROTOCOL) {
1532 l->rcv_nxt = peers_snd_nxt;
1533 l->state = LINK_ESTABLISHED;
1537 if (l->peer_caps & TIPC_BCAST_SYNCH)
1540 if (msg_peer_node_is_up(hdr))
1543 /* Compatibility: accept older, less safe initial synch data */
1544 if ((mtyp == RESET_MSG) || (mtyp == ACTIVATE_MSG))
1545 l->rcv_nxt = peers_snd_nxt;
1548 /* tipc_link_bc_sync_rcv - update rcv link according to peer's send state
1550 void tipc_link_bc_sync_rcv(struct tipc_link *l, struct tipc_msg *hdr,
1551 struct sk_buff_head *xmitq)
1553 u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
1558 if (!msg_peer_node_is_up(hdr))
1561 l->bc_peer_is_up = true;
1563 /* Ignore if peers_snd_nxt goes beyond receive window */
1564 if (more(peers_snd_nxt, l->rcv_nxt + l->window))
1567 if (!more(peers_snd_nxt, l->rcv_nxt)) {
1568 l->nack_state = BC_NACK_SND_CONDITIONAL;
1572 /* Don't NACK if one was recently sent or peeked */
1573 if (l->nack_state == BC_NACK_SND_SUPPRESS) {
1574 l->nack_state = BC_NACK_SND_UNCONDITIONAL;
1578 /* Conditionally delay NACK sending until next synch rcv */
1579 if (l->nack_state == BC_NACK_SND_CONDITIONAL) {
1580 l->nack_state = BC_NACK_SND_UNCONDITIONAL;
1581 if ((peers_snd_nxt - l->rcv_nxt) < TIPC_MIN_LINK_WIN)
1585 /* Send NACK now but suppress next one */
1586 tipc_link_build_bc_proto_msg(l, true, peers_snd_nxt, xmitq);
1587 l->nack_state = BC_NACK_SND_SUPPRESS;
1590 void tipc_link_bc_ack_rcv(struct tipc_link *l, u16 acked,
1591 struct sk_buff_head *xmitq)
1593 struct sk_buff *skb, *tmp;
1594 struct tipc_link *snd_l = l->bc_sndlink;
1596 if (!link_is_up(l) || !l->bc_peer_is_up)
1599 if (!more(acked, l->acked))
1602 /* Skip over packets peer has already acked */
1603 skb_queue_walk(&snd_l->transmq, skb) {
1604 if (more(buf_seqno(skb), l->acked))
1608 /* Update/release the packets peer is acking now */
1609 skb_queue_walk_from_safe(&snd_l->transmq, skb, tmp) {
1610 if (more(buf_seqno(skb), acked))
1612 if (!--TIPC_SKB_CB(skb)->ackers) {
1613 __skb_unlink(skb, &snd_l->transmq);
1618 tipc_link_advance_backlog(snd_l, xmitq);
1619 if (unlikely(!skb_queue_empty(&snd_l->wakeupq)))
1620 link_prepare_wakeup(snd_l);
1623 /* tipc_link_bc_nack_rcv(): receive broadcast nack message
1625 int tipc_link_bc_nack_rcv(struct tipc_link *l, struct sk_buff *skb,
1626 struct sk_buff_head *xmitq)
1628 struct tipc_msg *hdr = buf_msg(skb);
1629 u32 dnode = msg_destnode(hdr);
1630 int mtyp = msg_type(hdr);
1631 u16 acked = msg_bcast_ack(hdr);
1632 u16 from = acked + 1;
1633 u16 to = msg_bcgap_to(hdr);
1634 u16 peers_snd_nxt = to + 1;
1639 if (!tipc_link_is_up(l) || !l->bc_peer_is_up)
1642 if (mtyp != STATE_MSG)
1645 if (dnode == tipc_own_addr(l->net)) {
1646 tipc_link_bc_ack_rcv(l, acked, xmitq);
1647 rc = tipc_link_retrans(l->bc_sndlink, from, to, xmitq);
1648 l->stats.recv_nacks++;
1652 /* Msg for other node => suppress own NACK at next sync if applicable */
1653 if (more(peers_snd_nxt, l->rcv_nxt) && !less(l->rcv_nxt, from))
1654 l->nack_state = BC_NACK_SND_SUPPRESS;
1659 void tipc_link_set_queue_limits(struct tipc_link *l, u32 win)
1661 int max_bulk = TIPC_MAX_PUBLICATIONS / (l->mtu / ITEM_SIZE);
1664 l->backlog[TIPC_LOW_IMPORTANCE].limit = win / 2;
1665 l->backlog[TIPC_MEDIUM_IMPORTANCE].limit = win;
1666 l->backlog[TIPC_HIGH_IMPORTANCE].limit = win / 2 * 3;
1667 l->backlog[TIPC_CRITICAL_IMPORTANCE].limit = win * 2;
1668 l->backlog[TIPC_SYSTEM_IMPORTANCE].limit = max_bulk;
