Loading...
1/*
2 * net/tipc/bcast.c: TIPC broadcast code
3 *
4 * Copyright (c) 2004-2006, 2014-2017, Ericsson AB
5 * Copyright (c) 2004, Intel Corporation.
6 * Copyright (c) 2005, 2010-2011, Wind River Systems
7 * All rights reserved.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions are met:
11 *
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the names of the copyright holders nor the names of its
18 * contributors may be used to endorse or promote products derived from
19 * this software without specific prior written permission.
20 *
21 * Alternatively, this software may be distributed under the terms of the
22 * GNU General Public License ("GPL") version 2 as published by the Free
23 * Software Foundation.
24 *
25 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
26 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
29 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35 * POSSIBILITY OF SUCH DAMAGE.
36 */
37
38#include <linux/tipc_config.h>
39#include "socket.h"
40#include "msg.h"
41#include "bcast.h"
42#include "link.h"
43#include "name_table.h"
44
45#define BCLINK_WIN_DEFAULT 50 /* bcast link window size (default) */
46#define BCLINK_WIN_MIN 32 /* bcast minimum link window size */
47
48const char tipc_bclink_name[] = "broadcast-link";
49unsigned long sysctl_tipc_bc_retruni __read_mostly;
50
51/**
52 * struct tipc_bc_base - base structure for keeping broadcast send state
53 * @link: broadcast send link structure
54 * @inputq: data input queue; will only carry SOCK_WAKEUP messages
55 * @dests: array keeping number of reachable destinations per bearer
56 * @primary_bearer: a bearer having links to all broadcast destinations, if any
57 * @bcast_support: indicates if primary bearer, if any, supports broadcast
58 * @force_bcast: forces broadcast for multicast traffic
59 * @rcast_support: indicates if all peer nodes support replicast
60 * @force_rcast: forces replicast for multicast traffic
61 * @rc_ratio: dest count as percentage of cluster size where send method changes
62 * @bc_threshold: calculated from rc_ratio; if dests > threshold use broadcast
63 */
64struct tipc_bc_base {
65 struct tipc_link *link;
66 struct sk_buff_head inputq;
67 int dests[MAX_BEARERS];
68 int primary_bearer;
69 bool bcast_support;
70 bool force_bcast;
71 bool rcast_support;
72 bool force_rcast;
73 int rc_ratio;
74 int bc_threshold;
75};
76
77static struct tipc_bc_base *tipc_bc_base(struct net *net)
78{
79 return tipc_net(net)->bcbase;
80}
81
82/* tipc_bcast_get_mtu(): -get the MTU currently used by broadcast link
83 * Note: the MTU is decremented to give room for a tunnel header, in
84 * case the message needs to be sent as replicast
85 */
86int tipc_bcast_get_mtu(struct net *net)
87{
88 return tipc_link_mss(tipc_bc_sndlink(net));
89}
90
91void tipc_bcast_toggle_rcast(struct net *net, bool supp)
92{
93 tipc_bc_base(net)->rcast_support = supp;
94}
95
96static void tipc_bcbase_calc_bc_threshold(struct net *net)
97{
98 struct tipc_bc_base *bb = tipc_bc_base(net);
99 int cluster_size = tipc_link_bc_peers(tipc_bc_sndlink(net));
100
101 bb->bc_threshold = 1 + (cluster_size * bb->rc_ratio / 100);
102}
103
104/* tipc_bcbase_select_primary(): find a bearer with links to all destinations,
105 * if any, and make it primary bearer
106 */
107static void tipc_bcbase_select_primary(struct net *net)
108{
109 struct tipc_bc_base *bb = tipc_bc_base(net);
110 int all_dests = tipc_link_bc_peers(bb->link);
111 int i, mtu, prim;
112
113 bb->primary_bearer = INVALID_BEARER_ID;
114 bb->bcast_support = true;
115
116 if (!all_dests)
117 return;
118
119 for (i = 0; i < MAX_BEARERS; i++) {
120 if (!bb->dests[i])
121 continue;
122
123 mtu = tipc_bearer_mtu(net, i);
124 if (mtu < tipc_link_mtu(bb->link))
125 tipc_link_set_mtu(bb->link, mtu);
126 bb->bcast_support &= tipc_bearer_bcast_support(net, i);
127 if (bb->dests[i] < all_dests)
128 continue;
129
130 bb->primary_bearer = i;
131
132 /* Reduce risk that all nodes select same primary */
133 if ((i ^ tipc_own_addr(net)) & 1)
134 break;
135 }
136 prim = bb->primary_bearer;
137 if (prim != INVALID_BEARER_ID)
138 bb->bcast_support = tipc_bearer_bcast_support(net, prim);
139}
140
141void tipc_bcast_inc_bearer_dst_cnt(struct net *net, int bearer_id)
142{
143 struct tipc_bc_base *bb = tipc_bc_base(net);
144
145 tipc_bcast_lock(net);
146 bb->dests[bearer_id]++;
147 tipc_bcbase_select_primary(net);
148 tipc_bcast_unlock(net);
149}
150
151void tipc_bcast_dec_bearer_dst_cnt(struct net *net, int bearer_id)
152{
153 struct tipc_bc_base *bb = tipc_bc_base(net);
154
155 tipc_bcast_lock(net);
156 bb->dests[bearer_id]--;
157 tipc_bcbase_select_primary(net);
158 tipc_bcast_unlock(net);
159}
160
