net/tipc/bcast.c

   1 /*
   2  * net/tipc/bcast.c: TIPC broadcast code
   3  *
   4  * Copyright (c) 2004-2006, 2014, 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 "core.h"
  39 #include "link.h"
  40 #include "socket.h"
  41 #include "msg.h"
  42 #include "bcast.h"
  43 #include "name_distr.h"
  44
  45 #define MAX_PKT_DEFAULT_MCAST   1500    /* bcast link max packet size (fixed) */
  46 #define BCLINK_WIN_DEFAULT      20      /* bcast link window size (default) */
  47 #define BCBEARER                MAX_BEARERS
  48
  49 /**
  50  * struct tipc_bcbearer_pair - a pair of bearers used by broadcast link
  51  * @primary: pointer to primary bearer
  52  * @secondary: pointer to secondary bearer
  53  *
  54  * Bearers must have same priority and same set of reachable destinations
  55  * to be paired.
  56  */
  57
  58 struct tipc_bcbearer_pair {
  59         struct tipc_bearer *primary;
  60         struct tipc_bearer *secondary;
  61 };
  62
  63 /**
  64  * struct tipc_bcbearer - bearer used by broadcast link
  65  * @bearer: (non-standard) broadcast bearer structure
  66  * @media: (non-standard) broadcast media structure
  67  * @bpairs: array of bearer pairs
  68  * @bpairs_temp: temporary array of bearer pairs used by tipc_bcbearer_sort()
  69  * @remains: temporary node map used by tipc_bcbearer_send()
  70  * @remains_new: temporary node map used tipc_bcbearer_send()
  71  *
  72  * Note: The fields labelled "temporary" are incorporated into the bearer
  73  * to avoid consuming potentially limited stack space through the use of
  74  * large local variables within multicast routines.  Concurrent access is
  75  * prevented through use of the spinlock "bclink_lock".
  76  */
  77 struct tipc_bcbearer {
  78         struct tipc_bearer bearer;
  79         struct tipc_media media;
  80         struct tipc_bcbearer_pair bpairs[MAX_BEARERS];
  81         struct tipc_bcbearer_pair bpairs_temp[TIPC_MAX_LINK_PRI + 1];
  82         struct tipc_node_map remains;
  83         struct tipc_node_map remains_new;
  84 };
  85
  86 /**
  87  * struct tipc_bclink - link used for broadcast messages
  88  * @lock: spinlock governing access to structure
  89  * @link: (non-standard) broadcast link structure
  90  * @node: (non-standard) node structure representing b'cast link's peer node
  91  * @flags: represent bclink states
  92  * @bcast_nodes: map of broadcast-capable nodes
  93  * @retransmit_to: node that most recently requested a retransmit
  94  *
  95  * Handles sequence numbering, fragmentation, bundling, etc.
  96  */
  97 struct tipc_bclink {
  98         spinlock_t lock;
  99         struct tipc_link link;
 100         struct tipc_node node;
 101         unsigned int flags;
 102         struct tipc_node_map bcast_nodes;
 103         struct tipc_node *retransmit_to;
 104 };
 105
 106 static struct tipc_bcbearer *bcbearer;
 107 static struct tipc_bclink *bclink;
 108 static struct tipc_link *bcl;
 109
 110 const char tipc_bclink_name[] = "broadcast-link";
 111
 112 static void tipc_nmap_diff(struct tipc_node_map *nm_a,
 113                            struct tipc_node_map *nm_b,
 114                            struct tipc_node_map *nm_diff);
 115 static void tipc_nmap_add(struct tipc_node_map *nm_ptr, u32 node);
 116 static void tipc_nmap_remove(struct tipc_node_map *nm_ptr, u32 node);
 117
 118 static void tipc_bclink_lock(void)
 119 {
 120         spin_lock_bh(&bclink->lock);
 121 }
 122
 123 static void tipc_bclink_unlock(void)
 124 {
