Loading...
1/*
2 * net/tipc/link.c: TIPC link code
3 *
4 * Copyright (c) 1996-2007, 2012-2016, Ericsson AB
5 * Copyright (c) 2004-2007, 2010-2013, Wind River Systems
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions are met:
10 *
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the names of the copyright holders nor the names of its
17 * contributors may be used to endorse or promote products derived from
18 * this software without specific prior written permission.
19 *
20 * Alternatively, this software may be distributed under the terms of the
21 * GNU General Public License ("GPL") version 2 as published by the Free
22 * Software Foundation.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34 * POSSIBILITY OF SUCH DAMAGE.
35 */
36
37#include "core.h"
38#include "subscr.h"
39#include "link.h"
40#include "bcast.h"
41#include "socket.h"
42#include "name_distr.h"
43#include "discover.h"
44#include "netlink.h"
45#include "monitor.h"
46#include "trace.h"
47#include "crypto.h"
48
49#include <linux/pkt_sched.h>
50
51struct tipc_stats {
52 u32 sent_pkts;
53 u32 recv_pkts;
54 u32 sent_states;
55 u32 recv_states;
56 u32 sent_probes;
57 u32 recv_probes;
58 u32 sent_nacks;
59 u32 recv_nacks;
60 u32 sent_acks;
61 u32 sent_bundled;
62 u32 sent_bundles;
63 u32 recv_bundled;
64 u32 recv_bundles;
65 u32 retransmitted;
66 u32 sent_fragmented;
67 u32 sent_fragments;
68 u32 recv_fragmented;
69 u32 recv_fragments;
70 u32 link_congs; /* # port sends blocked by congestion */
71 u32 deferred_recv;
72 u32 duplicates;
73 u32 max_queue_sz; /* send queue size high water mark */
74 u32 accu_queue_sz; /* used for send queue size profiling */
75 u32 queue_sz_counts; /* used for send queue size profiling */
76 u32 msg_length_counts; /* used for message length profiling */
77 u32 msg_lengths_total; /* used for message length profiling */
78 u32 msg_length_profile[7]; /* used for msg. length profiling */
79};
80
81/**
82 * struct tipc_link - TIPC link data structure
83 * @addr: network address of link's peer node
84 * @name: link name character string
85 * @net: pointer to namespace struct
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 * @abort_limit: # of unacknowledged continuity probes needed to reset link
91 * @state: current state of link FSM
92 * @peer_caps: bitmap describing capabilities of peer node
93 * @silent_intv_cnt: # of timer intervals without any reception from peer
94 * @priority: current link priority
95 * @net_plane: current link network plane ('A' through 'H')
96 * @mon_state: cookie with information needed by link monitor
97 * @mtu: current maximum packet size for this link
98 * @advertised_mtu: advertised own mtu when link is being established
99 * @backlogq: queue for messages waiting to be sent
100 * @ackers: # of peers that needs to ack each packet before it can be released
101 * @acked: # last packet acked by a certain peer. Used for broadcast.
102 * @rcv_nxt: next sequence number to expect for inbound messages
103 * @inputq: buffer queue for messages to be delivered upwards
104 * @namedq: buffer queue for name table messages to be delivered upwards
105 * @wakeupq: linked list of wakeup msgs waiting for link congestion to abate
106 * @reasm_buf: head of partially reassembled inbound message fragments
107 * @stats: collects statistics regarding link activity
108 * @session: session to be used by link
109 * @snd_nxt_state: next send seq number
110 * @rcv_nxt_state: next rcv seq number
111 * @in_session: have received ACTIVATE_MSG from peer
112 * @active: link is active
113 * @if_name: associated interface name
114 * @rst_cnt: link reset counter
115 * @drop_point: seq number for failover handling (FIXME)
116 * @failover_reasm_skb: saved failover msg ptr (FIXME)
117 * @failover_deferdq: deferred message queue for failover processing (FIXME)
118 * @transmq: the link's transmit queue
119 * @backlog: link's backlog by priority (importance)
120 * @snd_nxt: next sequence number to be used
121 * @rcv_unacked: # messages read by user, but not yet acked back to peer
122 * @deferdq: deferred receive queue
123 * @window: sliding window size for congestion handling
124 * @min_win: minimal send window to be used by link
125 * @ssthresh: slow start threshold for congestion handling
126 * @max_win: maximal send window to be used by link
127 * @cong_acks: congestion acks for congestion avoidance (FIXME)
128 * @checkpoint: seq number for congestion window size handling
129 * @reasm_tnlmsg: fragmentation/reassembly area for tunnel protocol message
130 * @last_gap: last gap ack blocks for bcast (FIXME)
131 * @last_ga: ptr to gap ack blocks
132 * @bc_rcvlink: the peer specific link used for broadcast reception
133 * @bc_sndlink: the namespace global link used for broadcast sending
134 * @nack_state: bcast nack state
135 * @bc_peer_is_up: peer has acked the bcast init msg
136 */
137struct tipc_link {
138 u32 addr;
139 char name[TIPC_MAX_LINK_NAME];
140 struct net *net;
141
142 /* Management and link supervision data */
143 u16 peer_session;
144 u16 session;
145 u16 snd_nxt_state;
146 u16 rcv_nxt_state;
147 u32 peer_bearer_id;
148 u32 bearer_id;
149 u32 tolerance;
150 u32 abort_limit;
151 u32 state;
152 u16 peer_caps;
153 bool in_session;
154 bool active;
155 u32 silent_intv_cnt;
156 char if_name[TIPC_MAX_IF_NAME];
157 u32 priority;
158 char net_plane;
159 struct tipc_mon_state mon_state;
160 u16 rst_cnt;
161
162 /* Failover/synch */
163 u16 drop_point;
164 struct sk_buff *failover_reasm_skb;
165 struct sk_buff_head failover_deferdq;
166
167 /* Max packet negotiation */
168 u16 mtu;
169 u16 advertised_mtu;
170
171 /* Sending */
172 struct sk_buff_head transmq;
173 struct sk_buff_head backlogq;
174 struct {
175 u16 len;
176 u16 limit;
177 struct sk_buff *target_bskb;
178 } backlog[5];
179 u16 snd_nxt;
180
181 /* Reception */
182 u16 rcv_nxt;
183 u32 rcv_unacked;
184 struct sk_buff_head deferdq;
185 struct sk_buff_head *inputq;
186 struct sk_buff_head *namedq;
187
188 /* Congestion handling */
189 struct sk_buff_head wakeupq;
190 u16 window;
191 u16 min_win;
192 u16 ssthresh;
193 u16 max_win;
194 u16 cong_acks;
195 u16 checkpoint;
196
197 /* Fragmentation/reassembly */
198 struct sk_buff *reasm_buf;
199 struct sk_buff *reasm_tnlmsg;
200
201 /* Broadcast */
202 u16 ackers;
203 u16 acked;
204 u16 last_gap;
205 struct tipc_gap_ack_blks *last_ga;
206 struct tipc_link *bc_rcvlink;
207 struct tipc_link *bc_sndlink;
208 u8 nack_state;
209 bool bc_peer_is_up;
210
211 /* Statistics */
212 struct tipc_stats stats;
213};
214
215/*
216 * Error message prefixes
217 */
218static const char *link_co_err = "Link tunneling error, ";
219static const char *link_rst_msg = "Resetting link ";
220
221/* Send states for broadcast NACKs
222 */
223enum {
224 BC_NACK_SND_CONDITIONAL,
225 BC_NACK_SND_UNCONDITIONAL,
226 BC_NACK_SND_SUPPRESS,
227};
228
229#define TIPC_BC_RETR_LIM (jiffies + msecs_to_jiffies(10))
230#define TIPC_UC_RETR_TIME (jiffies + msecs_to_jiffies(1))
231
232/* Link FSM states:
233 */
234enum {
235 LINK_ESTABLISHED = 0xe,
236 LINK_ESTABLISHING = 0xe << 4,
237 LINK_RESET = 0x1 << 8,
238 LINK_RESETTING = 0x2 << 12,
239 LINK_PEER_RESET = 0xd << 16,
240 LINK_FAILINGOVER = 0xf << 20,
241 LINK_SYNCHING = 0xc << 24
242};
243
244static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
245 struct sk_buff_head *xmitq);
246static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
247 bool probe_reply, u16 rcvgap,
248 int tolerance, int priority,
249 struct sk_buff_head *xmitq);
250static void link_print(struct tipc_link *l, const char *str);
251static int tipc_link_build_nack_msg(struct tipc_link *l,
252 struct sk_buff_head *xmitq);
253static void tipc_link_build_bc_init_msg(struct tipc_link *l,
254 struct sk_buff_head *xmitq);
255static u8 __tipc_build_gap_ack_blks(struct tipc_gap_ack_blks *ga,
256 struct tipc_link *l, u8 start_index);
257static u16 tipc_build_gap_ack_blks(struct tipc_link *l, struct tipc_msg *hdr);
258static int tipc_link_advance_transmq(struct tipc_link *l, struct tipc_link *r,
259 u16 acked, u16 gap,
260 struct tipc_gap_ack_blks *ga,
261 struct sk_buff_head *xmitq,
262 bool *retransmitted, int *rc);
263static void tipc_link_update_cwin(struct tipc_link *l, int released,
264 bool retransmitted);
265/*
266 * Simple non-static link routines (i.e. referenced outside this file)
267 */
268bool tipc_link_is_up(struct tipc_link *l)
269{
270 return l->state & (LINK_ESTABLISHED | LINK_SYNCHING);
271}
272
273bool tipc_link_peer_is_down(struct tipc_link *l)
274{
275 return l->state == LINK_PEER_RESET;
276}
277
278bool tipc_link_is_reset(struct tipc_link *l)
279{
280 return l->state & (LINK_RESET | LINK_FAILINGOVER | LINK_ESTABLISHING);
281}
282
283bool tipc_link_is_establishing(struct tipc_link *l)
284{
285 return l->state == LINK_ESTABLISHING;
286}
287
288bool tipc_link_is_synching(struct tipc_link *l)
289{
290 return l->state == LINK_SYNCHING;
291}
292
293bool tipc_link_is_failingover(struct tipc_link *l)
294{
295 return l->state == LINK_FAILINGOVER;
296}
297
298bool tipc_link_is_blocked(struct tipc_link *l)
299{
300 return l->state & (LINK_RESETTING | LINK_PEER_RESET | LINK_FAILINGOVER);
301}
302
303static bool link_is_bc_sndlink(struct tipc_link *l)
304{
305 return !l->bc_sndlink;
306}
307
308static bool link_is_bc_rcvlink(struct tipc_link *l)
309{
310 return ((l->bc_rcvlink == l) && !link_is_bc_sndlink(l));
311}
312
313void tipc_link_set_active(struct tipc_link *l, bool active)
314{
315 l->active = active;
316}
317
318u32 tipc_link_id(struct tipc_link *l)
319{
320 return l->peer_bearer_id << 16 | l->bearer_id;
321}
322
323int tipc_link_min_win(struct tipc_link *l)
324{
325 return l->min_win;
326}
327
328int tipc_link_max_win(struct tipc_link *l)
329{
330 return l->max_win;
331}
332
333int tipc_link_prio(struct tipc_link *l)
334{
335 return l->priority;
336}
337
338unsigned long tipc_link_tolerance(struct tipc_link *l)
339{
340 return l->tolerance;
341}
342
343struct sk_buff_head *tipc_link_inputq(struct tipc_link *l)
344{
345 return l->inputq;
346}
347
348char tipc_link_plane(struct tipc_link *l)
349{
350 return l->net_plane;
351}
352
353struct net *tipc_link_net(struct tipc_link *l)
354{
355 return l->net;
356}
357
358void tipc_link_update_caps(struct tipc_link *l, u16 capabilities)
359{
360 l->peer_caps = capabilities;
361}
362
363void tipc_link_add_bc_peer(struct tipc_link *snd_l,
364 struct tipc_link *uc_l,
365 struct sk_buff_head *xmitq)
366{
367 struct tipc_link *rcv_l = uc_l->bc_rcvlink;
368
369 snd_l->ackers++;
370 rcv_l->acked = snd_l->snd_nxt - 1;
371 snd_l->state = LINK_ESTABLISHED;
372 tipc_link_build_bc_init_msg(uc_l, xmitq);
373}
374
375void tipc_link_remove_bc_peer(struct tipc_link *snd_l,
376 struct tipc_link *rcv_l,
377 struct sk_buff_head *xmitq)
378{
379 u16 ack = snd_l->snd_nxt - 1;
380
381 snd_l->ackers--;
382 rcv_l->bc_peer_is_up = true;
383 rcv_l->state = LINK_ESTABLISHED;
384 tipc_link_bc_ack_rcv(rcv_l, ack, 0, NULL, xmitq, NULL);
385 trace_tipc_link_reset(rcv_l, TIPC_DUMP_ALL, "bclink removed!");
386 tipc_link_reset(rcv_l);
387 rcv_l->state = LINK_RESET;
388 if (!snd_l->ackers) {
389 trace_tipc_link_reset(snd_l, TIPC_DUMP_ALL, "zero ackers!");
390 tipc_link_reset(snd_l);
391 snd_l->state = LINK_RESET;
392 __skb_queue_purge(xmitq);
393 }
394}
395
396int tipc_link_bc_peers(struct tipc_link *l)
397{
398 return l->ackers;
399}
400
401static u16 link_bc_rcv_gap(struct tipc_link *l)
402{
403 struct sk_buff *skb = skb_peek(&l->deferdq);
404 u16 gap = 0;
405
406 if (more(l->snd_nxt, l->rcv_nxt))
407 gap = l->snd_nxt - l->rcv_nxt;
408 if (skb)
409 gap = buf_seqno(skb) - l->rcv_nxt;
410 return gap;
411}
412
413void tipc_link_set_mtu(struct tipc_link *l, int mtu)
414{
415 l->mtu = mtu;
416}
417
418int tipc_link_mtu(struct tipc_link *l)
419{
420 return l->mtu;
421}
422
423int tipc_link_mss(struct tipc_link *l)
424{
425#ifdef CONFIG_TIPC_CRYPTO
426 return l->mtu - INT_H_SIZE - EMSG_OVERHEAD;
427#else
428 return l->mtu - INT_H_SIZE;
429#endif
430}
431
432u16 tipc_link_rcv_nxt(struct tipc_link *l)
433{
434 return l->rcv_nxt;
435}
436
437u16 tipc_link_acked(struct tipc_link *l)
438{
439 return l->acked;
440}
441
442char *tipc_link_name(struct tipc_link *l)
443{
444 return l->name;
445}
446
447u32 tipc_link_state(struct tipc_link *l)
448{
449 return l->state;
450}
451
452/**
453 * tipc_link_create - create a new link
454 * @net: pointer to associated network namespace
455 * @if_name: associated interface name
456 * @bearer_id: id (index) of associated bearer
457 * @tolerance: link tolerance to be used by link
458 * @net_plane: network plane (A,B,c..) this link belongs to
459 * @mtu: mtu to be advertised by link
460 * @priority: priority to be used by link
461 * @min_win: minimal send window to be used by link
462 * @max_win: maximal send window to be used by link
463 * @session: session to be used by link
464 * @peer: node id of peer node
465 * @peer_caps: bitmap describing peer node capabilities
466 * @bc_sndlink: the namespace global link used for broadcast sending
467 * @bc_rcvlink: the peer specific link used for broadcast reception
468 * @inputq: queue to put messages ready for delivery
469 * @namedq: queue to put binding table update messages ready for delivery
470 * @link: return value, pointer to put the created link
471 * @self: local unicast link id
472 * @peer_id: 128-bit ID of peer
473 *
474 * Return: true if link was created, otherwise false
475 */
476bool tipc_link_create(struct net *net, char *if_name, int bearer_id,
477 int tolerance, char net_plane, u32 mtu, int priority,
478 u32 min_win, u32 max_win, u32 session, u32 self,
479 u32 peer, u8 *peer_id, u16 peer_caps,
480 struct tipc_link *bc_sndlink,
481 struct tipc_link *bc_rcvlink,
482 struct sk_buff_head *inputq,
483 struct sk_buff_head *namedq,
484 struct tipc_link **link)
485{
486 char peer_str[NODE_ID_STR_LEN] = {0,};
487 char self_str[NODE_ID_STR_LEN] = {0,};
488 struct tipc_link *l;
489
490 l = kzalloc(sizeof(*l), GFP_ATOMIC);
491 if (!l)
492 return false;
493 *link = l;
494 l->session = session;
495
496 /* Set link name for unicast links only */
497 if (peer_id) {
498 tipc_nodeid2string(self_str, tipc_own_id(net));
499 if (strlen(self_str) > 16)
500 sprintf(self_str, "%x", self);
501 tipc_nodeid2string(peer_str, peer_id);
502 if (strlen(peer_str) > 16)
503 sprintf(peer_str, "%x", peer);
504 }
505 /* Peer i/f name will be completed by reset/activate message */
506 snprintf(l->name, sizeof(l->name), "%s:%s-%s:unknown",
507 self_str, if_name, peer_str);
508
509 strcpy(l->if_name, if_name);
510 l->addr = peer;
511 l->peer_caps = peer_caps;
512 l->net = net;
513 l->in_session = false;
514 l->bearer_id = bearer_id;
515 l->tolerance = tolerance;
516 if (bc_rcvlink)
517 bc_rcvlink->tolerance = tolerance;
518 l->net_plane = net_plane;
519 l->advertised_mtu = mtu;
520 l->mtu = mtu;
521 l->priority = priority;
522 tipc_link_set_queue_limits(l, min_win, max_win);
523 l->ackers = 1;
524 l->bc_sndlink = bc_sndlink;
525 l->bc_rcvlink = bc_rcvlink;
526 l->inputq = inputq;
527 l->namedq = namedq;
528 l->state = LINK_RESETTING;
529 __skb_queue_head_init(&l->transmq);
530 __skb_queue_head_init(&l->backlogq);
531 __skb_queue_head_init(&l->deferdq);
532 __skb_queue_head_init(&l->failover_deferdq);
533 skb_queue_head_init(&l->wakeupq);
534 skb_queue_head_init(l->inputq);
535 return true;
536}
537
538/**
539 * tipc_link_bc_create - create new link to be used for broadcast
540 * @net: pointer to associated network namespace
541 * @mtu: mtu to be used initially if no peers
542 * @min_win: minimal send window to be used by link
543 * @max_win: maximal send window to be used by link
544 * @inputq: queue to put messages ready for delivery
545 * @namedq: queue to put binding table update messages ready for delivery
546 * @link: return value, pointer to put the created link
547 * @ownnode: identity of own node
548 * @peer: node id of peer node
549 * @peer_id: 128-bit ID of peer
550 * @peer_caps: bitmap describing peer node capabilities
551 * @bc_sndlink: the namespace global link used for broadcast sending
552 *
553 * Return: true if link was created, otherwise false
554 */
555bool tipc_link_bc_create(struct net *net, u32 ownnode, u32 peer, u8 *peer_id,
556 int mtu, u32 min_win, u32 max_win, u16 peer_caps,
557 struct sk_buff_head *inputq,
558 struct sk_buff_head *namedq,
559 struct tipc_link *bc_sndlink,
560 struct tipc_link **link)
561{
562 struct tipc_link *l;
563
564 if (!tipc_link_create(net, "", MAX_BEARERS, 0, 'Z', mtu, 0, min_win,
565 max_win, 0, ownnode, peer, NULL, peer_caps,
566 bc_sndlink, NULL, inputq, namedq, link))
567 return false;
568
569 l = *link;
570 if (peer_id) {
571 char peer_str[NODE_ID_STR_LEN] = {0,};
572
573 tipc_nodeid2string(peer_str, peer_id);
574 if (strlen(peer_str) > 16)
575 sprintf(peer_str, "%x", peer);
576 /* Broadcast receiver link name: "broadcast-link:<peer>" */
577 snprintf(l->name, sizeof(l->name), "%s:%s", tipc_bclink_name,
578 peer_str);
579 } else {
580 strcpy(l->name, tipc_bclink_name);
581 }
582 trace_tipc_link_reset(l, TIPC_DUMP_ALL, "bclink created!");
583 tipc_link_reset(l);
584 l->state = LINK_RESET;
585 l->ackers = 0;
586 l->bc_rcvlink = l;
587
588 /* Broadcast send link is always up */
589 if (link_is_bc_sndlink(l))
590 l->state = LINK_ESTABLISHED;
591
592 /* Disable replicast if even a single peer doesn't support it */
593 if (link_is_bc_rcvlink(l) && !(peer_caps & TIPC_BCAST_RCAST))
594 tipc_bcast_toggle_rcast(net, false);
595
596 return true;
597}
598
599/**
600 * tipc_link_fsm_evt - link finite state machine
601 * @l: pointer to link
602 * @evt: state machine event to be processed
603 */
604int tipc_link_fsm_evt(struct tipc_link *l, int evt)
605{
606 int rc = 0;
607 int old_state = l->state;
608
609 switch (l->state) {
610 case LINK_RESETTING:
611 switch (evt) {
612 case LINK_PEER_RESET_EVT:
613 l->state = LINK_PEER_RESET;
614 break;
615 case LINK_RESET_EVT:
616 l->state = LINK_RESET;
617 break;
618 case LINK_FAILURE_EVT:
619 case LINK_FAILOVER_BEGIN_EVT:
620 case LINK_ESTABLISH_EVT:
621 case LINK_FAILOVER_END_EVT:
622 case LINK_SYNCH_BEGIN_EVT:
623 case LINK_SYNCH_END_EVT:
624 default:
625 goto illegal_evt;
626 }
627 break;
628 case LINK_RESET:
629 switch (evt) {
630 case LINK_PEER_RESET_EVT:
631 l->state = LINK_ESTABLISHING;
632 break;
633 case LINK_FAILOVER_BEGIN_EVT:
634 l->state = LINK_FAILINGOVER;
635 break;
636 case LINK_FAILURE_EVT:
637 case LINK_RESET_EVT:
638 case LINK_ESTABLISH_EVT:
639 case LINK_FAILOVER_END_EVT:
640 break;
641 case LINK_SYNCH_BEGIN_EVT:
642 case LINK_SYNCH_END_EVT:
643 default:
644 goto illegal_evt;
645 }
646 break;
647 case LINK_PEER_RESET:
648 switch (evt) {
649 case LINK_RESET_EVT:
650 l->state = LINK_ESTABLISHING;
651 break;
652 case LINK_PEER_RESET_EVT:
653 case LINK_ESTABLISH_EVT:
654 case LINK_FAILURE_EVT:
655 break;
656 case LINK_SYNCH_BEGIN_EVT:
657 case LINK_SYNCH_END_EVT:
658 case LINK_FAILOVER_BEGIN_EVT:
659 case LINK_FAILOVER_END_EVT:
660 default:
661 goto illegal_evt;
662 }
663 break;
664 case LINK_FAILINGOVER:
665 switch (evt) {
666 case LINK_FAILOVER_END_EVT:
667 l->state = LINK_RESET;
668 break;
669 case LINK_PEER_RESET_EVT:
670 case LINK_RESET_EVT:
671 case LINK_ESTABLISH_EVT:
672 case LINK_FAILURE_EVT:
673 break;
674 case LINK_FAILOVER_BEGIN_EVT:
675 case LINK_SYNCH_BEGIN_EVT:
676 case LINK_SYNCH_END_EVT:
677 default:
678 goto illegal_evt;
679 }
680 break;
681 case LINK_ESTABLISHING:
682 switch (evt) {
683 case LINK_ESTABLISH_EVT:
684 l->state = LINK_ESTABLISHED;
685 break;
686 case LINK_FAILOVER_BEGIN_EVT:
687 l->state = LINK_FAILINGOVER;
688 break;
689 case LINK_RESET_EVT:
690 l->state = LINK_RESET;
691 break;
692 case LINK_FAILURE_EVT:
693 case LINK_PEER_RESET_EVT:
694 case LINK_SYNCH_BEGIN_EVT:
695 case LINK_FAILOVER_END_EVT:
696 break;
697 case LINK_SYNCH_END_EVT:
698 default:
699 goto illegal_evt;
700 }
701 break;
702 case LINK_ESTABLISHED:
703 switch (evt) {
704 case LINK_PEER_RESET_EVT:
705 l->state = LINK_PEER_RESET;
706 rc |= TIPC_LINK_DOWN_EVT;
707 break;
708 case LINK_FAILURE_EVT:
709 l->state = LINK_RESETTING;
710 rc |= TIPC_LINK_DOWN_EVT;
711 break;
712 case LINK_RESET_EVT:
713 l->state = LINK_RESET;
714 break;
715 case LINK_ESTABLISH_EVT:
716 case LINK_SYNCH_END_EVT:
717 break;
718 case LINK_SYNCH_BEGIN_EVT:
719 l->state = LINK_SYNCHING;
720 break;
721 case LINK_FAILOVER_BEGIN_EVT:
722 case LINK_FAILOVER_END_EVT:
723 default:
724 goto illegal_evt;
725 }
726 break;
727 case LINK_SYNCHING:
728 switch (evt) {
729 case LINK_PEER_RESET_EVT:
730 l->state = LINK_PEER_RESET;
731 rc |= TIPC_LINK_DOWN_EVT;
732 break;
733 case LINK_FAILURE_EVT:
734 l->state = LINK_RESETTING;
735 rc |= TIPC_LINK_DOWN_EVT;
736 break;
737 case LINK_RESET_EVT:
738 l->state = LINK_RESET;
739 break;
740 case LINK_ESTABLISH_EVT:
741 case LINK_SYNCH_BEGIN_EVT:
742 break;
743 case LINK_SYNCH_END_EVT:
744 l->state = LINK_ESTABLISHED;
745 break;
746 case LINK_FAILOVER_BEGIN_EVT:
747 case LINK_FAILOVER_END_EVT:
748 default:
749 goto illegal_evt;
750 }
751 break;
752 default:
753 pr_err("Unknown FSM state %x in %s\n", l->state, l->name);
754 }
755 trace_tipc_link_fsm(l->name, old_state, l->state, evt);
756 return rc;
757illegal_evt:
758 pr_err("Illegal FSM event %x in state %x on link %s\n",
759 evt, l->state, l->name);
760 trace_tipc_link_fsm(l->name, old_state, l->state, evt);
761 return rc;
762}
763
764/* link_profile_stats - update statistical profiling of traffic
765 */
766static void link_profile_stats(struct tipc_link *l)
767{
768 struct sk_buff *skb;
769 struct tipc_msg *msg;
770 int length;
771
772 /* Update counters used in statistical profiling of send traffic */
773 l->stats.accu_queue_sz += skb_queue_len(&l->transmq);
774 l->stats.queue_sz_counts++;
775
776 skb = skb_peek(&l->transmq);
777 if (!skb)
778 return;
779 msg = buf_msg(skb);
780 length = msg_size(msg);
781
782 if (msg_user(msg) == MSG_FRAGMENTER) {
783 if (msg_type(msg) != FIRST_FRAGMENT)
784 return;
785 length = msg_size(msg_inner_hdr(msg));
786 }
787 l->stats.msg_lengths_total += length;
788 l->stats.msg_length_counts++;
789 if (length <= 64)
790 l->stats.msg_length_profile[0]++;
791 else if (length <= 256)
792 l->stats.msg_length_profile[1]++;
793 else if (length <= 1024)
794 l->stats.msg_length_profile[2]++;
795 else if (length <= 4096)
796 l->stats.msg_length_profile[3]++;
797 else if (length <= 16384)
798 l->stats.msg_length_profile[4]++;
799 else if (length <= 32768)
800 l->stats.msg_length_profile[5]++;
801 else
802 l->stats.msg_length_profile[6]++;
803}
804
805/**
806 * tipc_link_too_silent - check if link is "too silent"
807 * @l: tipc link to be checked
808 *
809 * Return: true if the link 'silent_intv_cnt' is about to reach the
810 * 'abort_limit' value, otherwise false
811 */
812bool tipc_link_too_silent(struct tipc_link *l)
813{
814 return (l->silent_intv_cnt + 2 > l->abort_limit);
815}
816
817/* tipc_link_timeout - perform periodic task as instructed from node timeout
818 */
819int tipc_link_timeout(struct tipc_link *l, struct sk_buff_head *xmitq)
820{
821 int mtyp = 0;
822 int rc = 0;
823 bool state = false;
824 bool probe = false;
825 bool setup = false;
826 u16 bc_snt = l->bc_sndlink->snd_nxt - 1;
827 u16 bc_acked = l->bc_rcvlink->acked;
828 struct tipc_mon_state *mstate = &l->mon_state;
829
830 trace_tipc_link_timeout(l, TIPC_DUMP_NONE, " ");
831 trace_tipc_link_too_silent(l, TIPC_DUMP_ALL, " ");
832 switch (l->state) {
833 case LINK_ESTABLISHED:
834 case LINK_SYNCHING:
835 mtyp = STATE_MSG;
836 link_profile_stats(l);
837 tipc_mon_get_state(l->net, l->addr, mstate, l->bearer_id);
838 if (mstate->reset || (l->silent_intv_cnt > l->abort_limit))
839 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
840 state = bc_acked != bc_snt;
841 state |= l->bc_rcvlink->rcv_unacked;
842 state |= l->rcv_unacked;
843 state |= !skb_queue_empty(&l->transmq);
844 probe = mstate->probing;
845 probe |= l->silent_intv_cnt;
846 if (probe || mstate->monitoring)
847 l->silent_intv_cnt++;
848 probe |= !skb_queue_empty(&l->deferdq);
849 if (l->snd_nxt == l->checkpoint) {
850 tipc_link_update_cwin(l, 0, 0);
851 probe = true;
852 }
853 l->checkpoint = l->snd_nxt;
854 break;
855 case LINK_RESET:
856 setup = l->rst_cnt++ <= 4;
857 setup |= !(l->rst_cnt % 16);
858 mtyp = RESET_MSG;
859 break;
860 case LINK_ESTABLISHING:
861 setup = true;
862 mtyp = ACTIVATE_MSG;
863 break;
864 case LINK_PEER_RESET:
865 case LINK_RESETTING:
866 case LINK_FAILINGOVER:
867 break;
868 default:
869 break;
870 }
871
872 if (state || probe || setup)
873 tipc_link_build_proto_msg(l, mtyp, probe, 0, 0, 0, 0, xmitq);
874
875 return rc;
876}
877
878/**
879 * link_schedule_user - schedule a message sender for wakeup after congestion
880 * @l: congested link
881 * @hdr: header of message that is being sent
882 * Create pseudo msg to send back to user when congestion abates
883 */
884static int link_schedule_user(struct tipc_link *l, struct tipc_msg *hdr)
885{
886 u32 dnode = tipc_own_addr(l->net);
887 u32 dport = msg_origport(hdr);
888 struct sk_buff *skb;
889
890 /* Create and schedule wakeup pseudo message */
891 skb = tipc_msg_create(SOCK_WAKEUP, 0, INT_H_SIZE, 0,
892 dnode, l->addr, dport, 0, 0);
893 if (!skb)
894 return -ENOBUFS;
895 msg_set_dest_droppable(buf_msg(skb), true);
896 TIPC_SKB_CB(skb)->chain_imp = msg_importance(hdr);
897 skb_queue_tail(&l->wakeupq, skb);
898 l->stats.link_congs++;
899 trace_tipc_link_conges(l, TIPC_DUMP_ALL, "wakeup scheduled!");
900 return -ELINKCONG;
901}
902
903/**
904 * link_prepare_wakeup - prepare users for wakeup after congestion
905 * @l: congested link
906 * Wake up a number of waiting users, as permitted by available space
907 * in the send queue
908 */
909static void link_prepare_wakeup(struct tipc_link *l)
910{
911 struct sk_buff_head *wakeupq = &l->wakeupq;
912 struct sk_buff_head *inputq = l->inputq;
913 struct sk_buff *skb, *tmp;
914 struct sk_buff_head tmpq;
915 int avail[5] = {0,};
916 int imp = 0;
917
918 __skb_queue_head_init(&tmpq);
919
920 for (; imp <= TIPC_SYSTEM_IMPORTANCE; imp++)
921 avail[imp] = l->backlog[imp].limit - l->backlog[imp].len;
922
923 skb_queue_walk_safe(wakeupq, skb, tmp) {
924 imp = TIPC_SKB_CB(skb)->chain_imp;
925 if (avail[imp] <= 0)
926 continue;
927 avail[imp]--;
928 __skb_unlink(skb, wakeupq);
929 __skb_queue_tail(&tmpq, skb);
930 }
931
932 spin_lock_bh(&inputq->lock);
933 skb_queue_splice_tail(&tmpq, inputq);
934 spin_unlock_bh(&inputq->lock);
935
936}
937
938/**
939 * tipc_link_set_skb_retransmit_time - set the time at which retransmission of
940 * the given skb should be next attempted
941 * @skb: skb to set a future retransmission time for
942 * @l: link the skb will be transmitted on
943 */
944static void tipc_link_set_skb_retransmit_time(struct sk_buff *skb,
945 struct tipc_link *l)
946{
947 if (link_is_bc_sndlink(l))
948 TIPC_SKB_CB(skb)->nxt_retr = TIPC_BC_RETR_LIM;
949 else
950 TIPC_SKB_CB(skb)->nxt_retr = TIPC_UC_RETR_TIME;
951}
952
953void tipc_link_reset(struct tipc_link *l)
954{
955 struct sk_buff_head list;
956 u32 imp;
957
958 __skb_queue_head_init(&list);
959
960 l->in_session = false;
961 /* Force re-synch of peer session number before establishing */
962 l->peer_session--;
963 l->session++;
964 l->mtu = l->advertised_mtu;
965
966 spin_lock_bh(&l->wakeupq.lock);
967 skb_queue_splice_init(&l->wakeupq, &list);
968 spin_unlock_bh(&l->wakeupq.lock);
969
970 spin_lock_bh(&l->inputq->lock);
971 skb_queue_splice_init(&list, l->inputq);
972 spin_unlock_bh(&l->inputq->lock);
973
974 __skb_queue_purge(&l->transmq);
975 __skb_queue_purge(&l->deferdq);
976 __skb_queue_purge(&l->backlogq);
977 __skb_queue_purge(&l->failover_deferdq);
978 for (imp = 0; imp <= TIPC_SYSTEM_IMPORTANCE; imp++) {
979 l->backlog[imp].len = 0;
980 l->backlog[imp].target_bskb = NULL;
981 }
982 kfree_skb(l->reasm_buf);
983 kfree_skb(l->reasm_tnlmsg);
984 kfree_skb(l->failover_reasm_skb);
985 l->reasm_buf = NULL;
986 l->reasm_tnlmsg = NULL;
987 l->failover_reasm_skb = NULL;
988 l->rcv_unacked = 0;
989 l->snd_nxt = 1;
990 l->rcv_nxt = 1;
991 l->snd_nxt_state = 1;
992 l->rcv_nxt_state = 1;
993 l->acked = 0;
994 l->last_gap = 0;
995 kfree(l->last_ga);
996 l->last_ga = NULL;
997 l->silent_intv_cnt = 0;
998 l->rst_cnt = 0;
999 l->bc_peer_is_up = false;
1000 memset(&l->mon_state, 0, sizeof(l->mon_state));
1001 tipc_link_reset_stats(l);
1002}
1003
1004/**
1005 * tipc_link_xmit(): enqueue buffer list according to queue situation
1006 * @l: link to use
1007 * @list: chain of buffers containing message
1008 * @xmitq: returned list of packets to be sent by caller
1009 *
1010 * Consumes the buffer chain.