1672 * link_reset_stats - reset link statistics
1673 * @l: pointer to link
1675 void tipc_link_reset_stats(struct tipc_link *l)
1677 memset(&l->stats, 0, sizeof(l->stats));
1678 if (!link_is_bc_sndlink(l)) {
1679 l->stats.sent_info = l->snd_nxt;
1680 l->stats.recv_info = l->rcv_nxt;
1684 static void link_print(struct tipc_link *l, const char *str)
1686 struct sk_buff *hskb = skb_peek(&l->transmq);
1687 u16 head = hskb ? msg_seqno(buf_msg(hskb)) : l->snd_nxt - 1;
1688 u16 tail = l->snd_nxt - 1;
1690 pr_info("%s Link <%s> state %x\n", str, l->name, l->state);
1691 pr_info("XMTQ: %u [%u-%u], BKLGQ: %u, SNDNX: %u, RCVNX: %u\n",
1692 skb_queue_len(&l->transmq), head, tail,
1693 skb_queue_len(&l->backlogq), l->snd_nxt, l->rcv_nxt);
1696 /* Parse and validate nested (link) properties valid for media, bearer and link
1698 int tipc_nl_parse_link_prop(struct nlattr *prop, struct nlattr *props[])
1702 err = nla_parse_nested(props, TIPC_NLA_PROP_MAX, prop,
1703 tipc_nl_prop_policy);
1707 if (props[TIPC_NLA_PROP_PRIO]) {
1710 prio = nla_get_u32(props[TIPC_NLA_PROP_PRIO]);
1711 if (prio > TIPC_MAX_LINK_PRI)
1715 if (props[TIPC_NLA_PROP_TOL]) {
1718 tol = nla_get_u32(props[TIPC_NLA_PROP_TOL]);
1719 if ((tol < TIPC_MIN_LINK_TOL) || (tol > TIPC_MAX_LINK_TOL))
1723 if (props[TIPC_NLA_PROP_WIN]) {
1726 win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
1727 if ((win < TIPC_MIN_LINK_WIN) || (win > TIPC_MAX_LINK_WIN))
1734 static int __tipc_nl_add_stats(struct sk_buff *skb, struct tipc_stats *s)
1737 struct nlattr *stats;
1744 struct nla_map map[] = {
1745 {TIPC_NLA_STATS_RX_INFO, s->recv_info},
1746 {TIPC_NLA_STATS_RX_FRAGMENTS, s->recv_fragments},
1747 {TIPC_NLA_STATS_RX_FRAGMENTED, s->recv_fragmented},
1748 {TIPC_NLA_STATS_RX_BUNDLES, s->recv_bundles},
1749 {TIPC_NLA_STATS_RX_BUNDLED, s->recv_bundled},
1750 {TIPC_NLA_STATS_TX_INFO, s->sent_info},
1751 {TIPC_NLA_STATS_TX_FRAGMENTS, s->sent_fragments},
1752 {TIPC_NLA_STATS_TX_FRAGMENTED, s->sent_fragmented},
1753 {TIPC_NLA_STATS_TX_BUNDLES, s->sent_bundles},
1754 {TIPC_NLA_STATS_TX_BUNDLED, s->sent_bundled},
1755 {TIPC_NLA_STATS_MSG_PROF_TOT, (s->msg_length_counts) ?
1756 s->msg_length_counts : 1},
1757 {TIPC_NLA_STATS_MSG_LEN_CNT, s->msg_length_counts},
1758 {TIPC_NLA_STATS_MSG_LEN_TOT, s->msg_lengths_total},
1759 {TIPC_NLA_STATS_MSG_LEN_P0, s->msg_length_profile[0]},
1760 {TIPC_NLA_STATS_MSG_LEN_P1, s->msg_length_profile[1]},
1761 {TIPC_NLA_STATS_MSG_LEN_P2, s->msg_length_profile[2]},
1762 {TIPC_NLA_STATS_MSG_LEN_P3, s->msg_length_profile[3]},
1763 {TIPC_NLA_STATS_MSG_LEN_P4, s->msg_length_profile[4]},
1764 {TIPC_NLA_STATS_MSG_LEN_P5, s->msg_length_profile[5]},
1765 {TIPC_NLA_STATS_MSG_LEN_P6, s->msg_length_profile[6]},
1766 {TIPC_NLA_STATS_RX_STATES, s->recv_states},
1767 {TIPC_NLA_STATS_RX_PROBES, s->recv_probes},
1768 {TIPC_NLA_STATS_RX_NACKS, s->recv_nacks},
1769 {TIPC_NLA_STATS_RX_DEFERRED, s->deferred_recv},
1770 {TIPC_NLA_STATS_TX_STATES, s->sent_states},
1771 {TIPC_NLA_STATS_TX_PROBES, s->sent_probes},
1772 {TIPC_NLA_STATS_TX_NACKS, s->sent_nacks},
1773 {TIPC_NLA_STATS_TX_ACKS, s->sent_acks},
1774 {TIPC_NLA_STATS_RETRANSMITTED, s->retransmitted},
1775 {TIPC_NLA_STATS_DUPLICATES, s->duplicates},
1776 {TIPC_NLA_STATS_LINK_CONGS, s->link_congs},
1777 {TIPC_NLA_STATS_MAX_QUEUE, s->max_queue_sz},
1778 {TIPC_NLA_STATS_AVG_QUEUE, s->queue_sz_counts ?