 125         struct tipc_node *node = NULL;
 126
 127         if (likely(!bclink->flags)) {
 128                 spin_unlock_bh(&bclink->lock);
 129                 return;
 130         }
 131
 132         if (bclink->flags & TIPC_BCLINK_RESET) {
 133                 bclink->flags &= ~TIPC_BCLINK_RESET;
 134                 node = tipc_bclink_retransmit_to();
 135         }
 136         spin_unlock_bh(&bclink->lock);
 137
 138         if (node)
 139                 tipc_link_reset_all(node);
 140 }
 141
 142 uint  tipc_bclink_get_mtu(void)
 143 {
 144         return MAX_PKT_DEFAULT_MCAST;
 145 }
 146
 147 void tipc_bclink_set_flags(unsigned int flags)
 148 {
 149         bclink->flags |= flags;
 150 }
 151
 152 static u32 bcbuf_acks(struct sk_buff *buf)
 153 {
 154         return (u32)(unsigned long)TIPC_SKB_CB(buf)->handle;
 155 }
 156
 157 static void bcbuf_set_acks(struct sk_buff *buf, u32 acks)
 158 {
 159         TIPC_SKB_CB(buf)->handle = (void *)(unsigned long)acks;
 160 }
 161
 162 static void bcbuf_decr_acks(struct sk_buff *buf)
 163 {
 164         bcbuf_set_acks(buf, bcbuf_acks(buf) - 1);
 165 }
 166
 167 void tipc_bclink_add_node(u32 addr)
 168 {
 169         tipc_bclink_lock();
 170         tipc_nmap_add(&bclink->bcast_nodes, addr);
 171         tipc_bclink_unlock();
 172 }
 173
 174 void tipc_bclink_remove_node(u32 addr)
 175 {
 176         tipc_bclink_lock();
 177         tipc_nmap_remove(&bclink->bcast_nodes, addr);
 178         tipc_bclink_unlock();
 179 }
 180
 181 static void bclink_set_last_sent(void)
 182 {
 183         if (bcl->next_out)
 184                 bcl->fsm_msg_cnt = mod(buf_seqno(bcl->next_out) - 1);
 185         else
 186                 bcl->fsm_msg_cnt = mod(bcl->next_out_no - 1);
 187 }
 188
 189 u32 tipc_bclink_get_last_sent(void)
 190 {
 191         return bcl->fsm_msg_cnt;
 192 }
 193
 194 static void bclink_update_last_sent(struct tipc_node *node, u32 seqno)
 195 {
 196         node->bclink.last_sent = less_eq(node->bclink.last_sent, seqno) ?
 197                                                 seqno : node->bclink.last_sent;
 198 }
 199
 200
 201 /**
 202  * tipc_bclink_retransmit_to - get most recent node to request retransmission
 203  *
 204  * Called with bclink_lock locked
 205  */
 206 struct tipc_node *tipc_bclink_retransmit_to(void)
 207 {
 208         return bclink->retransmit_to;
 209 }
 210
 211 /**
 212  * bclink_retransmit_pkt - retransmit broadcast packets
 213  * @after: sequence number of last packet to *not* retransmit
 214  * @to: sequence number of last packet to retransmit
 215  *
 216  * Called with bclink_lock locked
 217  */
 218 static void bclink_retransmit_pkt(u32 after, u32 to)
 219 {
 220         struct sk_buff *skb;
 221
 222         skb_queue_walk(&bcl->outqueue, skb) {
 223                 if (more(buf_seqno(skb), after))
 224                         break;
 225         }
 226         tipc_link_retransmit(bcl, skb, mod(to - after));
 227 }
 228
 229 /**
 230  * tipc_bclink_wakeup_users - wake up pending users
 231  *
 232  * Called with no locks taken
 233  */
 234 void tipc_bclink_wakeup_users(void)
 235 {
 236         struct sk_buff *skb;
 237
 238         while ((skb = skb_dequeue(&bclink->link.waiting_sks)))
 239                 tipc_sk_rcv(skb);
 240
 241 }
 242
 243 /**
 244  * tipc_bclink_acknowledge - handle acknowledgement of broadcast packets
 245  * @n_ptr: node that sent acknowledgement info
 246  * @acked: broadcast sequence # that has been acknowledged
 247  *
 248  * Node is locked, bclink_lock unlocked.