1011 * Messages at TIPC_SYSTEM_IMPORTANCE are always accepted
1012 * Return: 0 if success, or errno: -ELINKCONG, -EMSGSIZE or -ENOBUFS
1013 */
1014int tipc_link_xmit(struct tipc_link *l, struct sk_buff_head *list,
1015 struct sk_buff_head *xmitq)
1016{
1017 struct sk_buff_head *backlogq = &l->backlogq;
1018 struct sk_buff_head *transmq = &l->transmq;
1019 struct sk_buff *skb, *_skb;
1020 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1021 u16 ack = l->rcv_nxt - 1;
1022 u16 seqno = l->snd_nxt;
1023 int pkt_cnt = skb_queue_len(list);
1024 unsigned int mss = tipc_link_mss(l);
1025 unsigned int cwin = l->window;
1026 unsigned int mtu = l->mtu;
1027 struct tipc_msg *hdr;
1028 bool new_bundle;
1029 int rc = 0;
1030 int imp;
1031
1032 if (pkt_cnt <= 0)
1033 return 0;
1034
1035 hdr = buf_msg(skb_peek(list));
1036 if (unlikely(msg_size(hdr) > mtu)) {
1037 pr_warn("Too large msg, purging xmit list %d %d %d %d %d!\n",
1038 skb_queue_len(list), msg_user(hdr),
1039 msg_type(hdr), msg_size(hdr), mtu);
1040 __skb_queue_purge(list);
1041 return -EMSGSIZE;
1042 }
1043
1044 imp = msg_importance(hdr);
1045 /* Allow oversubscription of one data msg per source at congestion */
1046 if (unlikely(l->backlog[imp].len >= l->backlog[imp].limit)) {
1047 if (imp == TIPC_SYSTEM_IMPORTANCE) {
1048 pr_warn("%s<%s>, link overflow", link_rst_msg, l->name);
1049 return -ENOBUFS;
1050 }
1051 rc = link_schedule_user(l, hdr);
1052 }
1053
1054 if (pkt_cnt > 1) {
1055 l->stats.sent_fragmented++;
1056 l->stats.sent_fragments += pkt_cnt;
1057 }
1058
1059 /* Prepare each packet for sending, and add to relevant queue: */
1060 while ((skb = __skb_dequeue(list))) {
1061 if (likely(skb_queue_len(transmq) < cwin)) {
1062 hdr = buf_msg(skb);
1063 msg_set_seqno(hdr, seqno);
1064 msg_set_ack(hdr, ack);
1065 msg_set_bcast_ack(hdr, bc_ack);
1066 _skb = skb_clone(skb, GFP_ATOMIC);
1067 if (!_skb) {
1068 kfree_skb(skb);
1069 __skb_queue_purge(list);
1070 return -ENOBUFS;
1071 }
1072 __skb_queue_tail(transmq, skb);
1073 tipc_link_set_skb_retransmit_time(skb, l);
1074 __skb_queue_tail(xmitq, _skb);
1075 TIPC_SKB_CB(skb)->ackers = l->ackers;
1076 l->rcv_unacked = 0;
1077 l->stats.sent_pkts++;
1078 seqno++;
1079 continue;
1080 }
1081 if (tipc_msg_try_bundle(l->backlog[imp].target_bskb, &skb,
1082 mss, l->addr, &new_bundle)) {
1083 if (skb) {
1084 /* Keep a ref. to the skb for next try */
1085 l->backlog[imp].target_bskb = skb;
1086 l->backlog[imp].len++;
1087 __skb_queue_tail(backlogq, skb);
1088 } else {
1089 if (new_bundle) {
1090 l->stats.sent_bundles++;
1091 l->stats.sent_bundled++;
1092 }
1093 l->stats.sent_bundled++;
1094 }
1095 continue;
1096 }
1097 l->backlog[imp].target_bskb = NULL;
1098 l->backlog[imp].len += (1 + skb_queue_len(list));
1099 __skb_queue_tail(backlogq, skb);
1100 skb_queue_splice_tail_init(list, backlogq);
1101 }
1102 l->snd_nxt = seqno;
1103 return rc;
1104}
1105
1106static void tipc_link_update_cwin(struct tipc_link *l, int released,
1107 bool retransmitted)
1108{
1109 int bklog_len = skb_queue_len(&l->backlogq);
1110 struct sk_buff_head *txq = &l->transmq;
1111 int txq_len = skb_queue_len(txq);
1112 u16 cwin = l->window;
1113
1114 /* Enter fast recovery */
1115 if (unlikely(retransmitted)) {
1116 l->ssthresh = max_t(u16, l->window / 2, 300);
1117 l->window = min_t(u16, l->ssthresh, l->window);
1118 return;
1119 }
1120 /* Enter slow start */
1121 if (unlikely(!released)) {
1122 l->ssthresh = max_t(u16, l->window / 2, 300);
1123 l->window = l->min_win;
1124 return;
1125 }
1126 /* Don't increase window if no pressure on the transmit queue */
1127 if (txq_len + bklog_len < cwin)
1128 return;
1129
1130 /* Don't increase window if there are holes the transmit queue */
1131 if (txq_len && l->snd_nxt - buf_seqno(skb_peek(txq)) != txq_len)
1132 return;
1133
1134 l->cong_acks += released;
1135
1136 /* Slow start */
1137 if (cwin <= l->ssthresh) {
1138 l->window = min_t(u16, cwin + released, l->max_win);
1139 return;
1140 }
1141 /* Congestion avoidance */
1142 if (l->cong_acks < cwin)
1143 return;
1144 l->window = min_t(u16, ++cwin, l->max_win);
1145 l->cong_acks = 0;
1146}
1147
1148static void tipc_link_advance_backlog(struct tipc_link *l,
1149 struct sk_buff_head *xmitq)
1150{
1151 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1152 struct sk_buff_head *txq = &l->transmq;
1153 struct sk_buff *skb, *_skb;
1154 u16 ack = l->rcv_nxt - 1;
1155 u16 seqno = l->snd_nxt;
1156 struct tipc_msg *hdr;
1157 u16 cwin = l->window;
1158 u32 imp;
1159
1160 while (skb_queue_len(txq) < cwin) {
1161 skb = skb_peek(&l->backlogq);
1162 if (!skb)
1163 break;
1164 _skb = skb_clone(skb, GFP_ATOMIC);
1165 if (!_skb)
1166 break;
1167 __skb_dequeue(&l->backlogq);
1168 hdr = buf_msg(skb);
1169 imp = msg_importance(hdr);
1170 l->backlog[imp].len--;
1171 if (unlikely(skb == l->backlog[imp].target_bskb))
1172 l->backlog[imp].target_bskb = NULL;
1173 __skb_queue_tail(&l->transmq, skb);
1174 tipc_link_set_skb_retransmit_time(skb, l);
1175
1176 __skb_queue_tail(xmitq, _skb);
1177 TIPC_SKB_CB(skb)->ackers = l->ackers;
1178 msg_set_seqno(hdr, seqno);
1179 msg_set_ack(hdr, ack);
1180 msg_set_bcast_ack(hdr, bc_ack);
1181 l->rcv_unacked = 0;
1182 l->stats.sent_pkts++;
1183 seqno++;
1184 }
1185 l->snd_nxt = seqno;
1186}
1187
1188/**
1189 * link_retransmit_failure() - Detect repeated retransmit failures
1190 * @l: tipc link sender
1191 * @r: tipc link receiver (= l in case of unicast)
1192 * @rc: returned code
1193 *
1194 * Return: true if the repeated retransmit failures happens, otherwise
1195 * false
1196 */
1197static bool link_retransmit_failure(struct tipc_link *l, struct tipc_link *r,
1198 int *rc)
1199{
1200 struct sk_buff *skb = skb_peek(&l->transmq);
1201 struct tipc_msg *hdr;
1202
1203 if (!skb)
1204 return false;
1205
1206 if (!TIPC_SKB_CB(skb)->retr_cnt)
1207 return false;
1208
1209 if (!time_after(jiffies, TIPC_SKB_CB(skb)->retr_stamp +
1210 msecs_to_jiffies(r->tolerance * 10)))
1211 return false;
1212
1213 hdr = buf_msg(skb);
1214 if (link_is_bc_sndlink(l) && !less(r->acked, msg_seqno(hdr)))
1215 return false;
1216
1217 pr_warn("Retransmission failure on link <%s>\n", l->name);
1218 link_print(l, "State of link ");
1219 pr_info("Failed msg: usr %u, typ %u, len %u, err %u\n",
1220 msg_user(hdr), msg_type(hdr), msg_size(hdr), msg_errcode(hdr));
1221 pr_info("sqno %u, prev: %x, dest: %x\n",
1222 msg_seqno(hdr), msg_prevnode(hdr), msg_destnode(hdr));
1223 pr_info("retr_stamp %d, retr_cnt %d\n",
1224 jiffies_to_msecs(TIPC_SKB_CB(skb)->retr_stamp),
1225 TIPC_SKB_CB(skb)->retr_cnt);
1226
1227 trace_tipc_list_dump(&l->transmq, true, "retrans failure!");
1228 trace_tipc_link_dump(l, TIPC_DUMP_NONE, "retrans failure!");
1229 trace_tipc_link_dump(r, TIPC_DUMP_NONE, "retrans failure!");
1230
1231 if (link_is_bc_sndlink(l)) {
1232 r->state = LINK_RESET;
1233 *rc |= TIPC_LINK_DOWN_EVT;
1234 } else {
1235 *rc |= tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1236 }
1237
1238 return true;
1239}
1240
1241/* tipc_data_input - deliver data and name distr msgs to upper layer
1242 *
1243 * Consumes buffer if message is of right type
1244 * Node lock must be held
1245 */
1246static bool tipc_data_input(struct tipc_link *l, struct sk_buff *skb,
1247 struct sk_buff_head *inputq)
1248{
1249 struct sk_buff_head *mc_inputq = l->bc_rcvlink->inputq;
1250 struct tipc_msg *hdr = buf_msg(skb);
1251
1252 switch (msg_user(hdr)) {
1253 case TIPC_LOW_IMPORTANCE:
1254 case TIPC_MEDIUM_IMPORTANCE:
1255 case TIPC_HIGH_IMPORTANCE:
1256 case TIPC_CRITICAL_IMPORTANCE:
1257 if (unlikely(msg_in_group(hdr) || msg_mcast(hdr))) {
1258 skb_queue_tail(mc_inputq, skb);
1259 return true;
1260 }
1261 fallthrough;
1262 case CONN_MANAGER:
1263 skb_queue_tail(inputq, skb);
1264 return true;
1265 case GROUP_PROTOCOL:
1266 skb_queue_tail(mc_inputq, skb);
1267 return true;
1268 case NAME_DISTRIBUTOR:
1269 l->bc_rcvlink->state = LINK_ESTABLISHED;
1270 skb_queue_tail(l->namedq, skb);
1271 return true;
1272 case MSG_BUNDLER:
1273 case TUNNEL_PROTOCOL:
1274 case MSG_FRAGMENTER:
1275 case BCAST_PROTOCOL:
1276 return false;
1277#ifdef CONFIG_TIPC_CRYPTO
1278 case MSG_CRYPTO:
1279 if (sysctl_tipc_key_exchange_enabled &&
1280 TIPC_SKB_CB(skb)->decrypted) {
1281 tipc_crypto_msg_rcv(l->net, skb);
1282 return true;
1283 }
1284 fallthrough;
1285#endif
1286 default:
1287 pr_warn("Dropping received illegal msg type\n");
1288 kfree_skb(skb);
1289 return true;
1290 }
1291}
1292
1293/* tipc_link_input - process packet that has passed link protocol check
1294 *
1295 * Consumes buffer
1296 */
1297static int tipc_link_input(struct tipc_link *l, struct sk_buff *skb,
1298 struct sk_buff_head *inputq,
1299 struct sk_buff **reasm_skb)
1300{
1301 struct tipc_msg *hdr = buf_msg(skb);
1302 struct sk_buff *iskb;
1303 struct sk_buff_head tmpq;
1304 int usr = msg_user(hdr);
1305 int pos = 0;
1306
1307 if (usr == MSG_BUNDLER) {
1308 skb_queue_head_init(&tmpq);
1309 l->stats.recv_bundles++;
1310 l->stats.recv_bundled += msg_msgcnt(hdr);
1311 while (tipc_msg_extract(skb, &iskb, &pos))
1312 tipc_data_input(l, iskb, &tmpq);
1313 tipc_skb_queue_splice_tail(&tmpq, inputq);
1314 return 0;
1315 } else if (usr == MSG_FRAGMENTER) {
1316 l->stats.recv_fragments++;
1317 if (tipc_buf_append(reasm_skb, &skb)) {
1318 l->stats.recv_fragmented++;
1319 tipc_data_input(l, skb, inputq);
1320 } else if (!*reasm_skb && !link_is_bc_rcvlink(l)) {
1321 pr_warn_ratelimited("Unable to build fragment list\n");
1322 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1323 }
1324 return 0;
1325 } else if (usr == BCAST_PROTOCOL) {
1326 tipc_bcast_lock(l->net);
1327 tipc_link_bc_init_rcv(l->bc_rcvlink, hdr);
1328 tipc_bcast_unlock(l->net);
1329 }
1330
1331 kfree_skb(skb);
1332 return 0;
1333}
1334
1335/* tipc_link_tnl_rcv() - receive TUNNEL_PROTOCOL message, drop or process the
1336 * inner message along with the ones in the old link's
1337 * deferdq
1338 * @l: tunnel link
1339 * @skb: TUNNEL_PROTOCOL message
1340 * @inputq: queue to put messages ready for delivery
1341 */
1342static int tipc_link_tnl_rcv(struct tipc_link *l, struct sk_buff *skb,
1343 struct sk_buff_head *inputq)
1344{
1345 struct sk_buff **reasm_skb = &l->failover_reasm_skb;
1346 struct sk_buff **reasm_tnlmsg = &l->reasm_tnlmsg;
1347 struct sk_buff_head *fdefq = &l->failover_deferdq;
1348 struct tipc_msg *hdr = buf_msg(skb);
1349 struct sk_buff *iskb;
1350 int ipos = 0;
1351 int rc = 0;
1352 u16 seqno;
1353
1354 if (msg_type(hdr) == SYNCH_MSG) {
1355 kfree_skb(skb);
1356 return 0;
1357 }
1358
1359 /* Not a fragment? */
1360 if (likely(!msg_nof_fragms(hdr))) {
1361 if (unlikely(!tipc_msg_extract(skb, &iskb, &ipos))) {
1362 pr_warn_ratelimited("Unable to extract msg, defq: %d\n",
1363 skb_queue_len(fdefq));
1364 return 0;
1365 }
1366 kfree_skb(skb);
1367 } else {
1368 /* Set fragment type for buf_append */
1369 if (msg_fragm_no(hdr) == 1)
1370 msg_set_type(hdr, FIRST_FRAGMENT);
1371 else if (msg_fragm_no(hdr) < msg_nof_fragms(hdr))
1372 msg_set_type(hdr, FRAGMENT);
1373 else
1374 msg_set_type(hdr, LAST_FRAGMENT);
1375
1376 if (!tipc_buf_append(reasm_tnlmsg, &skb)) {
1377 /* Successful but non-complete reassembly? */
1378 if (*reasm_tnlmsg || link_is_bc_rcvlink(l))
1379 return 0;
1380 pr_warn_ratelimited("Unable to reassemble tunnel msg\n");
1381 return tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
1382 }
1383 iskb = skb;
1384 }
1385
1386 do {
1387 seqno = buf_seqno(iskb);
1388 if (unlikely(less(seqno, l->drop_point))) {
1389 kfree_skb(iskb);
1390 continue;
1391 }
1392 if (unlikely(seqno != l->drop_point)) {
1393 __tipc_skb_queue_sorted(fdefq, seqno, iskb);
1394 continue;
1395 }
1396
1397 l->drop_point++;
1398 if (!tipc_data_input(l, iskb, inputq))
1399 rc |= tipc_link_input(l, iskb, inputq, reasm_skb);
1400 if (unlikely(rc))
1401 break;
1402 } while ((iskb = __tipc_skb_dequeue(fdefq, l->drop_point)));
1403
1404 return rc;
1405}
1406
1407/**
1408 * tipc_get_gap_ack_blks - get Gap ACK blocks from PROTOCOL/STATE_MSG
1409 * @ga: returned pointer to the Gap ACK blocks if any
1410 * @l: the tipc link
1411 * @hdr: the PROTOCOL/STATE_MSG header
1412 * @uc: desired Gap ACK blocks type, i.e. unicast (= 1) or broadcast (= 0)
1413 *
1414 * Return: the total Gap ACK blocks size
1415 */
1416u16 tipc_get_gap_ack_blks(struct tipc_gap_ack_blks **ga, struct tipc_link *l,
1417 struct tipc_msg *hdr, bool uc)
1418{
1419 struct tipc_gap_ack_blks *p;
1420 u16 sz = 0;
1421
1422 /* Does peer support the Gap ACK blocks feature? */
1423 if (l->peer_caps & TIPC_GAP_ACK_BLOCK) {
1424 p = (struct tipc_gap_ack_blks *)msg_data(hdr);
1425 sz = ntohs(p->len);
1426 /* Sanity check */
1427 if (sz == struct_size(p, gacks, size_add(p->ugack_cnt, p->bgack_cnt))) {
1428 /* Good, check if the desired type exists */
1429 if ((uc && p->ugack_cnt) || (!uc && p->bgack_cnt))
1430 goto ok;
1431 /* Backward compatible: peer might not support bc, but uc? */
1432 } else if (uc && sz == struct_size(p, gacks, p->ugack_cnt)) {
1433 if (p->ugack_cnt) {
1434 p->bgack_cnt = 0;
1435 goto ok;
1436 }
1437 }
1438 }
1439 /* Other cases: ignore! */
1440 p = NULL;
1441
1442ok:
1443 *ga = p;
1444 return sz;
1445}
1446
1447static u8 __tipc_build_gap_ack_blks(struct tipc_gap_ack_blks *ga,
1448 struct tipc_link *l, u8 start_index)
1449{
1450 struct tipc_gap_ack *gacks = &ga->gacks[start_index];
1451 struct sk_buff *skb = skb_peek(&l->deferdq);
1452 u16 expect, seqno = 0;
1453 u8 n = 0;
1454
1455 if (!skb)
1456 return 0;
1457
1458 expect = buf_seqno(skb);
1459 skb_queue_walk(&l->deferdq, skb) {
1460 seqno = buf_seqno(skb);
1461 if (unlikely(more(seqno, expect))) {
1462 gacks[n].ack = htons(expect - 1);
1463 gacks[n].gap = htons(seqno - expect);
1464 if (++n >= MAX_GAP_ACK_BLKS / 2) {
1465 pr_info_ratelimited("Gacks on %s: %d, ql: %d!\n",
1466 l->name, n,
1467 skb_queue_len(&l->deferdq));
1468 return n;
1469 }
1470 } else if (unlikely(less(seqno, expect))) {
1471 pr_warn("Unexpected skb in deferdq!\n");
1472 continue;
1473 }
1474 expect = seqno + 1;
1475 }
1476
1477 /* last block */
1478 gacks[n].ack = htons(seqno);
1479 gacks[n].gap = 0;
1480 n++;
1481 return n;
1482}
1483
1484/* tipc_build_gap_ack_blks - build Gap ACK blocks
1485 * @l: tipc unicast link
1486 * @hdr: the tipc message buffer to store the Gap ACK blocks after built
1487 *
1488 * The function builds Gap ACK blocks for both the unicast & broadcast receiver
1489 * links of a certain peer, the buffer after built has the network data format
1490 * as found at the struct tipc_gap_ack_blks definition.
1491 *
1492 * returns the actual allocated memory size
1493 */
1494static u16 tipc_build_gap_ack_blks(struct tipc_link *l, struct tipc_msg *hdr)
1495{
1496 struct tipc_link *bcl = l->bc_rcvlink;
1497 struct tipc_gap_ack_blks *ga;
1498 u16 len;
1499
1500 ga = (struct tipc_gap_ack_blks *)msg_data(hdr);
1501
1502 /* Start with broadcast link first */
1503 tipc_bcast_lock(bcl->net);
1504 msg_set_bcast_ack(hdr, bcl->rcv_nxt - 1);
1505 msg_set_bc_gap(hdr, link_bc_rcv_gap(bcl));
1506 ga->bgack_cnt = __tipc_build_gap_ack_blks(ga, bcl, 0);
1507 tipc_bcast_unlock(bcl->net);
1508
1509 /* Now for unicast link, but an explicit NACK only (???) */
1510 ga->ugack_cnt = (msg_seq_gap(hdr)) ?
