net/tipc/bcast.c

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