161/* tipc_bcbase_xmit - broadcast a packet queue across one or more bearers
162 *
163 * Note that number of reachable destinations, as indicated in the dests[]
164 * array, may transitionally differ from the number of destinations indicated
165 * in each sent buffer. We can sustain this. Excess destination nodes will
166 * drop and never acknowledge the unexpected packets, and missing destinations
167 * will either require retransmission (if they are just about to be added to
168 * the bearer), or be removed from the buffer's 'ackers' counter (if they
169 * just went down)
170 */
171static void tipc_bcbase_xmit(struct net *net, struct sk_buff_head *xmitq)
172{
173 int bearer_id;
174 struct tipc_bc_base *bb = tipc_bc_base(net);
175 struct sk_buff *skb, *_skb;
176 struct sk_buff_head _xmitq;
177
178 if (skb_queue_empty(xmitq))
179 return;
180
181 /* The typical case: at least one bearer has links to all nodes */
182 bearer_id = bb->primary_bearer;
183 if (bearer_id >= 0) {
184 tipc_bearer_bc_xmit(net, bearer_id, xmitq);
185 return;
186 }
187
188 /* We have to transmit across all bearers */
189 __skb_queue_head_init(&_xmitq);
190 for (bearer_id = 0; bearer_id < MAX_BEARERS; bearer_id++) {
191 if (!bb->dests[bearer_id])
192 continue;
193
194 skb_queue_walk(xmitq, skb) {
195 _skb = pskb_copy_for_clone(skb, GFP_ATOMIC);
196 if (!_skb)
197 break;
198 __skb_queue_tail(&_xmitq, _skb);
199 }
200 tipc_bearer_bc_xmit(net, bearer_id, &_xmitq);
201 }
202 __skb_queue_purge(xmitq);
203 __skb_queue_purge(&_xmitq);
204}
205
206static void tipc_bcast_select_xmit_method(struct net *net, int dests,
207 struct tipc_mc_method *method)
208{
209 struct tipc_bc_base *bb = tipc_bc_base(net);
210 unsigned long exp = method->expires;
211
212 /* Broadcast supported by used bearer/bearers? */
213 if (!bb->bcast_support) {
214 method->rcast = true;
215 return;
216 }
217 /* Any destinations which don't support replicast ? */
218 if (!bb->rcast_support) {
219 method->rcast = false;
220 return;
221 }
222 /* Can current method be changed ? */
223 method->expires = jiffies + TIPC_METHOD_EXPIRE;
224 if (method->mandatory)
225 return;
226
227 if (!(tipc_net(net)->capabilities & TIPC_MCAST_RBCTL) &&
228 time_before(jiffies, exp))
229 return;
230
231 /* Configuration as force 'broadcast' method */
232 if (bb->force_bcast) {
233 method->rcast = false;
234 return;
235 }
236 /* Configuration as force 'replicast' method */
237 if (bb->force_rcast) {
238 method->rcast = true;
239 return;
240 }
241 /* Configuration as 'autoselect' or default method */
242 /* Determine method to use now */
243 method->rcast = dests <= bb->bc_threshold;
244}
245
246/* tipc_bcast_xmit - broadcast the buffer chain to all external nodes
247 * @net: the applicable net namespace
248 * @pkts: chain of buffers containing message
249 * @cong_link_cnt: set to 1 if broadcast link is congested, otherwise 0
250 * Consumes the buffer chain.
251 * Returns 0 if success, otherwise errno: -EHOSTUNREACH,-EMSGSIZE
252 */
253int tipc_bcast_xmit(struct net *net, struct sk_buff_head *pkts,
254 u16 *cong_link_cnt)
255{
256 struct tipc_link *l = tipc_bc_sndlink(net);
257 struct sk_buff_head xmitq;
258 int rc = 0;
259
260 __skb_queue_head_init(&xmitq);
261 tipc_bcast_lock(net);
262 if (tipc_link_bc_peers(l))
263 rc = tipc_link_xmit(l, pkts, &xmitq);
264 tipc_bcast_unlock(net);
265 tipc_bcbase_xmit(net, &xmitq);
266 __skb_queue_purge(pkts);
267 if (rc == -ELINKCONG) {
268 *cong_link_cnt = 1;
269 rc = 0;
270 }
271 return rc;
272}
273
274/* tipc_rcast_xmit - replicate and send a message to given destination nodes
275 * @net: the applicable net namespace
276 * @pkts: chain of buffers containing message
277 * @dests: list of destination nodes
278 * @cong_link_cnt: returns number of congested links
279 * @cong_links: returns identities of congested links
280 * Returns 0 if success, otherwise errno
281 */
282static int tipc_rcast_xmit(struct net *net, struct sk_buff_head *pkts,
283 struct tipc_nlist *dests, u16 *cong_link_cnt)
284{
285 struct tipc_dest *dst, *tmp;
286 struct sk_buff_head _pkts;
287 u32 dnode, selector;
288
289 selector = msg_link_selector(buf_msg(skb_peek(pkts)));
290 __skb_queue_head_init(&_pkts);
291
292 list_for_each_entry_safe(dst, tmp, &dests->list, list) {
293 dnode = dst->node;
294 if (!tipc_msg_pskb_copy(dnode, pkts, &_pkts))
295 return -ENOMEM;
296
297 /* Any other return value than -ELINKCONG is ignored */
298 if (tipc_node_xmit(net, &_pkts, dnode, selector) == -ELINKCONG)
299 (*cong_link_cnt)++;
300 }
301 return 0;
302}
303
304/* tipc_mcast_send_sync - deliver a dummy message with SYN bit
305 * @net: the applicable net namespace
306 * @skb: socket buffer to copy
307 * @method: send method to be used
308 * @dests: destination nodes for message.
