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"
48 #include <linux/pkt_sched.h>
69 u32 link_congs; /* # port sends blocked by congestion */
72 u32 max_queue_sz; /* send queue size high water mark */
73 u32 accu_queue_sz; /* used for send queue size profiling */
74 u32 queue_sz_counts; /* used for send queue size profiling */
75 u32 msg_length_counts; /* used for message length profiling */
76 u32 msg_lengths_total; /* used for message length profiling */
77 u32 msg_length_profile[7]; /* used for msg. length profiling */
81 * struct tipc_link - TIPC link data structure
82 * @addr: network address of link's peer node
83 * @name: link name character string
84 * @media_addr: media address to use when sending messages over link
86 * @net: pointer to namespace struct
87 * @refcnt: reference counter for permanent references (owner node & timer)
88 * @peer_session: link session # being used by peer end of link
89 * @peer_bearer_id: bearer id used by link's peer endpoint
90 * @bearer_id: local bearer id used by link
91 * @tolerance: minimum link continuity loss needed to reset link [in ms]
92 * @abort_limit: # of unacknowledged continuity probes needed to reset link
93 * @state: current state of link FSM
94 * @peer_caps: bitmap describing capabilities of peer node
95 * @silent_intv_cnt: # of timer intervals without any reception from peer
96 * @proto_msg: template for control messages generated by link
97 * @pmsg: convenience pointer to "proto_msg" field
98 * @priority: current link priority
99 * @net_plane: current link network plane ('A' through 'H')
100 * @mon_state: cookie with information needed by link monitor
101 * @backlog_limit: backlog queue congestion thresholds (indexed by importance)
102 * @exp_msg_count: # of tunnelled messages expected during link changeover
103 * @reset_rcv_checkpt: seq # of last acknowledged message at time of link reset
104 * @mtu: current maximum packet size for this link
105 * @advertised_mtu: advertised own mtu when link is being established
106 * @transmitq: queue for sent, non-acked messages
107 * @backlogq: queue for messages waiting to be sent
108 * @snt_nxt: next sequence number to use for outbound messages
109 * @prev_from: sequence number of most previous retransmission request
110 * @stale_cnt: counter for number of identical retransmit attempts
111 * @stale_limit: time when repeated identical retransmits must force link reset
112 * @ackers: # of peers that needs to ack each packet before it can be released
113 * @acked: # last packet acked by a certain peer. Used for broadcast.
114 * @rcv_nxt: next sequence number to expect for inbound messages
115 * @deferred_queue: deferred queue saved OOS b'cast message received from node
116 * @unacked_window: # of inbound messages rx'd without ack'ing back to peer
117 * @inputq: buffer queue for messages to be delivered upwards
118 * @namedq: buffer queue for name table messages to be delivered upwards
119 * @next_out: ptr to first unsent outbound message in queue
120 * @wakeupq: linked list of wakeup msgs waiting for link congestion to abate
121 * @long_msg_seq_no: next identifier to use for outbound fragmented messages
122 * @reasm_buf: head of partially reassembled inbound message fragments
123 * @bc_rcvr: marks that this is a broadcast receiver link
124 * @stats: collects statistics regarding link activity
128 char name[TIPC_MAX_LINK_NAME];
131 /* Management and link supervision data */
145 char if_name[TIPC_MAX_IF_NAME];
148 struct tipc_mon_state mon_state;
153 struct sk_buff *failover_reasm_skb;
154 struct sk_buff_head failover_deferdq;
156 /* Max packet negotiation */
161 struct sk_buff_head transmq;
162 struct sk_buff_head backlogq;
171 unsigned long stale_limit;
176 struct sk_buff_head deferdq;
177 struct sk_buff_head *inputq;
178 struct sk_buff_head *namedq;
180 /* Congestion handling */
181 struct sk_buff_head wakeupq;
183 /* Fragmentation/reassembly */
184 struct sk_buff *reasm_buf;
189 struct tipc_link *bc_rcvlink;
190 struct tipc_link *bc_sndlink;
195 struct tipc_stats stats;
199 * Error message prefixes
201 static const char *link_co_err = "Link tunneling error, ";
202 static const char *link_rst_msg = "Resetting link ";
204 /* Send states for broadcast NACKs
207 BC_NACK_SND_CONDITIONAL,
208 BC_NACK_SND_UNCONDITIONAL,
209 BC_NACK_SND_SUPPRESS,
212 #define TIPC_BC_RETR_LIM msecs_to_jiffies(10) /* [ms] */
213 #define TIPC_UC_RETR_TIME (jiffies + msecs_to_jiffies(1))
216 * Interval between NACKs when packets arrive out of order
218 #define TIPC_NACK_INTV (TIPC_MIN_LINK_WIN * 2)
223 LINK_ESTABLISHED = 0xe,
224 LINK_ESTABLISHING = 0xe << 4,
225 LINK_RESET = 0x1 << 8,
226 LINK_RESETTING = 0x2 << 12,
227 LINK_PEER_RESET = 0xd << 16,
228 LINK_FAILINGOVER = 0xf << 20,
229 LINK_SYNCHING = 0xc << 24
232 /* Link FSM state checking routines
234 static int link_is_up(struct tipc_link *l)
236 return l->state & (LINK_ESTABLISHED | LINK_SYNCHING);
239 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
240 struct sk_buff_head *xmitq);
241 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
242 bool probe_reply, u16 rcvgap,
243 int tolerance, int priority,
244 struct sk_buff_head *xmitq);
245 static void link_print(struct tipc_link *l, const char *str);
246 static int tipc_link_build_nack_msg(struct tipc_link *l,
247 struct sk_buff_head *xmitq);
248 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
249 struct sk_buff_head *xmitq);
250 static bool tipc_link_release_pkts(struct tipc_link *l, u16 to);
251 static u16 tipc_build_gap_ack_blks(struct tipc_link *l, void *data);
252 static void tipc_link_advance_transmq(struct tipc_link *l, u16 acked, u16 gap,
253 struct tipc_gap_ack_blks *ga,
254 struct sk_buff_head *xmitq);
257 * Simple non-static link routines (i.e. referenced outside this file)
259 bool tipc_link_is_up(struct tipc_link *l)
261 return link_is_up(l);
264 bool tipc_link_peer_is_down(struct tipc_link *l)
266 return l->state == LINK_PEER_RESET;
269 bool tipc_link_is_reset(struct tipc_link *l)
271 return l->state & (LINK_RESET | LINK_FAILINGOVER | LINK_ESTABLISHING);
274 bool tipc_link_is_establishing(struct tipc_link *l)
276 return l->state == LINK_ESTABLISHING;
279 bool tipc_link_is_synching(struct tipc_link *l)
281 return l->state == LINK_SYNCHING;
284 bool tipc_link_is_failingover(struct tipc_link *l)
286 return l->state == LINK_FAILINGOVER;
289 bool tipc_link_is_blocked(struct tipc_link *l)
291 return l->state & (LINK_RESETTING | LINK_PEER_RESET | LINK_FAILINGOVER);
294 static bool link_is_bc_sndlink(struct tipc_link *l)
296 return !l->bc_sndlink;
299 static bool link_is_bc_rcvlink(struct tipc_link *l)
301 return ((l->bc_rcvlink == l) && !link_is_bc_sndlink(l));
304 void tipc_link_set_active(struct tipc_link *l, bool active)
309 u32 tipc_link_id(struct tipc_link *l)
311 return l->peer_bearer_id << 16 | l->bearer_id;
314 int tipc_link_window(struct tipc_link *l)
319 int tipc_link_prio(struct tipc_link *l)
324 unsigned long tipc_link_tolerance(struct tipc_link *l)
329 struct sk_buff_head *tipc_link_inputq(struct tipc_link *l)
334 char tipc_link_plane(struct tipc_link *l)
339 void tipc_link_update_caps(struct tipc_link *l, u16 capabilities)
341 l->peer_caps = capabilities;
344 void tipc_link_add_bc_peer(struct tipc_link *snd_l,
345 struct tipc_link *uc_l,
346 struct sk_buff_head *xmitq)
348 struct tipc_link *rcv_l = uc_l->bc_rcvlink;
351 rcv_l->acked = snd_l->snd_nxt - 1;
352 snd_l->state = LINK_ESTABLISHED;
353 tipc_link_build_bc_init_msg(uc_l, xmitq);
356 void tipc_link_remove_bc_peer(struct tipc_link *snd_l,
357 struct tipc_link *rcv_l,
358 struct sk_buff_head *xmitq)
360 u16 ack = snd_l->snd_nxt - 1;
363 rcv_l->bc_peer_is_up = true;
364 rcv_l->state = LINK_ESTABLISHED;
365 tipc_link_bc_ack_rcv(rcv_l, ack, xmitq);
366 trace_tipc_link_reset(rcv_l, TIPC_DUMP_ALL, "bclink removed!");
367 tipc_link_reset(rcv_l);
368 rcv_l->state = LINK_RESET;
369 if (!snd_l->ackers) {
370 trace_tipc_link_reset(snd_l, TIPC_DUMP_ALL, "zero ackers!");
371 tipc_link_reset(snd_l);
372 snd_l->state = LINK_RESET;
373 __skb_queue_purge(xmitq);
377 int tipc_link_bc_peers(struct tipc_link *l)
382 static u16 link_bc_rcv_gap(struct tipc_link *l)
384 struct sk_buff *skb = skb_peek(&l->deferdq);
387 if (more(l->snd_nxt, l->rcv_nxt))
388 gap = l->snd_nxt - l->rcv_nxt;