1511 __tipc_build_gap_ack_blks(ga, l, ga->bgack_cnt) : 0;
1512
1513 /* Total len */
1514 len = struct_size(ga, gacks, size_add(ga->bgack_cnt, ga->ugack_cnt));
1515 ga->len = htons(len);
1516 return len;
1517}
1518
1519/* tipc_link_advance_transmq - advance TIPC link transmq queue by releasing
1520 * acked packets, also doing retransmissions if
1521 * gaps found
1522 * @l: tipc link with transmq queue to be advanced
1523 * @r: tipc link "receiver" i.e. in case of broadcast (= "l" if unicast)
1524 * @acked: seqno of last packet acked by peer without any gaps before
1525 * @gap: # of gap packets
1526 * @ga: buffer pointer to Gap ACK blocks from peer
1527 * @xmitq: queue for accumulating the retransmitted packets if any
1528 * @retransmitted: returned boolean value if a retransmission is really issued
1529 * @rc: returned code e.g. TIPC_LINK_DOWN_EVT if a repeated retransmit failures
1530 * happens (- unlikely case)
1531 *
1532 * Return: the number of packets released from the link transmq
1533 */
1534static int tipc_link_advance_transmq(struct tipc_link *l, struct tipc_link *r,
1535 u16 acked, u16 gap,
1536 struct tipc_gap_ack_blks *ga,
1537 struct sk_buff_head *xmitq,
1538 bool *retransmitted, int *rc)
1539{
1540 struct tipc_gap_ack_blks *last_ga = r->last_ga, *this_ga = NULL;
1541 struct tipc_gap_ack *gacks = NULL;
1542 struct sk_buff *skb, *_skb, *tmp;
1543 struct tipc_msg *hdr;
1544 u32 qlen = skb_queue_len(&l->transmq);
1545 u16 nacked = acked, ngap = gap, gack_cnt = 0;
1546 u16 bc_ack = l->bc_rcvlink->rcv_nxt - 1;
1547 u16 ack = l->rcv_nxt - 1;
1548 u16 seqno, n = 0;
1549 u16 end = r->acked, start = end, offset = r->last_gap;
1550 u16 si = (last_ga) ? last_ga->start_index : 0;
1551 bool is_uc = !link_is_bc_sndlink(l);
1552 bool bc_has_acked = false;
1553
1554 trace_tipc_link_retrans(r, acked + 1, acked + gap, &l->transmq);
1555
1556 /* Determine Gap ACK blocks if any for the particular link */
1557 if (ga && is_uc) {
1558 /* Get the Gap ACKs, uc part */
1559 gack_cnt = ga->ugack_cnt;
1560 gacks = &ga->gacks[ga->bgack_cnt];
1561 } else if (ga) {
1562 /* Copy the Gap ACKs, bc part, for later renewal if needed */
1563 this_ga = kmemdup(ga, struct_size(ga, gacks, ga->bgack_cnt),
1564 GFP_ATOMIC);
1565 if (likely(this_ga)) {
1566 this_ga->start_index = 0;
1567 /* Start with the bc Gap ACKs */
1568 gack_cnt = this_ga->bgack_cnt;
1569 gacks = &this_ga->gacks[0];
1570 } else {
1571 /* Hmm, we can get in trouble..., simply ignore it */
1572 pr_warn_ratelimited("Ignoring bc Gap ACKs, no memory\n");
1573 }
1574 }
1575
1576 /* Advance the link transmq */
1577 skb_queue_walk_safe(&l->transmq, skb, tmp) {
1578 seqno = buf_seqno(skb);
1579
1580next_gap_ack:
1581 if (less_eq(seqno, nacked)) {
1582 if (is_uc)
1583 goto release;
1584 /* Skip packets peer has already acked */
1585 if (!more(seqno, r->acked))
1586 continue;
1587 /* Get the next of last Gap ACK blocks */
1588 while (more(seqno, end)) {
1589 if (!last_ga || si >= last_ga->bgack_cnt)
1590 break;
1591 start = end + offset + 1;
1592 end = ntohs(last_ga->gacks[si].ack);
1593 offset = ntohs(last_ga->gacks[si].gap);
1594 si++;
1595 WARN_ONCE(more(start, end) ||
1596 (!offset &&
1597 si < last_ga->bgack_cnt) ||
1598 si > MAX_GAP_ACK_BLKS,
1599 "Corrupted Gap ACK: %d %d %d %d %d\n",
1600 start, end, offset, si,
1601 last_ga->bgack_cnt);
1602 }
1603 /* Check against the last Gap ACK block */
1604 if (tipc_in_range(seqno, start, end))
1605 continue;
1606 /* Update/release the packet peer is acking */
1607 bc_has_acked = true;
1608 if (--TIPC_SKB_CB(skb)->ackers)
1609 continue;
1610release:
1611 /* release skb */
1612 __skb_unlink(skb, &l->transmq);
1613 kfree_skb(skb);
1614 } else if (less_eq(seqno, nacked + ngap)) {
1615 /* First gap: check if repeated retrans failures? */
1616 if (unlikely(seqno == acked + 1 &&
1617 link_retransmit_failure(l, r, rc))) {
1618 /* Ignore this bc Gap ACKs if any */
1619 kfree(this_ga);
1620 this_ga = NULL;
1621 break;
1622 }
1623 /* retransmit skb if unrestricted*/
1624 if (time_before(jiffies, TIPC_SKB_CB(skb)->nxt_retr))
1625 continue;
1626 tipc_link_set_skb_retransmit_time(skb, l);
1627 _skb = pskb_copy(skb, GFP_ATOMIC);
1628 if (!_skb)
1629 continue;
1630 hdr = buf_msg(_skb);
1631 msg_set_ack(hdr, ack);
1632 msg_set_bcast_ack(hdr, bc_ack);
1633 _skb->priority = TC_PRIO_CONTROL;
1634 __skb_queue_tail(xmitq, _skb);
1635 l->stats.retransmitted++;
1636 if (!is_uc)
1637 r->stats.retransmitted++;
1638 *retransmitted = true;
1639 /* Increase actual retrans counter & mark first time */
1640 if (!TIPC_SKB_CB(skb)->retr_cnt++)
1641 TIPC_SKB_CB(skb)->retr_stamp = jiffies;
1642 } else {
1643 /* retry with Gap ACK blocks if any */
1644 if (n >= gack_cnt)
1645 break;
1646 nacked = ntohs(gacks[n].ack);
1647 ngap = ntohs(gacks[n].gap);
1648 n++;
1649 goto next_gap_ack;
1650 }
1651 }
1652
1653 /* Renew last Gap ACK blocks for bc if needed */
1654 if (bc_has_acked) {
1655 if (this_ga) {
1656 kfree(last_ga);
1657 r->last_ga = this_ga;
1658 r->last_gap = gap;
1659 } else if (last_ga) {
1660 if (less(acked, start)) {
1661 si--;
1662 offset = start - acked - 1;
1663 } else if (less(acked, end)) {
1664 acked = end;
1665 }
1666 if (si < last_ga->bgack_cnt) {
1667 last_ga->start_index = si;
1668 r->last_gap = offset;
1669 } else {
1670 kfree(last_ga);
1671 r->last_ga = NULL;
1672 r->last_gap = 0;
1673 }
1674 } else {
1675 r->last_gap = 0;
1676 }
1677 r->acked = acked;
1678 } else {
1679 kfree(this_ga);
1680 }
1681
1682 return qlen - skb_queue_len(&l->transmq);
1683}
1684
1685/* tipc_link_build_state_msg: prepare link state message for transmission
1686 *
1687 * Note that sending of broadcast ack is coordinated among nodes, to reduce
1688 * risk of ack storms towards the sender
1689 */
1690int tipc_link_build_state_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1691{
1692 if (!l)
1693 return 0;
1694
1695 /* Broadcast ACK must be sent via a unicast link => defer to caller */
1696 if (link_is_bc_rcvlink(l)) {
1697 if (((l->rcv_nxt ^ tipc_own_addr(l->net)) & 0xf) != 0xf)
1698 return 0;
1699 l->rcv_unacked = 0;
1700
1701 /* Use snd_nxt to store peer's snd_nxt in broadcast rcv link */
1702 l->snd_nxt = l->rcv_nxt;
1703 return TIPC_LINK_SND_STATE;
1704 }
1705 /* Unicast ACK */
1706 l->rcv_unacked = 0;
1707 l->stats.sent_acks++;
1708 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, 0, xmitq);
1709 return 0;
1710}
1711
1712/* tipc_link_build_reset_msg: prepare link RESET or ACTIVATE message
1713 */
1714void tipc_link_build_reset_msg(struct tipc_link *l, struct sk_buff_head *xmitq)
1715{
1716 int mtyp = RESET_MSG;
1717 struct sk_buff *skb;
1718
1719 if (l->state == LINK_ESTABLISHING)
1720 mtyp = ACTIVATE_MSG;
1721
1722 tipc_link_build_proto_msg(l, mtyp, 0, 0, 0, 0, 0, xmitq);
1723
1724 /* Inform peer that this endpoint is going down if applicable */
1725 skb = skb_peek_tail(xmitq);
1726 if (skb && (l->state == LINK_RESET))
1727 msg_set_peer_stopping(buf_msg(skb), 1);
1728}
1729
1730/* tipc_link_build_nack_msg: prepare link nack message for transmission
1731 * Note that sending of broadcast NACK is coordinated among nodes, to
1732 * reduce the risk of NACK storms towards the sender
1733 */
1734static int tipc_link_build_nack_msg(struct tipc_link *l,
1735 struct sk_buff_head *xmitq)
1736{
1737 u32 def_cnt = ++l->stats.deferred_recv;
1738 struct sk_buff_head *dfq = &l->deferdq;
1739 u32 defq_len = skb_queue_len(dfq);
1740 int match1, match2;
1741
1742 if (link_is_bc_rcvlink(l)) {
1743 match1 = def_cnt & 0xf;
1744 match2 = tipc_own_addr(l->net) & 0xf;
1745 if (match1 == match2)
1746 return TIPC_LINK_SND_STATE;
1747 return 0;
1748 }
1749
1750 if (defq_len >= 3 && !((defq_len - 3) % 16)) {
1751 u16 rcvgap = buf_seqno(skb_peek(dfq)) - l->rcv_nxt;
1752
1753 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0,
1754 rcvgap, 0, 0, xmitq);
1755 }
1756 return 0;
1757}
1758
1759/* tipc_link_rcv - process TIPC packets/messages arriving from off-node
1760 * @l: the link that should handle the message
1761 * @skb: TIPC packet
1762 * @xmitq: queue to place packets to be sent after this call
1763 */
1764int tipc_link_rcv(struct tipc_link *l, struct sk_buff *skb,
1765 struct sk_buff_head *xmitq)
1766{
1767 struct sk_buff_head *defq = &l->deferdq;
1768 struct tipc_msg *hdr = buf_msg(skb);
1769 u16 seqno, rcv_nxt, win_lim;
1770 int released = 0;
1771 int rc = 0;
1772
1773 /* Verify and update link state */
1774 if (unlikely(msg_user(hdr) == LINK_PROTOCOL))
1775 return tipc_link_proto_rcv(l, skb, xmitq);
1776
1777 /* Don't send probe at next timeout expiration */
1778 l->silent_intv_cnt = 0;
1779
1780 do {
1781 hdr = buf_msg(skb);
1782 seqno = msg_seqno(hdr);
1783 rcv_nxt = l->rcv_nxt;
1784 win_lim = rcv_nxt + TIPC_MAX_LINK_WIN;
1785
1786 if (unlikely(!tipc_link_is_up(l))) {
1787 if (l->state == LINK_ESTABLISHING)
1788 rc = TIPC_LINK_UP_EVT;
1789 kfree_skb(skb);
1790 break;
1791 }
1792
1793 /* Drop if outside receive window */
1794 if (unlikely(less(seqno, rcv_nxt) || more(seqno, win_lim))) {
1795 l->stats.duplicates++;
1796 kfree_skb(skb);
1797 break;
1798 }
1799 released += tipc_link_advance_transmq(l, l, msg_ack(hdr), 0,
1800 NULL, NULL, NULL, NULL);
1801
1802 /* Defer delivery if sequence gap */
1803 if (unlikely(seqno != rcv_nxt)) {
1804 if (!__tipc_skb_queue_sorted(defq, seqno, skb))
1805 l->stats.duplicates++;
1806 rc |= tipc_link_build_nack_msg(l, xmitq);
1807 break;
1808 }
1809
1810 /* Deliver packet */
1811 l->rcv_nxt++;
1812 l->stats.recv_pkts++;
1813
1814 if (unlikely(msg_user(hdr) == TUNNEL_PROTOCOL))
1815 rc |= tipc_link_tnl_rcv(l, skb, l->inputq);
1816 else if (!tipc_data_input(l, skb, l->inputq))
1817 rc |= tipc_link_input(l, skb, l->inputq, &l->reasm_buf);
1818 if (unlikely(++l->rcv_unacked >= TIPC_MIN_LINK_WIN))
1819 rc |= tipc_link_build_state_msg(l, xmitq);
1820 if (unlikely(rc & ~TIPC_LINK_SND_STATE))
1821 break;
1822 } while ((skb = __tipc_skb_dequeue(defq, l->rcv_nxt)));
1823
1824 /* Forward queues and wake up waiting users */
1825 if (released) {
1826 tipc_link_update_cwin(l, released, 0);
1827 tipc_link_advance_backlog(l, xmitq);
1828 if (unlikely(!skb_queue_empty(&l->wakeupq)))
1829 link_prepare_wakeup(l);
1830 }
1831 return rc;
1832}
1833
1834static void tipc_link_build_proto_msg(struct tipc_link *l, int mtyp, bool probe,
1835 bool probe_reply, u16 rcvgap,
1836 int tolerance, int priority,
1837 struct sk_buff_head *xmitq)
1838{
1839 struct tipc_mon_state *mstate = &l->mon_state;
1840 struct sk_buff_head *dfq = &l->deferdq;
1841 struct tipc_link *bcl = l->bc_rcvlink;
1842 struct tipc_msg *hdr;
1843 struct sk_buff *skb;
1844 bool node_up = tipc_link_is_up(bcl);
1845 u16 glen = 0, bc_rcvgap = 0;
1846 int dlen = 0;
1847 void *data;
1848
1849 /* Don't send protocol message during reset or link failover */
1850 if (tipc_link_is_blocked(l))
1851 return;
1852
1853 if (!tipc_link_is_up(l) && (mtyp == STATE_MSG))
1854 return;
1855
1856 if ((probe || probe_reply) && !skb_queue_empty(dfq))
1857 rcvgap = buf_seqno(skb_peek(dfq)) - l->rcv_nxt;
1858
1859 skb = tipc_msg_create(LINK_PROTOCOL, mtyp, INT_H_SIZE,
1860 tipc_max_domain_size + MAX_GAP_ACK_BLKS_SZ,
1861 l->addr, tipc_own_addr(l->net), 0, 0, 0);
1862 if (!skb)
1863 return;
1864
1865 hdr = buf_msg(skb);
1866 data = msg_data(hdr);
1867 msg_set_session(hdr, l->session);
1868 msg_set_bearer_id(hdr, l->bearer_id);
1869 msg_set_net_plane(hdr, l->net_plane);
1870 msg_set_next_sent(hdr, l->snd_nxt);
1871 msg_set_ack(hdr, l->rcv_nxt - 1);
1872 msg_set_bcast_ack(hdr, bcl->rcv_nxt - 1);
1873 msg_set_bc_ack_invalid(hdr, !node_up);
1874 msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
1875 msg_set_link_tolerance(hdr, tolerance);
1876 msg_set_linkprio(hdr, priority);
1877 msg_set_redundant_link(hdr, node_up);
1878 msg_set_seq_gap(hdr, 0);
1879 msg_set_seqno(hdr, l->snd_nxt + U16_MAX / 2);
1880
1881 if (mtyp == STATE_MSG) {
1882 if (l->peer_caps & TIPC_LINK_PROTO_SEQNO)
1883 msg_set_seqno(hdr, l->snd_nxt_state++);
1884 msg_set_seq_gap(hdr, rcvgap);
1885 bc_rcvgap = link_bc_rcv_gap(bcl);
1886 msg_set_bc_gap(hdr, bc_rcvgap);
1887 msg_set_probe(hdr, probe);
1888 msg_set_is_keepalive(hdr, probe || probe_reply);
1889 if (l->peer_caps & TIPC_GAP_ACK_BLOCK)
1890 glen = tipc_build_gap_ack_blks(l, hdr);
1891 tipc_mon_prep(l->net, data + glen, &dlen, mstate, l->bearer_id);
1892 msg_set_size(hdr, INT_H_SIZE + glen + dlen);
1893 skb_trim(skb, INT_H_SIZE + glen + dlen);
1894 l->stats.sent_states++;
1895 l->rcv_unacked = 0;
1896 } else {
1897 /* RESET_MSG or ACTIVATE_MSG */
1898 if (mtyp == ACTIVATE_MSG) {
1899 msg_set_dest_session_valid(hdr, 1);
1900 msg_set_dest_session(hdr, l->peer_session);
1901 }
1902 msg_set_max_pkt(hdr, l->advertised_mtu);
1903 strcpy(data, l->if_name);
1904 msg_set_size(hdr, INT_H_SIZE + TIPC_MAX_IF_NAME);
1905 skb_trim(skb, INT_H_SIZE + TIPC_MAX_IF_NAME);
1906 }
1907 if (probe)
1908 l->stats.sent_probes++;
1909 if (rcvgap)
1910 l->stats.sent_nacks++;
1911 if (bc_rcvgap)
1912 bcl->stats.sent_nacks++;
1913 skb->priority = TC_PRIO_CONTROL;
1914 __skb_queue_tail(xmitq, skb);
1915 trace_tipc_proto_build(skb, false, l->name);
1916}
1917
1918void tipc_link_create_dummy_tnl_msg(struct tipc_link *l,
1919 struct sk_buff_head *xmitq)
1920{
1921 u32 onode = tipc_own_addr(l->net);
1922 struct tipc_msg *hdr, *ihdr;
1923 struct sk_buff_head tnlq;
1924 struct sk_buff *skb;
1925 u32 dnode = l->addr;
1926
1927 __skb_queue_head_init(&tnlq);
1928 skb = tipc_msg_create(TUNNEL_PROTOCOL, FAILOVER_MSG,
1929 INT_H_SIZE, BASIC_H_SIZE,
1930 dnode, onode, 0, 0, 0);
1931 if (!skb) {
1932 pr_warn("%sunable to create tunnel packet\n", link_co_err);
1933 return;
1934 }
1935
1936 hdr = buf_msg(skb);
1937 msg_set_msgcnt(hdr, 1);
1938 msg_set_bearer_id(hdr, l->peer_bearer_id);
1939
1940 ihdr = (struct tipc_msg *)msg_data(hdr);
1941 tipc_msg_init(onode, ihdr, TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1942 BASIC_H_SIZE, dnode);
1943 msg_set_errcode(ihdr, TIPC_ERR_NO_PORT);
1944 __skb_queue_tail(&tnlq, skb);
1945 tipc_link_xmit(l, &tnlq, xmitq);
1946}
1947
1948/* tipc_link_tnl_prepare(): prepare and return a list of tunnel packets
1949 * with contents of the link's transmit and backlog queues.
1950 */
1951void tipc_link_tnl_prepare(struct tipc_link *l, struct tipc_link *tnl,
1952 int mtyp, struct sk_buff_head *xmitq)
1953{
1954 struct sk_buff_head *fdefq = &tnl->failover_deferdq;
1955 struct sk_buff *skb, *tnlskb;
1956 struct tipc_msg *hdr, tnlhdr;
1957 struct sk_buff_head *queue = &l->transmq;
1958 struct sk_buff_head tmpxq, tnlq, frags;
1959 u16 pktlen, pktcnt, seqno = l->snd_nxt;
1960 bool pktcnt_need_update = false;
1961 u16 syncpt;
1962 int rc;
1963
1964 if (!tnl)
1965 return;
1966
1967 __skb_queue_head_init(&tnlq);
1968 /* Link Synching:
1969 * From now on, send only one single ("dummy") SYNCH message
1970 * to peer. The SYNCH message does not contain any data, just
1971 * a header conveying the synch point to the peer.
1972 */
1973 if (mtyp == SYNCH_MSG && (tnl->peer_caps & TIPC_TUNNEL_ENHANCED)) {
1974 tnlskb = tipc_msg_create(TUNNEL_PROTOCOL, SYNCH_MSG,
1975 INT_H_SIZE, 0, l->addr,
1976 tipc_own_addr(l->net),
1977 0, 0, 0);
1978 if (!tnlskb) {
1979 pr_warn("%sunable to create dummy SYNCH_MSG\n",
1980 link_co_err);
1981 return;
1982 }
1983
1984 hdr = buf_msg(tnlskb);
1985 syncpt = l->snd_nxt + skb_queue_len(&l->backlogq) - 1;
1986 msg_set_syncpt(hdr, syncpt);
1987 msg_set_bearer_id(hdr, l->peer_bearer_id);
1988 __skb_queue_tail(&tnlq, tnlskb);
1989 tipc_link_xmit(tnl, &tnlq, xmitq);
1990 return;
1991 }
1992
1993 __skb_queue_head_init(&tmpxq);
1994 __skb_queue_head_init(&frags);
1995 /* At least one packet required for safe algorithm => add dummy */
1996 skb = tipc_msg_create(TIPC_LOW_IMPORTANCE, TIPC_DIRECT_MSG,
1997 BASIC_H_SIZE, 0, l->addr, tipc_own_addr(l->net),
1998 0, 0, TIPC_ERR_NO_PORT);
1999 if (!skb) {
2000 pr_warn("%sunable to create tunnel packet\n", link_co_err);
2001 return;
2002 }
2003 __skb_queue_tail(&tnlq, skb);
2004 tipc_link_xmit(l, &tnlq, &tmpxq);
2005 __skb_queue_purge(&tmpxq);
2006
2007 /* Initialize reusable tunnel packet header */
2008 tipc_msg_init(tipc_own_addr(l->net), &tnlhdr, TUNNEL_PROTOCOL,
2009 mtyp, INT_H_SIZE, l->addr);
2010 if (mtyp == SYNCH_MSG)
2011 pktcnt = l->snd_nxt - buf_seqno(skb_peek(&l->transmq));
2012 else
2013 pktcnt = skb_queue_len(&l->transmq);
2014 pktcnt += skb_queue_len(&l->backlogq);
2015 msg_set_msgcnt(&tnlhdr, pktcnt);
2016 msg_set_bearer_id(&tnlhdr, l->peer_bearer_id);
2017tnl:
2018 /* Wrap each packet into a tunnel packet */
2019 skb_queue_walk(queue, skb) {
2020 hdr = buf_msg(skb);
2021 if (queue == &l->backlogq)
2022 msg_set_seqno(hdr, seqno++);
2023 pktlen = msg_size(hdr);
2024
2025 /* Tunnel link MTU is not large enough? This could be
2026 * due to:
2027 * 1) Link MTU has just changed or set differently;
2028 * 2) Or FAILOVER on the top of a SYNCH message
2029 *
2030 * The 2nd case should not happen if peer supports
2031 * TIPC_TUNNEL_ENHANCED
2032 */
2033 if (pktlen > tnl->mtu - INT_H_SIZE) {
2034 if (mtyp == FAILOVER_MSG &&
2035 (tnl->peer_caps & TIPC_TUNNEL_ENHANCED)) {
2036 rc = tipc_msg_fragment(skb, &tnlhdr, tnl->mtu,
2037 &frags);
2038 if (rc) {
2039 pr_warn("%sunable to frag msg: rc %d\n",
2040 link_co_err, rc);
2041 return;
2042 }
2043 pktcnt += skb_queue_len(&frags) - 1;
2044 pktcnt_need_update = true;
2045 skb_queue_splice_tail_init(&frags, &tnlq);
2046 continue;
2047 }
2048 /* Unluckily, peer doesn't have TIPC_TUNNEL_ENHANCED
2049 * => Just warn it and return!
2050 */
2051 pr_warn_ratelimited("%stoo large msg <%d, %d>: %d!\n",
2052 link_co_err, msg_user(hdr),
2053 msg_type(hdr), msg_size(hdr));
2054 return;
2055 }
2056
2057 msg_set_size(&tnlhdr, pktlen + INT_H_SIZE);
2058 tnlskb = tipc_buf_acquire(pktlen + INT_H_SIZE, GFP_ATOMIC);
2059 if (!tnlskb) {
2060 pr_warn("%sunable to send packet\n", link_co_err);
2061 return;
2062 }
2063 skb_copy_to_linear_data(tnlskb, &tnlhdr, INT_H_SIZE);
2064 skb_copy_to_linear_data_offset(tnlskb, INT_H_SIZE, hdr, pktlen);
2065 __skb_queue_tail(&tnlq, tnlskb);
2066 }
2067 if (queue != &l->backlogq) {
2068 queue = &l->backlogq;
2069 goto tnl;
2070 }
2071
2072 if (pktcnt_need_update)
2073 skb_queue_walk(&tnlq, skb) {
2074 hdr = buf_msg(skb);
2075 msg_set_msgcnt(hdr, pktcnt);
2076 }
2077
2078 tipc_link_xmit(tnl, &tnlq, xmitq);
2079
2080 if (mtyp == FAILOVER_MSG) {
2081 tnl->drop_point = l->rcv_nxt;
2082 tnl->failover_reasm_skb = l->reasm_buf;
2083 l->reasm_buf = NULL;
2084
2085 /* Failover the link's deferdq */
2086 if (unlikely(!skb_queue_empty(fdefq))) {
2087 pr_warn("Link failover deferdq not empty: %d!\n",
2088 skb_queue_len(fdefq));
2089 __skb_queue_purge(fdefq);
2090 }
2091 skb_queue_splice_init(&l->deferdq, fdefq);
2092 }
2093}
2094
2095/**
2096 * tipc_link_failover_prepare() - prepare tnl for link failover
2097 *
2098 * This is a special version of the precursor - tipc_link_tnl_prepare(),
2099 * see the tipc_node_link_failover() for details
2100 *
2101 * @l: failover link
2102 * @tnl: tunnel link
2103 * @xmitq: queue for messages to be xmited
2104 */
2105void tipc_link_failover_prepare(struct tipc_link *l, struct tipc_link *tnl,
2106 struct sk_buff_head *xmitq)
2107{
2108 struct sk_buff_head *fdefq = &tnl->failover_deferdq;
2109
2110 tipc_link_create_dummy_tnl_msg(tnl, xmitq);
2111
2112 /* This failover link endpoint was never established before,
2113 * so it has not received anything from peer.
2114 * Otherwise, it must be a normal failover situation or the
2115 * node has entered SELF_DOWN_PEER_LEAVING and both peer nodes
2116 * would have to start over from scratch instead.
2117 */
2118 tnl->drop_point = 1;
2119 tnl->failover_reasm_skb = NULL;
2120
2121 /* Initiate the link's failover deferdq */
2122 if (unlikely(!skb_queue_empty(fdefq))) {
2123 pr_warn("Link failover deferdq not empty: %d!\n",
2124 skb_queue_len(fdefq));
2125 __skb_queue_purge(fdefq);
2126 }
2127}
2128
2129/* tipc_link_validate_msg(): validate message against current link state
2130 * Returns true if message should be accepted, otherwise false
2131 */
2132bool tipc_link_validate_msg(struct tipc_link *l, struct tipc_msg *hdr)
2133{
2134 u16 curr_session = l->peer_session;
2135 u16 session = msg_session(hdr);
2136 int mtyp = msg_type(hdr);
2137
2138 if (msg_user(hdr) != LINK_PROTOCOL)
2139 return true;
2140
2141 switch (mtyp) {
2142 case RESET_MSG:
2143 if (!l->in_session)
2144 return true;
2145 /* Accept only RESET with new session number */
2146 return more(session, curr_session);
2147 case ACTIVATE_MSG:
2148 if (!l->in_session)
2149 return true;
2150 /* Accept only ACTIVATE with new or current session number */
2151 return !less(session, curr_session);
2152 case STATE_MSG:
2153 /* Accept only STATE with current session number */
2154 if (!l->in_session)
2155 return false;
2156 if (session != curr_session)
2157 return false;
2158 /* Extra sanity check */
2159 if (!tipc_link_is_up(l) && msg_ack(hdr))
2160 return false;
2161 if (!(l->peer_caps & TIPC_LINK_PROTO_SEQNO))
2162 return true;
2163 /* Accept only STATE with new sequence number */
2164 return !less(msg_seqno(hdr), l->rcv_nxt_state);
2165 default:
2166 return false;
2167 }
2168}
2169
2170/* tipc_link_proto_rcv(): receive link level protocol message :
2171 * Note that network plane id propagates through the network, and may
2172 * change at any time. The node with lowest numerical id determines
2173 * network plane
2174 */
2175static int tipc_link_proto_rcv(struct tipc_link *l, struct sk_buff *skb,
2176 struct sk_buff_head *xmitq)
2177{
2178 struct tipc_msg *hdr = buf_msg(skb);
2179 struct tipc_gap_ack_blks *ga = NULL;
2180 bool reply = msg_probe(hdr), retransmitted = false;
2181 u32 dlen = msg_data_sz(hdr), glen = 0, msg_max;
2182 u16 peers_snd_nxt = msg_next_sent(hdr);
2183 u16 peers_tol = msg_link_tolerance(hdr);
2184 u16 peers_prio = msg_linkprio(hdr);
2185 u16 gap = msg_seq_gap(hdr);
2186 u16 ack = msg_ack(hdr);
2187 u16 rcv_nxt = l->rcv_nxt;
2188 u16 rcvgap = 0;
2189 int mtyp = msg_type(hdr);
2190 int rc = 0, released;
2191 char *if_name;
2192 void *data;
2193
2194 trace_tipc_proto_rcv(skb, false, l->name);
2195
2196 if (dlen > U16_MAX)
2197 goto exit;
2198
2199 if (tipc_link_is_blocked(l) || !xmitq)
2200 goto exit;
2201
2202 if (tipc_own_addr(l->net) > msg_prevnode(hdr))
2203 l->net_plane = msg_net_plane(hdr);
2204
2205 if (skb_linearize(skb))
2206 goto exit;
2207
2208 hdr = buf_msg(skb);
2209 data = msg_data(hdr);
2210
2211 if (!tipc_link_validate_msg(l, hdr)) {
2212 trace_tipc_skb_dump(skb, false, "PROTO invalid (1)!");
2213 trace_tipc_link_dump(l, TIPC_DUMP_NONE, "PROTO invalid (1)!");
2214 goto exit;
2215 }
2216
2217 switch (mtyp) {
2218 case RESET_MSG:
2219 case ACTIVATE_MSG:
2220 msg_max = msg_max_pkt(hdr);
2221 if (msg_max < tipc_bearer_min_mtu(l->net, l->bearer_id))
2222 break;
2223 /* Complete own link name with peer's interface name */
2224 if_name = strrchr(l->name, ':') + 1;
2225 if (sizeof(l->name) - (if_name - l->name) <= TIPC_MAX_IF_NAME)
2226 break;
2227 if (msg_data_sz(hdr) < TIPC_MAX_IF_NAME)
2228 break;
2229 strncpy(if_name, data, TIPC_MAX_IF_NAME);
2230
2231 /* Update own tolerance if peer indicates a non-zero value */
2232 if (tipc_in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
2233 l->tolerance = peers_tol;
2234 l->bc_rcvlink->tolerance = peers_tol;
2235 }
2236 /* Update own priority if peer's priority is higher */
2237 if (tipc_in_range(peers_prio, l->priority + 1, TIPC_MAX_LINK_PRI))
2238 l->priority = peers_prio;
2239
2240 /* If peer is going down we want full re-establish cycle */
2241 if (msg_peer_stopping(hdr)) {
2242 rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
2243 break;
2244 }
2245
2246 /* If this endpoint was re-created while peer was ESTABLISHING
2247 * it doesn't know current session number. Force re-synch.