309 * Returns 0 if success, otherwise errno
310 */
311static int tipc_mcast_send_sync(struct net *net, struct sk_buff *skb,
312 struct tipc_mc_method *method,
313 struct tipc_nlist *dests)
314{
315 struct tipc_msg *hdr, *_hdr;
316 struct sk_buff_head tmpq;
317 struct sk_buff *_skb;
318 u16 cong_link_cnt;
319 int rc = 0;
320
321 /* Is a cluster supporting with new capabilities ? */
322 if (!(tipc_net(net)->capabilities & TIPC_MCAST_RBCTL))
323 return 0;
324
325 hdr = buf_msg(skb);
326 if (msg_user(hdr) == MSG_FRAGMENTER)
327 hdr = msg_inner_hdr(hdr);
328 if (msg_type(hdr) != TIPC_MCAST_MSG)
329 return 0;
330
331 /* Allocate dummy message */
332 _skb = tipc_buf_acquire(MCAST_H_SIZE, GFP_KERNEL);
333 if (!_skb)
334 return -ENOMEM;
335
336 /* Preparing for 'synching' header */
337 msg_set_syn(hdr, 1);
338
339 /* Copy skb's header into a dummy header */
340 skb_copy_to_linear_data(_skb, hdr, MCAST_H_SIZE);
341 skb_orphan(_skb);
342
343 /* Reverse method for dummy message */
344 _hdr = buf_msg(_skb);
345 msg_set_size(_hdr, MCAST_H_SIZE);
346 msg_set_is_rcast(_hdr, !msg_is_rcast(hdr));
347 msg_set_errcode(_hdr, TIPC_ERR_NO_PORT);
348
349 __skb_queue_head_init(&tmpq);
350 __skb_queue_tail(&tmpq, _skb);
351 if (method->rcast)
352 rc = tipc_bcast_xmit(net, &tmpq, &cong_link_cnt);
353 else
354 rc = tipc_rcast_xmit(net, &tmpq, dests, &cong_link_cnt);
355
356 /* This queue should normally be empty by now */
357 __skb_queue_purge(&tmpq);
358
359 return rc;
360}
361
362/* tipc_mcast_xmit - deliver message to indicated destination nodes
363 * and to identified node local sockets
364 * @net: the applicable net namespace
365 * @pkts: chain of buffers containing message
366 * @method: send method to be used
367 * @dests: destination nodes for message.
368 * @cong_link_cnt: returns number of encountered congested destination links
369 * Consumes buffer chain.
370 * Returns 0 if success, otherwise errno
371 */
372int tipc_mcast_xmit(struct net *net, struct sk_buff_head *pkts,
373 struct tipc_mc_method *method, struct tipc_nlist *dests,
374 u16 *cong_link_cnt)
375{
376 struct sk_buff_head inputq, localq;
377 bool rcast = method->rcast;
378 struct tipc_msg *hdr;
379 struct sk_buff *skb;
380 int rc = 0;
381
382 skb_queue_head_init(&inputq);
383 __skb_queue_head_init(&localq);
384
385 /* Clone packets before they are consumed by next call */
386 if (dests->local && !tipc_msg_reassemble(pkts, &localq)) {
387 rc = -ENOMEM;
388 goto exit;
389 }
390 /* Send according to determined transmit method */
391 if (dests->remote) {
392 tipc_bcast_select_xmit_method(net, dests->remote, method);
393
394 skb = skb_peek(pkts);
395 hdr = buf_msg(skb);
396 if (msg_user(hdr) == MSG_FRAGMENTER)
397 hdr = msg_inner_hdr(hdr);
398 msg_set_is_rcast(hdr, method->rcast);
399
400 /* Switch method ? */
401 if (rcast != method->rcast) {
402 rc = tipc_mcast_send_sync(net, skb, method, dests);
403 if (unlikely(rc)) {
404 pr_err("Unable to send SYN: method %d, rc %d\n",
405 rcast, rc);
406 goto exit;
407 }
408 }
409
410 if (method->rcast)
411 rc = tipc_rcast_xmit(net, pkts, dests, cong_link_cnt);
412 else
413 rc = tipc_bcast_xmit(net, pkts, cong_link_cnt);
414 }
415
416 if (dests->local) {
417 tipc_loopback_trace(net, &localq);
418 tipc_sk_mcast_rcv(net, &localq, &inputq);
419 }
420exit:
421 /* This queue should normally be empty by now */
422 __skb_queue_purge(pkts);
423 return rc;
424}
425
426/* tipc_bcast_rcv - receive a broadcast packet, and deliver to rcv link
427 *
428 * RCU is locked, no other locks set
429 */
430int tipc_bcast_rcv(struct net *net, struct tipc_link *l, struct sk_buff *skb)
431{
432 struct tipc_msg *hdr = buf_msg(skb);
433 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
434 struct sk_buff_head xmitq;
435 int rc;
436
437 __skb_queue_head_init(&xmitq);
438
439 if (msg_mc_netid(hdr) != tipc_netid(net) || !tipc_link_is_up(l)) {
440 kfree_skb(skb);
441 return 0;
442 }
443
444 tipc_bcast_lock(net);
445 if (msg_user(hdr) == BCAST_PROTOCOL)
446 rc = tipc_link_bc_nack_rcv(l, skb, &xmitq);
447 else
448 rc = tipc_link_rcv(l, skb, NULL);
449 tipc_bcast_unlock(net);
450
451 tipc_bcbase_xmit(net, &xmitq);
452
453 /* Any socket wakeup messages ? */
454 if (!skb_queue_empty(inputq))
455 tipc_sk_rcv(net, inputq);
456