390 gap = buf_seqno(skb) - l->rcv_nxt;
394 void tipc_link_set_mtu(struct tipc_link *l, int mtu)
399 int tipc_link_mtu(struct tipc_link *l)
404 u16 tipc_link_rcv_nxt(struct tipc_link *l)
409 u16 tipc_link_acked(struct tipc_link *l)
414 char *tipc_link_name(struct tipc_link *l)
419 u32 tipc_link_state(struct tipc_link *l)
425 * tipc_link_create - create a new link
426 * @n: pointer to associated node
427 * @if_name: associated interface name
428 * @bearer_id: id (index) of associated bearer
429 * @tolerance: link tolerance to be used by link
430 * @net_plane: network plane (A,B,c..) this link belongs to
431 * @mtu: mtu to be advertised by link
432 * @priority: priority to be used by link
433 * @window: send window to be used by link
434 * @session: session to be used by link
435 * @ownnode: identity of own node
436 * @peer: node id of peer node
437 * @peer_caps: bitmap describing peer node capabilities
438 * @bc_sndlink: the namespace global link used for broadcast sending
439 * @bc_rcvlink: the peer specific link used for broadcast reception
440 * @inputq: queue to put messages ready for delivery
441 * @namedq: queue to put binding table update messages ready for delivery
442 * @link: return value, pointer to put the created link
444 * Returns true if link was created, otherwise false
446 bool tipc_link_create(struct net *net, char *if_name, int bearer_id,
447 int tolerance, char net_plane, u32 mtu, int priority,
448 int window, u32 session, u32 self,
449 u32 peer, u8 *peer_id, u16 peer_caps,
450 struct tipc_link *bc_sndlink,
451 struct tipc_link *bc_rcvlink,
452 struct sk_buff_head *inputq,
453 struct sk_buff_head *namedq,
454 struct tipc_link **link)
456 char peer_str[NODE_ID_STR_LEN] = {0,};
457 char self_str[NODE_ID_STR_LEN] = {0,};
460 l = kzalloc(sizeof(*l), GFP_ATOMIC);
464 l->session = session;
466 /* Set link name for unicast links only */
468 tipc_nodeid2string(self_str, tipc_own_id(net));
469 if (strlen(self_str) > 16)
470 sprintf(self_str, "%x", self);
471 tipc_nodeid2string(peer_str, peer_id);
472 if (strlen(peer_str) > 16)
473 sprintf(peer_str, "%x", peer);
475 /* Peer i/f name will be completed by reset/activate message */
476 snprintf(l->name, sizeof(l->name), "%s:%s-%s:unknown",
477 self_str, if_name, peer_str);
479 strcpy(l->if_name, if_name);
481 l->peer_caps = peer_caps;
483 l->in_session = false;
484 l->bearer_id = bearer_id;
485 l->tolerance = tolerance;
487 bc_rcvlink->tolerance = tolerance;
488 l->net_plane = net_plane;
489 l->advertised_mtu = mtu;
491 l->priority = priority;
492 tipc_link_set_queue_limits(l, window);
494 l->bc_sndlink = bc_sndlink;
495 l->bc_rcvlink = bc_rcvlink;
498 l->state = LINK_RESETTING;
499 __skb_queue_head_init(&l->transmq);
500 __skb_queue_head_init(&l->backlogq);
501 __skb_queue_head_init(&l->deferdq);
502 __skb_queue_head_init(&l->failover_deferdq);
503 skb_queue_head_init(&l->wakeupq);
504 skb_queue_head_init(l->inputq);
509 * tipc_link_bc_create - create new link to be used for broadcast
510 * @n: pointer to associated node
511 * @mtu: mtu to be used initially if no peers
512 * @window: send window to be used
513 * @inputq: queue to put messages ready for delivery
514 * @namedq: queue to put binding table update messages ready for delivery
515 * @link: return value, pointer to put the created link
517 * Returns true if link was created, otherwise false
519 bool tipc_link_bc_create(struct net *net, u32 ownnode, u32 peer,
520 int mtu, int window, u16 peer_caps,
521 struct sk_buff_head *inputq,
522 struct sk_buff_head *namedq,
523 struct tipc_link *bc_sndlink,
524 struct tipc_link **link)
528 if (!tipc_link_create(net, "", MAX_BEARERS, 0, 'Z', mtu, 0, window,
529 0, ownnode, peer, NULL, peer_caps, bc_sndlink,
530 NULL, inputq, namedq, link))
534 strcpy(l->name, tipc_bclink_name);
535 trace_tipc_link_reset(l, TIPC_DUMP_ALL, "bclink created!");
537 l->state = LINK_RESET;
541 /* Broadcast send link is always up */
542 if (link_is_bc_sndlink(l))
543 l->state = LINK_ESTABLISHED;
545 /* Disable replicast if even a single peer doesn't support it */
546 if (link_is_bc_rcvlink(l) && !(peer_caps & TIPC_BCAST_RCAST))
547 tipc_bcast_disable_rcast(net);
553 * tipc_link_fsm_evt - link finite state machine
554 * @l: pointer to link
555 * @evt: state machine event to be processed
557 int tipc_link_fsm_evt(struct tipc_link *l, int evt)
560 int old_state = l->state;
565 case LINK_PEER_RESET_EVT:
566 l->state = LINK_PEER_RESET;
569 l->state = LINK_RESET;
571 case LINK_FAILURE_EVT:
572 case LINK_FAILOVER_BEGIN_EVT:
573 case LINK_ESTABLISH_EVT:
574 case LINK_FAILOVER_END_EVT:
575 case LINK_SYNCH_BEGIN_EVT:
576 case LINK_SYNCH_END_EVT:
583 case LINK_PEER_RESET_EVT:
584 l->state = LINK_ESTABLISHING;
586 case LINK_FAILOVER_BEGIN_EVT:
587 l->state = LINK_FAILINGOVER;
588 case LINK_FAILURE_EVT:
590 case LINK_ESTABLISH_EVT:
591 case LINK_FAILOVER_END_EVT:
593 case LINK_SYNCH_BEGIN_EVT:
594 case LINK_SYNCH_END_EVT:
599 case LINK_PEER_RESET:
602 l->state = LINK_ESTABLISHING;
604 case LINK_PEER_RESET_EVT:
605 case LINK_ESTABLISH_EVT:
606 case LINK_FAILURE_EVT:
608 case LINK_SYNCH_BEGIN_EVT:
609 case LINK_SYNCH_END_EVT:
610 case LINK_FAILOVER_BEGIN_EVT:
611 case LINK_FAILOVER_END_EVT:
616 case LINK_FAILINGOVER:
618 case LINK_FAILOVER_END_EVT:
619 l->state = LINK_RESET;
621 case LINK_PEER_RESET_EVT:
623 case LINK_ESTABLISH_EVT:
624 case LINK_FAILURE_EVT:
626 case LINK_FAILOVER_BEGIN_EVT:
627 case LINK_SYNCH_BEGIN_EVT:
628 case LINK_SYNCH_END_EVT:
633 case LINK_ESTABLISHING:
635 case LINK_ESTABLISH_EVT:
636 l->state = LINK_ESTABLISHED;
638 case LINK_FAILOVER_BEGIN_EVT:
639 l->state = LINK_FAILINGOVER;
642 l->state = LINK_RESET;
644 case LINK_FAILURE_EVT:
645 case LINK_PEER_RESET_EVT:
646 case LINK_SYNCH_BEGIN_EVT:
647 case LINK_FAILOVER_END_EVT:
649 case LINK_SYNCH_END_EVT:
654 case LINK_ESTABLISHED:
656 case LINK_PEER_RESET_EVT:
657 l->state = LINK_PEER_RESET;
658 rc |= TIPC_LINK_DOWN_EVT;
660 case LINK_FAILURE_EVT:
661 l->state = LINK_RESETTING;
662 rc |= TIPC_LINK_DOWN_EVT;
665 l->state = LINK_RESET;
667 case LINK_ESTABLISH_EVT:
668 case LINK_SYNCH_END_EVT:
670 case LINK_SYNCH_BEGIN_EVT:
671 l->state = LINK_SYNCHING;
673 case LINK_FAILOVER_BEGIN_EVT:
674 case LINK_FAILOVER_END_EVT:
681 case LINK_PEER_RESET_EVT:
682 l->state = LINK_PEER_RESET;
683 rc |= TIPC_LINK_DOWN_EVT;
685 case LINK_FAILURE_EVT:
686 l->state = LINK_RESETTING;
687 rc |= TIPC_LINK_DOWN_EVT;
690 l->state = LINK_RESET;
692 case LINK_ESTABLISH_EVT:
693 case LINK_SYNCH_BEGIN_EVT:
695 case LINK_SYNCH_END_EVT:
696 l->state = LINK_ESTABLISHED;
698 case LINK_FAILOVER_BEGIN_EVT:
699 case LINK_FAILOVER_END_EVT:
705 pr_err("Unknown FSM state %x in %s\n", l->state, l->name);
707 trace_tipc_link_fsm(l->name, old_state, l->state, evt);
710 pr_err("Illegal FSM event %x in state %x on link %s\n",
711 evt, l->state, l->name);
712 trace_tipc_link_fsm(l->name, old_state, l->state, evt);
716 /* link_profile_stats - update statistical profiling of traffic
718 static void link_profile_stats(struct tipc_link *l)
721 struct tipc_msg *msg;
724 /* Update counters used in statistical profiling of send traffic */
725 l->stats.accu_queue_sz += skb_queue_len(&l->transmq);
726 l->stats.queue_sz_counts++;
728 skb = skb_peek(&l->transmq);
732 length = msg_size(msg);
734 if (msg_user(msg) == MSG_FRAGMENTER) {
735 if (msg_type(msg) != FIRST_FRAGMENT)
737 length = msg_size(msg_get_wrapped(msg));
739 l->stats.msg_lengths_total += length;
740 l->stats.msg_length_counts++;
742 l->stats.msg_length_profile[0]++;
743 else if (length <= 256)
744 l->stats.msg_length_profile[1]++;
745 else if (length <= 1024)
746 l->stats.msg_length_profile[2]++;
747 else if (length <= 4096)
748 l->stats.msg_length_profile[3]++;
749 else if (length <= 16384)
750 l->stats.msg_length_profile[4]++;
751 else if (length <= 32768)
752 l->stats.msg_length_profile[5]++;
754 l->stats.msg_length_profile[6]++;
758 * tipc_link_too_silent - check if link is "too silent"
759 * @l: tipc link to be checked
761 * Returns true if the link 'silent_intv_cnt' is about to reach the
762 * 'abort_limit' value, otherwise false
764 bool tipc_link_too_silent(struct tipc_link *l)
766 return (l->silent_intv_cnt + 2 > l->abort_limit);
769 /* tipc_link_timeout - perform periodic task as instructed from node timeout
771 int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq)