 249  */
 250 void tipc_bclink_acknowledge(struct tipc_node *n_ptr, u32 acked)
 251 {
 252         struct sk_buff *skb, *tmp;
 253         struct sk_buff *next;
 254         unsigned int released = 0;
 255
 256         tipc_bclink_lock();
 257         /* Bail out if tx queue is empty (no clean up is required) */
 258         skb = skb_peek(&bcl->outqueue);
 259         if (!skb)
 260                 goto exit;
 261
 262         /* Determine which messages need to be acknowledged */
 263         if (acked == INVALID_LINK_SEQ) {
 264                 /*
 265                  * Contact with specified node has been lost, so need to
 266                  * acknowledge sent messages only (if other nodes still exist)
 267                  * or both sent and unsent messages (otherwise)
 268                  */
 269                 if (bclink->bcast_nodes.count)
 270                         acked = bcl->fsm_msg_cnt;
 271                 else
 272                         acked = bcl->next_out_no;
 273         } else {
 274                 /*
 275                  * Bail out if specified sequence number does not correspond
 276                  * to a message that has been sent and not yet acknowledged
 277                  */
 278                 if (less(acked, buf_seqno(skb)) ||
 279                     less(bcl->fsm_msg_cnt, acked) ||
 280                     less_eq(acked, n_ptr->bclink.acked))
 281                         goto exit;
 282         }
 283
 284         /* Skip over packets that node has previously acknowledged */
 285         skb_queue_walk(&bcl->outqueue, skb) {
 286                 if (more(buf_seqno(skb), n_ptr->bclink.acked))
 287                         break;
 288         }
 289
 290         /* Update packets that node is now acknowledging */
 291         skb_queue_walk_from_safe(&bcl->outqueue, skb, tmp) {
 292                 if (more(buf_seqno(skb), acked))
 293                         break;
 294
 295                 next = tipc_skb_queue_next(&bcl->outqueue, skb);
 296                 if (skb != bcl->next_out) {
 297                         bcbuf_decr_acks(skb);
 298                 } else {
 299                         bcbuf_set_acks(skb, 0);
 300                         bcl->next_out = next;
 301                         bclink_set_last_sent();
 302                 }
 303
 304                 if (bcbuf_acks(skb) == 0) {
 305                         __skb_unlink(skb, &bcl->outqueue);
 306                         kfree_skb(skb);
 307                         released = 1;
 308                 }
 309         }
 310         n_ptr->bclink.acked = acked;
 311
 312         /* Try resolving broadcast link congestion, if necessary */
 313         if (unlikely(bcl->next_out)) {
 314                 tipc_link_push_packets(bcl);
 315                 bclink_set_last_sent();
 316         }
 317         if (unlikely(released && !skb_queue_empty(&bcl->waiting_sks)))
 318                 n_ptr->action_flags |= TIPC_WAKEUP_BCAST_USERS;
 319
 320 exit:
 321         tipc_bclink_unlock();
 322 }
 323
 324 /**
 325  * tipc_bclink_update_link_state - update broadcast link state
 326  *
 327  * RCU and node lock set
 328  */
 329 void tipc_bclink_update_link_state(struct tipc_node *n_ptr, u32 last_sent)
 330 {
 331         struct sk_buff *buf;
 332
 333         /* Ignore "stale" link state info */
 334         if (less_eq(last_sent, n_ptr->bclink.last_in))
 335                 return;
 336
 337         /* Update link synchronization state; quit if in sync */
 338         bclink_update_last_sent(n_ptr, last_sent);
 339
 340         if (n_ptr->bclink.last_sent == n_ptr->bclink.last_in)
 341                 return;
 342
 343         /* Update out-of-sync state; quit if loss is still unconfirmed */
 344         if ((++n_ptr->bclink.oos_state) == 1) {
 345                 if (n_ptr->bclink.deferred_size < (TIPC_MIN_LINK_WIN / 2))
 346                         return;
 347                 n_ptr->bclink.oos_state++;