457 return rc;
458}
459
460/* tipc_bcast_ack_rcv - receive and handle a broadcast acknowledge
461 *
462 * RCU is locked, no other locks set
463 */
464void tipc_bcast_ack_rcv(struct net *net, struct tipc_link *l,
465 struct tipc_msg *hdr)
466{
467 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
468 u16 acked = msg_bcast_ack(hdr);
469 struct sk_buff_head xmitq;
470
471 /* Ignore bc acks sent by peer before bcast synch point was received */
472 if (msg_bc_ack_invalid(hdr))
473 return;
474
475 __skb_queue_head_init(&xmitq);
476
477 tipc_bcast_lock(net);
478 tipc_link_bc_ack_rcv(l, acked, 0, NULL, &xmitq, NULL);
479 tipc_bcast_unlock(net);
480
481 tipc_bcbase_xmit(net, &xmitq);
482
483 /* Any socket wakeup messages ? */
484 if (!skb_queue_empty(inputq))
485 tipc_sk_rcv(net, inputq);
486}
487
488/* tipc_bcast_synch_rcv - check and update rcv link with peer's send state
489 *
490 * RCU is locked, no other locks set
491 */
492int tipc_bcast_sync_rcv(struct net *net, struct tipc_link *l,
493 struct tipc_msg *hdr,
494 struct sk_buff_head *retrq)
495{
496 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
497 struct tipc_gap_ack_blks *ga;
498 struct sk_buff_head xmitq;
499 int rc = 0;
500
501 __skb_queue_head_init(&xmitq);
502
503 tipc_bcast_lock(net);
504 if (msg_type(hdr) != STATE_MSG) {
505 tipc_link_bc_init_rcv(l, hdr);
506 } else if (!msg_bc_ack_invalid(hdr)) {
507 tipc_get_gap_ack_blks(&ga, l, hdr, false);
508 if (!sysctl_tipc_bc_retruni)
509 retrq = &xmitq;
510 rc = tipc_link_bc_ack_rcv(l, msg_bcast_ack(hdr),
511 msg_bc_gap(hdr), ga, &xmitq,
512 retrq);
513 rc |= tipc_link_bc_sync_rcv(l, hdr, &xmitq);
514 }
515 tipc_bcast_unlock(net);
516
517 tipc_bcbase_xmit(net, &xmitq);
518
519 /* Any socket wakeup messages ? */
520 if (!skb_queue_empty(inputq))
521 tipc_sk_rcv(net, inputq);
522 return rc;
523}
524
525/* tipc_bcast_add_peer - add a peer node to broadcast link and bearer
526 *
527 * RCU is locked, node lock is set
528 */
529void tipc_bcast_add_peer(struct net *net, struct tipc_link *uc_l,
530 struct sk_buff_head *xmitq)
531{
532 struct tipc_link *snd_l = tipc_bc_sndlink(net);
533
534 tipc_bcast_lock(net);
535 tipc_link_add_bc_peer(snd_l, uc_l, xmitq);
536 tipc_bcbase_select_primary(net);
537 tipc_bcbase_calc_bc_threshold(net);
538 tipc_bcast_unlock(net);
539}
540
541/* tipc_bcast_remove_peer - remove a peer node from broadcast link and bearer
542 *
543 * RCU is locked, node lock is set
544 */
545void tipc_bcast_remove_peer(struct net *net, struct tipc_link *rcv_l)
546{
547 struct tipc_link *snd_l = tipc_bc_sndlink(net);
548 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
549 struct sk_buff_head xmitq;
550
551 __skb_queue_head_init(&xmitq);
552
553 tipc_bcast_lock(net);
554 tipc_link_remove_bc_peer(snd_l, rcv_l, &xmitq);
555 tipc_bcbase_select_primary(net);
556 tipc_bcbase_calc_bc_threshold(net);
557 tipc_bcast_unlock(net);
558
559 tipc_bcbase_xmit(net, &xmitq);
560
561 /* Any socket wakeup messages ? */
562 if (!skb_queue_empty(inputq))
563 tipc_sk_rcv(net, inputq);
564}
565
566int tipc_bclink_reset_stats(struct net *net, struct tipc_link *l)
567{
568 if (!l)
569 return -ENOPROTOOPT;
570
571 tipc_bcast_lock(net);
572 tipc_link_reset_stats(l);
573 tipc_bcast_unlock(net);
574 return 0;
575}
576
577static int tipc_bc_link_set_queue_limits(struct net *net, u32 max_win)
578{
579 struct tipc_link *l = tipc_bc_sndlink(net);
580
581 if (!l)
582 return -ENOPROTOOPT;
583 if (max_win < BCLINK_WIN_MIN)
584 max_win = BCLINK_WIN_MIN;
585 if (max_win > TIPC_MAX_LINK_WIN)
586 return -EINVAL;
587 tipc_bcast_lock(net);
588 tipc_link_set_queue_limits(l, BCLINK_WIN_MIN, max_win);
589 tipc_bcast_unlock(net);
590 return 0;
591}
592
593static int tipc_bc_link_set_broadcast_mode(struct net *net, u32 bc_mode)
594{
595 struct tipc_bc_base *bb = tipc_bc_base(net);
596
597 switch (bc_mode) {
598 case BCLINK_MODE_BCAST:
599 if (!bb->bcast_support)
600 return -ENOPROTOOPT;
601
602 bb->force_bcast = true;
603 bb->force_rcast = false;
604 break;
605 case BCLINK_MODE_RCAST:
606 if (!bb->rcast_support)
607 return -ENOPROTOOPT;
608
609 bb->force_bcast = false;
610 bb->force_rcast = true;
611 break;
612 case BCLINK_MODE_SEL:
613 if (!bb->bcast_support || !bb->rcast_support)
614 return -ENOPROTOOPT;
615