778 u16 bc_snt = l->bc_sndlink->snd_nxt - 1;
779 u16 bc_acked = l->bc_rcvlink->acked;
780 struct tipc_mon_state *mstate = &l->mon_state;
782 trace_tipc_link_timeout(l, TIPC_DUMP_NONE, " ");
783 trace_tipc_link_too_silent(l, TIPC_DUMP_ALL, " ");
785 case LINK_ESTABLISHED:
788 link_profile_stats(l);
789 tipc_mon_get_state(l->net, l->addr, mstate, l->bearer_id);
790 if (mstate->reset || (l->silent_intv_cnt > l->abort_limit))
791 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
792 state = bc_acked != bc_snt;
793 state |= l->bc_rcvlink->rcv_unacked;
794 state |= l->rcv_unacked;
795 state |= !skb_queue_empty(&l->transmq);
796 state |= !skb_queue_empty(&l->deferdq);
797 probe = mstate->probing;
798 probe |= l->silent_intv_cnt;
799 if (probe || mstate->monitoring)
800 l->silent_intv_cnt++;
803 setup = l->rst_cnt++ <= 4;
804 setup |= !(l->rst_cnt % 16);
807 case LINK_ESTABLISHING:
811 case LINK_PEER_RESET:
813 case LINK_FAILINGOVER:
819 if (state || probe || setup)
820 tipc_link_build_proto_msg(l, mtyp, probe, 0, 0, 0, 0, xmitq);
826 * link_schedule_user - schedule a message sender for wakeup after congestion
828 * @hdr: header of message that is being sent
829 * Create pseudo msg to send back to user when congestion abates
831 static int link_schedule_user(struct tipc_link *l, struct tipc_msg *hdr)
833 u32 dnode = tipc_own_addr(l->net);
834 u32 dport = msg_origport(hdr);
837 /* Create and schedule wakeup pseudo message */
838 skb = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
839 dnode, l->addr, dport, 0, 0);
842 msg_set_dest_droppable(buf_msg(skb), true);
843 TIPC_SKB_CB(skb)->chain_imp = msg_importance(hdr);
844 skb_queue_tail(&l->wakeupq, skb);
845 l->stats.link_congs++;
846 trace_tipc_link_conges(l, TIPC_DUMP_ALL, "wakeup scheduled!");
851 * link_prepare_wakeup - prepare users for wakeup after congestion
853 * Wake up a number of waiting users, as permitted by available space
856 static void link_prepare_wakeup(struct tipc_link *l)
858 struct sk_buff *skb, *tmp;
861 skb_queue_walk_safe(&l->wakeupq, skb, tmp) {
862 imp = TIPC_SKB_CB(skb)->chain_imp;
863 if (l->backlog[imp].len < l->backlog[imp].limit) {
864 skb_unlink(skb, &l->wakeupq);
865 skb_queue_tail(l->inputq, skb);
866 } else if (i++ > 10) {
872 void tipc_link_reset(struct tipc_link *l)
874 struct sk_buff_head list;
876 __skb_queue_head_init(&list);
878 l->in_session = false;
879 /* Force re-synch of peer session number before establishing */
882 l->mtu = l->advertised_mtu;
884 spin_lock_bh(&l->wakeupq.lock);
885 skb_queue_splice_init(&l->wakeupq, &list);
886 spin_unlock_bh(&l->wakeupq.lock);
888 spin_lock_bh(&l->inputq->lock);
889 skb_queue_splice_init(&list, l->inputq);
890 spin_unlock_bh(&l->inputq->lock);
892 __skb_queue_purge(&l->transmq);
893 __skb_queue_purge(&l->deferdq);
894 __skb_queue_purge(&l->backlogq);
895 __skb_queue_purge(&l->failover_deferdq);
896 l->backlog[TIPC_LOW_IMPORTANCE].len = 0;
897 l->backlog[TIPC_MEDIUM_IMPORTANCE].len = 0;
898 l->backlog[TIPC_HIGH_IMPORTANCE].len = 0;
899 l->backlog[TIPC_CRITICAL_IMPORTANCE].len = 0;
900 l->backlog[TIPC_SYSTEM_IMPORTANCE].len = 0;
901 kfree_skb(l->reasm_buf);
902 kfree_skb(l->failover_reasm_skb);
904 l->failover_reasm_skb = NULL;
908 l->snd_nxt_state = 1;
909 l->rcv_nxt_state = 1;
911 l->silent_intv_cnt = 0;
914 l->bc_peer_is_up = false;
915 memset(&l->mon_state, 0, sizeof(l->mon_state));
916 tipc_link_reset_stats(l);
920 * tipc_link_xmit(): enqueue buffer list according to queue situation
922 * @list: chain of buffers containing message
923 * @xmitq: returned list of packets to be sent by caller
925 * Consumes the buffer chain.
926 * Returns 0 if success, or errno: -ELINKCONG, -EMSGSIZE or -ENOBUFS
927 * Messages at TIPC_SYSTEM_IMPORTANCE are always accepted
929 int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list,
930 struct sk_buff_head *xmitq)
932 struct tipc_msg *hdr = buf_msg(skb_peek(list));
933 unsigned int maxwin = l->window;
934 int imp = msg_importance(hdr);
935 unsigned int mtu = l->mtu;
936 u16 ack = l->rcv_nxt - 1;
937 u16 seqno = l->snd_nxt;
938 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
939 struct sk_buff_head *transmq = &l->transmq;
940 struct sk_buff_head *backlogq = &l->backlogq;
941 struct sk_buff *skb, *_skb, *bskb;
942 int pkt_cnt = skb_queue_len(list);
945 if (unlikely(msg_size(hdr) > mtu)) {
946 skb_queue_purge(list);
950 /* Allow oversubscription of one data msg per source at congestion */
951 if (unlikely(l->backlog[imp].len >= l->backlog[imp].limit)) {
952 if (imp == TIPC_SYSTEM_IMPORTANCE) {
953 pr_warn("%s<%s>, link overflow", link_rst_msg, l->name);
956 rc = link_schedule_user(l, hdr);
960 l->stats.sent_fragmented++;
961 l->stats.sent_fragments += pkt_cnt;
964 /* Prepare each packet for sending, and add to relevant queue: */
965 while (skb_queue_len(list)) {
966 skb = skb_peek(list);
968 msg_set_seqno(hdr, seqno);
969 msg_set_ack(hdr, ack);
970 msg_set_bcast_ack(hdr, bc_ack);
972 if (likely(skb_queue_len(transmq) < maxwin)) {
973 _skb = skb_clone(skb, GFP_ATOMIC);
975 skb_queue_purge(list);
979 __skb_queue_tail(transmq, skb);
980 /* next retransmit attempt */
981 if (link_is_bc_sndlink(l))
982 TIPC_SKB_CB(skb)->nxt_retr =
983 jiffies + TIPC_BC_RETR_LIM;
984 __skb_queue_tail(xmitq, _skb);
985 TIPC_SKB_CB(skb)->ackers = l->ackers;
987 l->stats.sent_pkts++;
991 if (tipc_msg_bundle(skb_peek_tail(backlogq), hdr, mtu)) {
992 kfree_skb(__skb_dequeue(list));
993 l->stats.sent_bundled++;
996 if (tipc_msg_make_bundle(&bskb, hdr, mtu, l->addr)) {
997 kfree_skb(__skb_dequeue(list));
998 __skb_queue_tail(backlogq, bskb);
999 l->backlog[msg_importance(buf_msg(bskb))].len++;
1000 l->stats.sent_bundled++;
1001 l->stats.sent_bundles++;
1004 l->backlog[imp].len += skb_queue_len(list);
1005 skb_queue_splice_tail_init(list, backlogq);
1011 static void tipc_link_advance_backlog(struct tipc_link *l,
1012 struct sk_buff_head *xmitq)
1014 struct sk_buff *skb, *_skb;
1015 struct tipc_msg *hdr;
1016 u16 seqno = l->snd_nxt;
1017 u16 ack = l->rcv_nxt - 1;
1018 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1020 while (skb_queue_len(&l->transmq) < l->window) {
1021 skb = skb_peek(&l->backlogq);
1024 _skb = skb_clone(skb, GFP_ATOMIC);
1027 __skb_dequeue(&l->backlogq);
1029 l->backlog[msg_importance(hdr)].len--;
1030 __skb_queue_tail(&l->transmq, skb);
1031 /* next retransmit attempt */
1032 if (link_is_bc_sndlink(l))
1033 TIPC_SKB_CB(skb)->nxt_retr = jiffies + TIPC_BC_RETR_LIM;
1035 __skb_queue_tail(xmitq, _skb);
1036 TIPC_SKB_CB(skb)->ackers = l->ackers;
1037 msg_set_seqno(hdr, seqno);
1038 msg_set_ack(hdr, ack);
1039 msg_set_bcast_ack(hdr, bc_ack);
1041 l->stats.sent_pkts++;
1047 static void link_retransmit_failure(struct tipc_link *l, struct sk_buff *skb)
1049 struct tipc_msg *hdr = buf_msg(skb);
1051 pr_warn("Retransmission failure on link <%s>\n", l->name);
1052 link_print(l, "State of link ");
1053 pr_info("Failed msg: usr %u, typ %u, len %u, err %u\n",
1054 msg_user(hdr), msg_type(hdr), msg_size(hdr), msg_errcode(hdr));
1055 pr_info("sqno %u, prev: %x, src: %x\n",
1056 msg_seqno(hdr), msg_prevnode(hdr), msg_orignode(hdr));
1059 /* tipc_link_retrans() - retransmit one or more packets
1060 * @l: the link to transmit on
1061 * @r: the receiving link ordering the retransmit. Same as l if unicast
1062 * @from: retransmit from (inclusive) this sequence number
1063 * @to: retransmit to (inclusive) this sequence number
1064 * xmitq: queue for accumulating the retransmitted packets
1066 static int tipc_link_retrans(struct tipc_link *l, struct tipc_link *r,
1067 u16 from, u16 to, struct sk_buff_head *xmitq)
1069 struct sk_buff *_skb, *skb = skb_peek(&l->transmq);
1070 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1071 u16 ack = l->rcv_nxt - 1;
1072 struct tipc_msg *hdr;
1079 trace_tipc_link_retrans(r, from, to, &l->transmq);
1080 /* Detect repeated retransmit failures on same packet */
1081 if (r->prev_from != from) {
1082 r->prev_from = from;
1083 r->stale_limit = jiffies + msecs_to_jiffies(r->tolerance);
1085 } else if (++r->stale_cnt > 99 && time_after(jiffies, r->stale_limit)) {
1086 link_retransmit_failure(l, skb);
1087 trace_tipc_list_dump(&l->transmq, true, "retrans failure!");
1088 trace_tipc_link_dump(l, TIPC_DUMP_NONE, "retrans failure!");