2248 */
2249 if (mtyp == ACTIVATE_MSG && msg_dest_session_valid(hdr) &&
2250 l->session != msg_dest_session(hdr)) {
2251 if (less(l->session, msg_dest_session(hdr)))
2252 l->session = msg_dest_session(hdr) + 1;
2253 break;
2254 }
2255
2256 /* ACTIVATE_MSG serves as PEER_RESET if link is already down */
2257 if (mtyp == RESET_MSG || !tipc_link_is_up(l))
2258 rc = tipc_link_fsm_evt(l, LINK_PEER_RESET_EVT);
2259
2260 /* ACTIVATE_MSG takes up link if it was already locally reset */
2261 if (mtyp == ACTIVATE_MSG && l->state == LINK_ESTABLISHING)
2262 rc = TIPC_LINK_UP_EVT;
2263
2264 l->peer_session = msg_session(hdr);
2265 l->in_session = true;
2266 l->peer_bearer_id = msg_bearer_id(hdr);
2267 if (l->mtu > msg_max)
2268 l->mtu = msg_max;
2269 break;
2270
2271 case STATE_MSG:
2272 /* Validate Gap ACK blocks, drop if invalid */
2273 glen = tipc_get_gap_ack_blks(&ga, l, hdr, true);
2274 if (glen > dlen)
2275 break;
2276
2277 l->rcv_nxt_state = msg_seqno(hdr) + 1;
2278
2279 /* Update own tolerance if peer indicates a non-zero value */
2280 if (tipc_in_range(peers_tol, TIPC_MIN_LINK_TOL, TIPC_MAX_LINK_TOL)) {
2281 l->tolerance = peers_tol;
2282 l->bc_rcvlink->tolerance = peers_tol;
2283 }
2284 /* Update own prio if peer indicates a different value */
2285 if ((peers_prio != l->priority) &&
2286 tipc_in_range(peers_prio, 1, TIPC_MAX_LINK_PRI)) {
2287 l->priority = peers_prio;
2288 rc = tipc_link_fsm_evt(l, LINK_FAILURE_EVT);
2289 }
2290
2291 l->silent_intv_cnt = 0;
2292 l->stats.recv_states++;
2293 if (msg_probe(hdr))
2294 l->stats.recv_probes++;
2295
2296 if (!tipc_link_is_up(l)) {
2297 if (l->state == LINK_ESTABLISHING)
2298 rc = TIPC_LINK_UP_EVT;
2299 break;
2300 }
2301
2302 tipc_mon_rcv(l->net, data + glen, dlen - glen, l->addr,
2303 &l->mon_state, l->bearer_id);
2304
2305 /* Send NACK if peer has sent pkts we haven't received yet */
2306 if ((reply || msg_is_keepalive(hdr)) &&
2307 more(peers_snd_nxt, rcv_nxt) &&
2308 !tipc_link_is_synching(l) &&
2309 skb_queue_empty(&l->deferdq))
2310 rcvgap = peers_snd_nxt - l->rcv_nxt;
2311 if (rcvgap || reply)
2312 tipc_link_build_proto_msg(l, STATE_MSG, 0, reply,
2313 rcvgap, 0, 0, xmitq);
2314
2315 released = tipc_link_advance_transmq(l, l, ack, gap, ga, xmitq,
2316 &retransmitted, &rc);
2317 if (gap)
2318 l->stats.recv_nacks++;
2319 if (released || retransmitted)
2320 tipc_link_update_cwin(l, released, retransmitted);
2321 if (released)
2322 tipc_link_advance_backlog(l, xmitq);
2323 if (unlikely(!skb_queue_empty(&l->wakeupq)))
2324 link_prepare_wakeup(l);
2325 }
2326exit:
2327 kfree_skb(skb);
2328 return rc;
2329}
2330
2331/* tipc_link_build_bc_proto_msg() - create broadcast protocol message
2332 */
2333static bool tipc_link_build_bc_proto_msg(struct tipc_link *l, bool bcast,
2334 u16 peers_snd_nxt,
2335 struct sk_buff_head *xmitq)
2336{
2337 struct sk_buff *skb;
2338 struct tipc_msg *hdr;
2339 struct sk_buff *dfrd_skb = skb_peek(&l->deferdq);
2340 u16 ack = l->rcv_nxt - 1;
2341 u16 gap_to = peers_snd_nxt - 1;
2342
2343 skb = tipc_msg_create(BCAST_PROTOCOL, STATE_MSG, INT_H_SIZE,
2344 0, l->addr, tipc_own_addr(l->net), 0, 0, 0);
2345 if (!skb)
2346 return false;
2347 hdr = buf_msg(skb);
2348 msg_set_last_bcast(hdr, l->bc_sndlink->snd_nxt - 1);
2349 msg_set_bcast_ack(hdr, ack);
2350 msg_set_bcgap_after(hdr, ack);
2351 if (dfrd_skb)
2352 gap_to = buf_seqno(dfrd_skb) - 1;
2353 msg_set_bcgap_to(hdr, gap_to);
2354 msg_set_non_seq(hdr, bcast);
2355 __skb_queue_tail(xmitq, skb);
2356 return true;
2357}
2358
2359/* tipc_link_build_bc_init_msg() - synchronize broadcast link endpoints.
2360 *
2361 * Give a newly added peer node the sequence number where it should
2362 * start receiving and acking broadcast packets.
2363 */
2364static void tipc_link_build_bc_init_msg(struct tipc_link *l,
2365 struct sk_buff_head *xmitq)
2366{
2367 struct sk_buff_head list;
2368
2369 __skb_queue_head_init(&list);
2370 if (!tipc_link_build_bc_proto_msg(l->bc_rcvlink, false, 0, &list))
2371 return;
2372 msg_set_bc_ack_invalid(buf_msg(skb_peek(&list)), true);
2373 tipc_link_xmit(l, &list, xmitq);
2374}
2375
2376/* tipc_link_bc_init_rcv - receive initial broadcast synch data from peer
2377 */
2378void tipc_link_bc_init_rcv(struct tipc_link *l, struct tipc_msg *hdr)
2379{
2380 int mtyp = msg_type(hdr);
2381 u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
2382
2383 if (tipc_link_is_up(l))
2384 return;
2385
2386 if (msg_user(hdr) == BCAST_PROTOCOL) {
2387 l->rcv_nxt = peers_snd_nxt;
2388 l->state = LINK_ESTABLISHED;
2389 return;
2390 }
2391
2392 if (l->peer_caps & TIPC_BCAST_SYNCH)
2393 return;
2394
2395 if (msg_peer_node_is_up(hdr))
2396 return;
2397
2398 /* Compatibility: accept older, less safe initial synch data */
2399 if ((mtyp == RESET_MSG) || (mtyp == ACTIVATE_MSG))
2400 l->rcv_nxt = peers_snd_nxt;
2401}
2402
2403/* tipc_link_bc_sync_rcv - update rcv link according to peer's send state
2404 */
2405int tipc_link_bc_sync_rcv(struct tipc_link *l, struct tipc_msg *hdr,
2406 struct sk_buff_head *xmitq)
2407{
2408 u16 peers_snd_nxt = msg_bc_snd_nxt(hdr);
2409 int rc = 0;
2410
2411 if (!tipc_link_is_up(l))
2412 return rc;
2413
2414 if (!msg_peer_node_is_up(hdr))
2415 return rc;
2416
2417 /* Open when peer acknowledges our bcast init msg (pkt #1) */
2418 if (msg_ack(hdr))
2419 l->bc_peer_is_up = true;
2420
2421 if (!l->bc_peer_is_up)
2422 return rc;
2423
2424 /* Ignore if peers_snd_nxt goes beyond receive window */
2425 if (more(peers_snd_nxt, l->rcv_nxt + l->window))
2426 return rc;
2427
2428 l->snd_nxt = peers_snd_nxt;
2429 if (link_bc_rcv_gap(l))
2430 rc |= TIPC_LINK_SND_STATE;
2431
2432 /* Return now if sender supports nack via STATE messages */
2433 if (l->peer_caps & TIPC_BCAST_STATE_NACK)
2434 return rc;
2435
2436 /* Otherwise, be backwards compatible */
2437
2438 if (!more(peers_snd_nxt, l->rcv_nxt)) {
2439 l->nack_state = BC_NACK_SND_CONDITIONAL;
2440 return 0;
2441 }
2442
2443 /* Don't NACK if one was recently sent or peeked */
2444 if (l->nack_state == BC_NACK_SND_SUPPRESS) {
2445 l->nack_state = BC_NACK_SND_UNCONDITIONAL;
2446 return 0;
2447 }
2448
2449 /* Conditionally delay NACK sending until next synch rcv */
2450 if (l->nack_state == BC_NACK_SND_CONDITIONAL) {
2451 l->nack_state = BC_NACK_SND_UNCONDITIONAL;
2452 if ((peers_snd_nxt - l->rcv_nxt) < TIPC_MIN_LINK_WIN)
2453 return 0;
2454 }
2455
2456 /* Send NACK now but suppress next one */
2457 tipc_link_build_bc_proto_msg(l, true, peers_snd_nxt, xmitq);
2458 l->nack_state = BC_NACK_SND_SUPPRESS;
2459 return 0;
2460}
2461
2462int tipc_link_bc_ack_rcv(struct tipc_link *r, u16 acked, u16 gap,
2463 struct tipc_gap_ack_blks *ga,
2464 struct sk_buff_head *xmitq,
2465 struct sk_buff_head *retrq)
2466{
2467 struct tipc_link *l = r->bc_sndlink;
2468 bool unused = false;
2469 int rc = 0;
2470
2471 if (!tipc_link_is_up(r) || !r->bc_peer_is_up)
2472 return 0;
2473
2474 if (gap) {
2475 l->stats.recv_nacks++;
2476 r->stats.recv_nacks++;
2477 }
2478
2479 if (less(acked, r->acked) || (acked == r->acked && !gap && !ga))
2480 return 0;
2481
2482 trace_tipc_link_bc_ack(r, acked, gap, &l->transmq);
2483 tipc_link_advance_transmq(l, r, acked, gap, ga, retrq, &unused, &rc);
2484
2485 tipc_link_advance_backlog(l, xmitq);
2486 if (unlikely(!skb_queue_empty(&l->wakeupq)))
2487 link_prepare_wakeup(l);
2488
2489 return rc;
2490}
2491
2492/* tipc_link_bc_nack_rcv(): receive broadcast nack message
2493 * This function is here for backwards compatibility, since
2494 * no BCAST_PROTOCOL/STATE messages occur from TIPC v2.5.
2495 */
2496int tipc_link_bc_nack_rcv(struct tipc_link *l, struct sk_buff *skb,
2497 struct sk_buff_head *xmitq)
2498{
2499 struct tipc_msg *hdr = buf_msg(skb);
2500 u32 dnode = msg_destnode(hdr);
2501 int mtyp = msg_type(hdr);
2502 u16 acked = msg_bcast_ack(hdr);
2503 u16 from = acked + 1;
2504 u16 to = msg_bcgap_to(hdr);
2505 u16 peers_snd_nxt = to + 1;
2506 int rc = 0;
2507
2508 kfree_skb(skb);
2509
2510 if (!tipc_link_is_up(l) || !l->bc_peer_is_up)
2511 return 0;
2512
2513 if (mtyp != STATE_MSG)
2514 return 0;
2515
2516 if (dnode == tipc_own_addr(l->net)) {
2517 rc = tipc_link_bc_ack_rcv(l, acked, to - acked, NULL, xmitq,
2518 xmitq);
2519 l->stats.recv_nacks++;
2520 return rc;
2521 }
2522
2523 /* Msg for other node => suppress own NACK at next sync if applicable */
2524 if (more(peers_snd_nxt, l->rcv_nxt) && !less(l->rcv_nxt, from))
2525 l->nack_state = BC_NACK_SND_SUPPRESS;
2526
2527 return 0;
2528}
2529
2530void tipc_link_set_queue_limits(struct tipc_link *l, u32 min_win, u32 max_win)
2531{
2532 int max_bulk = TIPC_MAX_PUBL / (l->mtu / ITEM_SIZE);
2533
2534 l->min_win = min_win;
2535 l->ssthresh = max_win;
2536 l->max_win = max_win;
2537 l->window = min_win;
2538 l->backlog[TIPC_LOW_IMPORTANCE].limit = min_win * 2;
2539 l->backlog[TIPC_MEDIUM_IMPORTANCE].limit = min_win * 4;
2540 l->backlog[TIPC_HIGH_IMPORTANCE].limit = min_win * 6;
2541 l->backlog[TIPC_CRITICAL_IMPORTANCE].limit = min_win * 8;
2542 l->backlog[TIPC_SYSTEM_IMPORTANCE].limit = max_bulk;
2543}
2544
2545/**
2546 * tipc_link_reset_stats - reset link statistics
2547 * @l: pointer to link
2548 */
2549void tipc_link_reset_stats(struct tipc_link *l)
2550{
2551 memset(&l->stats, 0, sizeof(l->stats));
2552}
2553
2554static void link_print(struct tipc_link *l, const char *str)
2555{
2556 struct sk_buff *hskb = skb_peek(&l->transmq);
2557 u16 head = hskb ? msg_seqno(buf_msg(hskb)) : l->snd_nxt - 1;
2558 u16 tail = l->snd_nxt - 1;
2559
2560 pr_info("%s Link <%s> state %x\n", str, l->name, l->state);
2561 pr_info("XMTQ: %u [%u-%u], BKLGQ: %u, SNDNX: %u, RCVNX: %u\n",
2562 skb_queue_len(&l->transmq), head, tail,
2563 skb_queue_len(&l->backlogq), l->snd_nxt, l->rcv_nxt);
2564}
2565
2566/* Parse and validate nested (link) properties valid for media, bearer and link
2567 */
2568int tipc_nl_parse_link_prop(struct nlattr *prop, struct nlattr *props[])
2569{
2570 int err;
2571
2572 err = nla_parse_nested_deprecated(props, TIPC_NLA_PROP_MAX, prop,
2573 tipc_nl_prop_policy, NULL);
2574 if (err)
2575 return err;
2576
2577 if (props[TIPC_NLA_PROP_PRIO]) {
2578 u32 prio;
2579
2580 prio = nla_get_u32(props[TIPC_NLA_PROP_PRIO]);
2581 if (prio > TIPC_MAX_LINK_PRI)
2582 return -EINVAL;
2583 }
2584
2585 if (props[TIPC_NLA_PROP_TOL]) {
2586 u32 tol;
2587
2588 tol = nla_get_u32(props[TIPC_NLA_PROP_TOL]);
2589 if ((tol < TIPC_MIN_LINK_TOL) || (tol > TIPC_MAX_LINK_TOL))
2590 return -EINVAL;
2591 }
2592
2593 if (props[TIPC_NLA_PROP_WIN]) {
2594 u32 max_win;
2595
2596 max_win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
2597 if (max_win < TIPC_DEF_LINK_WIN || max_win > TIPC_MAX_LINK_WIN)
2598 return -EINVAL;
2599 }
2600
2601 return 0;
2602}
2603
2604static int __tipc_nl_add_stats(struct sk_buff *skb, struct tipc_stats *s)
2605{
2606 int i;
2607 struct nlattr *stats;
2608
2609 struct nla_map {
2610 u32 key;
2611 u32 val;
2612 };
2613
2614 struct nla_map map[] = {
2615 {TIPC_NLA_STATS_RX_INFO, 0},
2616 {TIPC_NLA_STATS_RX_FRAGMENTS, s->recv_fragments},
2617 {TIPC_NLA_STATS_RX_FRAGMENTED, s->recv_fragmented},
2618 {TIPC_NLA_STATS_RX_BUNDLES, s->recv_bundles},
2619 {TIPC_NLA_STATS_RX_BUNDLED, s->recv_bundled},
2620 {TIPC_NLA_STATS_TX_INFO, 0},
2621 {TIPC_NLA_STATS_TX_FRAGMENTS, s->sent_fragments},
2622 {TIPC_NLA_STATS_TX_FRAGMENTED, s->sent_fragmented},
2623 {TIPC_NLA_STATS_TX_BUNDLES, s->sent_bundles},
2624 {TIPC_NLA_STATS_TX_BUNDLED, s->sent_bundled},
2625 {TIPC_NLA_STATS_MSG_PROF_TOT, (s->msg_length_counts) ?
2626 s->msg_length_counts : 1},
2627 {TIPC_NLA_STATS_MSG_LEN_CNT, s->msg_length_counts},
2628 {TIPC_NLA_STATS_MSG_LEN_TOT, s->msg_lengths_total},
2629 {TIPC_NLA_STATS_MSG_LEN_P0, s->msg_length_profile[0]},
2630 {TIPC_NLA_STATS_MSG_LEN_P1, s->msg_length_profile[1]},
2631 {TIPC_NLA_STATS_MSG_LEN_P2, s->msg_length_profile[2]},
2632 {TIPC_NLA_STATS_MSG_LEN_P3, s->msg_length_profile[3]},
2633 {TIPC_NLA_STATS_MSG_LEN_P4, s->msg_length_profile[4]},
2634 {TIPC_NLA_STATS_MSG_LEN_P5, s->msg_length_profile[5]},
2635 {TIPC_NLA_STATS_MSG_LEN_P6, s->msg_length_profile[6]},
2636 {TIPC_NLA_STATS_RX_STATES, s->recv_states},
2637 {TIPC_NLA_STATS_RX_PROBES, s->recv_probes},
2638 {TIPC_NLA_STATS_RX_NACKS, s->recv_nacks},
2639 {TIPC_NLA_STATS_RX_DEFERRED, s->deferred_recv},
2640 {TIPC_NLA_STATS_TX_STATES, s->sent_states},
2641 {TIPC_NLA_STATS_TX_PROBES, s->sent_probes},
2642 {TIPC_NLA_STATS_TX_NACKS, s->sent_nacks},
2643 {TIPC_NLA_STATS_TX_ACKS, s->sent_acks},
2644 {TIPC_NLA_STATS_RETRANSMITTED, s->retransmitted},
2645 {TIPC_NLA_STATS_DUPLICATES, s->duplicates},
2646 {TIPC_NLA_STATS_LINK_CONGS, s->link_congs},
2647 {TIPC_NLA_STATS_MAX_QUEUE, s->max_queue_sz},
2648 {TIPC_NLA_STATS_AVG_QUEUE, s->queue_sz_counts ?
2649 (s->accu_queue_sz / s->queue_sz_counts) : 0}
2650 };
2651
2652 stats = nla_nest_start_noflag(skb, TIPC_NLA_LINK_STATS);
2653 if (!stats)
2654 return -EMSGSIZE;
2655
2656 for (i = 0; i < ARRAY_SIZE(map); i++)
2657 if (nla_put_u32(skb, map[i].key, map[i].val))
2658 goto msg_full;
2659
2660 nla_nest_end(skb, stats);
2661
2662 return 0;
2663msg_full:
2664 nla_nest_cancel(skb, stats);
2665
2666 return -EMSGSIZE;
2667}
2668
2669/* Caller should hold appropriate locks to protect the link */
2670int __tipc_nl_add_link(struct net *net, struct tipc_nl_msg *msg,
2671 struct tipc_link *link, int nlflags)
2672{
2673 u32 self = tipc_own_addr(net);
2674 struct nlattr *attrs;
2675 struct nlattr *prop;
2676 void *hdr;
2677 int err;
2678
2679 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2680 nlflags, TIPC_NL_LINK_GET);
2681 if (!hdr)
2682 return -EMSGSIZE;
2683
2684 attrs = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK);
2685 if (!attrs)
2686 goto msg_full;
2687
2688 if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, link->name))
2689 goto attr_msg_full;
2690 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_DEST, tipc_cluster_mask(self)))
2691 goto attr_msg_full;
2692 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_MTU, link->mtu))
2693 goto attr_msg_full;
2694 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, link->stats.recv_pkts))
2695 goto attr_msg_full;
2696 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, link->stats.sent_pkts))
2697 goto attr_msg_full;
2698
2699 if (tipc_link_is_up(link))
2700 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2701 goto attr_msg_full;
2702 if (link->active)
2703 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_ACTIVE))
2704 goto attr_msg_full;
2705
2706 prop = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK_PROP);
2707 if (!prop)
2708 goto attr_msg_full;
2709 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2710 goto prop_msg_full;
2711 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_TOL, link->tolerance))
2712 goto prop_msg_full;
2713 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN,
2714 link->window))
2715 goto prop_msg_full;
2716 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_PRIO, link->priority))
2717 goto prop_msg_full;
2718 nla_nest_end(msg->skb, prop);
2719
2720 err = __tipc_nl_add_stats(msg->skb, &link->stats);
2721 if (err)
2722 goto attr_msg_full;
2723
2724 nla_nest_end(msg->skb, attrs);
2725 genlmsg_end(msg->skb, hdr);
2726
2727 return 0;
2728
2729prop_msg_full:
2730 nla_nest_cancel(msg->skb, prop);
2731attr_msg_full:
2732 nla_nest_cancel(msg->skb, attrs);
2733msg_full:
2734 genlmsg_cancel(msg->skb, hdr);
2735
2736 return -EMSGSIZE;
2737}
2738
2739static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
2740 struct tipc_stats *stats)
2741{
2742 int i;
2743 struct nlattr *nest;
2744
2745 struct nla_map {
2746 __u32 key;
2747 __u32 val;
2748 };
2749
2750 struct nla_map map[] = {
2751 {TIPC_NLA_STATS_RX_INFO, stats->recv_pkts},
2752 {TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
2753 {TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
2754 {TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
2755 {TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
2756 {TIPC_NLA_STATS_TX_INFO, stats->sent_pkts},
2757 {TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
2758 {TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
2759 {TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
2760 {TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
2761 {TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
2762 {TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
2763 {TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
2764 {TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
2765 {TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
2766 {TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
2767 {TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
2768 {TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
2769 {TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
2770 (stats->accu_queue_sz / stats->queue_sz_counts) : 0}
2771 };
2772
2773 nest = nla_nest_start_noflag(skb, TIPC_NLA_LINK_STATS);
2774 if (!nest)
2775 return -EMSGSIZE;
2776
2777 for (i = 0; i < ARRAY_SIZE(map); i++)
2778 if (nla_put_u32(skb, map[i].key, map[i].val))
2779 goto msg_full;
2780
2781 nla_nest_end(skb, nest);
2782
2783 return 0;
2784msg_full:
2785 nla_nest_cancel(skb, nest);
2786
2787 return -EMSGSIZE;
2788}
2789
2790int tipc_nl_add_bc_link(struct net *net, struct tipc_nl_msg *msg,
2791 struct tipc_link *bcl)
2792{
2793 int err;
2794 void *hdr;
2795 struct nlattr *attrs;
2796 struct nlattr *prop;
2797 u32 bc_mode = tipc_bcast_get_mode(net);
2798 u32 bc_ratio = tipc_bcast_get_broadcast_ratio(net);
2799
2800 if (!bcl)
2801 return 0;
2802
2803 tipc_bcast_lock(net);
2804
2805 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
2806 NLM_F_MULTI, TIPC_NL_LINK_GET);
2807 if (!hdr) {
2808 tipc_bcast_unlock(net);
2809 return -EMSGSIZE;
2810 }
2811
2812 attrs = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK);
2813 if (!attrs)
2814 goto msg_full;
2815
2816 /* The broadcast link is always up */
2817 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
2818 goto attr_msg_full;
2819
2820 if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
2821 goto attr_msg_full;
2822 if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
2823 goto attr_msg_full;
2824 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, 0))
2825 goto attr_msg_full;
2826 if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, 0))
2827 goto attr_msg_full;
2828
2829 prop = nla_nest_start_noflag(msg->skb, TIPC_NLA_LINK_PROP);
2830 if (!prop)
2831 goto attr_msg_full;
2832 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->max_win))
2833 goto prop_msg_full;
2834 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_BROADCAST, bc_mode))
2835 goto prop_msg_full;
2836 if (bc_mode & BCLINK_MODE_SEL)
2837 if (nla_put_u32(msg->skb, TIPC_NLA_PROP_BROADCAST_RATIO,
2838 bc_ratio))
2839 goto prop_msg_full;
2840 nla_nest_end(msg->skb, prop);
2841
2842 err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
2843 if (err)
2844 goto attr_msg_full;
2845
2846 tipc_bcast_unlock(net);
2847 nla_nest_end(msg->skb, attrs);
2848 genlmsg_end(msg->skb, hdr);
2849
2850 return 0;
2851
2852prop_msg_full:
2853 nla_nest_cancel(msg->skb, prop);
2854attr_msg_full:
2855 nla_nest_cancel(msg->skb, attrs);
2856msg_full:
2857 tipc_bcast_unlock(net);
2858 genlmsg_cancel(msg->skb, hdr);
2859
2860 return -EMSGSIZE;
2861}
2862
2863void tipc_link_set_tolerance(struct tipc_link *l, u32 tol,
2864 struct sk_buff_head *xmitq)
2865{
2866 l->tolerance = tol;
2867 if (l->bc_rcvlink)
2868 l->bc_rcvlink->tolerance = tol;
2869 if (tipc_link_is_up(l))
2870 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, tol, 0, xmitq);
2871}
2872
2873void tipc_link_set_prio(struct tipc_link *l, u32 prio,
2874 struct sk_buff_head *xmitq)
2875{
2876 l->priority = prio;
2877 tipc_link_build_proto_msg(l, STATE_MSG, 0, 0, 0, 0, prio, xmitq);
2878}
2879
2880void tipc_link_set_abort_limit(struct tipc_link *l, u32 limit)
2881{
2882 l->abort_limit = limit;
2883}
2884
2885/**
2886 * tipc_link_dump - dump TIPC link data
2887 * @l: tipc link to be dumped
2888 * @dqueues: bitmask to decide if any link queue to be dumped?
2889 * - TIPC_DUMP_NONE: don't dump link queues
2890 * - TIPC_DUMP_TRANSMQ: dump link transmq queue
2891 * - TIPC_DUMP_BACKLOGQ: dump link backlog queue
2892 * - TIPC_DUMP_DEFERDQ: dump link deferd queue
2893 * - TIPC_DUMP_INPUTQ: dump link input queue
2894 * - TIPC_DUMP_WAKEUP: dump link wakeup queue
2895 * - TIPC_DUMP_ALL: dump all the link queues above
2896 * @buf: returned buffer of dump data in format
2897 */
2898int tipc_link_dump(struct tipc_link *l, u16 dqueues, char *buf)
2899{
2900 int i = 0;
2901 size_t sz = (dqueues) ? LINK_LMAX : LINK_LMIN;
2902 struct sk_buff_head *list;
2903 struct sk_buff *hskb, *tskb;
2904 u32 len;
2905
2906 if (!l) {
2907 i += scnprintf(buf, sz, "link data: (null)\n");
2908 return i;
2909 }
2910
2911 i += scnprintf(buf, sz, "link data: %x", l->addr);
2912 i += scnprintf(buf + i, sz - i, " %x", l->state);
2913 i += scnprintf(buf + i, sz - i, " %u", l->in_session);
2914 i += scnprintf(buf + i, sz - i, " %u", l->session);
2915 i += scnprintf(buf + i, sz - i, " %u", l->peer_session);
2916 i += scnprintf(buf + i, sz - i, " %u", l->snd_nxt);
2917 i += scnprintf(buf + i, sz - i, " %u", l->rcv_nxt);
2918 i += scnprintf(buf + i, sz - i, " %u", l->snd_nxt_state);
2919 i += scnprintf(buf + i, sz - i, " %u", l->rcv_nxt_state);
2920 i += scnprintf(buf + i, sz - i, " %x", l->peer_caps);
2921 i += scnprintf(buf + i, sz - i, " %u", l->silent_intv_cnt);
2922 i += scnprintf(buf + i, sz - i, " %u", l->rst_cnt);
2923 i += scnprintf(buf + i, sz - i, " %u", 0);
2924 i += scnprintf(buf + i, sz - i, " %u", 0);
2925 i += scnprintf(buf + i, sz - i, " %u", l->acked);
2926
2927 list = &l->transmq;
2928 len = skb_queue_len(list);
2929 hskb = skb_peek(list);
2930 tskb = skb_peek_tail(list);
2931 i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2932 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2933 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2934
2935 list = &l->deferdq;
2936 len = skb_queue_len(list);
2937 hskb = skb_peek(list);
2938 tskb = skb_peek_tail(list);
2939 i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2940 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2941 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2942
2943 list = &l->backlogq;
2944 len = skb_queue_len(list);
2945 hskb = skb_peek(list);
2946 tskb = skb_peek_tail(list);
2947 i += scnprintf(buf + i, sz - i, " | %u %u %u", len,
2948 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2949 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2950
2951 list = l->inputq;
2952 len = skb_queue_len(list);
2953 hskb = skb_peek(list);
2954 tskb = skb_peek_tail(list);
2955 i += scnprintf(buf + i, sz - i, " | %u %u %u\n", len,
2956 (hskb) ? msg_seqno(buf_msg(hskb)) : 0,
2957 (tskb) ? msg_seqno(buf_msg(tskb)) : 0);
2958
2959 if (dqueues & TIPC_DUMP_TRANSMQ) {
2960 i += scnprintf(buf + i, sz - i, "transmq: ");
2961 i += tipc_list_dump(&l->transmq, false, buf + i);
2962 }
2963 if (dqueues & TIPC_DUMP_BACKLOGQ) {
2964 i += scnprintf(buf + i, sz - i,
2965 "backlogq: <%u %u %u %u %u>, ",
2966 l->backlog[TIPC_LOW_IMPORTANCE].len,
2967 l->backlog[TIPC_MEDIUM_IMPORTANCE].len,
2968 l->backlog[TIPC_HIGH_IMPORTANCE].len,
2969 l->backlog[TIPC_CRITICAL_IMPORTANCE].len,
2970 l->backlog[TIPC_SYSTEM_IMPORTANCE].len);
2971 i += tipc_list_dump(&l->backlogq, false, buf + i);
2972 }
2973 if (dqueues & TIPC_DUMP_DEFERDQ) {
2974 i += scnprintf(buf + i, sz - i, "deferdq: ");
2975 i += tipc_list_dump(&l->deferdq, false, buf + i);
2976 }
2977 if (dqueues & TIPC_DUMP_INPUTQ) {
2978 i += scnprintf(buf + i, sz - i, "inputq: ");
2979 i += tipc_list_dump(l->inputq, false, buf + i);
2980 }
2981 if (dqueues & TIPC_DUMP_WAKEUP) {
2982 i += scnprintf(buf + i, sz - i, "wakeup: ");
2983 i += tipc_list_dump(&l->wakeupq, false, buf + i);
2984 }
2985
2986 return i;
2987}
1/*
2 * net/tipc/link.c: TIPC link code
3 *
4 * Copyright (c) 1996-2007, Ericsson AB
5 * Copyright (c) 2004-2007, 2010-2011, Wind River Systems
6 * All rights reserved.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions are met:
10 *
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the names of the copyright holders nor the names of its
17 * contributors may be used to endorse or promote products derived from
18 * this software without specific prior written permission.