616 bb->force_bcast = false;
617 bb->force_rcast = false;
618 break;
619 default:
620 return -EINVAL;
621 }
622
623 return 0;
624}
625
626static int tipc_bc_link_set_broadcast_ratio(struct net *net, u32 bc_ratio)
627{
628 struct tipc_bc_base *bb = tipc_bc_base(net);
629
630 if (!bb->bcast_support || !bb->rcast_support)
631 return -ENOPROTOOPT;
632
633 if (bc_ratio > 100 || bc_ratio <= 0)
634 return -EINVAL;
635
636 bb->rc_ratio = bc_ratio;
637 tipc_bcast_lock(net);
638 tipc_bcbase_calc_bc_threshold(net);
639 tipc_bcast_unlock(net);
640
641 return 0;
642}
643
644int tipc_nl_bc_link_set(struct net *net, struct nlattr *attrs[])
645{
646 int err;
647 u32 win;
648 u32 bc_mode;
649 u32 bc_ratio;
650 struct nlattr *props[TIPC_NLA_PROP_MAX + 1];
651
652 if (!attrs[TIPC_NLA_LINK_PROP])
653 return -EINVAL;
654
655 err = tipc_nl_parse_link_prop(attrs[TIPC_NLA_LINK_PROP], props);
656 if (err)
657 return err;
658
659 if (!props[TIPC_NLA_PROP_WIN] &&
660 !props[TIPC_NLA_PROP_BROADCAST] &&
661 !props[TIPC_NLA_PROP_BROADCAST_RATIO]) {
662 return -EOPNOTSUPP;
663 }
664
665 if (props[TIPC_NLA_PROP_BROADCAST]) {
666 bc_mode = nla_get_u32(props[TIPC_NLA_PROP_BROADCAST]);
667 err = tipc_bc_link_set_broadcast_mode(net, bc_mode);
668 }
669
670 if (!err && props[TIPC_NLA_PROP_BROADCAST_RATIO]) {
671 bc_ratio = nla_get_u32(props[TIPC_NLA_PROP_BROADCAST_RATIO]);
672 err = tipc_bc_link_set_broadcast_ratio(net, bc_ratio);
673 }
674
675 if (!err && props[TIPC_NLA_PROP_WIN]) {
676 win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
677 err = tipc_bc_link_set_queue_limits(net, win);
678 }
679
680 return err;
681}
682
683int tipc_bcast_init(struct net *net)
684{
685 struct tipc_net *tn = tipc_net(net);
686 struct tipc_bc_base *bb = NULL;
687 struct tipc_link *l = NULL;
688
689 bb = kzalloc(sizeof(*bb), GFP_KERNEL);
690 if (!bb)
691 goto enomem;
692 tn->bcbase = bb;
693 spin_lock_init(&tipc_net(net)->bclock);
694
695 if (!tipc_link_bc_create(net, 0, 0, NULL,
696 FB_MTU,
697 BCLINK_WIN_DEFAULT,
698 BCLINK_WIN_DEFAULT,
699 0,
700 &bb->inputq,
701 NULL,
702 NULL,
703 &l))
704 goto enomem;
705 bb->link = l;
706 tn->bcl = l;
707 bb->rc_ratio = 10;
708 bb->rcast_support = true;
709 return 0;
710enomem:
711 kfree(bb);
712 kfree(l);
713 return -ENOMEM;
714}
715
716void tipc_bcast_stop(struct net *net)
717{
718 struct tipc_net *tn = net_generic(net, tipc_net_id);
719
720 synchronize_net();
721 kfree(tn->bcbase);
722 kfree(tn->bcl);
723}
724
725void tipc_nlist_init(struct tipc_nlist *nl, u32 self)
726{
727 memset(nl, 0, sizeof(*nl));
728 INIT_LIST_HEAD(&nl->list);
729 nl->self = self;
730}
731
732void tipc_nlist_add(struct tipc_nlist *nl, u32 node)
733{
734 if (node == nl->self)
735 nl->local = true;
736 else if (tipc_dest_push(&nl->list, node, 0))
737 nl->remote++;
738}
739
740void tipc_nlist_del(struct tipc_nlist *nl, u32 node)
741{
742 if (node == nl->self)
743 nl->local = false;
744 else if (tipc_dest_del(&nl->list, node, 0))
745 nl->remote--;
746}
747
748void tipc_nlist_purge(struct tipc_nlist *nl)
749{
750 tipc_dest_list_purge(&nl->list);
751 nl->remote = 0;
752 nl->local = false;
753}
754
755u32 tipc_bcast_get_mode(struct net *net)
756{
757 struct tipc_bc_base *bb = tipc_bc_base(net);
758
759 if (bb->force_bcast)
760 return BCLINK_MODE_BCAST;
761
762 if (bb->force_rcast)
763 return BCLINK_MODE_RCAST;
764
765 if (bb->bcast_support && bb->rcast_support)
766 return BCLINK_MODE_SEL;
767
768 return 0;
769}
770
771u32 tipc_bcast_get_broadcast_ratio(struct net *net)
772{
773 struct tipc_bc_base *bb = tipc_bc_base(net);
774
775 return bb->rc_ratio;
776}
777
778void tipc_mcast_filter_msg(struct net *net, struct sk_buff_head *defq,
779 struct sk_buff_head *inputq)
780{
781 struct sk_buff *skb, *_skb, *tmp;
782 struct tipc_msg *hdr, *_hdr;
783 bool match = false;
784 u32 node, port;
785
786 skb = skb_peek(inputq);
787 if (!skb)
788 return;
789
790 hdr = buf_msg(skb);
791
792 if (likely(!msg_is_syn(hdr) && skb_queue_empty(defq)))
793 return;
794
795 node = msg_orignode(hdr);
796 if (node == tipc_own_addr(net))
797 return;
798
799 port = msg_origport(hdr);
800
801 /* Has the twin SYN message already arrived ? */
802 skb_queue_walk(defq, _skb) {
803 _hdr = buf_msg(_skb);
804 if (msg_orignode(_hdr) != node)
805 continue;