1089 trace_tipc_link_dump(r, TIPC_DUMP_NONE, "retrans failure!");
1090 if (link_is_bc_sndlink(l))
1091 return TIPC_LINK_DOWN_EVT;
1092 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1095 skb_queue_walk(&l->transmq, skb) {
1097 if (less(msg_seqno(hdr), from))
1099 if (more(msg_seqno(hdr), to))
1101 if (link_is_bc_sndlink(l)) {
1102 if (time_before(jiffies, TIPC_SKB_CB(skb)->nxt_retr))
1104 TIPC_SKB_CB(skb)->nxt_retr = jiffies + TIPC_BC_RETR_LIM;
1106 _skb = __pskb_copy(skb, MIN_H_SIZE, GFP_ATOMIC);
1109 hdr = buf_msg(_skb);
1110 msg_set_ack(hdr, ack);
1111 msg_set_bcast_ack(hdr, bc_ack);
1112 _skb->priority = TC_PRIO_CONTROL;
1113 __skb_queue_tail(xmitq, _skb);
1114 l->stats.retransmitted++;
1119 /* tipc_data_input - deliver data and name distr msgs to upper layer
1121 * Consumes buffer if message is of right type
1122 * Node lock must be held
1124 static bool tipc_data_input(struct tipc_link *l, struct sk_buff *skb,
1125 struct sk_buff_head *inputq)
1127 struct sk_buff_head *mc_inputq = l->bc_rcvlink->inputq;
1128 struct tipc_msg *hdr = buf_msg(skb);
1130 switch (msg_user(hdr)) {
1131 case TIPC_LOW_IMPORTANCE:
1132 case TIPC_MEDIUM_IMPORTANCE:
1133 case TIPC_HIGH_IMPORTANCE:
1134 case TIPC_CRITICAL_IMPORTANCE:
1135 if (unlikely(msg_in_group(hdr) || msg_mcast(hdr))) {
1136 skb_queue_tail(mc_inputq, skb);
1141 skb_queue_tail(inputq, skb);
1143 case GROUP_PROTOCOL:
1144 skb_queue_tail(mc_inputq, skb);
1146 case NAME_DISTRIBUTOR:
1147 l->bc_rcvlink->state = LINK_ESTABLISHED;
1148 skb_queue_tail(l->namedq, skb);
1151 case TUNNEL_PROTOCOL:
1152 case MSG_FRAGMENTER:
1153 case BCAST_PROTOCOL:
1156 pr_warn("Dropping received illegal msg type\n");
1162 /* tipc_link_input - process packet that has passed link protocol check
1166 static int tipc_link_input(struct tipc_link *l, struct sk_buff *skb,
1167 struct sk_buff_head *inputq,
1168 struct sk_buff **reasm_skb)
1170 struct tipc_msg *hdr = buf_msg(skb);
1171 struct sk_buff *iskb;
1172 struct sk_buff_head tmpq;
1173 int usr = msg_user(hdr);
1176 if (usr == MSG_BUNDLER) {
1177 skb_queue_head_init(&tmpq);
1178 l->stats.recv_bundles++;
1179 l->stats.recv_bundled += msg_msgcnt(hdr);
1180 while (tipc_msg_extract(skb, &iskb, &pos))
1181 tipc_data_input(l, iskb, &tmpq);
1182 tipc_skb_queue_splice_tail(&tmpq, inputq);
1184 } else if (usr == MSG_FRAGMENTER) {
1185 l->stats.recv_fragments++;
1186 if (tipc_buf_append(reasm_skb, &skb)) {
1187 l->stats.recv_fragmented++;
1188 tipc_data_input(l, skb, inputq);
1189 } else if (!*reasm_skb && !link_is_bc_rcvlink(l)) {
1190 pr_warn_ratelimited("Unable to build fragment list\n");
1191 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1194 } else if (usr == BCAST_PROTOCOL) {
1195 tipc_bcast_lock(l->net);
1196 tipc_link_bc_init_rcv(l->bc_rcvlink, hdr);
1197 tipc_bcast_unlock(l->net);
1204 /* tipc_link_tnl_rcv() - receive TUNNEL_PROTOCOL message, drop or process the
1205 * inner message along with the ones in the old link's
1208 * @skb: TUNNEL_PROTOCOL message
1209 * @inputq: queue to put messages ready for delivery
1211 static int tipc_link_tnl_rcv(struct tipc_link *l, struct sk_buff *skb,
1212 struct sk_buff_head *inputq)
1214 struct sk_buff **reasm_skb = &l->failover_reasm_skb;
1215 struct sk_buff_head *fdefq = &l->failover_deferdq;
1216 struct tipc_msg *hdr = buf_msg(skb);
1217 struct sk_buff *iskb;
1223 if (msg_type(hdr) == SYNCH_MSG)
1227 if (!tipc_msg_extract(skb, &iskb, &ipos)) {
1228 pr_warn_ratelimited("Cannot extract FAILOVER_MSG, defq: %d\n",
1229 skb_queue_len(fdefq));
1234 seqno = buf_seqno(iskb);
1236 if (unlikely(less(seqno, l->drop_point))) {
1241 if (unlikely(seqno != l->drop_point)) {
1242 __tipc_skb_queue_sorted(fdefq, seqno, iskb);
1248 if (!tipc_data_input(l, iskb, inputq))
1249 rc |= tipc_link_input(l, iskb, inputq, reasm_skb);
1252 } while ((iskb = __tipc_skb_dequeue(fdefq, l->drop_point)));
1259 static bool tipc_link_release_pkts(struct tipc_link *l, u16 acked)
1261 bool released = false;
1262 struct sk_buff *skb, *tmp;
1264 skb_queue_walk_safe(&l->transmq, skb, tmp) {
1265 if (more(buf_seqno(skb), acked))
1267 __skb_unlink(skb, &l->transmq);
1274 /* tipc_build_gap_ack_blks - build Gap ACK blocks
1275 * @l: tipc link that data have come with gaps in sequence if any
1276 * @data: data buffer to store the Gap ACK blocks after built
1278 * returns the actual allocated memory size
1280 static u16 tipc_build_gap_ack_blks(struct tipc_link *l, void *data)
1282 struct sk_buff *skb = skb_peek(&l->deferdq);
1283 struct tipc_gap_ack_blks *ga = data;
1284 u16 len, expect, seqno = 0;
1290 expect = buf_seqno(skb);
1291 skb_queue_walk(&l->deferdq, skb) {
1292 seqno = buf_seqno(skb);
1293 if (unlikely(more(seqno, expect))) {
1294 ga->gacks[n].ack = htons(expect - 1);
1295 ga->gacks[n].gap = htons(seqno - expect);
1296 if (++n >= MAX_GAP_ACK_BLKS) {
1297 pr_info_ratelimited("Too few Gap ACK blocks!\n");
1300 } else if (unlikely(less(seqno, expect))) {
1301 pr_warn("Unexpected skb in deferdq!\n");
1308 ga->gacks[n].ack = htons(seqno);
1309 ga->gacks[n].gap = 0;
1313 len = tipc_gap_ack_blks_sz(n);
1314 ga->len = htons(len);
1319 /* tipc_link_advance_transmq - advance TIPC link transmq queue by releasing
1320 * acked packets, also doing retransmissions if
1322 * @l: tipc link with transmq queue to be advanced
1323 * @acked: seqno of last packet acked by peer without any gaps before
1324 * @gap: # of gap packets
1325 * @ga: buffer pointer to Gap ACK blocks from peer
1326 * @xmitq: queue for accumulating the retransmitted packets if any
1328 static void tipc_link_advance_transmq(struct tipc_link *l, u16 acked, u16 gap,
1329 struct tipc_gap_ack_blks *ga,
1330 struct sk_buff_head *xmitq)
1332 struct sk_buff *skb, *_skb, *tmp;
1333 struct tipc_msg *hdr;
1334 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1335 u16 ack = l->rcv_nxt - 1;
1339 skb_queue_walk_safe(&l->transmq, skb, tmp) {
1340 seqno = buf_seqno(skb);
1343 if (less_eq(seqno, acked)) {
1345 __skb_unlink(skb, &l->transmq);
1347 } else if (less_eq(seqno, acked + gap)) {
1348 /* retransmit skb */
1349 if (time_before(jiffies, TIPC_SKB_CB(skb)->nxt_retr))
1351 TIPC_SKB_CB(skb)->nxt_retr = TIPC_UC_RETR_TIME;
1353 _skb = __pskb_copy(skb, MIN_H_SIZE, GFP_ATOMIC);
1356 hdr = buf_msg(_skb);
1357 msg_set_ack(hdr, ack);
1358 msg_set_bcast_ack(hdr, bc_ack);
1359 _skb->priority = TC_PRIO_CONTROL;
1360 __skb_queue_tail(xmitq, _skb);
1361 l->stats.retransmitted++;
1363 /* retry with Gap ACK blocks if any */
1364 if (!ga || n >= ga->gack_cnt)
1366 acked = ntohs(ga->gacks[n].ack);
1367 gap = ntohs(ga->gacks[n].gap);
1374 /* tipc_link_build_state_msg: prepare link state message for transmission
1376 * Note that sending of broadcast ack is coordinated among nodes, to reduce
1377 * risk of ack storms towards the sender
1379 int tipc_link_build_state_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1384 /* Broadcast ACK must be sent via a unicast link => defer to caller */
1385 if (link_is_bc_rcvlink(l)) {
1386 if (((l->rcv_nxt ^ tipc_own_addr(l->net)) & 0xf) != 0xf)
1390 /* Use snd_nxt to store peer's snd_nxt in broadcast rcv link */
1391 l->snd_nxt = l->rcv_nxt;
1392 return TIPC_LINK_SND_STATE;
1397 l->stats.sent_acks++;
1398 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, 0, xmitq);
1402 /* tipc_link_build_reset_msg: prepare link RESET or ACTIVATE message
1404 void tipc_link_build_reset_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1406 int mtyp = RESET_MSG;
1407 struct sk_buff *skb;
1409 if (l->state == LINK_ESTABLISHING)
1410 mtyp = ACTIVATE_MSG;
1412 tipc_link_build_proto_msg(l, mtyp, 0, 0, 0, 0, 0, xmitq);
1414 /* Inform peer that this endpoint is going down if applicable */
1415 skb = skb_peek_tail(xmitq);
1416 if (skb && (l->state == LINK_RESET))
1417 msg_set_peer_stopping(buf_msg(skb), 1);
1420 /* tipc_link_build_nack_msg: prepare link nack message for transmission
1421 * Note that sending of broadcast NACK is coordinated among nodes, to
1422 * reduce the risk of NACK storms towards the sender
1424 static int tipc_link_build_nack_msg(struct tipc_link *l,
1425 struct sk_buff_head *xmitq)
1427 u32 def_cnt = ++l->stats.deferred_recv;
1428 u32 defq_len = skb_queue_len(&l->deferdq);
1431 if (link_is_bc_rcvlink(l)) {
1432 match1 = def_cnt & 0xf;
1433 match2 = tipc_own_addr(l->net) & 0xf;
1434 if (match1 == match2)
1435 return TIPC_LINK_SND_STATE;