19 *
20 * Alternatively, this software may be distributed under the terms of the
21 * GNU General Public License ("GPL") version 2 as published by the Free
22 * Software Foundation.
23 *
24 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34 * POSSIBILITY OF SUCH DAMAGE.
35 */
36
37#include "core.h"
38#include "link.h"
39#include "port.h"
40#include "name_distr.h"
41#include "discover.h"
42#include "config.h"
43
44
45/*
46 * Out-of-range value for link session numbers
47 */
48#define INVALID_SESSION 0x10000
49
50/*
51 * Link state events:
52 */
53#define STARTING_EVT 856384768 /* link processing trigger */
54#define TRAFFIC_MSG_EVT 560815u /* rx'd ??? */
55#define TIMEOUT_EVT 560817u /* link timer expired */
56
57/*
58 * The following two 'message types' is really just implementation
59 * data conveniently stored in the message header.
60 * They must not be considered part of the protocol
61 */
62#define OPEN_MSG 0
63#define CLOSED_MSG 1
64
65/*
66 * State value stored in 'exp_msg_count'
67 */
68#define START_CHANGEOVER 100000u
69
70/**
71 * struct tipc_link_name - deconstructed link name
72 * @addr_local: network address of node at this end
73 * @if_local: name of interface at this end
74 * @addr_peer: network address of node at far end
75 * @if_peer: name of interface at far end
76 */
77struct tipc_link_name {
78 u32 addr_local;
79 char if_local[TIPC_MAX_IF_NAME];
80 u32 addr_peer;
81 char if_peer[TIPC_MAX_IF_NAME];
82};
83
84static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
85 struct sk_buff *buf);
86static void link_recv_proto_msg(struct tipc_link *l_ptr, struct sk_buff *buf);
87static int link_recv_changeover_msg(struct tipc_link **l_ptr,
88 struct sk_buff **buf);
89static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance);
90static int link_send_sections_long(struct tipc_port *sender,
91 struct iovec const *msg_sect,
92 u32 num_sect, unsigned int total_len,
93 u32 destnode);
94static void link_check_defragm_bufs(struct tipc_link *l_ptr);
95static void link_state_event(struct tipc_link *l_ptr, u32 event);
96static void link_reset_statistics(struct tipc_link *l_ptr);
97static void link_print(struct tipc_link *l_ptr, const char *str);
98static void link_start(struct tipc_link *l_ptr);
99static int link_send_long_buf(struct tipc_link *l_ptr, struct sk_buff *buf);
100
101/*
102 * Simple link routines
103 */
104static unsigned int align(unsigned int i)
105{
106 return (i + 3) & ~3u;
107}
108
109static void link_init_max_pkt(struct tipc_link *l_ptr)
110{
111 u32 max_pkt;
112
113 max_pkt = (l_ptr->b_ptr->mtu & ~3);
114 if (max_pkt > MAX_MSG_SIZE)
115 max_pkt = MAX_MSG_SIZE;
116
117 l_ptr->max_pkt_target = max_pkt;
118 if (l_ptr->max_pkt_target < MAX_PKT_DEFAULT)
119 l_ptr->max_pkt = l_ptr->max_pkt_target;
120 else
121 l_ptr->max_pkt = MAX_PKT_DEFAULT;
122
123 l_ptr->max_pkt_probes = 0;
124}
125
126static u32 link_next_sent(struct tipc_link *l_ptr)
127{
128 if (l_ptr->next_out)
129 return buf_seqno(l_ptr->next_out);
130 return mod(l_ptr->next_out_no);
131}
132
133static u32 link_last_sent(struct tipc_link *l_ptr)
134{
135 return mod(link_next_sent(l_ptr) - 1);
136}
137
138/*
139 * Simple non-static link routines (i.e. referenced outside this file)
140 */
141int tipc_link_is_up(struct tipc_link *l_ptr)
142{
143 if (!l_ptr)
144 return 0;
145 return link_working_working(l_ptr) || link_working_unknown(l_ptr);
146}
147
148int tipc_link_is_active(struct tipc_link *l_ptr)
149{
150 return (l_ptr->owner->active_links[0] == l_ptr) ||
151 (l_ptr->owner->active_links[1] == l_ptr);
152}
153
154/**
155 * link_name_validate - validate & (optionally) deconstruct tipc_link name
156 * @name - ptr to link name string
157 * @name_parts - ptr to area for link name components (or NULL if not needed)
158 *
159 * Returns 1 if link name is valid, otherwise 0.
160 */
161static int link_name_validate(const char *name,
162 struct tipc_link_name *name_parts)
163{
164 char name_copy[TIPC_MAX_LINK_NAME];
165 char *addr_local;
166 char *if_local;
167 char *addr_peer;
168 char *if_peer;
169 char dummy;
170 u32 z_local, c_local, n_local;
171 u32 z_peer, c_peer, n_peer;
172 u32 if_local_len;
173 u32 if_peer_len;
174
175 /* copy link name & ensure length is OK */
176 name_copy[TIPC_MAX_LINK_NAME - 1] = 0;
177 /* need above in case non-Posix strncpy() doesn't pad with nulls */
178 strncpy(name_copy, name, TIPC_MAX_LINK_NAME);
179 if (name_copy[TIPC_MAX_LINK_NAME - 1] != 0)
180 return 0;
181
182 /* ensure all component parts of link name are present */
183 addr_local = name_copy;
184 if_local = strchr(addr_local, ':');
185 if (if_local == NULL)
186 return 0;
187 *(if_local++) = 0;
188 addr_peer = strchr(if_local, '-');
189 if (addr_peer == NULL)
190 return 0;
191 *(addr_peer++) = 0;
192 if_local_len = addr_peer - if_local;
193 if_peer = strchr(addr_peer, ':');
194 if (if_peer == NULL)
195 return 0;
196 *(if_peer++) = 0;
197 if_peer_len = strlen(if_peer) + 1;
198
199 /* validate component parts of link name */
200 if ((sscanf(addr_local, "%u.%u.%u%c",
201 &z_local, &c_local, &n_local, &dummy) != 3) ||
202 (sscanf(addr_peer, "%u.%u.%u%c",
203 &z_peer, &c_peer, &n_peer, &dummy) != 3) ||
204 (z_local > 255) || (c_local > 4095) || (n_local > 4095) ||
205 (z_peer > 255) || (c_peer > 4095) || (n_peer > 4095) ||
206 (if_local_len <= 1) || (if_local_len > TIPC_MAX_IF_NAME) ||
207 (if_peer_len <= 1) || (if_peer_len > TIPC_MAX_IF_NAME) ||
208 (strspn(if_local, tipc_alphabet) != (if_local_len - 1)) ||
209 (strspn(if_peer, tipc_alphabet) != (if_peer_len - 1)))
210 return 0;
211
212 /* return link name components, if necessary */
213 if (name_parts) {
214 name_parts->addr_local = tipc_addr(z_local, c_local, n_local);
215 strcpy(name_parts->if_local, if_local);
216 name_parts->addr_peer = tipc_addr(z_peer, c_peer, n_peer);
217 strcpy(name_parts->if_peer, if_peer);
218 }
219 return 1;
220}
221
222/**
223 * link_timeout - handle expiration of link timer
224 * @l_ptr: pointer to link
225 *
226 * This routine must not grab "tipc_net_lock" to avoid a potential deadlock conflict
227 * with tipc_link_delete(). (There is no risk that the node will be deleted by
228 * another thread because tipc_link_delete() always cancels the link timer before
229 * tipc_node_delete() is called.)
230 */
231static void link_timeout(struct tipc_link *l_ptr)
232{
233 tipc_node_lock(l_ptr->owner);
234
235 /* update counters used in statistical profiling of send traffic */
236 l_ptr->stats.accu_queue_sz += l_ptr->out_queue_size;
237 l_ptr->stats.queue_sz_counts++;
238
239 if (l_ptr->first_out) {
240 struct tipc_msg *msg = buf_msg(l_ptr->first_out);
241 u32 length = msg_size(msg);
242
243 if ((msg_user(msg) == MSG_FRAGMENTER) &&
244 (msg_type(msg) == FIRST_FRAGMENT)) {
245 length = msg_size(msg_get_wrapped(msg));
246 }
247 if (length) {
248 l_ptr->stats.msg_lengths_total += length;
249 l_ptr->stats.msg_length_counts++;
250 if (length <= 64)
251 l_ptr->stats.msg_length_profile[0]++;
252 else if (length <= 256)
253 l_ptr->stats.msg_length_profile[1]++;
254 else if (length <= 1024)
255 l_ptr->stats.msg_length_profile[2]++;
256 else if (length <= 4096)
257 l_ptr->stats.msg_length_profile[3]++;
258 else if (length <= 16384)
259 l_ptr->stats.msg_length_profile[4]++;
260 else if (length <= 32768)
261 l_ptr->stats.msg_length_profile[5]++;
262 else
263 l_ptr->stats.msg_length_profile[6]++;
264 }
265 }
266
267 /* do all other link processing performed on a periodic basis */
268 link_check_defragm_bufs(l_ptr);
269
270 link_state_event(l_ptr, TIMEOUT_EVT);
271
272 if (l_ptr->next_out)
273 tipc_link_push_queue(l_ptr);
274
275 tipc_node_unlock(l_ptr->owner);
276}
277
278static void link_set_timer(struct tipc_link *l_ptr, u32 time)
279{
280 k_start_timer(&l_ptr->timer, time);
281}
282
283/**
284 * tipc_link_create - create a new link
285 * @n_ptr: pointer to associated node
286 * @b_ptr: pointer to associated bearer
287 * @media_addr: media address to use when sending messages over link
288 *
289 * Returns pointer to link.
290 */
291struct tipc_link *tipc_link_create(struct tipc_node *n_ptr,
292 struct tipc_bearer *b_ptr,
293 const struct tipc_media_addr *media_addr)
294{
295 struct tipc_link *l_ptr;
296 struct tipc_msg *msg;
297 char *if_name;
298 char addr_string[16];
299 u32 peer = n_ptr->addr;
300
301 if (n_ptr->link_cnt >= 2) {
302 tipc_addr_string_fill(addr_string, n_ptr->addr);
303 err("Attempt to establish third link to %s\n", addr_string);
304 return NULL;
305 }
306
307 if (n_ptr->links[b_ptr->identity]) {
308 tipc_addr_string_fill(addr_string, n_ptr->addr);
309 err("Attempt to establish second link on <%s> to %s\n",
310 b_ptr->name, addr_string);
311 return NULL;
312 }
313
314 l_ptr = kzalloc(sizeof(*l_ptr), GFP_ATOMIC);
315 if (!l_ptr) {
316 warn("Link creation failed, no memory\n");
317 return NULL;
318 }
319
320 l_ptr->addr = peer;
321 if_name = strchr(b_ptr->name, ':') + 1;
322 sprintf(l_ptr->name, "%u.%u.%u:%s-%u.%u.%u:unknown",
323 tipc_zone(tipc_own_addr), tipc_cluster(tipc_own_addr),
324 tipc_node(tipc_own_addr),
325 if_name,
326 tipc_zone(peer), tipc_cluster(peer), tipc_node(peer));
327 /* note: peer i/f name is updated by reset/activate message */
328 memcpy(&l_ptr->media_addr, media_addr, sizeof(*media_addr));
329 l_ptr->owner = n_ptr;
330 l_ptr->checkpoint = 1;
331 l_ptr->peer_session = INVALID_SESSION;
332 l_ptr->b_ptr = b_ptr;
333 link_set_supervision_props(l_ptr, b_ptr->tolerance);
334 l_ptr->state = RESET_UNKNOWN;
335
336 l_ptr->pmsg = (struct tipc_msg *)&l_ptr->proto_msg;
337 msg = l_ptr->pmsg;
338 tipc_msg_init(msg, LINK_PROTOCOL, RESET_MSG, INT_H_SIZE, l_ptr->addr);
339 msg_set_size(msg, sizeof(l_ptr->proto_msg));
340 msg_set_session(msg, (tipc_random & 0xffff));
341 msg_set_bearer_id(msg, b_ptr->identity);
342 strcpy((char *)msg_data(msg), if_name);
343
344 l_ptr->priority = b_ptr->priority;
345 tipc_link_set_queue_limits(l_ptr, b_ptr->window);
346
347 link_init_max_pkt(l_ptr);
348
349 l_ptr->next_out_no = 1;
350 INIT_LIST_HEAD(&l_ptr->waiting_ports);
351
352 link_reset_statistics(l_ptr);
353
354 tipc_node_attach_link(n_ptr, l_ptr);
355
356 k_init_timer(&l_ptr->timer, (Handler)link_timeout, (unsigned long)l_ptr);
357 list_add_tail(&l_ptr->link_list, &b_ptr->links);
358 tipc_k_signal((Handler)link_start, (unsigned long)l_ptr);
359
360 return l_ptr;
361}
362
363/**
364 * tipc_link_delete - delete a link
365 * @l_ptr: pointer to link
366 *
367 * Note: 'tipc_net_lock' is write_locked, bearer is locked.
368 * This routine must not grab the node lock until after link timer cancellation
369 * to avoid a potential deadlock situation.
370 */
371void tipc_link_delete(struct tipc_link *l_ptr)
372{
373 if (!l_ptr) {
374 err("Attempt to delete non-existent link\n");
375 return;
376 }
377
378 k_cancel_timer(&l_ptr->timer);
379
380 tipc_node_lock(l_ptr->owner);
381 tipc_link_reset(l_ptr);
382 tipc_node_detach_link(l_ptr->owner, l_ptr);
383 tipc_link_stop(l_ptr);
384 list_del_init(&l_ptr->link_list);
385 tipc_node_unlock(l_ptr->owner);
386 k_term_timer(&l_ptr->timer);
387 kfree(l_ptr);
388}
389
390static void link_start(struct tipc_link *l_ptr)
391{
392 tipc_node_lock(l_ptr->owner);
393 link_state_event(l_ptr, STARTING_EVT);
394 tipc_node_unlock(l_ptr->owner);
395}
396
397/**
398 * link_schedule_port - schedule port for deferred sending
399 * @l_ptr: pointer to link
400 * @origport: reference to sending port
401 * @sz: amount of data to be sent
402 *
403 * Schedules port for renewed sending of messages after link congestion
404 * has abated.
405 */
406static int link_schedule_port(struct tipc_link *l_ptr, u32 origport, u32 sz)
407{
408 struct tipc_port *p_ptr;
409
410 spin_lock_bh(&tipc_port_list_lock);
411 p_ptr = tipc_port_lock(origport);
412 if (p_ptr) {
413 if (!p_ptr->wakeup)
414 goto exit;
415 if (!list_empty(&p_ptr->wait_list))
416 goto exit;
417 p_ptr->congested = 1;
418 p_ptr->waiting_pkts = 1 + ((sz - 1) / l_ptr->max_pkt);
419 list_add_tail(&p_ptr->wait_list, &l_ptr->waiting_ports);
420 l_ptr->stats.link_congs++;
421exit:
422 tipc_port_unlock(p_ptr);
423 }
424 spin_unlock_bh(&tipc_port_list_lock);
425 return -ELINKCONG;
426}
427
428void tipc_link_wakeup_ports(struct tipc_link *l_ptr, int all)
429{
430 struct tipc_port *p_ptr;
431 struct tipc_port *temp_p_ptr;
432 int win = l_ptr->queue_limit[0] - l_ptr->out_queue_size;
433
434 if (all)
435 win = 100000;
436 if (win <= 0)
437 return;
438 if (!spin_trylock_bh(&tipc_port_list_lock))
439 return;
440 if (link_congested(l_ptr))
441 goto exit;
442 list_for_each_entry_safe(p_ptr, temp_p_ptr, &l_ptr->waiting_ports,
443 wait_list) {
444 if (win <= 0)
445 break;
446 list_del_init(&p_ptr->wait_list);
447 spin_lock_bh(p_ptr->lock);
448 p_ptr->congested = 0;
449 p_ptr->wakeup(p_ptr);
450 win -= p_ptr->waiting_pkts;
451 spin_unlock_bh(p_ptr->lock);
452 }
453
454exit:
455 spin_unlock_bh(&tipc_port_list_lock);
456}
457
458/**
459 * link_release_outqueue - purge link's outbound message queue
460 * @l_ptr: pointer to link
461 */
462static void link_release_outqueue(struct tipc_link *l_ptr)
463{
464 struct sk_buff *buf = l_ptr->first_out;
465 struct sk_buff *next;
466
467 while (buf) {
468 next = buf->next;
469 kfree_skb(buf);
470 buf = next;
471 }
472 l_ptr->first_out = NULL;
473 l_ptr->out_queue_size = 0;
474}
475
476/**
477 * tipc_link_reset_fragments - purge link's inbound message fragments queue
478 * @l_ptr: pointer to link
479 */
480void tipc_link_reset_fragments(struct tipc_link *l_ptr)
481{
482 struct sk_buff *buf = l_ptr->defragm_buf;
483 struct sk_buff *next;
484
485 while (buf) {
486 next = buf->next;
487 kfree_skb(buf);
488 buf = next;
489 }
490 l_ptr->defragm_buf = NULL;
491}
492
493/**
494 * tipc_link_stop - purge all inbound and outbound messages associated with link
495 * @l_ptr: pointer to link
496 */
497void tipc_link_stop(struct tipc_link *l_ptr)
498{
499 struct sk_buff *buf;
500 struct sk_buff *next;
501
502 buf = l_ptr->oldest_deferred_in;
503 while (buf) {
504 next = buf->next;
505 kfree_skb(buf);
506 buf = next;
507 }
508
509 buf = l_ptr->first_out;
510 while (buf) {
511 next = buf->next;
512 kfree_skb(buf);
513 buf = next;
514 }
515
516 tipc_link_reset_fragments(l_ptr);
517
518 kfree_skb(l_ptr->proto_msg_queue);
519 l_ptr->proto_msg_queue = NULL;
520}
521
522void tipc_link_reset(struct tipc_link *l_ptr)
523{
524 struct sk_buff *buf;
525 u32 prev_state = l_ptr->state;
526 u32 checkpoint = l_ptr->next_in_no;
527 int was_active_link = tipc_link_is_active(l_ptr);
528
529 msg_set_session(l_ptr->pmsg, ((msg_session(l_ptr->pmsg) + 1) & 0xffff));
530
531 /* Link is down, accept any session */
532 l_ptr->peer_session = INVALID_SESSION;
533
534 /* Prepare for max packet size negotiation */
535 link_init_max_pkt(l_ptr);
536
537 l_ptr->state = RESET_UNKNOWN;
538
539 if ((prev_state == RESET_UNKNOWN) || (prev_state == RESET_RESET))
540 return;
541
542 tipc_node_link_down(l_ptr->owner, l_ptr);
543 tipc_bearer_remove_dest(l_ptr->b_ptr, l_ptr->addr);
544
545 if (was_active_link && tipc_node_active_links(l_ptr->owner) &&
546 l_ptr->owner->permit_changeover) {
547 l_ptr->reset_checkpoint = checkpoint;
548 l_ptr->exp_msg_count = START_CHANGEOVER;
549 }
550
551 /* Clean up all queues: */
552 link_release_outqueue(l_ptr);
553 kfree_skb(l_ptr->proto_msg_queue);
554 l_ptr->proto_msg_queue = NULL;
555 buf = l_ptr->oldest_deferred_in;
556 while (buf) {
557 struct sk_buff *next = buf->next;
558 kfree_skb(buf);
559 buf = next;
560 }
561 if (!list_empty(&l_ptr->waiting_ports))
562 tipc_link_wakeup_ports(l_ptr, 1);
563
564 l_ptr->retransm_queue_head = 0;
565 l_ptr->retransm_queue_size = 0;
566 l_ptr->last_out = NULL;
567 l_ptr->first_out = NULL;
568 l_ptr->next_out = NULL;
569 l_ptr->unacked_window = 0;
570 l_ptr->checkpoint = 1;
571 l_ptr->next_out_no = 1;
572 l_ptr->deferred_inqueue_sz = 0;
573 l_ptr->oldest_deferred_in = NULL;
574 l_ptr->newest_deferred_in = NULL;
575 l_ptr->fsm_msg_cnt = 0;
576 l_ptr->stale_count = 0;
577 link_reset_statistics(l_ptr);
578}
579
580
581static void link_activate(struct tipc_link *l_ptr)
582{
583 l_ptr->next_in_no = l_ptr->stats.recv_info = 1;
584 tipc_node_link_up(l_ptr->owner, l_ptr);
585 tipc_bearer_add_dest(l_ptr->b_ptr, l_ptr->addr);
586}
587
588/**
589 * link_state_event - link finite state machine
590 * @l_ptr: pointer to link
591 * @event: state machine event to process
592 */
593static void link_state_event(struct tipc_link *l_ptr, unsigned int event)
594{
595 struct tipc_link *other;
596 u32 cont_intv = l_ptr->continuity_interval;
597
598 if (!l_ptr->started && (event != STARTING_EVT))
599 return; /* Not yet. */
600
601 if (link_blocked(l_ptr)) {
602 if (event == TIMEOUT_EVT)
603 link_set_timer(l_ptr, cont_intv);
604 return; /* Changeover going on */
605 }
606
607 switch (l_ptr->state) {
608 case WORKING_WORKING:
609 switch (event) {
610 case TRAFFIC_MSG_EVT:
611 case ACTIVATE_MSG:
612 break;
613 case TIMEOUT_EVT:
614 if (l_ptr->next_in_no != l_ptr->checkpoint) {
615 l_ptr->checkpoint = l_ptr->next_in_no;
616 if (tipc_bclink_acks_missing(l_ptr->owner)) {
617 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
618 0, 0, 0, 0, 0);
619 l_ptr->fsm_msg_cnt++;
620 } else if (l_ptr->max_pkt < l_ptr->max_pkt_target) {
621 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
622 1, 0, 0, 0, 0);
623 l_ptr->fsm_msg_cnt++;
624 }
625 link_set_timer(l_ptr, cont_intv);
626 break;
627 }
628 l_ptr->state = WORKING_UNKNOWN;
629 l_ptr->fsm_msg_cnt = 0;
630 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
631 l_ptr->fsm_msg_cnt++;
632 link_set_timer(l_ptr, cont_intv / 4);
633 break;
634 case RESET_MSG:
635 info("Resetting link <%s>, requested by peer\n",
636 l_ptr->name);
637 tipc_link_reset(l_ptr);
638 l_ptr->state = RESET_RESET;
639 l_ptr->fsm_msg_cnt = 0;
640 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 0, 0, 0, 0, 0);
641 l_ptr->fsm_msg_cnt++;
642 link_set_timer(l_ptr, cont_intv);
643 break;
644 default:
645 err("Unknown link event %u in WW state\n", event);
646 }
647 break;
648 case WORKING_UNKNOWN:
649 switch (event) {
650 case TRAFFIC_MSG_EVT:
651 case ACTIVATE_MSG:
652 l_ptr->state = WORKING_WORKING;
653 l_ptr->fsm_msg_cnt = 0;
654 link_set_timer(l_ptr, cont_intv);
655 break;
656 case RESET_MSG:
657 info("Resetting link <%s>, requested by peer "
658 "while probing\n", l_ptr->name);
659 tipc_link_reset(l_ptr);
660 l_ptr->state = RESET_RESET;
661 l_ptr->fsm_msg_cnt = 0;
662 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 0, 0, 0, 0, 0);
663 l_ptr->fsm_msg_cnt++;
664 link_set_timer(l_ptr, cont_intv);
665 break;
666 case TIMEOUT_EVT:
667 if (l_ptr->next_in_no != l_ptr->checkpoint) {
668 l_ptr->state = WORKING_WORKING;
669 l_ptr->fsm_msg_cnt = 0;
670 l_ptr->checkpoint = l_ptr->next_in_no;
671 if (tipc_bclink_acks_missing(l_ptr->owner)) {
672 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
673 0, 0, 0, 0, 0);
674 l_ptr->fsm_msg_cnt++;
675 }
676 link_set_timer(l_ptr, cont_intv);
677 } else if (l_ptr->fsm_msg_cnt < l_ptr->abort_limit) {
678 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
679 1, 0, 0, 0, 0);
680 l_ptr->fsm_msg_cnt++;
681 link_set_timer(l_ptr, cont_intv / 4);
682 } else { /* Link has failed */
683 warn("Resetting link <%s>, peer not responding\n",
684 l_ptr->name);
685 tipc_link_reset(l_ptr);
686 l_ptr->state = RESET_UNKNOWN;
687 l_ptr->fsm_msg_cnt = 0;
688 tipc_link_send_proto_msg(l_ptr, RESET_MSG,
689 0, 0, 0, 0, 0);
690 l_ptr->fsm_msg_cnt++;
691 link_set_timer(l_ptr, cont_intv);
692 }
693 break;
694 default:
695 err("Unknown link event %u in WU state\n", event);
696 }
697 break;
698 case RESET_UNKNOWN:
699 switch (event) {
700 case TRAFFIC_MSG_EVT:
701 break;
702 case ACTIVATE_MSG:
703 other = l_ptr->owner->active_links[0];
704 if (other && link_working_unknown(other))
705 break;
706 l_ptr->state = WORKING_WORKING;
707 l_ptr->fsm_msg_cnt = 0;
708 link_activate(l_ptr);
709 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
710 l_ptr->fsm_msg_cnt++;
711 link_set_timer(l_ptr, cont_intv);
712 break;
713 case RESET_MSG:
714 l_ptr->state = RESET_RESET;
715 l_ptr->fsm_msg_cnt = 0;
716 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 1, 0, 0, 0, 0);
717 l_ptr->fsm_msg_cnt++;
718 link_set_timer(l_ptr, cont_intv);
719 break;
720 case STARTING_EVT:
721 l_ptr->started = 1;
722 /* fall through */
723 case TIMEOUT_EVT:
724 tipc_link_send_proto_msg(l_ptr, RESET_MSG, 0, 0, 0, 0, 0);
725 l_ptr->fsm_msg_cnt++;
726 link_set_timer(l_ptr, cont_intv);
727 break;
728 default:
729 err("Unknown link event %u in RU state\n", event);
730 }
731 break;
732 case RESET_RESET:
733 switch (event) {
734 case TRAFFIC_MSG_EVT:
735 case ACTIVATE_MSG:
736 other = l_ptr->owner->active_links[0];
737 if (other && link_working_unknown(other))
738 break;
739 l_ptr->state = WORKING_WORKING;
740 l_ptr->fsm_msg_cnt = 0;
741 link_activate(l_ptr);
742 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 1, 0, 0, 0, 0);
743 l_ptr->fsm_msg_cnt++;
744 link_set_timer(l_ptr, cont_intv);
745 break;
746 case RESET_MSG:
747 break;
748 case TIMEOUT_EVT:
749 tipc_link_send_proto_msg(l_ptr, ACTIVATE_MSG, 0, 0, 0, 0, 0);
750 l_ptr->fsm_msg_cnt++;
751 link_set_timer(l_ptr, cont_intv);
752 break;
753 default:
754 err("Unknown link event %u in RR state\n", event);
755 }
756 break;
757 default:
758 err("Unknown link state %u/%u\n", l_ptr->state, event);
759 }
760}
761
762/*
763 * link_bundle_buf(): Append contents of a buffer to
764 * the tail of an existing one.