 348         }
 349
 350         /* Don't NACK if one has been recently sent (or seen) */
 351         if (n_ptr->bclink.oos_state & 0x1)
 352                 return;
 353
 354         /* Send NACK */
 355         buf = tipc_buf_acquire(INT_H_SIZE);
 356         if (buf) {
 357                 struct tipc_msg *msg = buf_msg(buf);
 358                 struct sk_buff *skb = skb_peek(&n_ptr->bclink.deferred_queue);
 359                 u32 to = skb ? buf_seqno(skb) - 1 : n_ptr->bclink.last_sent;
 360
 361                 tipc_msg_init(msg, BCAST_PROTOCOL, STATE_MSG,
 362                               INT_H_SIZE, n_ptr->addr);
 363                 msg_set_non_seq(msg, 1);
 364                 msg_set_mc_netid(msg, tipc_net_id);
 365                 msg_set_bcast_ack(msg, n_ptr->bclink.last_in);
 366                 msg_set_bcgap_after(msg, n_ptr->bclink.last_in);
 367                 msg_set_bcgap_to(msg, to);
 368
 369                 tipc_bclink_lock();
 370                 tipc_bearer_send(MAX_BEARERS, buf, NULL);
 371                 bcl->stats.sent_nacks++;
 372                 tipc_bclink_unlock();
 373                 kfree_skb(buf);
 374
 375                 n_ptr->bclink.oos_state++;
 376         }
 377 }
 378
 379 /**
 380  * bclink_peek_nack - monitor retransmission requests sent by other nodes
 381  *
 382  * Delay any upcoming NACK by this node if another node has already
 383  * requested the first message this node is going to ask for.
 384  */
 385 static void bclink_peek_nack(struct tipc_msg *msg)
 386 {
 387         struct tipc_node *n_ptr = tipc_node_find(msg_destnode(msg));
 388
 389         if (unlikely(!n_ptr))
 390                 return;
 391
 392         tipc_node_lock(n_ptr);
 393
 394         if (n_ptr->bclink.recv_permitted &&
 395             (n_ptr->bclink.last_in != n_ptr->bclink.last_sent) &&
 396             (n_ptr->bclink.last_in == msg_bcgap_after(msg)))
 397                 n_ptr->bclink.oos_state = 2;
 398
 399         tipc_node_unlock(n_ptr);
 400 }
 401
 402 /* tipc_bclink_xmit - broadcast buffer chain to all nodes in cluster
 403  *                    and to identified node local sockets
 404  * @list: chain of buffers containing message
 405  * Consumes the buffer chain, except when returning -ELINKCONG
 406  * Returns 0 if success, otherwise errno: -ELINKCONG,-EHOSTUNREACH,-EMSGSIZE
 407  */
 408 int tipc_bclink_xmit(struct sk_buff_head *list)
 409 {
 410         int rc = 0;
 411         int bc = 0;
 412         struct sk_buff *skb;
 413
 414         /* Prepare clone of message for local node */
 415         skb = tipc_msg_reassemble(list);
 416         if (unlikely(!skb)) {
 417                 __skb_queue_purge(list);
 418                 return -EHOSTUNREACH;
 419         }
 420
 421         /* Broadcast to all other nodes */
 422         if (likely(bclink)) {
 423                 tipc_bclink_lock();
 424                 if (likely(bclink->bcast_nodes.count)) {
 425                         rc = __tipc_link_xmit(bcl, list);
 426                         if (likely(!rc)) {
 427                                 u32 len = skb_queue_len(&bcl->outqueue);
 428
 429                                 bclink_set_last_sent();
 430                                 bcl->stats.queue_sz_counts++;
 431                                 bcl->stats.accu_queue_sz += len;
 432                         }
 433                         bc = 1;
 434                 }
 435                 tipc_bclink_unlock();
 436         }
 437
 438         if (unlikely(!bc))
 439                 __skb_queue_purge(list);
 440
 441         /* Deliver message clone */
 442         if (likely(!rc))
 443                 tipc_sk_mcast_rcv(skb);
 444         else
 445                 kfree_skb(skb);
 446
 447         return rc;
 448 }
 449
 450 /**
 451  * bclink_accept_pkt - accept an incoming, in-sequence broadcast packet
 452  *
 453  * Called with both sending node's lock and bclink_lock taken.