1439 if (defq_len >= 3 && !((defq_len - 3) % 16))
1440 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, 0, xmitq);
1444 /* tipc_link_rcv - process TIPC packets/messages arriving from off-node
1445 * @l: the link that should handle the message
1447 * @xmitq: queue to place packets to be sent after this call
1449 int tipc_link_rcv(struct tipc_link *l, struct sk_buff *skb,
1450 struct sk_buff_head *xmitq)
1452 struct sk_buff_head *defq = &l->deferdq;
1453 struct tipc_msg *hdr = buf_msg(skb);
1454 u16 seqno, rcv_nxt, win_lim;
1457 /* Verify and update link state */
1458 if (unlikely(msg_user(hdr) == LINK_PROTOCOL))
1459 return tipc_link_proto_rcv(l, skb, xmitq);
1461 /* Don't send probe at next timeout expiration */
1462 l->silent_intv_cnt = 0;
1466 seqno = msg_seqno(hdr);
1467 rcv_nxt = l->rcv_nxt;
1468 win_lim = rcv_nxt + TIPC_MAX_LINK_WIN;
1470 if (unlikely(!link_is_up(l))) {
1471 if (l->state == LINK_ESTABLISHING)
1472 rc = TIPC_LINK_UP_EVT;
1476 /* Drop if outside receive window */
1477 if (unlikely(less(seqno, rcv_nxt) || more(seqno, win_lim))) {
1478 l->stats.duplicates++;
1482 /* Forward queues and wake up waiting users */
1483 if (likely(tipc_link_release_pkts(l, msg_ack(hdr)))) {
1485 tipc_link_advance_backlog(l, xmitq);
1486 if (unlikely(!skb_queue_empty(&l->wakeupq)))
1487 link_prepare_wakeup(l);
1490 /* Defer delivery if sequence gap */
1491 if (unlikely(seqno != rcv_nxt)) {
1492 __tipc_skb_queue_sorted(defq, seqno, skb);
1493 rc |= tipc_link_build_nack_msg(l, xmitq);
1497 /* Deliver packet */
1499 l->stats.recv_pkts++;
1501 if (unlikely(msg_user(hdr) == TUNNEL_PROTOCOL))
1502 rc |= tipc_link_tnl_rcv(l, skb, l->inputq);
1503 else if (!tipc_data_input(l, skb, l->inputq))
1504 rc |= tipc_link_input(l, skb, l->inputq, &l->reasm_buf);
1505 if (unlikely(++l->rcv_unacked >= TIPC_MIN_LINK_WIN))
1506 rc |= tipc_link_build_state_msg(l, xmitq);
1507 if (unlikely(rc & ~TIPC_LINK_SND_STATE))
1509 } while ((skb = __tipc_skb_dequeue(defq, l->rcv_nxt)));
1517 static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
1518 bool probe_reply, u16 rcvgap,
1519 int tolerance, int priority,
1520 struct sk_buff_head *xmitq)
1522 struct tipc_link *bcl = l->bc_rcvlink;
1523 struct sk_buff *skb;
1524 struct tipc_msg *hdr;
1525 struct sk_buff_head *dfq = &l->deferdq;
1526 bool node_up = link_is_up(bcl);
1527 struct tipc_mon_state *mstate = &l->mon_state;
1532 /* Don't send protocol message during reset or link failover */
1533 if (tipc_link_is_blocked(l))
1536 if (!tipc_link_is_up(l) && (mtyp == STATE_MSG))
1539 if (!skb_queue_empty(dfq))
1540 rcvgap = buf_seqno(skb_peek(dfq)) - l->rcv_nxt;
1542 skb = tipc_msg_create(LINK_PROTOCOL, mtyp, INT_H_SIZE,
1543 tipc_max_domain_size + MAX_GAP_ACK_BLKS_SZ,
1544 l->addr, tipc_own_addr(l->net), 0, 0, 0);
1549 data = msg_data(hdr);
1550 msg_set_session(hdr, l->session);
1551 msg_set_bearer_id(hdr, l->bearer_id);
1552 msg_set_net_plane(hdr, l->net_plane);
1553 msg_set_next_sent(hdr, l->snd_nxt);
1554 msg_set_ack(hdr, l->rcv_nxt - 1);
1555 msg_set_bcast_ack(hdr, bcl->rcv_nxt - 1);
1556 msg_set_bc_ack_invalid(hdr, !node_up);
1557 msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1558 msg_set_link_tolerance(hdr, tolerance);
1559 msg_set_linkprio(hdr, priority);
1560 msg_set_redundant_link(hdr, node_up);
1561 msg_set_seq_gap(hdr, 0);
1562 msg_set_seqno(hdr, l->snd_nxt + U16_MAX / 2);
1564 if (mtyp == STATE_MSG) {
1565 if (l->peer_caps & TIPC_LINK_PROTO_SEQNO)
1566 msg_set_seqno(hdr, l->snd_nxt_state++);
1567 msg_set_seq_gap(hdr, rcvgap);
1568 msg_set_bc_gap(hdr, link_bc_rcv_gap(bcl));
1569 msg_set_probe(hdr, probe);
1570 msg_set_is_keepalive(hdr, probe || probe_reply);
1571 if (l->peer_caps & TIPC_GAP_ACK_BLOCK)
1572 glen = tipc_build_gap_ack_blks(l, data);
1573 tipc_mon_prep(l->net, data + glen, &dlen, mstate, l->bearer_id);
1574 msg_set_size(hdr, INT_H_SIZE + glen + dlen);
1575 skb_trim(skb, INT_H_SIZE + glen + dlen);
1576 l->stats.sent_states++;
1579 /* RESET_MSG or ACTIVATE_MSG */
1580 if (mtyp == ACTIVATE_MSG) {
1581 msg_set_dest_session_valid(hdr, 1);
1582 msg_set_dest_session(hdr, l->peer_session);
1584 msg_set_max_pkt(hdr, l->advertised_mtu);
1585 strcpy(data, l->if_name);
1586 msg_set_size(hdr, INT_H_SIZE + TIPC_MAX_IF_NAME);
1587 skb_trim(skb, INT_H_SIZE + TIPC_MAX_IF_NAME);
1590 l->stats.sent_probes++;
1592 l->stats.sent_nacks++;
1593 skb->priority = TC_PRIO_CONTROL;
1594 __skb_queue_tail(xmitq, skb);
1595 trace_tipc_proto_build(skb, false, l->name);
1598 void tipc_link_create_dummy_tnl_msg(struct tipc_link *l,
1599 struct sk_buff_head *xmitq)
1601 u32 onode = tipc_own_addr(l->net);
1602 struct tipc_msg *hdr, *ihdr;
1603 struct sk_buff_head tnlq;
1604 struct sk_buff *skb;
1605 u32 dnode = l->addr;
1607 skb_queue_head_init(&tnlq);
1608 skb = tipc_msg_create(TUNNEL_PROTOCOL, FAILOVER_MSG,
1609 INT_H_SIZE, BASIC_H_SIZE,
1610 dnode, onode, 0, 0, 0);
1612 pr_warn("%sunable to create tunnel packet\n", link_co_err);
1617 msg_set_msgcnt(hdr, 1);
1618 msg_set_bearer_id(hdr, l->peer_bearer_id);
1620 ihdr = (struct tipc_msg *)msg_data(hdr);
1621 tipc_msg_init(onode, ihdr, TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1622 BASIC_H_SIZE, dnode);
1623 msg_set_errcode(ihdr, TIPC_ERR_NO_PORT);
1624 __skb_queue_tail(&tnlq, skb);
1625 tipc_link_xmit(l, &tnlq, xmitq);
1628 /* tipc_link_tnl_prepare(): prepare and return a list of tunnel packets
1629 * with contents of the link's transmit and backlog queues.
1631 void tipc_link_tnl_prepare(struct tipc_link *l, struct tipc_link *tnl,
1632 int mtyp, struct sk_buff_head *xmitq)
1634 struct sk_buff_head *fdefq = &tnl->failover_deferdq;
1635 struct sk_buff *skb, *tnlskb;
1636 struct tipc_msg *hdr, tnlhdr;
1637 struct sk_buff_head *queue = &l->transmq;
1638 struct sk_buff_head tmpxq, tnlq;
1639 u16 pktlen, pktcnt, seqno = l->snd_nxt;
1644 skb_queue_head_init(&tnlq);
1645 skb_queue_head_init(&tmpxq);
1647 /* At least one packet required for safe algorithm => add dummy */
1648 skb = tipc_msg_create(TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1649 BASIC_H_SIZE, 0, l->addr, tipc_own_addr(l->net),
1650 0, 0, TIPC_ERR_NO_PORT);
1652 pr_warn("%sunable to create tunnel packet\n", link_co_err);
1655 skb_queue_tail(&tnlq, skb);
1656 tipc_link_xmit(l, &tnlq, &tmpxq);
1657 __skb_queue_purge(&tmpxq);
1659 /* Initialize reusable tunnel packet header */
1660 tipc_msg_init(tipc_own_addr(l->net), &tnlhdr, TUNNEL_PROTOCOL,
1661 mtyp, INT_H_SIZE, l->addr);
1662 if (mtyp == SYNCH_MSG)
1663 pktcnt = l->snd_nxt - buf_seqno(skb_peek(&l->transmq));
1665 pktcnt = skb_queue_len(&l->transmq);
1666 pktcnt += skb_queue_len(&l->backlogq);
1667 msg_set_msgcnt(&tnlhdr, pktcnt);
1668 msg_set_bearer_id(&tnlhdr, l->peer_bearer_id);
1670 /* Wrap each packet into a tunnel packet */
1671 skb_queue_walk(queue, skb) {
1673 if (queue == &l->backlogq)
1674 msg_set_seqno(hdr, seqno++);
1675 pktlen = msg_size(hdr);
1676 msg_set_size(&tnlhdr, pktlen + INT_H_SIZE);
1677 tnlskb = tipc_buf_acquire(pktlen + INT_H_SIZE, GFP_ATOMIC);
1679 pr_warn("%sunable to send packet\n", link_co_err);
1682 skb_copy_to_linear_data(tnlskb, &tnlhdr, INT_H_SIZE);
1683 skb_copy_to_linear_data_offset(tnlskb, INT_H_SIZE, hdr, pktlen);
1684 __skb_queue_tail(&tnlq, tnlskb);
1686 if (queue != &l->backlogq) {
1687 queue = &l->backlogq;
1691 tipc_link_xmit(tnl, &tnlq, xmitq);
1693 if (mtyp == FAILOVER_MSG) {
1694 tnl->drop_point = l->rcv_nxt;
1695 tnl->failover_reasm_skb = l->reasm_buf;
1696 l->reasm_buf = NULL;
1698 /* Failover the link's deferdq */
1699 if (unlikely(!skb_queue_empty(fdefq))) {
1700 pr_warn("Link failover deferdq not empty: %d!\n",
1701 skb_queue_len(fdefq));
1702 __skb_queue_purge(fdefq);
1704 skb_queue_splice_init(&l->deferdq, fdefq);
1709 * tipc_link_failover_prepare() - prepare tnl for link failover
1711 * This is a special version of the precursor - tipc_link_tnl_prepare(),
1712 * see the tipc_node_link_failover() for details
1716 * @xmitq: queue for messages to be xmited
1718 void tipc_link_failover_prepare(struct tipc_link *l, struct tipc_link *tnl,
1719 struct sk_buff_head *xmitq)
1721 struct sk_buff_head *fdefq = &tnl->failover_deferdq;
1723 tipc_link_create_dummy_tnl_msg(tnl, xmitq);
1725 /* This failover link enpoint was never established before,
1726 * so it has not received anything from peer.