806 if (msg_origport(_hdr) != port)
807 continue;
808 match = true;
809 break;
810 }
811
812 if (!match) {
813 if (!msg_is_syn(hdr))
814 return;
815 __skb_dequeue(inputq);
816 __skb_queue_tail(defq, skb);
817 return;
818 }
819
820 /* Deliver non-SYN message from other link, otherwise queue it */
821 if (!msg_is_syn(hdr)) {
822 if (msg_is_rcast(hdr) != msg_is_rcast(_hdr))
823 return;
824 __skb_dequeue(inputq);
825 __skb_queue_tail(defq, skb);
826 return;
827 }
828
829 /* Queue non-SYN/SYN message from same link */
830 if (msg_is_rcast(hdr) == msg_is_rcast(_hdr)) {
831 __skb_dequeue(inputq);
832 __skb_queue_tail(defq, skb);
833 return;
834 }
835
836 /* Matching SYN messages => return the one with data, if any */
837 __skb_unlink(_skb, defq);
838 if (msg_data_sz(hdr)) {
839 kfree_skb(_skb);
840 } else {
841 __skb_dequeue(inputq);
842 kfree_skb(skb);
843 __skb_queue_tail(inputq, _skb);
844 }
845
846 /* Deliver subsequent non-SYN messages from same peer */
847 skb_queue_walk_safe(defq, _skb, tmp) {
848 _hdr = buf_msg(_skb);
849 if (msg_orignode(_hdr) != node)
850 continue;
851 if (msg_origport(_hdr) != port)
852 continue;
853 if (msg_is_syn(_hdr))
854 break;
855 __skb_unlink(_skb, defq);
856 __skb_queue_tail(inputq, _skb);
857 }
858}
1/*
2 * net/tipc/bcast.c: TIPC broadcast code
3 *
4 * Copyright (c) 2004-2006, 2014-2015, Ericsson AB
5 * Copyright (c) 2004, Intel Corporation.
6 * Copyright (c) 2005, 2010-2011, Wind River Systems
7 * All rights reserved.
8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions are met:
11 *
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the names of the copyright holders nor the names of its
18 * contributors may be used to endorse or promote products derived from
19 * this software without specific prior written permission.
20 *
21 * Alternatively, this software may be distributed under the terms of the
22 * GNU General Public License ("GPL") version 2 as published by the Free
23 * Software Foundation.
24 *
25 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
26 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
29 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35 * POSSIBILITY OF SUCH DAMAGE.
36 */
37
38#include <linux/tipc_config.h>
39#include "socket.h"
40#include "msg.h"
41#include "bcast.h"
42#include "name_distr.h"
43#include "link.h"
44#include "node.h"
45
46#define BCLINK_WIN_DEFAULT 50 /* bcast link window size (default) */
47#define BCLINK_WIN_MIN 32 /* bcast minimum link window size */
48
49const char tipc_bclink_name[] = "broadcast-link";
50
51/**
52 * struct tipc_bc_base - base structure for keeping broadcast send state
53 * @link: broadcast send link structure
54 * @inputq: data input queue; will only carry SOCK_WAKEUP messages
55 * @dest: array keeping number of reachable destinations per bearer
56 * @primary_bearer: a bearer having links to all broadcast destinations, if any
57 */
58struct tipc_bc_base {
59 struct tipc_link *link;
60 struct sk_buff_head inputq;
61 int dests[MAX_BEARERS];
62 int primary_bearer;
63};
64
65static struct tipc_bc_base *tipc_bc_base(struct net *net)
66{
67 return tipc_net(net)->bcbase;
68}
69
70int tipc_bcast_get_mtu(struct net *net)
71{
72 return tipc_link_mtu(tipc_bc_sndlink(net));
73}
74
75/* tipc_bcbase_select_primary(): find a bearer with links to all destinations,
76 * if any, and make it primary bearer
77 */
78static void tipc_bcbase_select_primary(struct net *net)
79{
80 struct tipc_bc_base *bb = tipc_bc_base(net);
81 int all_dests = tipc_link_bc_peers(bb->link);
82 int i, mtu;
83
84 bb->primary_bearer = INVALID_BEARER_ID;
85
86 if (!all_dests)
87 return;
88
89 for (i = 0; i < MAX_BEARERS; i++) {
90 if (!bb->dests[i])
91 continue;
92
93 mtu = tipc_bearer_mtu(net, i);
94 if (mtu < tipc_link_mtu(bb->link))
95 tipc_link_set_mtu(bb->link, mtu);
96
97 if (bb->dests[i] < all_dests)
98 continue;
99
100 bb->primary_bearer = i;
101
102 /* Reduce risk that all nodes select same primary */
103 if ((i ^ tipc_own_addr(net)) & 1)
104 break;
105 }
106}
107
108void tipc_bcast_inc_bearer_dst_cnt(struct net *net, int bearer_id)
109{
110 struct tipc_bc_base *bb = tipc_bc_base(net);
111
112 tipc_bcast_lock(net);
113 bb->dests[bearer_id]++;
114 tipc_bcbase_select_primary(net);
115 tipc_bcast_unlock(net);
