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