765 */
766static int link_bundle_buf(struct tipc_link *l_ptr,
767 struct sk_buff *bundler,
768 struct sk_buff *buf)
769{
770 struct tipc_msg *bundler_msg = buf_msg(bundler);
771 struct tipc_msg *msg = buf_msg(buf);
772 u32 size = msg_size(msg);
773 u32 bundle_size = msg_size(bundler_msg);
774 u32 to_pos = align(bundle_size);
775 u32 pad = to_pos - bundle_size;
776
777 if (msg_user(bundler_msg) != MSG_BUNDLER)
778 return 0;
779 if (msg_type(bundler_msg) != OPEN_MSG)
780 return 0;
781 if (skb_tailroom(bundler) < (pad + size))
782 return 0;
783 if (l_ptr->max_pkt < (to_pos + size))
784 return 0;
785
786 skb_put(bundler, pad + size);
787 skb_copy_to_linear_data_offset(bundler, to_pos, buf->data, size);
788 msg_set_size(bundler_msg, to_pos + size);
789 msg_set_msgcnt(bundler_msg, msg_msgcnt(bundler_msg) + 1);
790 kfree_skb(buf);
791 l_ptr->stats.sent_bundled++;
792 return 1;
793}
794
795static void link_add_to_outqueue(struct tipc_link *l_ptr,
796 struct sk_buff *buf,
797 struct tipc_msg *msg)
798{
799 u32 ack = mod(l_ptr->next_in_no - 1);
800 u32 seqno = mod(l_ptr->next_out_no++);
801
802 msg_set_word(msg, 2, ((ack << 16) | seqno));
803 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
804 buf->next = NULL;
805 if (l_ptr->first_out) {
806 l_ptr->last_out->next = buf;
807 l_ptr->last_out = buf;
808 } else
809 l_ptr->first_out = l_ptr->last_out = buf;
810
811 l_ptr->out_queue_size++;
812 if (l_ptr->out_queue_size > l_ptr->stats.max_queue_sz)
813 l_ptr->stats.max_queue_sz = l_ptr->out_queue_size;
814}
815
816static void link_add_chain_to_outqueue(struct tipc_link *l_ptr,
817 struct sk_buff *buf_chain,
818 u32 long_msgno)
819{
820 struct sk_buff *buf;
821 struct tipc_msg *msg;
822
823 if (!l_ptr->next_out)
824 l_ptr->next_out = buf_chain;
825 while (buf_chain) {
826 buf = buf_chain;
827 buf_chain = buf_chain->next;
828
829 msg = buf_msg(buf);
830 msg_set_long_msgno(msg, long_msgno);
831 link_add_to_outqueue(l_ptr, buf, msg);
832 }
833}
834
835/*
836 * tipc_link_send_buf() is the 'full path' for messages, called from
837 * inside TIPC when the 'fast path' in tipc_send_buf
838 * has failed, and from link_send()
839 */
840int tipc_link_send_buf(struct tipc_link *l_ptr, struct sk_buff *buf)
841{
842 struct tipc_msg *msg = buf_msg(buf);
843 u32 size = msg_size(msg);
844 u32 dsz = msg_data_sz(msg);
845 u32 queue_size = l_ptr->out_queue_size;
846 u32 imp = tipc_msg_tot_importance(msg);
847 u32 queue_limit = l_ptr->queue_limit[imp];
848 u32 max_packet = l_ptr->max_pkt;
849
850 /* Match msg importance against queue limits: */
851 if (unlikely(queue_size >= queue_limit)) {
852 if (imp <= TIPC_CRITICAL_IMPORTANCE) {
853 link_schedule_port(l_ptr, msg_origport(msg), size);
854 kfree_skb(buf);
855 return -ELINKCONG;
856 }
857 kfree_skb(buf);
858 if (imp > CONN_MANAGER) {
859 warn("Resetting link <%s>, send queue full", l_ptr->name);
860 tipc_link_reset(l_ptr);
861 }
862 return dsz;
863 }
864
865 /* Fragmentation needed ? */
866 if (size > max_packet)
867 return link_send_long_buf(l_ptr, buf);
868
869 /* Packet can be queued or sent. */
870 if (likely(!tipc_bearer_congested(l_ptr->b_ptr, l_ptr) &&
871 !link_congested(l_ptr))) {
872 link_add_to_outqueue(l_ptr, buf, msg);
873
874 if (likely(tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr))) {
875 l_ptr->unacked_window = 0;
876 } else {
877 tipc_bearer_schedule(l_ptr->b_ptr, l_ptr);
878 l_ptr->stats.bearer_congs++;
879 l_ptr->next_out = buf;
880 }
881 return dsz;
882 }
883 /* Congestion: can message be bundled ? */
884 if ((msg_user(msg) != CHANGEOVER_PROTOCOL) &&
885 (msg_user(msg) != MSG_FRAGMENTER)) {
886
887 /* Try adding message to an existing bundle */
888 if (l_ptr->next_out &&
889 link_bundle_buf(l_ptr, l_ptr->last_out, buf)) {
890 tipc_bearer_resolve_congestion(l_ptr->b_ptr, l_ptr);
891 return dsz;
892 }
893
894 /* Try creating a new bundle */
895 if (size <= max_packet * 2 / 3) {
896 struct sk_buff *bundler = tipc_buf_acquire(max_packet);
897 struct tipc_msg bundler_hdr;
898
899 if (bundler) {
900 tipc_msg_init(&bundler_hdr, MSG_BUNDLER, OPEN_MSG,
901 INT_H_SIZE, l_ptr->addr);
902 skb_copy_to_linear_data(bundler, &bundler_hdr,
903 INT_H_SIZE);
904 skb_trim(bundler, INT_H_SIZE);
905 link_bundle_buf(l_ptr, bundler, buf);
906 buf = bundler;
907 msg = buf_msg(buf);
908 l_ptr->stats.sent_bundles++;
909 }
910 }
911 }
912 if (!l_ptr->next_out)
913 l_ptr->next_out = buf;
914 link_add_to_outqueue(l_ptr, buf, msg);
915 tipc_bearer_resolve_congestion(l_ptr->b_ptr, l_ptr);
916 return dsz;
917}
918
919/*
920 * tipc_link_send(): same as tipc_link_send_buf(), but the link to use has
921 * not been selected yet, and the the owner node is not locked
922 * Called by TIPC internal users, e.g. the name distributor
923 */
924int tipc_link_send(struct sk_buff *buf, u32 dest, u32 selector)
925{
926 struct tipc_link *l_ptr;
927 struct tipc_node *n_ptr;
928 int res = -ELINKCONG;
929
930 read_lock_bh(&tipc_net_lock);
931 n_ptr = tipc_node_find(dest);
932 if (n_ptr) {
933 tipc_node_lock(n_ptr);
934 l_ptr = n_ptr->active_links[selector & 1];
935 if (l_ptr)
936 res = tipc_link_send_buf(l_ptr, buf);
937 else
938 kfree_skb(buf);
939 tipc_node_unlock(n_ptr);
940 } else {
941 kfree_skb(buf);
942 }
943 read_unlock_bh(&tipc_net_lock);
944 return res;
945}
946
947/*
948 * tipc_link_send_names - send name table entries to new neighbor
949 *
950 * Send routine for bulk delivery of name table messages when contact
951 * with a new neighbor occurs. No link congestion checking is performed
952 * because name table messages *must* be delivered. The messages must be
953 * small enough not to require fragmentation.
954 * Called without any locks held.
955 */
956void tipc_link_send_names(struct list_head *message_list, u32 dest)
957{
958 struct tipc_node *n_ptr;
959 struct tipc_link *l_ptr;
960 struct sk_buff *buf;
961 struct sk_buff *temp_buf;
962
963 if (list_empty(message_list))
964 return;
965
966 read_lock_bh(&tipc_net_lock);
967 n_ptr = tipc_node_find(dest);
968 if (n_ptr) {
969 tipc_node_lock(n_ptr);
970 l_ptr = n_ptr->active_links[0];
971 if (l_ptr) {
972 /* convert circular list to linear list */
973 ((struct sk_buff *)message_list->prev)->next = NULL;
974 link_add_chain_to_outqueue(l_ptr,
975 (struct sk_buff *)message_list->next, 0);
976 tipc_link_push_queue(l_ptr);
977 INIT_LIST_HEAD(message_list);
978 }
979 tipc_node_unlock(n_ptr);
980 }
981 read_unlock_bh(&tipc_net_lock);
982
983 /* discard the messages if they couldn't be sent */
984 list_for_each_safe(buf, temp_buf, ((struct sk_buff *)message_list)) {
985 list_del((struct list_head *)buf);
986 kfree_skb(buf);
987 }
988}
989
990/*
991 * link_send_buf_fast: Entry for data messages where the
992 * destination link is known and the header is complete,
993 * inclusive total message length. Very time critical.
994 * Link is locked. Returns user data length.
995 */
996static int link_send_buf_fast(struct tipc_link *l_ptr, struct sk_buff *buf,
997 u32 *used_max_pkt)
998{
999 struct tipc_msg *msg = buf_msg(buf);
1000 int res = msg_data_sz(msg);
1001
1002 if (likely(!link_congested(l_ptr))) {
1003 if (likely(msg_size(msg) <= l_ptr->max_pkt)) {
1004 if (likely(list_empty(&l_ptr->b_ptr->cong_links))) {
1005 link_add_to_outqueue(l_ptr, buf, msg);
1006 if (likely(tipc_bearer_send(l_ptr->b_ptr, buf,
1007 &l_ptr->media_addr))) {
1008 l_ptr->unacked_window = 0;
1009 return res;
1010 }
1011 tipc_bearer_schedule(l_ptr->b_ptr, l_ptr);
1012 l_ptr->stats.bearer_congs++;
1013 l_ptr->next_out = buf;
1014 return res;
1015 }
1016 } else
1017 *used_max_pkt = l_ptr->max_pkt;
1018 }
1019 return tipc_link_send_buf(l_ptr, buf); /* All other cases */
1020}
1021
1022/*
1023 * tipc_send_buf_fast: Entry for data messages where the
1024 * destination node is known and the header is complete,
1025 * inclusive total message length.
1026 * Returns user data length.
1027 */
1028int tipc_send_buf_fast(struct sk_buff *buf, u32 destnode)
1029{
1030 struct tipc_link *l_ptr;
1031 struct tipc_node *n_ptr;
1032 int res;
1033 u32 selector = msg_origport(buf_msg(buf)) & 1;
1034 u32 dummy;
1035
1036 read_lock_bh(&tipc_net_lock);
1037 n_ptr = tipc_node_find(destnode);
1038 if (likely(n_ptr)) {
1039 tipc_node_lock(n_ptr);
1040 l_ptr = n_ptr->active_links[selector];
1041 if (likely(l_ptr)) {
1042 res = link_send_buf_fast(l_ptr, buf, &dummy);
1043 tipc_node_unlock(n_ptr);
1044 read_unlock_bh(&tipc_net_lock);
1045 return res;
1046 }
1047 tipc_node_unlock(n_ptr);
1048 }
1049 read_unlock_bh(&tipc_net_lock);
1050 res = msg_data_sz(buf_msg(buf));
1051 tipc_reject_msg(buf, TIPC_ERR_NO_NODE);
1052 return res;
1053}
1054
1055
1056/*
1057 * tipc_link_send_sections_fast: Entry for messages where the
1058 * destination processor is known and the header is complete,
1059 * except for total message length.
1060 * Returns user data length or errno.
1061 */
1062int tipc_link_send_sections_fast(struct tipc_port *sender,
1063 struct iovec const *msg_sect,
1064 const u32 num_sect,
1065 unsigned int total_len,
1066 u32 destaddr)
1067{
1068 struct tipc_msg *hdr = &sender->phdr;
1069 struct tipc_link *l_ptr;
1070 struct sk_buff *buf;
1071 struct tipc_node *node;
1072 int res;
1073 u32 selector = msg_origport(hdr) & 1;
1074
1075again:
1076 /*
1077 * Try building message using port's max_pkt hint.
1078 * (Must not hold any locks while building message.)
1079 */
1080 res = tipc_msg_build(hdr, msg_sect, num_sect, total_len,
1081 sender->max_pkt, !sender->user_port, &buf);
1082
1083 read_lock_bh(&tipc_net_lock);
1084 node = tipc_node_find(destaddr);
1085 if (likely(node)) {
1086 tipc_node_lock(node);
1087 l_ptr = node->active_links[selector];
1088 if (likely(l_ptr)) {
1089 if (likely(buf)) {
1090 res = link_send_buf_fast(l_ptr, buf,
1091 &sender->max_pkt);
1092exit:
1093 tipc_node_unlock(node);
1094 read_unlock_bh(&tipc_net_lock);
1095 return res;
1096 }
1097
1098 /* Exit if build request was invalid */
1099 if (unlikely(res < 0))
1100 goto exit;
1101
1102 /* Exit if link (or bearer) is congested */
1103 if (link_congested(l_ptr) ||
1104 !list_empty(&l_ptr->b_ptr->cong_links)) {
1105 res = link_schedule_port(l_ptr,
1106 sender->ref, res);
1107 goto exit;
1108 }
1109
1110 /*
1111 * Message size exceeds max_pkt hint; update hint,
1112 * then re-try fast path or fragment the message
1113 */
1114 sender->max_pkt = l_ptr->max_pkt;
1115 tipc_node_unlock(node);
1116 read_unlock_bh(&tipc_net_lock);
1117
1118
1119 if ((msg_hdr_sz(hdr) + res) <= sender->max_pkt)
1120 goto again;
1121
1122 return link_send_sections_long(sender, msg_sect,
1123 num_sect, total_len,
1124 destaddr);
1125 }
1126 tipc_node_unlock(node);
1127 }
1128 read_unlock_bh(&tipc_net_lock);
1129
1130 /* Couldn't find a link to the destination node */
1131 if (buf)
1132 return tipc_reject_msg(buf, TIPC_ERR_NO_NODE);
1133 if (res >= 0)
1134 return tipc_port_reject_sections(sender, hdr, msg_sect, num_sect,
1135 total_len, TIPC_ERR_NO_NODE);
1136 return res;
1137}
1138
1139/*
1140 * link_send_sections_long(): Entry for long messages where the
1141 * destination node is known and the header is complete,
1142 * inclusive total message length.
1143 * Link and bearer congestion status have been checked to be ok,
1144 * and are ignored if they change.
1145 *
1146 * Note that fragments do not use the full link MTU so that they won't have
1147 * to undergo refragmentation if link changeover causes them to be sent
1148 * over another link with an additional tunnel header added as prefix.
1149 * (Refragmentation will still occur if the other link has a smaller MTU.)
1150 *
1151 * Returns user data length or errno.
1152 */
1153static int link_send_sections_long(struct tipc_port *sender,
1154 struct iovec const *msg_sect,
1155 u32 num_sect,
1156 unsigned int total_len,
1157 u32 destaddr)
1158{
1159 struct tipc_link *l_ptr;
1160 struct tipc_node *node;
1161 struct tipc_msg *hdr = &sender->phdr;
1162 u32 dsz = total_len;
1163 u32 max_pkt, fragm_sz, rest;
1164 struct tipc_msg fragm_hdr;
1165 struct sk_buff *buf, *buf_chain, *prev;
1166 u32 fragm_crs, fragm_rest, hsz, sect_rest;
1167 const unchar *sect_crs;
1168 int curr_sect;
1169 u32 fragm_no;
1170
1171again:
1172 fragm_no = 1;
1173 max_pkt = sender->max_pkt - INT_H_SIZE;
1174 /* leave room for tunnel header in case of link changeover */
1175 fragm_sz = max_pkt - INT_H_SIZE;
1176 /* leave room for fragmentation header in each fragment */
1177 rest = dsz;
1178 fragm_crs = 0;
1179 fragm_rest = 0;
1180 sect_rest = 0;
1181 sect_crs = NULL;
1182 curr_sect = -1;
1183
1184 /* Prepare reusable fragment header */
1185 tipc_msg_init(&fragm_hdr, MSG_FRAGMENTER, FIRST_FRAGMENT,
1186 INT_H_SIZE, msg_destnode(hdr));
1187 msg_set_size(&fragm_hdr, max_pkt);
1188 msg_set_fragm_no(&fragm_hdr, 1);
1189
1190 /* Prepare header of first fragment */
1191 buf_chain = buf = tipc_buf_acquire(max_pkt);
1192 if (!buf)
1193 return -ENOMEM;
1194 buf->next = NULL;
1195 skb_copy_to_linear_data(buf, &fragm_hdr, INT_H_SIZE);
1196 hsz = msg_hdr_sz(hdr);
1197 skb_copy_to_linear_data_offset(buf, INT_H_SIZE, hdr, hsz);
1198
1199 /* Chop up message */
1200 fragm_crs = INT_H_SIZE + hsz;
1201 fragm_rest = fragm_sz - hsz;
1202
1203 do { /* For all sections */
1204 u32 sz;
1205
1206 if (!sect_rest) {
1207 sect_rest = msg_sect[++curr_sect].iov_len;
1208 sect_crs = (const unchar *)msg_sect[curr_sect].iov_base;
1209 }
1210
1211 if (sect_rest < fragm_rest)
1212 sz = sect_rest;
1213 else
1214 sz = fragm_rest;
1215
1216 if (likely(!sender->user_port)) {
1217 if (copy_from_user(buf->data + fragm_crs, sect_crs, sz)) {
1218error:
1219 for (; buf_chain; buf_chain = buf) {
1220 buf = buf_chain->next;
1221 kfree_skb(buf_chain);
1222 }
1223 return -EFAULT;
1224 }
1225 } else
1226 skb_copy_to_linear_data_offset(buf, fragm_crs,
1227 sect_crs, sz);
1228 sect_crs += sz;
1229 sect_rest -= sz;
1230 fragm_crs += sz;
1231 fragm_rest -= sz;
1232 rest -= sz;
1233
1234 if (!fragm_rest && rest) {
1235
1236 /* Initiate new fragment: */
1237 if (rest <= fragm_sz) {
1238 fragm_sz = rest;
1239 msg_set_type(&fragm_hdr, LAST_FRAGMENT);
1240 } else {
1241 msg_set_type(&fragm_hdr, FRAGMENT);
1242 }
1243 msg_set_size(&fragm_hdr, fragm_sz + INT_H_SIZE);
1244 msg_set_fragm_no(&fragm_hdr, ++fragm_no);
1245 prev = buf;
1246 buf = tipc_buf_acquire(fragm_sz + INT_H_SIZE);
1247 if (!buf)
1248 goto error;
1249
1250 buf->next = NULL;
1251 prev->next = buf;
1252 skb_copy_to_linear_data(buf, &fragm_hdr, INT_H_SIZE);
1253 fragm_crs = INT_H_SIZE;
1254 fragm_rest = fragm_sz;
1255 }
1256 } while (rest > 0);
1257
1258 /*
1259 * Now we have a buffer chain. Select a link and check
1260 * that packet size is still OK
1261 */
1262 node = tipc_node_find(destaddr);
1263 if (likely(node)) {
1264 tipc_node_lock(node);
1265 l_ptr = node->active_links[sender->ref & 1];
1266 if (!l_ptr) {
1267 tipc_node_unlock(node);
1268 goto reject;
1269 }
1270 if (l_ptr->max_pkt < max_pkt) {
1271 sender->max_pkt = l_ptr->max_pkt;
1272 tipc_node_unlock(node);
1273 for (; buf_chain; buf_chain = buf) {
1274 buf = buf_chain->next;
1275 kfree_skb(buf_chain);
1276 }
1277 goto again;
1278 }
1279 } else {
1280reject:
1281 for (; buf_chain; buf_chain = buf) {
1282 buf = buf_chain->next;
1283 kfree_skb(buf_chain);
1284 }
1285 return tipc_port_reject_sections(sender, hdr, msg_sect, num_sect,
1286 total_len, TIPC_ERR_NO_NODE);
1287 }
1288
1289 /* Append chain of fragments to send queue & send them */
1290 l_ptr->long_msg_seq_no++;
1291 link_add_chain_to_outqueue(l_ptr, buf_chain, l_ptr->long_msg_seq_no);
1292 l_ptr->stats.sent_fragments += fragm_no;
1293 l_ptr->stats.sent_fragmented++;
1294 tipc_link_push_queue(l_ptr);
1295 tipc_node_unlock(node);
1296 return dsz;
1297}
1298
1299/*
1300 * tipc_link_push_packet: Push one unsent packet to the media
1301 */
1302u32 tipc_link_push_packet(struct tipc_link *l_ptr)
1303{
1304 struct sk_buff *buf = l_ptr->first_out;
1305 u32 r_q_size = l_ptr->retransm_queue_size;
1306 u32 r_q_head = l_ptr->retransm_queue_head;
1307
1308 /* Step to position where retransmission failed, if any, */
1309 /* consider that buffers may have been released in meantime */
1310 if (r_q_size && buf) {
1311 u32 last = lesser(mod(r_q_head + r_q_size),
1312 link_last_sent(l_ptr));
1313 u32 first = buf_seqno(buf);
1314
1315 while (buf && less(first, r_q_head)) {
1316 first = mod(first + 1);
1317 buf = buf->next;
1318 }
1319 l_ptr->retransm_queue_head = r_q_head = first;
1320 l_ptr->retransm_queue_size = r_q_size = mod(last - first);
1321 }
1322
1323 /* Continue retransmission now, if there is anything: */
1324 if (r_q_size && buf) {
1325 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
1326 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
1327 if (tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr)) {
1328 l_ptr->retransm_queue_head = mod(++r_q_head);
1329 l_ptr->retransm_queue_size = --r_q_size;
1330 l_ptr->stats.retransmitted++;
1331 return 0;
1332 } else {
1333 l_ptr->stats.bearer_congs++;
1334 return PUSH_FAILED;
1335 }
1336 }
1337
1338 /* Send deferred protocol message, if any: */
1339 buf = l_ptr->proto_msg_queue;
1340 if (buf) {
1341 msg_set_ack(buf_msg(buf), mod(l_ptr->next_in_no - 1));
1342 msg_set_bcast_ack(buf_msg(buf), l_ptr->owner->bclink.last_in);
1343 if (tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr)) {
1344 l_ptr->unacked_window = 0;
1345 kfree_skb(buf);
1346 l_ptr->proto_msg_queue = NULL;
1347 return 0;
1348 } else {
1349 l_ptr->stats.bearer_congs++;
1350 return PUSH_FAILED;
1351 }
1352 }
1353
1354 /* Send one deferred data message, if send window not full: */
1355 buf = l_ptr->next_out;
1356 if (buf) {
1357 struct tipc_msg *msg = buf_msg(buf);
1358 u32 next = msg_seqno(msg);
1359 u32 first = buf_seqno(l_ptr->first_out);
1360
1361 if (mod(next - first) < l_ptr->queue_limit[0]) {
1362 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1363 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1364 if (tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr)) {
1365 if (msg_user(msg) == MSG_BUNDLER)
1366 msg_set_type(msg, CLOSED_MSG);
1367 l_ptr->next_out = buf->next;
1368 return 0;
1369 } else {
1370 l_ptr->stats.bearer_congs++;
1371 return PUSH_FAILED;
1372 }
1373 }
1374 }
1375 return PUSH_FINISHED;
1376}
1377
1378/*
1379 * push_queue(): push out the unsent messages of a link where
1380 * congestion has abated. Node is locked
1381 */
1382void tipc_link_push_queue(struct tipc_link *l_ptr)
1383{
1384 u32 res;
1385
1386 if (tipc_bearer_congested(l_ptr->b_ptr, l_ptr))
1387 return;
1388
1389 do {
1390 res = tipc_link_push_packet(l_ptr);
1391 } while (!res);
1392
1393 if (res == PUSH_FAILED)
1394 tipc_bearer_schedule(l_ptr->b_ptr, l_ptr);
1395}
1396
1397static void link_reset_all(unsigned long addr)
1398{
1399 struct tipc_node *n_ptr;
1400 char addr_string[16];
1401 u32 i;
1402
1403 read_lock_bh(&tipc_net_lock);
1404 n_ptr = tipc_node_find((u32)addr);
1405 if (!n_ptr) {
1406 read_unlock_bh(&tipc_net_lock);
1407 return; /* node no longer exists */
1408 }
1409
1410 tipc_node_lock(n_ptr);
1411
1412 warn("Resetting all links to %s\n",
1413 tipc_addr_string_fill(addr_string, n_ptr->addr));
1414
1415 for (i = 0; i < MAX_BEARERS; i++) {
1416 if (n_ptr->links[i]) {
1417 link_print(n_ptr->links[i], "Resetting link\n");
1418 tipc_link_reset(n_ptr->links[i]);
1419 }
1420 }
1421
1422 tipc_node_unlock(n_ptr);
1423 read_unlock_bh(&tipc_net_lock);
1424}
1425
1426static void link_retransmit_failure(struct tipc_link *l_ptr,
1427 struct sk_buff *buf)
1428{
1429 struct tipc_msg *msg = buf_msg(buf);
1430
1431 warn("Retransmission failure on link <%s>\n", l_ptr->name);
1432
1433 if (l_ptr->addr) {
1434 /* Handle failure on standard link */
1435 link_print(l_ptr, "Resetting link\n");
1436 tipc_link_reset(l_ptr);
1437
1438 } else {
1439 /* Handle failure on broadcast link */
1440 struct tipc_node *n_ptr;
1441 char addr_string[16];
1442
1443 info("Msg seq number: %u, ", msg_seqno(msg));
1444 info("Outstanding acks: %lu\n",
1445 (unsigned long) TIPC_SKB_CB(buf)->handle);
1446
1447 n_ptr = tipc_bclink_retransmit_to();
1448 tipc_node_lock(n_ptr);
1449
1450 tipc_addr_string_fill(addr_string, n_ptr->addr);
1451 info("Broadcast link info for %s\n", addr_string);
1452 info("Supportable: %d, ", n_ptr->bclink.supportable);
1453 info("Supported: %d, ", n_ptr->bclink.supported);
1454 info("Acked: %u\n", n_ptr->bclink.acked);
1455 info("Last in: %u, ", n_ptr->bclink.last_in);
1456 info("Oos state: %u, ", n_ptr->bclink.oos_state);
1457 info("Last sent: %u\n", n_ptr->bclink.last_sent);
1458
1459 tipc_k_signal((Handler)link_reset_all, (unsigned long)n_ptr->addr);
1460
1461 tipc_node_unlock(n_ptr);
1462
1463 l_ptr->stale_count = 0;
1464 }
1465}
1466
1467void tipc_link_retransmit(struct tipc_link *l_ptr, struct sk_buff *buf,
1468 u32 retransmits)
1469{
1470 struct tipc_msg *msg;
1471
1472 if (!buf)
1473 return;
1474
1475 msg = buf_msg(buf);
1476
1477 if (tipc_bearer_congested(l_ptr->b_ptr, l_ptr)) {
1478 if (l_ptr->retransm_queue_size == 0) {
1479 l_ptr->retransm_queue_head = msg_seqno(msg);
1480 l_ptr->retransm_queue_size = retransmits;
1481 } else {
1482 err("Unexpected retransmit on link %s (qsize=%d)\n",
1483 l_ptr->name, l_ptr->retransm_queue_size);
1484 }
1485 return;
1486 } else {
1487 /* Detect repeated retransmit failures on uncongested bearer */
1488 if (l_ptr->last_retransmitted == msg_seqno(msg)) {
1489 if (++l_ptr->stale_count > 100) {
1490 link_retransmit_failure(l_ptr, buf);
1491 return;
1492 }
1493 } else {
1494 l_ptr->last_retransmitted = msg_seqno(msg);
1495 l_ptr->stale_count = 1;
1496 }
1497 }
1498
1499 while (retransmits && (buf != l_ptr->next_out) && buf) {
1500 msg = buf_msg(buf);
1501 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1502 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1503 if (tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr)) {
1504 buf = buf->next;
1505 retransmits--;
1506 l_ptr->stats.retransmitted++;
1507 } else {
1508 tipc_bearer_schedule(l_ptr->b_ptr, l_ptr);
1509 l_ptr->stats.bearer_congs++;
1510 l_ptr->retransm_queue_head = buf_seqno(buf);
1511 l_ptr->retransm_queue_size = retransmits;
1512 return;
1513 }
1514 }
1515
1516 l_ptr->retransm_queue_head = l_ptr->retransm_queue_size = 0;
1517}
1518
1519/**
1520 * link_insert_deferred_queue - insert deferred messages back into receive chain
1521 */
1522static struct sk_buff *link_insert_deferred_queue(struct tipc_link *l_ptr,
1523 struct sk_buff *buf)
1524{
1525 u32 seq_no;
1526
1527 if (l_ptr->oldest_deferred_in == NULL)
1528 return buf;
1529
1530 seq_no = buf_seqno(l_ptr->oldest_deferred_in);
1531 if (seq_no == mod(l_ptr->next_in_no)) {
1532 l_ptr->newest_deferred_in->next = buf;
1533 buf = l_ptr->oldest_deferred_in;
1534 l_ptr->oldest_deferred_in = NULL;
1535 l_ptr->deferred_inqueue_sz = 0;
1536 }
1537 return buf;
1538}
1539
1540/**
1541 * link_recv_buf_validate - validate basic format of received message
1542 *
1543 * This routine ensures a TIPC message has an acceptable header, and at least
1544 * as much data as the header indicates it should. The routine also ensures
1545 * that the entire message header is stored in the main fragment of the message
1546 * buffer, to simplify future access to message header fields.
1547 *
1548 * Note: Having extra info present in the message header or data areas is OK.
1549 * TIPC will ignore the excess, under the assumption that it is optional info
1550 * introduced by a later release of the protocol.
1551 */
1552static int link_recv_buf_validate(struct sk_buff *buf)
1553{
1554 static u32 min_data_hdr_size[8] = {
1555 SHORT_H_SIZE, MCAST_H_SIZE, NAMED_H_SIZE, BASIC_H_SIZE,
1556 MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE, MAX_H_SIZE
1557 };
1558
1559 struct tipc_msg *msg;
1560 u32 tipc_hdr[2];
1561 u32 size;
1562 u32 hdr_size;
1563 u32 min_hdr_size;
1564
1565 if (unlikely(buf->len < MIN_H_SIZE))
1566 return 0;
1567
1568 msg = skb_header_pointer(buf, 0, sizeof(tipc_hdr), tipc_hdr);
1569 if (msg == NULL)
1570 return 0;
1571
1572 if (unlikely(msg_version(msg) != TIPC_VERSION))
1573 return 0;
1574
1575 size = msg_size(msg);
1576 hdr_size = msg_hdr_sz(msg);
1577 min_hdr_size = msg_isdata(msg) ?
1578 min_data_hdr_size[msg_type(msg)] : INT_H_SIZE;
1579
1580 if (unlikely((hdr_size < min_hdr_size) ||
1581 (size < hdr_size) ||
1582 (buf->len < size) ||
1583 (size - hdr_size > TIPC_MAX_USER_MSG_SIZE)))
1584 return 0;
1585
1586 return pskb_may_pull(buf, hdr_size);
1587}
1588
1589/**
1590 * tipc_recv_msg - process TIPC messages arriving from off-node
1591 * @head: pointer to message buffer chain
1592 * @tb_ptr: pointer to bearer message arrived on
1593 *
1594 * Invoked with no locks held. Bearer pointer must point to a valid bearer
1595 * structure (i.e. cannot be NULL), but bearer can be inactive.