 454  */
 455 static void bclink_accept_pkt(struct tipc_node *node, u32 seqno)
 456 {
 457         bclink_update_last_sent(node, seqno);
 458         node->bclink.last_in = seqno;
 459         node->bclink.oos_state = 0;
 460         bcl->stats.recv_info++;
 461
 462         /*
 463          * Unicast an ACK periodically, ensuring that
 464          * all nodes in the cluster don't ACK at the same time
 465          */
 466         if (((seqno - tipc_own_addr) % TIPC_MIN_LINK_WIN) == 0) {
 467                 tipc_link_proto_xmit(node->active_links[node->addr & 1],
 468                                      STATE_MSG, 0, 0, 0, 0, 0);
 469                 bcl->stats.sent_acks++;
 470         }
 471 }
 472
 473 /**
 474  * tipc_bclink_rcv - receive a broadcast packet, and deliver upwards
 475  *
 476  * RCU is locked, no other locks set
 477  */
 478 void tipc_bclink_rcv(struct sk_buff *buf)
 479 {
 480         struct tipc_msg *msg = buf_msg(buf);
 481         struct tipc_node *node;
 482         u32 next_in;
 483         u32 seqno;
 484         int deferred = 0;
 485
 486         /* Screen out unwanted broadcast messages */
 487         if (msg_mc_netid(msg) != tipc_net_id)
 488                 goto exit;
 489
 490         node = tipc_node_find(msg_prevnode(msg));
 491         if (unlikely(!node))
 492                 goto exit;
 493
 494         tipc_node_lock(node);
 495         if (unlikely(!node->bclink.recv_permitted))
 496                 goto unlock;
 497
 498         /* Handle broadcast protocol message */
 499         if (unlikely(msg_user(msg) == BCAST_PROTOCOL)) {
 500                 if (msg_type(msg) != STATE_MSG)
 501                         goto unlock;
 502                 if (msg_destnode(msg) == tipc_own_addr) {
 503                         tipc_bclink_acknowledge(node, msg_bcast_ack(msg));
 504                         tipc_node_unlock(node);
 505                         tipc_bclink_lock();
 506                         bcl->stats.recv_nacks++;
 507                         bclink->retransmit_to = node;
 508                         bclink_retransmit_pkt(msg_bcgap_after(msg),
 509                                               msg_bcgap_to(msg));
 510                         tipc_bclink_unlock();
 511                 } else {
 512                         tipc_node_unlock(node);
 513                         bclink_peek_nack(msg);
 514                 }
 515                 goto exit;
 516         }
 517
 518         /* Handle in-sequence broadcast message */
 519         seqno = msg_seqno(msg);
 520         next_in = mod(node->bclink.last_in + 1);
 521
 522         if (likely(seqno == next_in)) {
 523 receive:
 524                 /* Deliver message to destination */
 525                 if (likely(msg_isdata(msg))) {
 526                         tipc_bclink_lock();
 527                         bclink_accept_pkt(node, seqno);
 528                         tipc_bclink_unlock();
 529                         tipc_node_unlock(node);
 530                         if (likely(msg_mcast(msg)))
 531                                 tipc_sk_mcast_rcv(buf);
 532                         else
 533                                 kfree_skb(buf);
 534                 } else if (msg_user(msg) == MSG_BUNDLER) {
 535                         tipc_bclink_lock();
 536                         bclink_accept_pkt(node, seqno);
 537                         bcl->stats.recv_bundles++;
 538                         bcl->stats.recv_bundled += msg_msgcnt(msg);
 539                         tipc_bclink_unlock();
 540                         tipc_node_unlock(node);
 541                         tipc_link_bundle_rcv(buf);