1727 * Otherwise, it must be a normal failover situation or the
1728 * node has entered SELF_DOWN_PEER_LEAVING and both peer nodes
1729 * would have to start over from scratch instead.
1731 tnl->drop_point = 1;
1732 tnl->failover_reasm_skb = NULL;
1734 /* Initiate the link's failover deferdq */
1735 if (unlikely(!skb_queue_empty(fdefq))) {
1736 pr_warn("Link failover deferdq not empty: %d!\n",
1737 skb_queue_len(fdefq));
1738 __skb_queue_purge(fdefq);
1742 /* tipc_link_validate_msg(): validate message against current link state
1743 * Returns true if message should be accepted, otherwise false
1745 bool tipc_link_validate_msg(struct tipc_link *l, struct tipc_msg *hdr)
1747 u16 curr_session = l->peer_session;
1748 u16 session = msg_session(hdr);
1749 int mtyp = msg_type(hdr);
1751 if (msg_user(hdr) != LINK_PROTOCOL)
1758 /* Accept only RESET with new session number */
1759 return more(session, curr_session);
1763 /* Accept only ACTIVATE with new or current session number */
1764 return !less(session, curr_session);
1766 /* Accept only STATE with current session number */
1769 if (session != curr_session)
1771 /* Extra sanity check */
1772 if (!link_is_up(l) && msg_ack(hdr))
1774 if (!(l->peer_caps & TIPC_LINK_PROTO_SEQNO))
1776 /* Accept only STATE with new sequence number */
1777 return !less(msg_seqno(hdr), l->rcv_nxt_state);
1783 /* tipc_link_proto_rcv(): receive link level protocol message :
1784 * Note that network plane id propagates through the network, and may
1785 * change at any time. The node with lowest numerical id determines
1788 static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
1789 struct sk_buff_head *xmitq)
1791 struct tipc_msg *hdr = buf_msg(skb);
1792 struct tipc_gap_ack_blks *ga = NULL;
1794 u16 ack = msg_ack(hdr);
1795 u16 gap = msg_seq_gap(hdr);
1796 u16 peers_snd_nxt = msg_next_sent(hdr);
1797 u16 peers_tol = msg_link_tolerance(hdr);
1798 u16 peers_prio = msg_linkprio(hdr);
1799 u16 rcv_nxt = l->rcv_nxt;
1800 u16 dlen = msg_data_sz(hdr);
1801 int mtyp = msg_type(hdr);
1802 bool reply = msg_probe(hdr);
1808 trace_tipc_proto_rcv(skb, false, l->name);
1809 if (tipc_link_is_blocked(l) || !xmitq)
1812 if (tipc_own_addr(l->net) > msg_prevnode(hdr))
1813 l->net_plane = msg_net_plane(hdr);
1817 data = msg_data(hdr);
1819 if (!tipc_link_validate_msg(l, hdr)) {
1820 trace_tipc_skb_dump(skb, false, "PROTO invalid (1)!");
1821 trace_tipc_link_dump(l, TIPC_DUMP_NONE, "PROTO invalid (1)!");
1828 /* Complete own link name with peer's interface name */
1829 if_name = strrchr(l->name, ':') + 1;
1830 if (sizeof(l->name) - (if_name - l->name) <= TIPC_MAX_IF_NAME)
1832 if (msg_data_sz(hdr) < TIPC_MAX_IF_NAME)
1834 strncpy(if_name, data, TIPC_MAX_IF_NAME);
1836 /* Update own tolerance if peer indicates a non-zero value */
1837 if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
1838 l->tolerance = peers_tol;
1839 l->bc_rcvlink->tolerance = peers_tol;
1841 /* Update own priority if peer's priority is higher */
1842 if (in_range(peers_prio, l->priority + 1, TIPC_MAX_LINK_PRI))
1843 l->priority = peers_prio;
1845 /* If peer is going down we want full re-establish cycle */
1846 if (msg_peer_stopping(hdr)) {
1847 rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1851 /* If this endpoint was re-created while peer was ESTABLISHING
1852 * it doesn't know current session number. Force re-synch.
1854 if (mtyp == ACTIVATE_MSG && msg_dest_session_valid(hdr) &&
1855 l->session != msg_dest_session(hdr)) {
1856 if (less(l->session, msg_dest_session(hdr)))
1857 l->session = msg_dest_session(hdr) + 1;
1861 /* ACTIVATE_MSG serves as PEER_RESET if link is already down */
1862 if (mtyp == RESET_MSG || !link_is_up(l))
1863 rc = tipc_link_fsm_evt(l, LINK_PEER_RESET_EVT);
1865 /* ACTIVATE_MSG takes up link if it was already locally reset */
1866 if (mtyp == ACTIVATE_MSG && l->state == LINK_ESTABLISHING)
1867 rc = TIPC_LINK_UP_EVT;
1869 l->peer_session = msg_session(hdr);
1870 l->in_session = true;
1871 l->peer_bearer_id = msg_bearer_id(hdr);
1872 if (l->mtu > msg_max_pkt(hdr))
1873 l->mtu = msg_max_pkt(hdr);
1877 l->rcv_nxt_state = msg_seqno(hdr) + 1;
1879 /* Update own tolerance if peer indicates a non-zero value */
1880 if (in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
1881 l->tolerance = peers_tol;
1882 l->bc_rcvlink->tolerance = peers_tol;
1884 /* Update own prio if peer indicates a different value */
1885 if ((peers_prio != l->priority) &&
1886 in_range(peers_prio, 1, TIPC_MAX_LINK_PRI)) {
1887 l->priority = peers_prio;
1888 rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1891 l->silent_intv_cnt = 0;
1892 l->stats.recv_states++;
1894 l->stats.recv_probes++;
1896 if (!link_is_up(l)) {
1897 if (l->state == LINK_ESTABLISHING)
1898 rc = TIPC_LINK_UP_EVT;
1902 /* Receive Gap ACK blocks from peer if any */
1903 if (l->peer_caps & TIPC_GAP_ACK_BLOCK) {
1904 ga = (struct tipc_gap_ack_blks *)data;
1905 glen = ntohs(ga->len);
1906 /* sanity check: if failed, ignore Gap ACK blocks */
1907 if (glen != tipc_gap_ack_blks_sz(ga->gack_cnt))
1911 tipc_mon_rcv(l->net, data + glen, dlen - glen, l->addr,
1912 &l->mon_state, l->bearer_id);
1914 /* Send NACK if peer has sent pkts we haven't received yet */
1915 if (more(peers_snd_nxt, rcv_nxt) && !tipc_link_is_synching(l))
1916 rcvgap = peers_snd_nxt - l->rcv_nxt;
1917 if (rcvgap || reply)
1918 tipc_link_build_proto_msg(l, STATE_MSG, 0, reply,
1919 rcvgap, 0, 0, xmitq);
1921 tipc_link_advance_transmq(l, ack, gap, ga, xmitq);
1923 /* If NACK, retransmit will now start at right position */
1925 l->stats.recv_nacks++;
1927 tipc_link_advance_backlog(l, xmitq);
1928 if (unlikely(!skb_queue_empty(&l->wakeupq)))
1929 link_prepare_wakeup(l);
1936 /* tipc_link_build_bc_proto_msg() - create broadcast protocol message
1938 static bool tipc_link_build_bc_proto_msg(struct tipc_link *l, bool bcast,
1940 struct sk_buff_head *xmitq)
1942 struct sk_buff *skb;
1943 struct tipc_msg *hdr;
1944 struct sk_buff *dfrd_skb = skb_peek(&l->deferdq);
1945 u16 ack = l->rcv_nxt - 1;
1946 u16 gap_to = peers_snd_nxt - 1;
1948 skb = tipc_msg_create(BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE,
1949 0, l->addr, tipc_own_addr(l->net), 0, 0, 0);
1953 msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1954 msg_set_bcast_ack(hdr, ack);
1955 msg_set_bcgap_after(hdr, ack);
1957 gap_to = buf_seqno(dfrd_skb) - 1;
1958 msg_set_bcgap_to(hdr, gap_to);
1959 msg_set_non_seq(hdr, bcast);
1960 __skb_queue_tail(xmitq, skb);
1964 /* tipc_link_build_bc_init_msg() - synchronize broadcast link endpoints.
1966 * Give a newly added peer node the sequence number where it should
1967 * start receiving and acking broadcast packets.