1596 */
1597void tipc_recv_msg(struct sk_buff *head, struct tipc_bearer *b_ptr)
1598{
1599 read_lock_bh(&tipc_net_lock);
1600 while (head) {
1601 struct tipc_node *n_ptr;
1602 struct tipc_link *l_ptr;
1603 struct sk_buff *crs;
1604 struct sk_buff *buf = head;
1605 struct tipc_msg *msg;
1606 u32 seq_no;
1607 u32 ackd;
1608 u32 released = 0;
1609 int type;
1610
1611 head = head->next;
1612
1613 /* Ensure bearer is still enabled */
1614 if (unlikely(!b_ptr->active))
1615 goto cont;
1616
1617 /* Ensure message is well-formed */
1618 if (unlikely(!link_recv_buf_validate(buf)))
1619 goto cont;
1620
1621 /* Ensure message data is a single contiguous unit */
1622 if (unlikely(skb_linearize(buf)))
1623 goto cont;
1624
1625 /* Handle arrival of a non-unicast link message */
1626 msg = buf_msg(buf);
1627
1628 if (unlikely(msg_non_seq(msg))) {
1629 if (msg_user(msg) == LINK_CONFIG)
1630 tipc_disc_recv_msg(buf, b_ptr);
1631 else
1632 tipc_bclink_recv_pkt(buf);
1633 continue;
1634 }
1635
1636 /* Discard unicast link messages destined for another node */
1637 if (unlikely(!msg_short(msg) &&
1638 (msg_destnode(msg) != tipc_own_addr)))
1639 goto cont;
1640
1641 /* Locate neighboring node that sent message */
1642 n_ptr = tipc_node_find(msg_prevnode(msg));
1643 if (unlikely(!n_ptr))
1644 goto cont;
1645 tipc_node_lock(n_ptr);
1646
1647 /* Locate unicast link endpoint that should handle message */
1648 l_ptr = n_ptr->links[b_ptr->identity];
1649 if (unlikely(!l_ptr)) {
1650 tipc_node_unlock(n_ptr);
1651 goto cont;
1652 }
1653
1654 /* Verify that communication with node is currently allowed */
1655 if ((n_ptr->block_setup & WAIT_PEER_DOWN) &&
1656 msg_user(msg) == LINK_PROTOCOL &&
1657 (msg_type(msg) == RESET_MSG ||
1658 msg_type(msg) == ACTIVATE_MSG) &&
1659 !msg_redundant_link(msg))
1660 n_ptr->block_setup &= ~WAIT_PEER_DOWN;
1661
1662 if (n_ptr->block_setup) {
1663 tipc_node_unlock(n_ptr);
1664 goto cont;
1665 }
1666
1667 /* Validate message sequence number info */
1668 seq_no = msg_seqno(msg);
1669 ackd = msg_ack(msg);
1670
1671 /* Release acked messages */
1672 if (n_ptr->bclink.supported)
1673 tipc_bclink_acknowledge(n_ptr, msg_bcast_ack(msg));
1674
1675 crs = l_ptr->first_out;
1676 while ((crs != l_ptr->next_out) &&
1677 less_eq(buf_seqno(crs), ackd)) {
1678 struct sk_buff *next = crs->next;
1679
1680 kfree_skb(crs);
1681 crs = next;
1682 released++;
1683 }
1684 if (released) {
1685 l_ptr->first_out = crs;
1686 l_ptr->out_queue_size -= released;
1687 }
1688
1689 /* Try sending any messages link endpoint has pending */
1690 if (unlikely(l_ptr->next_out))
1691 tipc_link_push_queue(l_ptr);
1692 if (unlikely(!list_empty(&l_ptr->waiting_ports)))
1693 tipc_link_wakeup_ports(l_ptr, 0);
1694 if (unlikely(++l_ptr->unacked_window >= TIPC_MIN_LINK_WIN)) {
1695 l_ptr->stats.sent_acks++;
1696 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1697 }
1698
1699 /* Now (finally!) process the incoming message */
1700protocol_check:
1701 if (likely(link_working_working(l_ptr))) {
1702 if (likely(seq_no == mod(l_ptr->next_in_no))) {
1703 l_ptr->next_in_no++;
1704 if (unlikely(l_ptr->oldest_deferred_in))
1705 head = link_insert_deferred_queue(l_ptr,
1706 head);
1707deliver:
1708 if (likely(msg_isdata(msg))) {
1709 tipc_node_unlock(n_ptr);
1710 tipc_port_recv_msg(buf);
1711 continue;
1712 }
1713 switch (msg_user(msg)) {
1714 int ret;
1715 case MSG_BUNDLER:
1716 l_ptr->stats.recv_bundles++;
1717 l_ptr->stats.recv_bundled +=
1718 msg_msgcnt(msg);
1719 tipc_node_unlock(n_ptr);
1720 tipc_link_recv_bundle(buf);
1721 continue;
1722 case NAME_DISTRIBUTOR:
1723 tipc_node_unlock(n_ptr);
1724 tipc_named_recv(buf);
1725 continue;
1726 case CONN_MANAGER:
1727 tipc_node_unlock(n_ptr);
1728 tipc_port_recv_proto_msg(buf);
1729 continue;
1730 case MSG_FRAGMENTER:
1731 l_ptr->stats.recv_fragments++;
1732 ret = tipc_link_recv_fragment(
1733 &l_ptr->defragm_buf,
1734 &buf, &msg);
1735 if (ret == 1) {
1736 l_ptr->stats.recv_fragmented++;
1737 goto deliver;
1738 }
1739 if (ret == -1)
1740 l_ptr->next_in_no--;
1741 break;
1742 case CHANGEOVER_PROTOCOL:
1743 type = msg_type(msg);
1744 if (link_recv_changeover_msg(&l_ptr,
1745 &buf)) {
1746 msg = buf_msg(buf);
1747 seq_no = msg_seqno(msg);
1748 if (type == ORIGINAL_MSG)
1749 goto deliver;
1750 goto protocol_check;
1751 }
1752 break;
1753 default:
1754 kfree_skb(buf);
1755 buf = NULL;
1756 break;
1757 }
1758 tipc_node_unlock(n_ptr);
1759 tipc_net_route_msg(buf);
1760 continue;
1761 }
1762 link_handle_out_of_seq_msg(l_ptr, buf);
1763 head = link_insert_deferred_queue(l_ptr, head);
1764 tipc_node_unlock(n_ptr);
1765 continue;
1766 }
1767
1768 if (msg_user(msg) == LINK_PROTOCOL) {
1769 link_recv_proto_msg(l_ptr, buf);
1770 head = link_insert_deferred_queue(l_ptr, head);
1771 tipc_node_unlock(n_ptr);
1772 continue;
1773 }
1774 link_state_event(l_ptr, TRAFFIC_MSG_EVT);
1775
1776 if (link_working_working(l_ptr)) {
1777 /* Re-insert in front of queue */
1778 buf->next = head;
1779 head = buf;
1780 tipc_node_unlock(n_ptr);
1781 continue;
1782 }
1783 tipc_node_unlock(n_ptr);
1784cont:
1785 kfree_skb(buf);
1786 }
1787 read_unlock_bh(&tipc_net_lock);
1788}
1789
1790/*
1791 * tipc_link_defer_pkt - Add out-of-sequence message to deferred reception queue
1792 *
1793 * Returns increase in queue length (i.e. 0 or 1)
1794 */
1795u32 tipc_link_defer_pkt(struct sk_buff **head, struct sk_buff **tail,
1796 struct sk_buff *buf)
1797{
1798 struct sk_buff *queue_buf;
1799 struct sk_buff **prev;
1800 u32 seq_no = buf_seqno(buf);
1801
1802 buf->next = NULL;
1803
1804 /* Empty queue ? */
1805 if (*head == NULL) {
1806 *head = *tail = buf;
1807 return 1;
1808 }
1809
1810 /* Last ? */
1811 if (less(buf_seqno(*tail), seq_no)) {
1812 (*tail)->next = buf;
1813 *tail = buf;
1814 return 1;
1815 }
1816
1817 /* Locate insertion point in queue, then insert; discard if duplicate */
1818 prev = head;
1819 queue_buf = *head;
1820 for (;;) {
1821 u32 curr_seqno = buf_seqno(queue_buf);
1822
1823 if (seq_no == curr_seqno) {
1824 kfree_skb(buf);
1825 return 0;
1826 }
1827
1828 if (less(seq_no, curr_seqno))
1829 break;
1830
1831 prev = &queue_buf->next;
1832 queue_buf = queue_buf->next;
1833 }
1834
1835 buf->next = queue_buf;
1836 *prev = buf;
1837 return 1;
1838}
1839
1840/*
1841 * link_handle_out_of_seq_msg - handle arrival of out-of-sequence packet
1842 */
1843static void link_handle_out_of_seq_msg(struct tipc_link *l_ptr,
1844 struct sk_buff *buf)
1845{
1846 u32 seq_no = buf_seqno(buf);
1847
1848 if (likely(msg_user(buf_msg(buf)) == LINK_PROTOCOL)) {
1849 link_recv_proto_msg(l_ptr, buf);
1850 return;
1851 }
1852
1853 /* Record OOS packet arrival (force mismatch on next timeout) */
1854 l_ptr->checkpoint--;
1855
1856 /*
1857 * Discard packet if a duplicate; otherwise add it to deferred queue
1858 * and notify peer of gap as per protocol specification
1859 */
1860 if (less(seq_no, mod(l_ptr->next_in_no))) {
1861 l_ptr->stats.duplicates++;
1862 kfree_skb(buf);
1863 return;
1864 }
1865
1866 if (tipc_link_defer_pkt(&l_ptr->oldest_deferred_in,
1867 &l_ptr->newest_deferred_in, buf)) {
1868 l_ptr->deferred_inqueue_sz++;
1869 l_ptr->stats.deferred_recv++;
1870 if ((l_ptr->deferred_inqueue_sz % 16) == 1)
1871 tipc_link_send_proto_msg(l_ptr, STATE_MSG, 0, 0, 0, 0, 0);
1872 } else
1873 l_ptr->stats.duplicates++;
1874}
1875
1876/*
1877 * Send protocol message to the other endpoint.
1878 */
1879void tipc_link_send_proto_msg(struct tipc_link *l_ptr, u32 msg_typ,
1880 int probe_msg, u32 gap, u32 tolerance,
1881 u32 priority, u32 ack_mtu)
1882{
1883 struct sk_buff *buf = NULL;
1884 struct tipc_msg *msg = l_ptr->pmsg;
1885 u32 msg_size = sizeof(l_ptr->proto_msg);
1886 int r_flag;
1887
1888 /* Discard any previous message that was deferred due to congestion */
1889 if (l_ptr->proto_msg_queue) {
1890 kfree_skb(l_ptr->proto_msg_queue);
1891 l_ptr->proto_msg_queue = NULL;
1892 }
1893
1894 if (link_blocked(l_ptr))
1895 return;
1896
1897 /* Abort non-RESET send if communication with node is prohibited */
1898 if ((l_ptr->owner->block_setup) && (msg_typ != RESET_MSG))
1899 return;
1900
1901 /* Create protocol message with "out-of-sequence" sequence number */
1902 msg_set_type(msg, msg_typ);
1903 msg_set_net_plane(msg, l_ptr->b_ptr->net_plane);
1904 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
1905 msg_set_last_bcast(msg, tipc_bclink_get_last_sent());
1906
1907 if (msg_typ == STATE_MSG) {
1908 u32 next_sent = mod(l_ptr->next_out_no);
1909
1910 if (!tipc_link_is_up(l_ptr))
1911 return;
1912 if (l_ptr->next_out)
1913 next_sent = buf_seqno(l_ptr->next_out);
1914 msg_set_next_sent(msg, next_sent);
1915 if (l_ptr->oldest_deferred_in) {
1916 u32 rec = buf_seqno(l_ptr->oldest_deferred_in);
1917 gap = mod(rec - mod(l_ptr->next_in_no));
1918 }
1919 msg_set_seq_gap(msg, gap);
1920 if (gap)
1921 l_ptr->stats.sent_nacks++;
1922 msg_set_link_tolerance(msg, tolerance);
1923 msg_set_linkprio(msg, priority);
1924 msg_set_max_pkt(msg, ack_mtu);
1925 msg_set_ack(msg, mod(l_ptr->next_in_no - 1));
1926 msg_set_probe(msg, probe_msg != 0);
1927 if (probe_msg) {
1928 u32 mtu = l_ptr->max_pkt;
1929
1930 if ((mtu < l_ptr->max_pkt_target) &&
1931 link_working_working(l_ptr) &&
1932 l_ptr->fsm_msg_cnt) {
1933 msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1934 if (l_ptr->max_pkt_probes == 10) {
1935 l_ptr->max_pkt_target = (msg_size - 4);
1936 l_ptr->max_pkt_probes = 0;
1937 msg_size = (mtu + (l_ptr->max_pkt_target - mtu)/2 + 2) & ~3;
1938 }
1939 l_ptr->max_pkt_probes++;
1940 }
1941
1942 l_ptr->stats.sent_probes++;
1943 }
1944 l_ptr->stats.sent_states++;
1945 } else { /* RESET_MSG or ACTIVATE_MSG */
1946 msg_set_ack(msg, mod(l_ptr->reset_checkpoint - 1));
1947 msg_set_seq_gap(msg, 0);
1948 msg_set_next_sent(msg, 1);
1949 msg_set_probe(msg, 0);
1950 msg_set_link_tolerance(msg, l_ptr->tolerance);
1951 msg_set_linkprio(msg, l_ptr->priority);
1952 msg_set_max_pkt(msg, l_ptr->max_pkt_target);
1953 }
1954
1955 r_flag = (l_ptr->owner->working_links > tipc_link_is_up(l_ptr));
1956 msg_set_redundant_link(msg, r_flag);
1957 msg_set_linkprio(msg, l_ptr->priority);
1958 msg_set_size(msg, msg_size);
1959
1960 msg_set_seqno(msg, mod(l_ptr->next_out_no + (0xffff/2)));
1961
1962 buf = tipc_buf_acquire(msg_size);
1963 if (!buf)
1964 return;
1965
1966 skb_copy_to_linear_data(buf, msg, sizeof(l_ptr->proto_msg));
1967
1968 /* Defer message if bearer is already congested */
1969 if (tipc_bearer_congested(l_ptr->b_ptr, l_ptr)) {
1970 l_ptr->proto_msg_queue = buf;
1971 return;
1972 }
1973
1974 /* Defer message if attempting to send results in bearer congestion */
1975 if (!tipc_bearer_send(l_ptr->b_ptr, buf, &l_ptr->media_addr)) {
1976 tipc_bearer_schedule(l_ptr->b_ptr, l_ptr);
1977 l_ptr->proto_msg_queue = buf;
1978 l_ptr->stats.bearer_congs++;
1979 return;
1980 }
1981
1982 /* Discard message if it was sent successfully */
1983 l_ptr->unacked_window = 0;
1984 kfree_skb(buf);
1985}
1986
1987/*
1988 * Receive protocol message :
1989 * Note that network plane id propagates through the network, and may
1990 * change at any time. The node with lowest address rules
1991 */
1992static void link_recv_proto_msg(struct tipc_link *l_ptr, struct sk_buff *buf)
1993{
1994 u32 rec_gap = 0;
1995 u32 max_pkt_info;
1996 u32 max_pkt_ack;
1997 u32 msg_tol;
1998 struct tipc_msg *msg = buf_msg(buf);
1999
2000 if (link_blocked(l_ptr))
2001 goto exit;
2002
2003 /* record unnumbered packet arrival (force mismatch on next timeout) */
2004 l_ptr->checkpoint--;
2005
2006 if (l_ptr->b_ptr->net_plane != msg_net_plane(msg))
2007 if (tipc_own_addr > msg_prevnode(msg))
2008 l_ptr->b_ptr->net_plane = msg_net_plane(msg);
2009
2010 l_ptr->owner->permit_changeover = msg_redundant_link(msg);
2011
2012 switch (msg_type(msg)) {
2013
2014 case RESET_MSG:
2015 if (!link_working_unknown(l_ptr) &&
2016 (l_ptr->peer_session != INVALID_SESSION)) {
2017 if (less_eq(msg_session(msg), l_ptr->peer_session))
2018 break; /* duplicate or old reset: ignore */
2019 }
2020
2021 if (!msg_redundant_link(msg) && (link_working_working(l_ptr) ||
2022 link_working_unknown(l_ptr))) {
2023 /*
2024 * peer has lost contact -- don't allow peer's links
2025 * to reactivate before we recognize loss & clean up
2026 */
2027 l_ptr->owner->block_setup = WAIT_NODE_DOWN;
2028 }
2029
2030 link_state_event(l_ptr, RESET_MSG);
2031
2032 /* fall thru' */
2033 case ACTIVATE_MSG:
2034 /* Update link settings according other endpoint's values */
2035 strcpy((strrchr(l_ptr->name, ':') + 1), (char *)msg_data(msg));
2036
2037 msg_tol = msg_link_tolerance(msg);
2038 if (msg_tol > l_ptr->tolerance)
2039 link_set_supervision_props(l_ptr, msg_tol);
2040
2041 if (msg_linkprio(msg) > l_ptr->priority)
2042 l_ptr->priority = msg_linkprio(msg);
2043
2044 max_pkt_info = msg_max_pkt(msg);
2045 if (max_pkt_info) {
2046 if (max_pkt_info < l_ptr->max_pkt_target)
2047 l_ptr->max_pkt_target = max_pkt_info;
2048 if (l_ptr->max_pkt > l_ptr->max_pkt_target)
2049 l_ptr->max_pkt = l_ptr->max_pkt_target;
2050 } else {
2051 l_ptr->max_pkt = l_ptr->max_pkt_target;
2052 }
2053 l_ptr->owner->bclink.supportable = (max_pkt_info != 0);
2054
2055 /* Synchronize broadcast link info, if not done previously */
2056 if (!tipc_node_is_up(l_ptr->owner)) {
2057 l_ptr->owner->bclink.last_sent =
2058 l_ptr->owner->bclink.last_in =
2059 msg_last_bcast(msg);
2060 l_ptr->owner->bclink.oos_state = 0;
2061 }
2062
2063 l_ptr->peer_session = msg_session(msg);
2064 l_ptr->peer_bearer_id = msg_bearer_id(msg);
2065
2066 if (msg_type(msg) == ACTIVATE_MSG)
2067 link_state_event(l_ptr, ACTIVATE_MSG);
2068 break;
2069 case STATE_MSG:
2070
2071 msg_tol = msg_link_tolerance(msg);
2072 if (msg_tol)
2073 link_set_supervision_props(l_ptr, msg_tol);
2074
2075 if (msg_linkprio(msg) &&
2076 (msg_linkprio(msg) != l_ptr->priority)) {
2077 warn("Resetting link <%s>, priority change %u->%u\n",
2078 l_ptr->name, l_ptr->priority, msg_linkprio(msg));
2079 l_ptr->priority = msg_linkprio(msg);
2080 tipc_link_reset(l_ptr); /* Enforce change to take effect */
2081 break;
2082 }
2083 link_state_event(l_ptr, TRAFFIC_MSG_EVT);
2084 l_ptr->stats.recv_states++;
2085 if (link_reset_unknown(l_ptr))
2086 break;
2087
2088 if (less_eq(mod(l_ptr->next_in_no), msg_next_sent(msg))) {
2089 rec_gap = mod(msg_next_sent(msg) -
2090 mod(l_ptr->next_in_no));
2091 }
2092
2093 max_pkt_ack = msg_max_pkt(msg);
2094 if (max_pkt_ack > l_ptr->max_pkt) {
2095 l_ptr->max_pkt = max_pkt_ack;
2096 l_ptr->max_pkt_probes = 0;
2097 }
2098
2099 max_pkt_ack = 0;
2100 if (msg_probe(msg)) {
2101 l_ptr->stats.recv_probes++;
2102 if (msg_size(msg) > sizeof(l_ptr->proto_msg))
2103 max_pkt_ack = msg_size(msg);
2104 }
2105
2106 /* Protocol message before retransmits, reduce loss risk */
2107 if (l_ptr->owner->bclink.supported)
2108 tipc_bclink_update_link_state(l_ptr->owner,
2109 msg_last_bcast(msg));
2110
2111 if (rec_gap || (msg_probe(msg))) {
2112 tipc_link_send_proto_msg(l_ptr, STATE_MSG,
2113 0, rec_gap, 0, 0, max_pkt_ack);
2114 }
2115 if (msg_seq_gap(msg)) {
2116 l_ptr->stats.recv_nacks++;
2117 tipc_link_retransmit(l_ptr, l_ptr->first_out,
2118 msg_seq_gap(msg));
2119 }
2120 break;
2121 }
2122exit:
2123 kfree_skb(buf);
2124}
2125
2126
2127/*
2128 * tipc_link_tunnel(): Send one message via a link belonging to
2129 * another bearer. Owner node is locked.
2130 */
2131static void tipc_link_tunnel(struct tipc_link *l_ptr,
2132 struct tipc_msg *tunnel_hdr,
2133 struct tipc_msg *msg,
2134 u32 selector)
2135{
2136 struct tipc_link *tunnel;
2137 struct sk_buff *buf;
2138 u32 length = msg_size(msg);
2139
2140 tunnel = l_ptr->owner->active_links[selector & 1];
2141 if (!tipc_link_is_up(tunnel)) {
2142 warn("Link changeover error, "
2143 "tunnel link no longer available\n");
2144 return;
2145 }
2146 msg_set_size(tunnel_hdr, length + INT_H_SIZE);
2147 buf = tipc_buf_acquire(length + INT_H_SIZE);
2148 if (!buf) {
2149 warn("Link changeover error, "
2150 "unable to send tunnel msg\n");
2151 return;
2152 }
2153 skb_copy_to_linear_data(buf, tunnel_hdr, INT_H_SIZE);
2154 skb_copy_to_linear_data_offset(buf, INT_H_SIZE, msg, length);
2155 tipc_link_send_buf(tunnel, buf);
2156}
2157
2158
2159
2160/*
2161 * changeover(): Send whole message queue via the remaining link
2162 * Owner node is locked.
2163 */
2164void tipc_link_changeover(struct tipc_link *l_ptr)
2165{
2166 u32 msgcount = l_ptr->out_queue_size;
2167 struct sk_buff *crs = l_ptr->first_out;
2168 struct tipc_link *tunnel = l_ptr->owner->active_links[0];
2169 struct tipc_msg tunnel_hdr;
2170 int split_bundles;
2171
2172 if (!tunnel)
2173 return;
2174
2175 if (!l_ptr->owner->permit_changeover) {
2176 warn("Link changeover error, "
2177 "peer did not permit changeover\n");
2178 return;
2179 }
2180
2181 tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
2182 ORIGINAL_MSG, INT_H_SIZE, l_ptr->addr);
2183 msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
2184 msg_set_msgcnt(&tunnel_hdr, msgcount);
2185
2186 if (!l_ptr->first_out) {
2187 struct sk_buff *buf;
2188
2189 buf = tipc_buf_acquire(INT_H_SIZE);
2190 if (buf) {
2191 skb_copy_to_linear_data(buf, &tunnel_hdr, INT_H_SIZE);
2192 msg_set_size(&tunnel_hdr, INT_H_SIZE);
2193 tipc_link_send_buf(tunnel, buf);
2194 } else {
2195 warn("Link changeover error, "
2196 "unable to send changeover msg\n");
2197 }
2198 return;
2199 }
2200
2201 split_bundles = (l_ptr->owner->active_links[0] !=
2202 l_ptr->owner->active_links[1]);
2203
2204 while (crs) {
2205 struct tipc_msg *msg = buf_msg(crs);
2206
2207 if ((msg_user(msg) == MSG_BUNDLER) && split_bundles) {
2208 struct tipc_msg *m = msg_get_wrapped(msg);
2209 unchar *pos = (unchar *)m;
2210
2211 msgcount = msg_msgcnt(msg);
2212 while (msgcount--) {
2213 msg_set_seqno(m, msg_seqno(msg));
2214 tipc_link_tunnel(l_ptr, &tunnel_hdr, m,
2215 msg_link_selector(m));
2216 pos += align(msg_size(m));
2217 m = (struct tipc_msg *)pos;
2218 }
2219 } else {
2220 tipc_link_tunnel(l_ptr, &tunnel_hdr, msg,
2221 msg_link_selector(msg));
2222 }
2223 crs = crs->next;
2224 }
2225}
2226
2227void tipc_link_send_duplicate(struct tipc_link *l_ptr, struct tipc_link *tunnel)
2228{
2229 struct sk_buff *iter;
2230 struct tipc_msg tunnel_hdr;
2231
2232 tipc_msg_init(&tunnel_hdr, CHANGEOVER_PROTOCOL,
2233 DUPLICATE_MSG, INT_H_SIZE, l_ptr->addr);
2234 msg_set_msgcnt(&tunnel_hdr, l_ptr->out_queue_size);
2235 msg_set_bearer_id(&tunnel_hdr, l_ptr->peer_bearer_id);
2236 iter = l_ptr->first_out;
2237 while (iter) {
2238 struct sk_buff *outbuf;
2239 struct tipc_msg *msg = buf_msg(iter);
2240 u32 length = msg_size(msg);
2241
2242 if (msg_user(msg) == MSG_BUNDLER)
2243 msg_set_type(msg, CLOSED_MSG);
2244 msg_set_ack(msg, mod(l_ptr->next_in_no - 1)); /* Update */
2245 msg_set_bcast_ack(msg, l_ptr->owner->bclink.last_in);
2246 msg_set_size(&tunnel_hdr, length + INT_H_SIZE);
2247 outbuf = tipc_buf_acquire(length + INT_H_SIZE);
2248 if (outbuf == NULL) {
2249 warn("Link changeover error, "
2250 "unable to send duplicate msg\n");
2251 return;
2252 }
2253 skb_copy_to_linear_data(outbuf, &tunnel_hdr, INT_H_SIZE);
2254 skb_copy_to_linear_data_offset(outbuf, INT_H_SIZE, iter->data,
2255 length);
2256 tipc_link_send_buf(tunnel, outbuf);
2257 if (!tipc_link_is_up(l_ptr))
2258 return;
2259 iter = iter->next;
2260 }
2261}
2262
2263/**
2264 * buf_extract - extracts embedded TIPC message from another message
2265 * @skb: encapsulating message buffer
2266 * @from_pos: offset to extract from
2267 *
2268 * Returns a new message buffer containing an embedded message. The
2269 * encapsulating message itself is left unchanged.
2270 */
2271static struct sk_buff *buf_extract(struct sk_buff *skb, u32 from_pos)
2272{
2273 struct tipc_msg *msg = (struct tipc_msg *)(skb->data + from_pos);
2274 u32 size = msg_size(msg);
2275 struct sk_buff *eb;
2276
2277 eb = tipc_buf_acquire(size);
2278 if (eb)
2279 skb_copy_to_linear_data(eb, msg, size);
2280 return eb;
2281}
2282
2283/*
2284 * link_recv_changeover_msg(): Receive tunneled packet sent
2285 * via other link. Node is locked. Return extracted buffer.
2286 */
2287static int link_recv_changeover_msg(struct tipc_link **l_ptr,
2288 struct sk_buff **buf)
2289{
2290 struct sk_buff *tunnel_buf = *buf;
2291 struct tipc_link *dest_link;
2292 struct tipc_msg *msg;
2293 struct tipc_msg *tunnel_msg = buf_msg(tunnel_buf);
2294 u32 msg_typ = msg_type(tunnel_msg);
2295 u32 msg_count = msg_msgcnt(tunnel_msg);
2296
2297 dest_link = (*l_ptr)->owner->links[msg_bearer_id(tunnel_msg)];
2298 if (!dest_link)
2299 goto exit;
2300 if (dest_link == *l_ptr) {
2301 err("Unexpected changeover message on link <%s>\n",
2302 (*l_ptr)->name);
2303 goto exit;
2304 }
2305 *l_ptr = dest_link;
2306 msg = msg_get_wrapped(tunnel_msg);
2307
2308 if (msg_typ == DUPLICATE_MSG) {
2309 if (less(msg_seqno(msg), mod(dest_link->next_in_no)))
2310 goto exit;
2311 *buf = buf_extract(tunnel_buf, INT_H_SIZE);
2312 if (*buf == NULL) {
2313 warn("Link changeover error, duplicate msg dropped\n");
2314 goto exit;
2315 }
2316 kfree_skb(tunnel_buf);
2317 return 1;
2318 }
2319
2320 /* First original message ?: */
2321 if (tipc_link_is_up(dest_link)) {
2322 info("Resetting link <%s>, changeover initiated by peer\n",
2323 dest_link->name);
2324 tipc_link_reset(dest_link);
2325 dest_link->exp_msg_count = msg_count;
2326 if (!msg_count)
2327 goto exit;
2328 } else if (dest_link->exp_msg_count == START_CHANGEOVER) {
2329 dest_link->exp_msg_count = msg_count;
2330 if (!msg_count)
2331 goto exit;
2332 }
2333
2334 /* Receive original message */
2335 if (dest_link->exp_msg_count == 0) {
2336 warn("Link switchover error, "
2337 "got too many tunnelled messages\n");
2338 goto exit;
2339 }
2340 dest_link->exp_msg_count--;
2341 if (less(msg_seqno(msg), dest_link->reset_checkpoint)) {
2342 goto exit;
2343 } else {
2344 *buf = buf_extract(tunnel_buf, INT_H_SIZE);
2345 if (*buf != NULL) {
2346 kfree_skb(tunnel_buf);
2347 return 1;
2348 } else {
2349 warn("Link changeover error, original msg dropped\n");
2350 }
2351 }
2352exit:
2353 *buf = NULL;
2354 kfree_skb(tunnel_buf);
2355 return 0;
2356}
2357
2358/*
2359 * Bundler functionality:
2360 */
2361void tipc_link_recv_bundle(struct sk_buff *buf)
2362{
2363 u32 msgcount = msg_msgcnt(buf_msg(buf));
2364 u32 pos = INT_H_SIZE;
2365 struct sk_buff *obuf;
2366
2367 while (msgcount--) {
2368 obuf = buf_extract(buf, pos);
2369 if (obuf == NULL) {
2370 warn("Link unable to unbundle message(s)\n");
2371 break;
2372 }
2373 pos += align(msg_size(buf_msg(obuf)));
2374 tipc_net_route_msg(obuf);
2375 }
2376 kfree_skb(buf);
2377}
2378
2379/*
2380 * Fragmentation/defragmentation:
2381 */
2382
2383/*
2384 * link_send_long_buf: Entry for buffers needing fragmentation.