116}
117
118void tipc_bcast_dec_bearer_dst_cnt(struct net *net, int bearer_id)
119{
120 struct tipc_bc_base *bb = tipc_bc_base(net);
121
122 tipc_bcast_lock(net);
123 bb->dests[bearer_id]--;
124 tipc_bcbase_select_primary(net);
125 tipc_bcast_unlock(net);
126}
127
128/* tipc_bcbase_xmit - broadcast a packet queue across one or more bearers
129 *
130 * Note that number of reachable destinations, as indicated in the dests[]
131 * array, may transitionally differ from the number of destinations indicated
132 * in each sent buffer. We can sustain this. Excess destination nodes will
133 * drop and never acknowledge the unexpected packets, and missing destinations
134 * will either require retransmission (if they are just about to be added to
135 * the bearer), or be removed from the buffer's 'ackers' counter (if they
136 * just went down)
137 */
138static void tipc_bcbase_xmit(struct net *net, struct sk_buff_head *xmitq)
139{
140 int bearer_id;
141 struct tipc_bc_base *bb = tipc_bc_base(net);
142 struct sk_buff *skb, *_skb;
143 struct sk_buff_head _xmitq;
144
145 if (skb_queue_empty(xmitq))
146 return;
147
148 /* The typical case: at least one bearer has links to all nodes */
149 bearer_id = bb->primary_bearer;
150 if (bearer_id >= 0) {
151 tipc_bearer_bc_xmit(net, bearer_id, xmitq);
152 return;
153 }
154
155 /* We have to transmit across all bearers */
156 skb_queue_head_init(&_xmitq);
157 for (bearer_id = 0; bearer_id < MAX_BEARERS; bearer_id++) {
158 if (!bb->dests[bearer_id])
159 continue;
160
161 skb_queue_walk(xmitq, skb) {
162 _skb = pskb_copy_for_clone(skb, GFP_ATOMIC);
163 if (!_skb)
164 break;
165 __skb_queue_tail(&_xmitq, _skb);
166 }
167 tipc_bearer_bc_xmit(net, bearer_id, &_xmitq);
168 }
169 __skb_queue_purge(xmitq);
170 __skb_queue_purge(&_xmitq);
171}
172
173/* tipc_bcast_xmit - deliver buffer chain to all nodes in cluster
174 * and to identified node local sockets
175 * @net: the applicable net namespace
176 * @list: chain of buffers containing message
177 * Consumes the buffer chain, except when returning -ELINKCONG
178 * Returns 0 if success, otherwise errno: -ELINKCONG,-EHOSTUNREACH,-EMSGSIZE
179 */
180int tipc_bcast_xmit(struct net *net, struct sk_buff_head *list)
181{
182 struct tipc_link *l = tipc_bc_sndlink(net);
183 struct sk_buff_head xmitq, inputq, rcvq;
184 int rc = 0;
185
186 __skb_queue_head_init(&rcvq);
187 __skb_queue_head_init(&xmitq);
188 skb_queue_head_init(&inputq);
189
190 /* Prepare message clone for local node */
191 if (unlikely(!tipc_msg_reassemble(list, &rcvq)))
192 return -EHOSTUNREACH;
193
194 tipc_bcast_lock(net);
195 if (tipc_link_bc_peers(l))
196 rc = tipc_link_xmit(l, list, &xmitq);
197 tipc_bcast_unlock(net);
198
199 /* Don't send to local node if adding to link failed */
200 if (unlikely(rc)) {
201 __skb_queue_purge(&rcvq);
202 return rc;
203 }
204
205 /* Broadcast to all nodes, inluding local node */
206 tipc_bcbase_xmit(net, &xmitq);
207 tipc_sk_mcast_rcv(net, &rcvq, &inputq);
208 __skb_queue_purge(list);
209 return 0;
210}
211
212/* tipc_bcast_rcv - receive a broadcast packet, and deliver to rcv link
213 *
214 * RCU is locked, no other locks set
215 */
216int tipc_bcast_rcv(struct net *net, struct tipc_link *l, struct sk_buff *skb)
217{
218 struct tipc_msg *hdr = buf_msg(skb);
219 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
220 struct sk_buff_head xmitq;
221 int rc;
222
223 __skb_queue_head_init(&xmitq);
224
225 if (msg_mc_netid(hdr) != tipc_netid(net) || !tipc_link_is_up(l)) {
226 kfree_skb(skb);
227 return 0;
228 }
229
230 tipc_bcast_lock(net);
231 if (msg_user(hdr) == BCAST_PROTOCOL)
232 rc = tipc_link_bc_nack_rcv(l, skb, &xmitq);
233 else
234 rc = tipc_link_rcv(l, skb, NULL);
235 tipc_bcast_unlock(net);
236
237 tipc_bcbase_xmit(net, &xmitq);
238
239 /* Any socket wakeup messages ? */
240 if (!skb_queue_empty(inputq))
241 tipc_sk_rcv(net, inputq);
242
243 return rc;
244}
245
246/* tipc_bcast_ack_rcv - receive and handle a broadcast acknowledge
247 *
248 * RCU is locked, no other locks set
249 */
250void tipc_bcast_ack_rcv(struct net *net, struct tipc_link *l,
251 struct tipc_msg *hdr)
252{
253 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
254 u16 acked = msg_bcast_ack(hdr);
255 struct sk_buff_head xmitq;
256