1779 (s->accu_queue_sz / s->queue_sz_counts) : 0}
1782 stats = nla_nest_start(skb, TIPC_NLA_LINK_STATS);
1786 for (i = 0; i < ARRAY_SIZE(map); i++)
1787 if (nla_put_u32(skb, map[i].key, map[i].val))
1790 nla_nest_end(skb, stats);
1794 nla_nest_cancel(skb, stats);
1799 /* Caller should hold appropriate locks to protect the link */
1800 int __tipc_nl_add_link(struct net *net, struct tipc_nl_msg *msg,
1801 struct tipc_link *link, int nlflags)
1805 struct nlattr *attrs;
1806 struct nlattr *prop;
1807 struct tipc_net *tn = net_generic(net, tipc_net_id);
1809 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
1810 nlflags, TIPC_NL_LINK_GET);
1814 attrs = nla_nest_start(msg->skb, TIPC_NLA_LINK);
1818 if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, link->name))
1820 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_DEST,
1821 tipc_cluster_mask(tn->own_addr)))
1823 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_MTU, link->mtu))
1825 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, link->rcv_nxt))
1827 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, link->snd_nxt))
1830 if (tipc_link_is_up(link))
1831 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
1834 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_ACTIVE))
1837 prop = nla_nest_start(msg->skb, TIPC_NLA_LINK_PROP);
1840 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
1842 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_TOL, link->tolerance))
1844 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN,
1847 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
1849 nla_nest_end(msg->skb, prop);
1851 err = __tipc_nl_add_stats(msg->skb, &link->stats);
1855 nla_nest_end(msg->skb, attrs);
1856 genlmsg_end(msg->skb, hdr);
1861 nla_nest_cancel(msg->skb, prop);
1863 nla_nest_cancel(msg->skb, attrs);
1865 genlmsg_cancel(msg->skb, hdr);
1870 static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
1871 struct tipc_stats *stats)
1874 struct nlattr *nest;
1881 struct nla_map map[] = {
1882 {TIPC_NLA_STATS_RX_INFO, stats->recv_info},
1883 {TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
1884 {TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
1885 {TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
1886 {TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
1887 {TIPC_NLA_STATS_TX_INFO, stats->sent_info},
1888 {TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
1889 {TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
1890 {TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
1891 {TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
1892 {TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
1893 {TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
1894 {TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
1895 {TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
1896 {TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
1897 {TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
1898 {TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
1899 {TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
1900 {TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
1901 (stats->accu_queue_sz / stats->queue_sz_counts) : 0}
1904 nest = nla_nest_start(skb, TIPC_NLA_LINK_STATS);
1908 for (i = 0; i < ARRAY_SIZE(map); i++)
1909 if (nla_put_u32(skb, map[i].key, map[i].val))
1912 nla_nest_end(skb, nest);
1916 nla_nest_cancel(skb, nest);
1921 int tipc_nl_add_bc_link(struct net *net, struct tipc_nl_msg *msg)
1925 struct nlattr *attrs;
1926 struct nlattr *prop;
1927 struct tipc_net *tn = net_generic(net, tipc_net_id);
1928 struct tipc_link *bcl = tn->bcl;
1933 tipc_bcast_lock(net);
1935 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
1936 NLM_F_MULTI, TIPC_NL_LINK_GET);
1938 tipc_bcast_unlock(net);
1942 attrs = nla_nest_start(msg->skb, TIPC_NLA_LINK);
1946 /* The broadcast link is always up */
1947 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
1950 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
1952 if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
1954 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, bcl->rcv_nxt))
1956 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, bcl->snd_nxt))
1959 prop = nla_nest_start(msg->skb, TIPC_NLA_LINK_PROP);
1962 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->window))
1964 nla_nest_end(msg->skb, prop);
1966 err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
1970 tipc_bcast_unlock(net);
1971 nla_nest_end(msg->skb, attrs);
1972 genlmsg_end(msg->skb, hdr);
1977 nla_nest_cancel(msg->skb, prop);
1979 nla_nest_cancel(msg->skb, attrs);
1981 tipc_bcast_unlock(net);
1982 genlmsg_cancel(msg->skb, hdr);
1987 void tipc_link_set_tolerance(struct tipc_link *l, u32 tol,
1988 struct sk_buff_head *xmitq)
1991 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, tol, 0, xmitq);
1994 void tipc_link_set_prio(struct tipc_link *l, u32 prio,
1995 struct sk_buff_head *xmitq)
1998 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, prio, xmitq);
2001 void tipc_link_set_abort_limit(struct tipc_link *l, u32 limit)
2003 l->abort_limit = limit;