 542                 } else if (msg_user(msg) == MSG_FRAGMENTER) {
 543                         tipc_buf_append(&node->bclink.reasm_buf, &buf);
 544                         if (unlikely(!buf && !node->bclink.reasm_buf))
 545                                 goto unlock;
 546                         tipc_bclink_lock();
 547                         bclink_accept_pkt(node, seqno);
 548                         bcl->stats.recv_fragments++;
 549                         if (buf) {
 550                                 bcl->stats.recv_fragmented++;
 551                                 msg = buf_msg(buf);
 552                                 tipc_bclink_unlock();
 553                                 goto receive;
 554                         }
 555                         tipc_bclink_unlock();
 556                         tipc_node_unlock(node);
 557                 } else if (msg_user(msg) == NAME_DISTRIBUTOR) {
 558                         tipc_bclink_lock();
 559                         bclink_accept_pkt(node, seqno);
 560                         tipc_bclink_unlock();
 561                         tipc_node_unlock(node);
 562                         tipc_named_rcv(buf);
 563                 } else {
 564                         tipc_bclink_lock();
 565                         bclink_accept_pkt(node, seqno);
 566                         tipc_bclink_unlock();
 567                         tipc_node_unlock(node);
 568                         kfree_skb(buf);
 569                 }
 570                 buf = NULL;
 571
 572                 /* Determine new synchronization state */
 573                 tipc_node_lock(node);
 574                 if (unlikely(!tipc_node_is_up(node)))
 575                         goto unlock;
 576
 577                 if (node->bclink.last_in == node->bclink.last_sent)
 578                         goto unlock;
 579
 580                 if (skb_queue_empty(&node->bclink.deferred_queue)) {
 581                         node->bclink.oos_state = 1;
 582                         goto unlock;
 583                 }
 584
 585                 msg = buf_msg(skb_peek(&node->bclink.deferred_queue));
 586                 seqno = msg_seqno(msg);
 587                 next_in = mod(next_in + 1);
 588                 if (seqno != next_in)
 589                         goto unlock;
 590
 591                 /* Take in-sequence message from deferred queue & deliver it */
 592                 buf = __skb_dequeue(&node->bclink.deferred_queue);
 593                 goto receive;
 594         }
 595
 596         /* Handle out-of-sequence broadcast message */
 597         if (less(next_in, seqno)) {
 598                 deferred = tipc_link_defer_pkt(&node->bclink.deferred_queue,
 599                                                buf);
 600                 bclink_update_last_sent(node, seqno);
 601                 buf = NULL;
 602         }
 603
 604         tipc_bclink_lock();
 605
 606         if (deferred)
 607                 bcl->stats.deferred_recv++;
 608         else
 609                 bcl->stats.duplicates++;
 610
 611         tipc_bclink_unlock();
 612
 613 unlock:
 614         tipc_node_unlock(node);
 615 exit:
 616         kfree_skb(buf);
 617 }
 618
 619 u32 tipc_bclink_acks_missing(struct tipc_node *n_ptr)
 620 {
 621         return (n_ptr->bclink.recv_permitted &&
 622                 (tipc_bclink_get_last_sent() != n_ptr->bclink.acked));
 623 }
 624
 625
 626 /**
 627  * tipc_bcbearer_send - send a packet through the broadcast pseudo-bearer
 628  *
 629  * Send packet over as many bearers as necessary to reach all nodes
 630  * that have joined the broadcast link.