1969 static void tipc_link_build_bc_init_msg(struct tipc_link *l,
1970 struct sk_buff_head *xmitq)
1972 struct sk_buff_head list;
1974 __skb_queue_head_init(&list);
1975 if (!tipc_link_build_bc_proto_msg(l->bc_rcvlink, false, 0, &list))
1977 msg_set_bc_ack_invalid(buf_msg(skb_peek(&list)), true);
1978 tipc_link_xmit(l, &list, xmitq);
1981 /* tipc_link_bc_init_rcv - receive initial broadcast synch data from peer
1983 void tipc_link_bc_init_rcv(struct tipc_link *l, struct tipc_msg *hdr)
1985 int mtyp = msg_type(hdr);
1986 u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
1991 if (msg_user(hdr) == BCAST_PROTOCOL) {
1992 l->rcv_nxt = peers_snd_nxt;
1993 l->state = LINK_ESTABLISHED;
1997 if (l->peer_caps & TIPC_BCAST_SYNCH)
2000 if (msg_peer_node_is_up(hdr))
2003 /* Compatibility: accept older, less safe initial synch data */
2004 if ((mtyp == RESET_MSG) || (mtyp == ACTIVATE_MSG))
2005 l->rcv_nxt = peers_snd_nxt;
2008 /* tipc_link_bc_sync_rcv - update rcv link according to peer's send state
2010 int tipc_link_bc_sync_rcv(struct tipc_link *l, struct tipc_msg *hdr,
2011 struct sk_buff_head *xmitq)
2013 struct tipc_link *snd_l = l->bc_sndlink;
2014 u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
2015 u16 from = msg_bcast_ack(hdr) + 1;
2016 u16 to = from + msg_bc_gap(hdr) - 1;
2022 if (!msg_peer_node_is_up(hdr))
2025 /* Open when peer ackowledges our bcast init msg (pkt #1) */
2027 l->bc_peer_is_up = true;
2029 if (!l->bc_peer_is_up)
2032 l->stats.recv_nacks++;
2034 /* Ignore if peers_snd_nxt goes beyond receive window */
2035 if (more(peers_snd_nxt, l->rcv_nxt + l->window))
2038 rc = tipc_link_retrans(snd_l, l, from, to, xmitq);
2040 l->snd_nxt = peers_snd_nxt;
2041 if (link_bc_rcv_gap(l))
2042 rc |= TIPC_LINK_SND_STATE;
2044 /* Return now if sender supports nack via STATE messages */
2045 if (l->peer_caps & TIPC_BCAST_STATE_NACK)
2048 /* Otherwise, be backwards compatible */
2050 if (!more(peers_snd_nxt, l->rcv_nxt)) {
2051 l->nack_state = BC_NACK_SND_CONDITIONAL;
2055 /* Don't NACK if one was recently sent or peeked */
2056 if (l->nack_state == BC_NACK_SND_SUPPRESS) {
2057 l->nack_state = BC_NACK_SND_UNCONDITIONAL;
2061 /* Conditionally delay NACK sending until next synch rcv */
2062 if (l->nack_state == BC_NACK_SND_CONDITIONAL) {
2063 l->nack_state = BC_NACK_SND_UNCONDITIONAL;
2064 if ((peers_snd_nxt - l->rcv_nxt) < TIPC_MIN_LINK_WIN)
2068 /* Send NACK now but suppress next one */
2069 tipc_link_build_bc_proto_msg(l, true, peers_snd_nxt, xmitq);
2070 l->nack_state = BC_NACK_SND_SUPPRESS;
2074 void tipc_link_bc_ack_rcv(struct tipc_link *l, u16 acked,
2075 struct sk_buff_head *xmitq)
2077 struct sk_buff *skb, *tmp;
2078 struct tipc_link *snd_l = l->bc_sndlink;
2080 if (!link_is_up(l) || !l->bc_peer_is_up)
2083 if (!more(acked, l->acked))
2086 trace_tipc_link_bc_ack(l, l->acked, acked, &snd_l->transmq);
2087 /* Skip over packets peer has already acked */
2088 skb_queue_walk(&snd_l->transmq, skb) {
2089 if (more(buf_seqno(skb), l->acked))
2093 /* Update/release the packets peer is acking now */
2094 skb_queue_walk_from_safe(&snd_l->transmq, skb, tmp) {
2095 if (more(buf_seqno(skb), acked))
2097 if (!--TIPC_SKB_CB(skb)->ackers) {
2098 __skb_unlink(skb, &snd_l->transmq);
2103 tipc_link_advance_backlog(snd_l, xmitq);
2104 if (unlikely(!skb_queue_empty(&snd_l->wakeupq)))
2105 link_prepare_wakeup(snd_l);
2108 /* tipc_link_bc_nack_rcv(): receive broadcast nack message
2109 * This function is here for backwards compatibility, since
2110 * no BCAST_PROTOCOL/STATE messages occur from TIPC v2.5.
2112 int tipc_link_bc_nack_rcv(struct tipc_link *l, struct sk_buff *skb,
2113 struct sk_buff_head *xmitq)
2115 struct tipc_msg *hdr = buf_msg(skb);
2116 u32 dnode = msg_destnode(hdr);
2117 int mtyp = msg_type(hdr);
2118 u16 acked = msg_bcast_ack(hdr);
2119 u16 from = acked + 1;
2120 u16 to = msg_bcgap_to(hdr);
2121 u16 peers_snd_nxt = to + 1;
2126 if (!tipc_link_is_up(l) || !l->bc_peer_is_up)
2129 if (mtyp != STATE_MSG)
2132 if (dnode == tipc_own_addr(l->net)) {
2133 tipc_link_bc_ack_rcv(l, acked, xmitq);
2134 rc = tipc_link_retrans(l->bc_sndlink, l, from, to, xmitq);
2135 l->stats.recv_nacks++;
2139 /* Msg for other node => suppress own NACK at next sync if applicable */
2140 if (more(peers_snd_nxt, l->rcv_nxt) && !less(l->rcv_nxt, from))
2141 l->nack_state = BC_NACK_SND_SUPPRESS;
2146 void tipc_link_set_queue_limits(struct tipc_link *l, u32 win)
2148 int max_bulk = TIPC_MAX_PUBL / (l->mtu / ITEM_SIZE);
2151 l->backlog[TIPC_LOW_IMPORTANCE].limit = max_t(u16, 50, win);
2152 l->backlog[TIPC_MEDIUM_IMPORTANCE].limit = max_t(u16, 100, win * 2);
2153 l->backlog[TIPC_HIGH_IMPORTANCE].limit = max_t(u16, 150, win * 3);
2154 l->backlog[TIPC_CRITICAL_IMPORTANCE].limit = max_t(u16, 200, win * 4);
2155 l->backlog[TIPC_SYSTEM_IMPORTANCE].limit = max_bulk;
2159 * link_reset_stats - reset link statistics
2160 * @l: pointer to link
2162 void tipc_link_reset_stats(struct tipc_link *l)
2164 memset(&l->stats, 0, sizeof(l->stats));
2167 static void link_print(struct tipc_link *l, const char *str)
2169 struct sk_buff *hskb = skb_peek(&l->transmq);
2170 u16 head = hskb ? msg_seqno(buf_msg(hskb)) : l->snd_nxt - 1;
2171 u16 tail = l->snd_nxt - 1;
2173 pr_info("%s Link <%s> state %x\n", str, l->name, l->state);
2174 pr_info("XMTQ: %u [%u-%u], BKLGQ: %u, SNDNX: %u, RCVNX: %u\n",
2175 skb_queue_len(&l->transmq), head, tail,
2176 skb_queue_len(&l->backlogq), l->snd_nxt, l->rcv_nxt);
2179 /* Parse and validate nested (link) properties valid for media, bearer and link
2181 int tipc_nl_parse_link_prop(struct nlattr *prop, struct nlattr *props[])
2185 err = nla_parse_nested_deprecated(props, TIPC_NLA_PROP_MAX, prop,
2186 tipc_nl_prop_policy, NULL);
2190 if (props[TIPC_NLA_PROP_PRIO]) {
2193 prio = nla_get_u32(props[TIPC_NLA_PROP_PRIO]);
2194 if (prio > TIPC_MAX_LINK_PRI)
2198 if (props[TIPC_NLA_PROP_TOL]) {
2201 tol = nla_get_u32(props[TIPC_NLA_PROP_TOL]);
2202 if ((tol < TIPC_MIN_LINK_TOL) || (tol > TIPC_MAX_LINK_TOL))
2206 if (props[TIPC_NLA_PROP_WIN]) {
2209 win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
2210 if ((win < TIPC_MIN_LINK_WIN) || (win > TIPC_MAX_LINK_WIN))
2217 static int __tipc_nl_add_stats(struct sk_buff *skb, struct tipc_stats *s)
2220 struct nlattr *stats;
2227 struct nla_map map[] = {
2228 {TIPC_NLA_STATS_RX_INFO, 0},
2229 {TIPC_NLA_STATS_RX_FRAGMENTS, s->recv_fragments},
2230 {TIPC_NLA_STATS_RX_FRAGMENTED, s->recv_fragmented},
2231 {TIPC_NLA_STATS_RX_BUNDLES, s->recv_bundles},
2232 {TIPC_NLA_STATS_RX_BUNDLED, s->recv_bundled},
2233 {TIPC_NLA_STATS_TX_INFO, 0},
2234 {TIPC_NLA_STATS_TX_FRAGMENTS, s->sent_fragments},
2235 {TIPC_NLA_STATS_TX_FRAGMENTED, s->sent_fragmented},
2236 {TIPC_NLA_STATS_TX_BUNDLES, s->sent_bundles},
2237 {TIPC_NLA_STATS_TX_BUNDLED, s->sent_bundled},
2238 {TIPC_NLA_STATS_MSG_PROF_TOT, (s->msg_length_counts) ?
2239 s->msg_length_counts : 1},
2240 {TIPC_NLA_STATS_MSG_LEN_CNT, s->msg_length_counts},
2241 {TIPC_NLA_STATS_MSG_LEN_TOT, s->msg_lengths_total},
2242 {TIPC_NLA_STATS_MSG_LEN_P0, s->msg_length_profile[0]},
2243 {TIPC_NLA_STATS_MSG_LEN_P1, s->msg_length_profile[1]},
2244 {TIPC_NLA_STATS_MSG_LEN_P2, s->msg_length_profile[2]},
2245 {TIPC_NLA_STATS_MSG_LEN_P3, s->msg_length_profile[3]},
2246 {TIPC_NLA_STATS_MSG_LEN_P4, s->msg_length_profile[4]},
2247 {TIPC_NLA_STATS_MSG_LEN_P5, s->msg_length_profile[5]},
2248 {TIPC_NLA_STATS_MSG_LEN_P6, s->msg_length_profile[6]},
2249 {TIPC_NLA_STATS_RX_STATES, s->recv_states},
2250 {TIPC_NLA_STATS_RX_PROBES, s->recv_probes},
2251 {TIPC_NLA_STATS_RX_NACKS, s->recv_nacks},
2252 {TIPC_NLA_STATS_RX_DEFERRED, s->deferred_recv},
2253 {TIPC_NLA_STATS_TX_STATES, s->sent_states},
2254 {TIPC_NLA_STATS_TX_PROBES, s->sent_probes},
2255 {TIPC_NLA_STATS_TX_NACKS, s->sent_nacks},
2256 {TIPC_NLA_STATS_TX_ACKS, s->sent_acks},
2257 {TIPC_NLA_STATS_RETRANSMITTED, s->retransmitted},
2258 {TIPC_NLA_STATS_DUPLICATES, s->duplicates},
2259 {TIPC_NLA_STATS_LINK_CONGS, s->link_congs},
2260 {TIPC_NLA_STATS_MAX_QUEUE, s->max_queue_sz},
2261 {TIPC_NLA_STATS_AVG_QUEUE, s->queue_sz_counts ?