2385 * The buffer is complete, inclusive total message length.
2386 * Returns user data length.
2387 */
2388static int link_send_long_buf(struct tipc_link *l_ptr, struct sk_buff *buf)
2389{
2390 struct sk_buff *buf_chain = NULL;
2391 struct sk_buff *buf_chain_tail = (struct sk_buff *)&buf_chain;
2392 struct tipc_msg *inmsg = buf_msg(buf);
2393 struct tipc_msg fragm_hdr;
2394 u32 insize = msg_size(inmsg);
2395 u32 dsz = msg_data_sz(inmsg);
2396 unchar *crs = buf->data;
2397 u32 rest = insize;
2398 u32 pack_sz = l_ptr->max_pkt;
2399 u32 fragm_sz = pack_sz - INT_H_SIZE;
2400 u32 fragm_no = 0;
2401 u32 destaddr;
2402
2403 if (msg_short(inmsg))
2404 destaddr = l_ptr->addr;
2405 else
2406 destaddr = msg_destnode(inmsg);
2407
2408 /* Prepare reusable fragment header: */
2409 tipc_msg_init(&fragm_hdr, MSG_FRAGMENTER, FIRST_FRAGMENT,
2410 INT_H_SIZE, destaddr);
2411
2412 /* Chop up message: */
2413 while (rest > 0) {
2414 struct sk_buff *fragm;
2415
2416 if (rest <= fragm_sz) {
2417 fragm_sz = rest;
2418 msg_set_type(&fragm_hdr, LAST_FRAGMENT);
2419 }
2420 fragm = tipc_buf_acquire(fragm_sz + INT_H_SIZE);
2421 if (fragm == NULL) {
2422 kfree_skb(buf);
2423 while (buf_chain) {
2424 buf = buf_chain;
2425 buf_chain = buf_chain->next;
2426 kfree_skb(buf);
2427 }
2428 return -ENOMEM;
2429 }
2430 msg_set_size(&fragm_hdr, fragm_sz + INT_H_SIZE);
2431 fragm_no++;
2432 msg_set_fragm_no(&fragm_hdr, fragm_no);
2433 skb_copy_to_linear_data(fragm, &fragm_hdr, INT_H_SIZE);
2434 skb_copy_to_linear_data_offset(fragm, INT_H_SIZE, crs,
2435 fragm_sz);
2436 buf_chain_tail->next = fragm;
2437 buf_chain_tail = fragm;
2438
2439 rest -= fragm_sz;
2440 crs += fragm_sz;
2441 msg_set_type(&fragm_hdr, FRAGMENT);
2442 }
2443 kfree_skb(buf);
2444
2445 /* Append chain of fragments to send queue & send them */
2446 l_ptr->long_msg_seq_no++;
2447 link_add_chain_to_outqueue(l_ptr, buf_chain, l_ptr->long_msg_seq_no);
2448 l_ptr->stats.sent_fragments += fragm_no;
2449 l_ptr->stats.sent_fragmented++;
2450 tipc_link_push_queue(l_ptr);
2451
2452 return dsz;
2453}
2454
2455/*
2456 * A pending message being re-assembled must store certain values
2457 * to handle subsequent fragments correctly. The following functions
2458 * help storing these values in unused, available fields in the
2459 * pending message. This makes dynamic memory allocation unnecessary.
2460 */
2461static void set_long_msg_seqno(struct sk_buff *buf, u32 seqno)
2462{
2463 msg_set_seqno(buf_msg(buf), seqno);
2464}
2465
2466static u32 get_fragm_size(struct sk_buff *buf)
2467{
2468 return msg_ack(buf_msg(buf));
2469}
2470
2471static void set_fragm_size(struct sk_buff *buf, u32 sz)
2472{
2473 msg_set_ack(buf_msg(buf), sz);
2474}
2475
2476static u32 get_expected_frags(struct sk_buff *buf)
2477{
2478 return msg_bcast_ack(buf_msg(buf));
2479}
2480
2481static void set_expected_frags(struct sk_buff *buf, u32 exp)
2482{
2483 msg_set_bcast_ack(buf_msg(buf), exp);
2484}
2485
2486static u32 get_timer_cnt(struct sk_buff *buf)
2487{
2488 return msg_reroute_cnt(buf_msg(buf));
2489}
2490
2491static void incr_timer_cnt(struct sk_buff *buf)
2492{
2493 msg_incr_reroute_cnt(buf_msg(buf));
2494}
2495
2496/*
2497 * tipc_link_recv_fragment(): Called with node lock on. Returns
2498 * the reassembled buffer if message is complete.
2499 */
2500int tipc_link_recv_fragment(struct sk_buff **pending, struct sk_buff **fb,
2501 struct tipc_msg **m)
2502{
2503 struct sk_buff *prev = NULL;
2504 struct sk_buff *fbuf = *fb;
2505 struct tipc_msg *fragm = buf_msg(fbuf);
2506 struct sk_buff *pbuf = *pending;
2507 u32 long_msg_seq_no = msg_long_msgno(fragm);
2508
2509 *fb = NULL;
2510
2511 /* Is there an incomplete message waiting for this fragment? */
2512 while (pbuf && ((buf_seqno(pbuf) != long_msg_seq_no) ||
2513 (msg_orignode(fragm) != msg_orignode(buf_msg(pbuf))))) {
2514 prev = pbuf;
2515 pbuf = pbuf->next;
2516 }
2517
2518 if (!pbuf && (msg_type(fragm) == FIRST_FRAGMENT)) {
2519 struct tipc_msg *imsg = (struct tipc_msg *)msg_data(fragm);
2520 u32 msg_sz = msg_size(imsg);
2521 u32 fragm_sz = msg_data_sz(fragm);
2522 u32 exp_fragm_cnt = msg_sz/fragm_sz + !!(msg_sz % fragm_sz);
2523 u32 max = TIPC_MAX_USER_MSG_SIZE + NAMED_H_SIZE;
2524 if (msg_type(imsg) == TIPC_MCAST_MSG)
2525 max = TIPC_MAX_USER_MSG_SIZE + MCAST_H_SIZE;
2526 if (msg_size(imsg) > max) {
2527 kfree_skb(fbuf);
2528 return 0;
2529 }
2530 pbuf = tipc_buf_acquire(msg_size(imsg));
2531 if (pbuf != NULL) {
2532 pbuf->next = *pending;
2533 *pending = pbuf;
2534 skb_copy_to_linear_data(pbuf, imsg,
2535 msg_data_sz(fragm));
2536 /* Prepare buffer for subsequent fragments. */
2537 set_long_msg_seqno(pbuf, long_msg_seq_no);
2538 set_fragm_size(pbuf, fragm_sz);
2539 set_expected_frags(pbuf, exp_fragm_cnt - 1);
2540 } else {
2541 dbg("Link unable to reassemble fragmented message\n");
2542 kfree_skb(fbuf);
2543 return -1;
2544 }
2545 kfree_skb(fbuf);
2546 return 0;
2547 } else if (pbuf && (msg_type(fragm) != FIRST_FRAGMENT)) {
2548 u32 dsz = msg_data_sz(fragm);
2549 u32 fsz = get_fragm_size(pbuf);
2550 u32 crs = ((msg_fragm_no(fragm) - 1) * fsz);
2551 u32 exp_frags = get_expected_frags(pbuf) - 1;
2552 skb_copy_to_linear_data_offset(pbuf, crs,
2553 msg_data(fragm), dsz);
2554 kfree_skb(fbuf);
2555
2556 /* Is message complete? */
2557 if (exp_frags == 0) {
2558 if (prev)
2559 prev->next = pbuf->next;
2560 else
2561 *pending = pbuf->next;
2562 msg_reset_reroute_cnt(buf_msg(pbuf));
2563 *fb = pbuf;
2564 *m = buf_msg(pbuf);
2565 return 1;
2566 }
2567 set_expected_frags(pbuf, exp_frags);
2568 return 0;
2569 }
2570 kfree_skb(fbuf);
2571 return 0;
2572}
2573
2574/**
2575 * link_check_defragm_bufs - flush stale incoming message fragments
2576 * @l_ptr: pointer to link
2577 */
2578static void link_check_defragm_bufs(struct tipc_link *l_ptr)
2579{
2580 struct sk_buff *prev = NULL;
2581 struct sk_buff *next = NULL;
2582 struct sk_buff *buf = l_ptr->defragm_buf;
2583
2584 if (!buf)
2585 return;
2586 if (!link_working_working(l_ptr))
2587 return;
2588 while (buf) {
2589 u32 cnt = get_timer_cnt(buf);
2590
2591 next = buf->next;
2592 if (cnt < 4) {
2593 incr_timer_cnt(buf);
2594 prev = buf;
2595 } else {
2596 if (prev)
2597 prev->next = buf->next;
2598 else
2599 l_ptr->defragm_buf = buf->next;
2600 kfree_skb(buf);
2601 }
2602 buf = next;
2603 }
2604}
2605
2606static void link_set_supervision_props(struct tipc_link *l_ptr, u32 tolerance)
2607{
2608 if ((tolerance < TIPC_MIN_LINK_TOL) || (tolerance > TIPC_MAX_LINK_TOL))
2609 return;
2610
2611 l_ptr->tolerance = tolerance;
2612 l_ptr->continuity_interval =
2613 ((tolerance / 4) > 500) ? 500 : tolerance / 4;
2614 l_ptr->abort_limit = tolerance / (l_ptr->continuity_interval / 4);
2615}
2616
2617void tipc_link_set_queue_limits(struct tipc_link *l_ptr, u32 window)
2618{
2619 /* Data messages from this node, inclusive FIRST_FRAGM */
2620 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE] = window;
2621 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE] = (window / 3) * 4;
2622 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE] = (window / 3) * 5;
2623 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE] = (window / 3) * 6;
2624 /* Transiting data messages,inclusive FIRST_FRAGM */
2625 l_ptr->queue_limit[TIPC_LOW_IMPORTANCE + 4] = 300;
2626 l_ptr->queue_limit[TIPC_MEDIUM_IMPORTANCE + 4] = 600;
2627 l_ptr->queue_limit[TIPC_HIGH_IMPORTANCE + 4] = 900;
2628 l_ptr->queue_limit[TIPC_CRITICAL_IMPORTANCE + 4] = 1200;
2629 l_ptr->queue_limit[CONN_MANAGER] = 1200;
2630 l_ptr->queue_limit[CHANGEOVER_PROTOCOL] = 2500;
2631 l_ptr->queue_limit[NAME_DISTRIBUTOR] = 3000;
2632 /* FRAGMENT and LAST_FRAGMENT packets */
2633 l_ptr->queue_limit[MSG_FRAGMENTER] = 4000;
2634}
2635
2636/**
2637 * link_find_link - locate link by name
2638 * @name - ptr to link name string
2639 * @node - ptr to area to be filled with ptr to associated node
2640 *
2641 * Caller must hold 'tipc_net_lock' to ensure node and bearer are not deleted;
2642 * this also prevents link deletion.
2643 *
2644 * Returns pointer to link (or 0 if invalid link name).
2645 */
2646static struct tipc_link *link_find_link(const char *name,
2647 struct tipc_node **node)
2648{
2649 struct tipc_link_name link_name_parts;
2650 struct tipc_bearer *b_ptr;
2651 struct tipc_link *l_ptr;
2652
2653 if (!link_name_validate(name, &link_name_parts))
2654 return NULL;
2655
2656 b_ptr = tipc_bearer_find_interface(link_name_parts.if_local);
2657 if (!b_ptr)
2658 return NULL;
2659
2660 *node = tipc_node_find(link_name_parts.addr_peer);
2661 if (!*node)
2662 return NULL;
2663
2664 l_ptr = (*node)->links[b_ptr->identity];
2665 if (!l_ptr || strcmp(l_ptr->name, name))
2666 return NULL;
2667
2668 return l_ptr;
2669}
2670
2671/**
2672 * link_value_is_valid -- validate proposed link tolerance/priority/window
2673 *
2674 * @cmd - value type (TIPC_CMD_SET_LINK_*)
2675 * @new_value - the new value
2676 *
2677 * Returns 1 if value is within range, 0 if not.
2678 */
2679static int link_value_is_valid(u16 cmd, u32 new_value)
2680{
2681 switch (cmd) {
2682 case TIPC_CMD_SET_LINK_TOL:
2683 return (new_value >= TIPC_MIN_LINK_TOL) &&
2684 (new_value <= TIPC_MAX_LINK_TOL);
2685 case TIPC_CMD_SET_LINK_PRI:
2686 return (new_value <= TIPC_MAX_LINK_PRI);
2687 case TIPC_CMD_SET_LINK_WINDOW:
2688 return (new_value >= TIPC_MIN_LINK_WIN) &&
2689 (new_value <= TIPC_MAX_LINK_WIN);
2690 }
2691 return 0;
2692}
2693
2694/**
2695 * link_cmd_set_value - change priority/tolerance/window for link/bearer/media
2696 * @name - ptr to link, bearer, or media name
2697 * @new_value - new value of link, bearer, or media setting
2698 * @cmd - which link, bearer, or media attribute to set (TIPC_CMD_SET_LINK_*)
2699 *
2700 * Caller must hold 'tipc_net_lock' to ensure link/bearer/media is not deleted.
2701 *
2702 * Returns 0 if value updated and negative value on error.
2703 */
2704static int link_cmd_set_value(const char *name, u32 new_value, u16 cmd)
2705{
2706 struct tipc_node *node;
2707 struct tipc_link *l_ptr;
2708 struct tipc_bearer *b_ptr;
2709 struct tipc_media *m_ptr;
2710
2711 l_ptr = link_find_link(name, &node);
2712 if (l_ptr) {
2713 /*
2714 * acquire node lock for tipc_link_send_proto_msg().
2715 * see "TIPC locking policy" in net.c.
2716 */
2717 tipc_node_lock(node);
2718 switch (cmd) {
2719 case TIPC_CMD_SET_LINK_TOL:
2720 link_set_supervision_props(l_ptr, new_value);
2721 tipc_link_send_proto_msg(l_ptr,
2722 STATE_MSG, 0, 0, new_value, 0, 0);
2723 break;
2724 case TIPC_CMD_SET_LINK_PRI:
2725 l_ptr->priority = new_value;
2726 tipc_link_send_proto_msg(l_ptr,
2727 STATE_MSG, 0, 0, 0, new_value, 0);
2728 break;
2729 case TIPC_CMD_SET_LINK_WINDOW:
2730 tipc_link_set_queue_limits(l_ptr, new_value);
2731 break;
2732 }
2733 tipc_node_unlock(node);
2734 return 0;
2735 }
2736
2737 b_ptr = tipc_bearer_find(name);
2738 if (b_ptr) {
2739 switch (cmd) {
2740 case TIPC_CMD_SET_LINK_TOL:
2741 b_ptr->tolerance = new_value;
2742 return 0;
2743 case TIPC_CMD_SET_LINK_PRI:
2744 b_ptr->priority = new_value;
2745 return 0;
2746 case TIPC_CMD_SET_LINK_WINDOW:
2747 b_ptr->window = new_value;
2748 return 0;
2749 }
2750 return -EINVAL;
2751 }
2752
2753 m_ptr = tipc_media_find(name);
2754 if (!m_ptr)
2755 return -ENODEV;
2756 switch (cmd) {
2757 case TIPC_CMD_SET_LINK_TOL:
2758 m_ptr->tolerance = new_value;
2759 return 0;
2760 case TIPC_CMD_SET_LINK_PRI:
2761 m_ptr->priority = new_value;
2762 return 0;
2763 case TIPC_CMD_SET_LINK_WINDOW:
2764 m_ptr->window = new_value;
2765 return 0;
2766 }
2767 return -EINVAL;
2768}
2769
2770struct sk_buff *tipc_link_cmd_config(const void *req_tlv_area, int req_tlv_space,
2771 u16 cmd)
2772{
2773 struct tipc_link_config *args;
2774 u32 new_value;
2775 int res;
2776
2777 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_CONFIG))
2778 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2779
2780 args = (struct tipc_link_config *)TLV_DATA(req_tlv_area);
2781 new_value = ntohl(args->value);
2782
2783 if (!link_value_is_valid(cmd, new_value))
2784 return tipc_cfg_reply_error_string(
2785 "cannot change, value invalid");
2786
2787 if (!strcmp(args->name, tipc_bclink_name)) {
2788 if ((cmd == TIPC_CMD_SET_LINK_WINDOW) &&
2789 (tipc_bclink_set_queue_limits(new_value) == 0))
2790 return tipc_cfg_reply_none();
2791 return tipc_cfg_reply_error_string(TIPC_CFG_NOT_SUPPORTED
2792 " (cannot change setting on broadcast link)");
2793 }
2794
2795 read_lock_bh(&tipc_net_lock);
2796 res = link_cmd_set_value(args->name, new_value, cmd);
2797 read_unlock_bh(&tipc_net_lock);
2798 if (res)
2799 return tipc_cfg_reply_error_string("cannot change link setting");
2800
2801 return tipc_cfg_reply_none();
2802}
2803
2804/**
2805 * link_reset_statistics - reset link statistics
2806 * @l_ptr: pointer to link
2807 */
2808static void link_reset_statistics(struct tipc_link *l_ptr)
2809{
2810 memset(&l_ptr->stats, 0, sizeof(l_ptr->stats));
2811 l_ptr->stats.sent_info = l_ptr->next_out_no;
2812 l_ptr->stats.recv_info = l_ptr->next_in_no;
2813}
2814
2815struct sk_buff *tipc_link_cmd_reset_stats(const void *req_tlv_area, int req_tlv_space)
2816{
2817 char *link_name;
2818 struct tipc_link *l_ptr;
2819 struct tipc_node *node;
2820
2821 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2822 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2823
2824 link_name = (char *)TLV_DATA(req_tlv_area);
2825 if (!strcmp(link_name, tipc_bclink_name)) {
2826 if (tipc_bclink_reset_stats())
2827 return tipc_cfg_reply_error_string("link not found");
2828 return tipc_cfg_reply_none();
2829 }
2830
2831 read_lock_bh(&tipc_net_lock);
2832 l_ptr = link_find_link(link_name, &node);
2833 if (!l_ptr) {
2834 read_unlock_bh(&tipc_net_lock);
2835 return tipc_cfg_reply_error_string("link not found");
2836 }
2837
2838 tipc_node_lock(node);
2839 link_reset_statistics(l_ptr);
2840 tipc_node_unlock(node);
2841 read_unlock_bh(&tipc_net_lock);
2842 return tipc_cfg_reply_none();
2843}
2844
2845/**
2846 * percent - convert count to a percentage of total (rounding up or down)
2847 */
2848static u32 percent(u32 count, u32 total)
2849{
2850 return (count * 100 + (total / 2)) / total;
2851}
2852
2853/**
2854 * tipc_link_stats - print link statistics
2855 * @name: link name
2856 * @buf: print buffer area
2857 * @buf_size: size of print buffer area
2858 *
2859 * Returns length of print buffer data string (or 0 if error)
2860 */
2861static int tipc_link_stats(const char *name, char *buf, const u32 buf_size)
2862{
2863 struct print_buf pb;
2864 struct tipc_link *l_ptr;
2865 struct tipc_node *node;
2866 char *status;
2867 u32 profile_total = 0;
2868
2869 if (!strcmp(name, tipc_bclink_name))
2870 return tipc_bclink_stats(buf, buf_size);
2871
2872 tipc_printbuf_init(&pb, buf, buf_size);
2873
2874 read_lock_bh(&tipc_net_lock);
2875 l_ptr = link_find_link(name, &node);
2876 if (!l_ptr) {
2877 read_unlock_bh(&tipc_net_lock);
2878 return 0;
2879 }
2880 tipc_node_lock(node);
2881
2882 if (tipc_link_is_active(l_ptr))
2883 status = "ACTIVE";
2884 else if (tipc_link_is_up(l_ptr))
2885 status = "STANDBY";
2886 else
2887 status = "DEFUNCT";
2888 tipc_printf(&pb, "Link <%s>\n"
2889 " %s MTU:%u Priority:%u Tolerance:%u ms"
2890 " Window:%u packets\n",
2891 l_ptr->name, status, l_ptr->max_pkt,
2892 l_ptr->priority, l_ptr->tolerance, l_ptr->queue_limit[0]);
2893 tipc_printf(&pb, " RX packets:%u fragments:%u/%u bundles:%u/%u\n",
2894 l_ptr->next_in_no - l_ptr->stats.recv_info,
2895 l_ptr->stats.recv_fragments,
2896 l_ptr->stats.recv_fragmented,
2897 l_ptr->stats.recv_bundles,
2898 l_ptr->stats.recv_bundled);
2899 tipc_printf(&pb, " TX packets:%u fragments:%u/%u bundles:%u/%u\n",
2900 l_ptr->next_out_no - l_ptr->stats.sent_info,
2901 l_ptr->stats.sent_fragments,
2902 l_ptr->stats.sent_fragmented,
2903 l_ptr->stats.sent_bundles,
2904 l_ptr->stats.sent_bundled);
2905 profile_total = l_ptr->stats.msg_length_counts;
2906 if (!profile_total)
2907 profile_total = 1;
2908 tipc_printf(&pb, " TX profile sample:%u packets average:%u octets\n"
2909 " 0-64:%u%% -256:%u%% -1024:%u%% -4096:%u%% "
2910 "-16384:%u%% -32768:%u%% -66000:%u%%\n",
2911 l_ptr->stats.msg_length_counts,
2912 l_ptr->stats.msg_lengths_total / profile_total,
2913 percent(l_ptr->stats.msg_length_profile[0], profile_total),
2914 percent(l_ptr->stats.msg_length_profile[1], profile_total),
2915 percent(l_ptr->stats.msg_length_profile[2], profile_total),
2916 percent(l_ptr->stats.msg_length_profile[3], profile_total),
2917 percent(l_ptr->stats.msg_length_profile[4], profile_total),
2918 percent(l_ptr->stats.msg_length_profile[5], profile_total),
2919 percent(l_ptr->stats.msg_length_profile[6], profile_total));
2920 tipc_printf(&pb, " RX states:%u probes:%u naks:%u defs:%u dups:%u\n",
2921 l_ptr->stats.recv_states,
2922 l_ptr->stats.recv_probes,
2923 l_ptr->stats.recv_nacks,
2924 l_ptr->stats.deferred_recv,
2925 l_ptr->stats.duplicates);
2926 tipc_printf(&pb, " TX states:%u probes:%u naks:%u acks:%u dups:%u\n",
2927 l_ptr->stats.sent_states,
2928 l_ptr->stats.sent_probes,
2929 l_ptr->stats.sent_nacks,
2930 l_ptr->stats.sent_acks,
2931 l_ptr->stats.retransmitted);
2932 tipc_printf(&pb, " Congestion bearer:%u link:%u Send queue max:%u avg:%u\n",
2933 l_ptr->stats.bearer_congs,
2934 l_ptr->stats.link_congs,
2935 l_ptr->stats.max_queue_sz,
2936 l_ptr->stats.queue_sz_counts
2937 ? (l_ptr->stats.accu_queue_sz / l_ptr->stats.queue_sz_counts)
2938 : 0);
2939
2940 tipc_node_unlock(node);
2941 read_unlock_bh(&tipc_net_lock);
2942 return tipc_printbuf_validate(&pb);
2943}
2944
2945#define MAX_LINK_STATS_INFO 2000
2946
2947struct sk_buff *tipc_link_cmd_show_stats(const void *req_tlv_area, int req_tlv_space)
2948{
2949 struct sk_buff *buf;
2950 struct tlv_desc *rep_tlv;
2951 int str_len;
2952
2953 if (!TLV_CHECK(req_tlv_area, req_tlv_space, TIPC_TLV_LINK_NAME))
2954 return tipc_cfg_reply_error_string(TIPC_CFG_TLV_ERROR);
2955
2956 buf = tipc_cfg_reply_alloc(TLV_SPACE(MAX_LINK_STATS_INFO));
2957 if (!buf)
2958 return NULL;
2959
2960 rep_tlv = (struct tlv_desc *)buf->data;
2961
2962 str_len = tipc_link_stats((char *)TLV_DATA(req_tlv_area),
2963 (char *)TLV_DATA(rep_tlv), MAX_LINK_STATS_INFO);
2964 if (!str_len) {
2965 kfree_skb(buf);
2966 return tipc_cfg_reply_error_string("link not found");
2967 }
2968
2969 skb_put(buf, TLV_SPACE(str_len));
2970 TLV_SET(rep_tlv, TIPC_TLV_ULTRA_STRING, NULL, str_len);
2971
2972 return buf;
2973}
2974
2975/**
2976 * tipc_link_get_max_pkt - get maximum packet size to use when sending to destination
2977 * @dest: network address of destination node
2978 * @selector: used to select from set of active links
2979 *
2980 * If no active link can be found, uses default maximum packet size.
2981 */
2982u32 tipc_link_get_max_pkt(u32 dest, u32 selector)
2983{
2984 struct tipc_node *n_ptr;
2985 struct tipc_link *l_ptr;
2986 u32 res = MAX_PKT_DEFAULT;
2987
2988 if (dest == tipc_own_addr)
2989 return MAX_MSG_SIZE;
2990
2991 read_lock_bh(&tipc_net_lock);
2992 n_ptr = tipc_node_find(dest);
2993 if (n_ptr) {
2994 tipc_node_lock(n_ptr);
2995 l_ptr = n_ptr->active_links[selector & 1];
2996 if (l_ptr)
2997 res = l_ptr->max_pkt;
2998 tipc_node_unlock(n_ptr);
2999 }
3000 read_unlock_bh(&tipc_net_lock);
3001 return res;
3002}
3003
3004static void link_print(struct tipc_link *l_ptr, const char *str)
3005{
3006 char print_area[256];
3007 struct print_buf pb;
3008 struct print_buf *buf = &pb;
3009
3010 tipc_printbuf_init(buf, print_area, sizeof(print_area));
3011
3012 tipc_printf(buf, str);
3013 tipc_printf(buf, "Link %x<%s>:",
3014 l_ptr->addr, l_ptr->b_ptr->name);
3015
3016#ifdef CONFIG_TIPC_DEBUG
3017 if (link_reset_reset(l_ptr) || link_reset_unknown(l_ptr))
3018 goto print_state;
3019
3020 tipc_printf(buf, ": NXO(%u):", mod(l_ptr->next_out_no));
3021 tipc_printf(buf, "NXI(%u):", mod(l_ptr->next_in_no));
3022 tipc_printf(buf, "SQUE");
3023 if (l_ptr->first_out) {
3024 tipc_printf(buf, "[%u..", buf_seqno(l_ptr->first_out));
3025 if (l_ptr->next_out)
3026 tipc_printf(buf, "%u..", buf_seqno(l_ptr->next_out));
3027 tipc_printf(buf, "%u]", buf_seqno(l_ptr->last_out));
3028 if ((mod(buf_seqno(l_ptr->last_out) -
3029 buf_seqno(l_ptr->first_out))
3030 != (l_ptr->out_queue_size - 1)) ||
3031 (l_ptr->last_out->next != NULL)) {
3032 tipc_printf(buf, "\nSend queue inconsistency\n");
3033 tipc_printf(buf, "first_out= %p ", l_ptr->first_out);
3034 tipc_printf(buf, "next_out= %p ", l_ptr->next_out);
3035 tipc_printf(buf, "last_out= %p ", l_ptr->last_out);
3036 }
3037 } else
3038 tipc_printf(buf, "[]");
3039 tipc_printf(buf, "SQSIZ(%u)", l_ptr->out_queue_size);
3040 if (l_ptr->oldest_deferred_in) {
3041 u32 o = buf_seqno(l_ptr->oldest_deferred_in);
3042 u32 n = buf_seqno(l_ptr->newest_deferred_in);
3043 tipc_printf(buf, ":RQUE[%u..%u]", o, n);
3044 if (l_ptr->deferred_inqueue_sz != mod((n + 1) - o)) {
3045 tipc_printf(buf, ":RQSIZ(%u)",
3046 l_ptr->deferred_inqueue_sz);
3047 }
3048 }
3049print_state:
3050#endif
3051
3052 if (link_working_unknown(l_ptr))
3053 tipc_printf(buf, ":WU");
3054 else if (link_reset_reset(l_ptr))
3055 tipc_printf(buf, ":RR");
3056 else if (link_reset_unknown(l_ptr))
3057 tipc_printf(buf, ":RU");
3058 else if (link_working_working(l_ptr))
3059 tipc_printf(buf, ":WW");
3060 tipc_printf(buf, "\n");
3061
3062 tipc_printbuf_validate(buf);
3063 info("%s", print_area);
3064}