 631  *
 632  * Returns 0 (packet sent successfully) under all circumstances,
 633  * since the broadcast link's pseudo-bearer never blocks
 634  */
 635 static int tipc_bcbearer_send(struct sk_buff *buf, struct tipc_bearer *unused1,
 636                               struct tipc_media_addr *unused2)
 637 {
 638         int bp_index;
 639         struct tipc_msg *msg = buf_msg(buf);
 640
 641         /* Prepare broadcast link message for reliable transmission,
 642          * if first time trying to send it;
 643          * preparation is skipped for broadcast link protocol messages
 644          * since they are sent in an unreliable manner and don't need it
 645          */
 646         if (likely(!msg_non_seq(buf_msg(buf)))) {
 647                 bcbuf_set_acks(buf, bclink->bcast_nodes.count);
 648                 msg_set_non_seq(msg, 1);
 649                 msg_set_mc_netid(msg, tipc_net_id);
 650                 bcl->stats.sent_info++;
 651
 652                 if (WARN_ON(!bclink->bcast_nodes.count)) {
 653                         dump_stack();
 654                         return 0;
 655                 }
 656         }
 657
 658         /* Send buffer over bearers until all targets reached */
 659         bcbearer->remains = bclink->bcast_nodes;
 660
 661         for (bp_index = 0; bp_index < MAX_BEARERS; bp_index++) {
 662                 struct tipc_bearer *p = bcbearer->bpairs[bp_index].primary;
 663                 struct tipc_bearer *s = bcbearer->bpairs[bp_index].secondary;
 664                 struct tipc_bearer *bp[2] = {p, s};
 665                 struct tipc_bearer *b = bp[msg_link_selector(msg)];
 666                 struct sk_buff *tbuf;
 667
 668                 if (!p)
 669                         break; /* No more bearers to try */
 670                 if (!b)
 671                         b = p;
 672                 tipc_nmap_diff(&bcbearer->remains, &b->nodes,
 673                                &bcbearer->remains_new);
 674                 if (bcbearer->remains_new.count == bcbearer->remains.count)
 675                         continue; /* Nothing added by bearer pair */
 676
 677                 if (bp_index == 0) {
 678                         /* Use original buffer for first bearer */
 679                         tipc_bearer_send(b->identity, buf, &b->bcast_addr);
 680                 } else {
 681                         /* Avoid concurrent buffer access */
 682                         tbuf = pskb_copy_for_clone(buf, GFP_ATOMIC);
 683                         if (!tbuf)
 684                                 break;
 685                         tipc_bearer_send(b->identity, tbuf, &b->bcast_addr);
 686                         kfree_skb(tbuf); /* Bearer keeps a clone */
 687                 }
 688                 if (bcbearer->remains_new.count == 0)
 689                         break; /* All targets reached */
 690
 691                 bcbearer->remains = bcbearer->remains_new;
 692         }
 693
 694         return 0;
 695 }
 696
 697 /**
 698  * tipc_bcbearer_sort - create sets of bearer pairs used by broadcast bearer
 699  */
 700 void tipc_bcbearer_sort(struct tipc_node_map *nm_ptr, u32 node, bool action)
 701 {
 702         struct tipc_bcbearer_pair *bp_temp = bcbearer->bpairs_temp;
 703         struct tipc_bcbearer_pair *bp_curr;
 704         struct tipc_bearer *b;
 705         int b_index;
 706         int pri;
 707
 708         tipc_bclink_lock();
 709
 710         if (action)
 711                 tipc_nmap_add(nm_ptr, node);
 712         else
 713                 tipc_nmap_remove(nm_ptr, node);
 714
 715         /* Group bearers by priority (can assume max of two per priority) */
 716         memset(bp_temp, 0, sizeof(bcbearer->bpairs_temp));
 717
 718         rcu_read_lock();
 719         for (b_index = 0; b_index < MAX_BEARERS; b_index++) {
 720                 b = rcu_dereference_rtnl(bearer_list[b_index]);
 721                 if (!b || !b->nodes.count)
 722                         continue;
 723
 724                 if (!bp_temp[b->priority].primary)
 725                         bp_temp[b->priority].primary = b;
 726                 else
 727                         bp_temp[b->priority].secondary = b;
 728         }
 729         rcu_read_unlock();
 730
 731         /* Create array of bearer pairs for broadcasting */