2262 (s->accu_queue_sz / s->queue_sz_counts) : 0}
2265 stats = nla_nest_start_noflag(skb, TIPC_NLA_LINK_STATS);
2269 for (i = 0; i < ARRAY_SIZE(map); i++)
2270 if (nla_put_u32(skb, map[i].key, map[i].val))
2273 nla_nest_end(skb, stats);
2277 nla_nest_cancel(skb, stats);
2282 /* Caller should hold appropriate locks to protect the link */
2283 int __tipc_nl_add_link(struct net *net, struct tipc_nl_msg *msg,
2284 struct tipc_link *link, int nlflags)
2286 u32 self = tipc_own_addr(net);
2287 struct nlattr *attrs;
2288 struct nlattr *prop;
2292 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2293 nlflags, TIPC_NL_LINK_GET);
2297 attrs = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK);
2301 if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, link->name))
2303 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_DEST, tipc_cluster_mask(self)))
2305 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_MTU, link->mtu))
2307 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, link->stats.recv_pkts))
2309 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, link->stats.sent_pkts))
2312 if (tipc_link_is_up(link))
2313 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2316 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_ACTIVE))
2319 prop = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK_PROP);
2322 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2324 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_TOL, link->tolerance))
2326 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN,
2329 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2331 nla_nest_end(msg->skb, prop);
2333 err = __tipc_nl_add_stats(msg->skb, &link->stats);
2337 nla_nest_end(msg->skb, attrs);
2338 genlmsg_end(msg->skb, hdr);
2343 nla_nest_cancel(msg->skb, prop);
2345 nla_nest_cancel(msg->skb, attrs);
2347 genlmsg_cancel(msg->skb, hdr);
2352 static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
2353 struct tipc_stats *stats)
2356 struct nlattr *nest;
2363 struct nla_map map[] = {
2364 {TIPC_NLA_STATS_RX_INFO, stats->recv_pkts},
2365 {TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
2366 {TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
2367 {TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
2368 {TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
2369 {TIPC_NLA_STATS_TX_INFO, stats->sent_pkts},
2370 {TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
2371 {TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
2372 {TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
2373 {TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
2374 {TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
2375 {TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
2376 {TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
2377 {TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
2378 {TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
2379 {TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
2380 {TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
2381 {TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
2382 {TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
2383 (stats->accu_queue_sz / stats->queue_sz_counts) : 0}
2386 nest = nla_nest_start_noflag(skb, TIPC_NLA_LINK_STATS);
2390 for (i = 0; i < ARRAY_SIZE(map); i++)
2391 if (nla_put_u32(skb, map[i].key, map[i].val))
2394 nla_nest_end(skb, nest);
2398 nla_nest_cancel(skb, nest);
2403 int tipc_nl_add_bc_link(struct net *net, struct tipc_nl_msg *msg)
2407 struct nlattr *attrs;
2408 struct nlattr *prop;
2409 struct tipc_net *tn = net_generic(net, tipc_net_id);
2410 u32 bc_mode = tipc_bcast_get_broadcast_mode(net);
2411 u32 bc_ratio = tipc_bcast_get_broadcast_ratio(net);
2412 struct tipc_link *bcl = tn->bcl;
2417 tipc_bcast_lock(net);
2419 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2420 NLM_F_MULTI, TIPC_NL_LINK_GET);
2422 tipc_bcast_unlock(net);
2426 attrs = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK);
2430 /* The broadcast link is always up */
2431 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2434 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
2436 if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
2438 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, 0))
2440 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, 0))
2443 prop = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK_PROP);
2446 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->window))
2448 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_BROADCAST, bc_mode))
2450 if (bc_mode & BCLINK_MODE_SEL)
2451 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_BROADCAST_RATIO,
2454 nla_nest_end(msg->skb, prop);
2456 err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
2460 tipc_bcast_unlock(net);
2461 nla_nest_end(msg->skb, attrs);
2462 genlmsg_end(msg->skb, hdr);
2467 nla_nest_cancel(msg->skb, prop);
2469 nla_nest_cancel(msg->skb, attrs);
2471 tipc_bcast_unlock(net);
2472 genlmsg_cancel(msg->skb, hdr);
2477 void tipc_link_set_tolerance(struct tipc_link *l, u32 tol,
2478 struct sk_buff_head *xmitq)
2482 l->bc_rcvlink->tolerance = tol;
2484 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, tol, 0, xmitq);
2487 void tipc_link_set_prio(struct tipc_link *l, u32 prio,
2488 struct sk_buff_head *xmitq)
2491 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, prio, xmitq);
2494 void tipc_link_set_abort_limit(struct tipc_link *l, u32 limit)
2496 l->abort_limit = limit;
2499 char *tipc_link_name_ext(struct tipc_link *l, char *buf)
2502 scnprintf(buf, TIPC_MAX_LINK_NAME, "null");
2503 else if (link_is_bc_sndlink(l))
2504 scnprintf(buf, TIPC_MAX_LINK_NAME, "broadcast-sender");
2505 else if (link_is_bc_rcvlink(l))
2506 scnprintf(buf, TIPC_MAX_LINK_NAME,
2507 "broadcast-receiver, peer %x", l->addr);
2509 memcpy(buf, l->name, TIPC_MAX_LINK_NAME);
2515 * tipc_link_dump - dump TIPC link data
2516 * @l: tipc link to be dumped
2517 * @dqueues: bitmask to decide if any link queue to be dumped?
2518 * - TIPC_DUMP_NONE: don't dump link queues
2519 * - TIPC_DUMP_TRANSMQ: dump link transmq queue
2520 * - TIPC_DUMP_BACKLOGQ: dump link backlog queue
2521 * - TIPC_DUMP_DEFERDQ: dump link deferd queue
2522 * - TIPC_DUMP_INPUTQ: dump link input queue
2523 * - TIPC_DUMP_WAKEUP: dump link wakeup queue
2524 * - TIPC_DUMP_ALL: dump all the link queues above
2525 * @buf: returned buffer of dump data in format
2527 int tipc_link_dump(struct tipc_link *l, u16 dqueues, char *buf)
2530 size_t sz = (dqueues) ? LINK_LMAX : LINK_LMIN;
2531 struct sk_buff_head *list;
2532 struct sk_buff *hskb, *tskb;
2536 i += scnprintf(buf, sz, "link data: (null)\n");
2540 i += scnprintf(buf, sz, "link data: %x", l->addr);
2541 i += scnprintf(buf + i, sz - i, " %x", l->state);
2542 i += scnprintf(buf + i, sz - i, " %u", l->in_session);
2543 i += scnprintf(buf + i, sz - i, " %u", l->session);
2544 i += scnprintf(buf + i, sz - i, " %u", l->peer_session);
2545 i += scnprintf(buf + i, sz - i, " %u", l->snd_nxt);
2546 i += scnprintf(buf + i, sz - i, " %u", l->rcv_nxt);
2547 i += scnprintf(buf + i, sz - i, " %u", l->snd_nxt_state);
2548 i += scnprintf(buf + i, sz - i, " %u", l->rcv_nxt_state);
2549 i += scnprintf(buf + i, sz - i, " %x", l->peer_caps);
2550 i += scnprintf(buf + i, sz - i, " %u", l->silent_intv_cnt);
2551 i += scnprintf(buf + i, sz - i, " %u", l->rst_cnt);
2552 i += scnprintf(buf + i, sz - i, " %u", l->prev_from);
2553 i += scnprintf(buf + i, sz - i, " %u", l->stale_cnt);
2554 i += scnprintf(buf + i, sz - i, " %u", l->acked);
2557 len = skb_queue_len(list);
2558 hskb = skb_peek(list);
2559 tskb = skb_peek_tail(list);
2560 i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2561 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2562 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2565 len = skb_queue_len(list);
2566 hskb = skb_peek(list);
2567 tskb = skb_peek_tail(list);
2568 i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2569 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2570 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2572 list = &l->backlogq;
2573 len = skb_queue_len(list);
2574 hskb = skb_peek(list);
2575 tskb = skb_peek_tail(list);
2576 i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2577 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2578 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2581 len = skb_queue_len(list);
2582 hskb = skb_peek(list);
2583 tskb = skb_peek_tail(list);
2584 i += scnprintf(buf + i, sz - i, " | %u %u %u\n", len,
2585 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2586 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2588 if (dqueues & TIPC_DUMP_TRANSMQ) {
2589 i += scnprintf(buf + i, sz - i, "transmq: ");
2590 i += tipc_list_dump(&l->transmq, false, buf + i);
2592 if (dqueues & TIPC_DUMP_BACKLOGQ) {
2593 i += scnprintf(buf + i, sz - i,
2594 "backlogq: <%u %u %u %u %u>, ",
2595 l->backlog[TIPC_LOW_IMPORTANCE].len,
2596 l->backlog[TIPC_MEDIUM_IMPORTANCE].len,
2597 l->backlog[TIPC_HIGH_IMPORTANCE].len,
2598 l->backlog[TIPC_CRITICAL_IMPORTANCE].len,
2599 l->backlog[TIPC_SYSTEM_IMPORTANCE].len);
2600 i += tipc_list_dump(&l->backlogq, false, buf + i);
2602 if (dqueues & TIPC_DUMP_DEFERDQ) {
2603 i += scnprintf(buf + i, sz - i, "deferdq: ");
2604 i += tipc_list_dump(&l->deferdq, false, buf + i);
2606 if (dqueues & TIPC_DUMP_INPUTQ) {
2607 i += scnprintf(buf + i, sz - i, "inputq: ");
2608 i += tipc_list_dump(l->inputq, false, buf + i);
2610 if (dqueues & TIPC_DUMP_WAKEUP) {
2611 i += scnprintf(buf + i, sz - i, "wakeup: ");
2612 i += tipc_list_dump(&l->wakeupq, false, buf + i);