257 /* Ignore bc acks sent by peer before bcast synch point was received */
258 if (msg_bc_ack_invalid(hdr))
259 return;
260
261 __skb_queue_head_init(&xmitq);
262
263 tipc_bcast_lock(net);
264 tipc_link_bc_ack_rcv(l, acked, &xmitq);
265 tipc_bcast_unlock(net);
266
267 tipc_bcbase_xmit(net, &xmitq);
268
269 /* Any socket wakeup messages ? */
270 if (!skb_queue_empty(inputq))
271 tipc_sk_rcv(net, inputq);
272}
273
274/* tipc_bcast_synch_rcv - check and update rcv link with peer's send state
275 *
276 * RCU is locked, no other locks set
277 */
278int tipc_bcast_sync_rcv(struct net *net, struct tipc_link *l,
279 struct tipc_msg *hdr)
280{
281 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
282 struct sk_buff_head xmitq;
283 int rc = 0;
284
285 __skb_queue_head_init(&xmitq);
286
287 tipc_bcast_lock(net);
288 if (msg_type(hdr) != STATE_MSG) {
289 tipc_link_bc_init_rcv(l, hdr);
290 } else if (!msg_bc_ack_invalid(hdr)) {
291 tipc_link_bc_ack_rcv(l, msg_bcast_ack(hdr), &xmitq);
292 rc = tipc_link_bc_sync_rcv(l, hdr, &xmitq);
293 }
294 tipc_bcast_unlock(net);
295
296 tipc_bcbase_xmit(net, &xmitq);
297
298 /* Any socket wakeup messages ? */
299 if (!skb_queue_empty(inputq))
300 tipc_sk_rcv(net, inputq);
301 return rc;
302}
303
304/* tipc_bcast_add_peer - add a peer node to broadcast link and bearer
305 *
306 * RCU is locked, node lock is set
307 */
308void tipc_bcast_add_peer(struct net *net, struct tipc_link *uc_l,
309 struct sk_buff_head *xmitq)
310{
311 struct tipc_link *snd_l = tipc_bc_sndlink(net);
312
313 tipc_bcast_lock(net);
314 tipc_link_add_bc_peer(snd_l, uc_l, xmitq);
315 tipc_bcbase_select_primary(net);
316 tipc_bcast_unlock(net);
317}
318
319/* tipc_bcast_remove_peer - remove a peer node from broadcast link and bearer
320 *
321 * RCU is locked, node lock is set
322 */
323void tipc_bcast_remove_peer(struct net *net, struct tipc_link *rcv_l)
324{
325 struct tipc_link *snd_l = tipc_bc_sndlink(net);
326 struct sk_buff_head *inputq = &tipc_bc_base(net)->inputq;
327 struct sk_buff_head xmitq;
328
329 __skb_queue_head_init(&xmitq);
330
331 tipc_bcast_lock(net);
332 tipc_link_remove_bc_peer(snd_l, rcv_l, &xmitq);
333 tipc_bcbase_select_primary(net);
334 tipc_bcast_unlock(net);
335
336 tipc_bcbase_xmit(net, &xmitq);
337
338 /* Any socket wakeup messages ? */
339 if (!skb_queue_empty(inputq))
340 tipc_sk_rcv(net, inputq);
341}
342
343int tipc_bclink_reset_stats(struct net *net)
344{
345 struct tipc_link *l = tipc_bc_sndlink(net);
346
347 if (!l)
348 return -ENOPROTOOPT;
349
350 tipc_bcast_lock(net);
351 tipc_link_reset_stats(l);
352 tipc_bcast_unlock(net);
353 return 0;
354}
355
356static int tipc_bc_link_set_queue_limits(struct net *net, u32 limit)
357{
358 struct tipc_link *l = tipc_bc_sndlink(net);
359
360 if (!l)
361 return -ENOPROTOOPT;
362 if (limit < BCLINK_WIN_MIN)
363 limit = BCLINK_WIN_MIN;
364 if (limit > TIPC_MAX_LINK_WIN)
365 return -EINVAL;
366 tipc_bcast_lock(net);
367 tipc_link_set_queue_limits(l, limit);
368 tipc_bcast_unlock(net);
369 return 0;
370}
371
372int tipc_nl_bc_link_set(struct net *net, struct nlattr *attrs[])
373{
374 int err;
375 u32 win;
376 struct nlattr *props[TIPC_NLA_PROP_MAX + 1];
377
378 if (!attrs[TIPC_NLA_LINK_PROP])
379 return -EINVAL;
380
381 err = tipc_nl_parse_link_prop(attrs[TIPC_NLA_LINK_PROP], props);
382 if (err)
383 return err;
384
385 if (!props[TIPC_NLA_PROP_WIN])
386 return -EOPNOTSUPP;
387
388 win = nla_get_u32(props[TIPC_NLA_PROP_WIN]);
389
390 return tipc_bc_link_set_queue_limits(net, win);
391}
392
393int tipc_bcast_init(struct net *net)
394{
395 struct tipc_net *tn = tipc_net(net);
396 struct tipc_bc_base *bb = NULL;
397 struct tipc_link *l = NULL;
398
399 bb = kzalloc(sizeof(*bb), GFP_ATOMIC);
400 if (!bb)
401 goto enomem;
402 tn->bcbase = bb;
403 spin_lock_init(&tipc_net(net)->bclock);
404
405 if (!tipc_link_bc_create(net, 0, 0,
406 U16_MAX,
407 BCLINK_WIN_DEFAULT,
408 0,
409 &bb->inputq,
410 NULL,
411 NULL,
412 &l))
413 goto enomem;
414 bb->link = l;
415 tn->bcl = l;
416 return 0;
417enomem:
418 kfree(bb);
419 kfree(l);
420 return -ENOMEM;
421}
422
423void tipc_bcast_stop(struct net *net)
424{
425 struct tipc_net *tn = net_generic(net, tipc_net_id);
426
427 synchronize_net();
428 kfree(tn->bcbase);
429 kfree(tn->bcl);
430}