 732         bp_curr = bcbearer->bpairs;
 733         memset(bcbearer->bpairs, 0, sizeof(bcbearer->bpairs));
 734
 735         for (pri = TIPC_MAX_LINK_PRI; pri >= 0; pri--) {
 736
 737                 if (!bp_temp[pri].primary)
 738                         continue;
 739
 740                 bp_curr->primary = bp_temp[pri].primary;
 741
 742                 if (bp_temp[pri].secondary) {
 743                         if (tipc_nmap_equal(&bp_temp[pri].primary->nodes,
 744                                             &bp_temp[pri].secondary->nodes)) {
 745                                 bp_curr->secondary = bp_temp[pri].secondary;
 746                         } else {
 747                                 bp_curr++;
 748                                 bp_curr->primary = bp_temp[pri].secondary;
 749                         }
 750                 }
 751
 752                 bp_curr++;
 753         }
 754
 755         tipc_bclink_unlock();
 756 }
 757
 758 static int __tipc_nl_add_bc_link_stat(struct sk_buff *skb,
 759                                       struct tipc_stats *stats)
 760 {
 761         int i;
 762         struct nlattr *nest;
 763
 764         struct nla_map {
 765                 __u32 key;
 766                 __u32 val;
 767         };
 768
 769         struct nla_map map[] = {
 770                 {TIPC_NLA_STATS_RX_INFO, stats->recv_info},
 771                 {TIPC_NLA_STATS_RX_FRAGMENTS, stats->recv_fragments},
 772                 {TIPC_NLA_STATS_RX_FRAGMENTED, stats->recv_fragmented},
 773                 {TIPC_NLA_STATS_RX_BUNDLES, stats->recv_bundles},
 774                 {TIPC_NLA_STATS_RX_BUNDLED, stats->recv_bundled},
 775                 {TIPC_NLA_STATS_TX_INFO, stats->sent_info},
 776                 {TIPC_NLA_STATS_TX_FRAGMENTS, stats->sent_fragments},
 777                 {TIPC_NLA_STATS_TX_FRAGMENTED, stats->sent_fragmented},
 778                 {TIPC_NLA_STATS_TX_BUNDLES, stats->sent_bundles},
 779                 {TIPC_NLA_STATS_TX_BUNDLED, stats->sent_bundled},
 780                 {TIPC_NLA_STATS_RX_NACKS, stats->recv_nacks},
 781                 {TIPC_NLA_STATS_RX_DEFERRED, stats->deferred_recv},
 782                 {TIPC_NLA_STATS_TX_NACKS, stats->sent_nacks},
 783                 {TIPC_NLA_STATS_TX_ACKS, stats->sent_acks},
 784                 {TIPC_NLA_STATS_RETRANSMITTED, stats->retransmitted},
 785                 {TIPC_NLA_STATS_DUPLICATES, stats->duplicates},
 786                 {TIPC_NLA_STATS_LINK_CONGS, stats->link_congs},
 787                 {TIPC_NLA_STATS_MAX_QUEUE, stats->max_queue_sz},
 788                 {TIPC_NLA_STATS_AVG_QUEUE, stats->queue_sz_counts ?
 789                         (stats->accu_queue_sz / stats->queue_sz_counts) : 0}
 790         };
 791
 792         nest = nla_nest_start(skb, TIPC_NLA_LINK_STATS);
 793         if (!nest)
 794                 return -EMSGSIZE;
 795
 796         for (i = 0; i <  ARRAY_SIZE(map); i++)
 797                 if (nla_put_u32(skb, map[i].key, map[i].val))
 798                         goto msg_full;
 799
 800         nla_nest_end(skb, nest);
 801
 802         return 0;
 803 msg_full:
 804         nla_nest_cancel(skb, nest);
 805
 806         return -EMSGSIZE;
 807 }
 808
 809 int tipc_nl_add_bc_link(struct tipc_nl_msg *msg)
 810 {
 811         int err;
 812         void *hdr;
 813         struct nlattr *attrs;
 814         struct nlattr *prop;
 815
 816         if (!bcl)
 817                 return 0;
 818
 819         tipc_bclink_lock();
 820
 821         hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_v2_family,
 822                           NLM_F_MULTI, TIPC_NL_LINK_GET);
 823         if (!hdr)
 824                 return -EMSGSIZE;
 825
 826         attrs = nla_nest_start(msg->skb, TIPC_NLA_LINK);
 827         if (!attrs)
 828                 goto msg_full;
 829
 830         /* The broadcast link is always up */
 831         if (nla_put_flag(msg->skb, TIPC_NLA_LINK_UP))
 832                 goto attr_msg_full;
 833
 834         if (nla_put_flag(msg->skb, TIPC_NLA_LINK_BROADCAST))
 835                 goto attr_msg_full;
 836         if (nla_put_string(msg->skb, TIPC_NLA_LINK_NAME, bcl->name))
 837                 goto attr_msg_full;
 838         if (nla_put_u32(msg->skb, TIPC_NLA_LINK_RX, bcl->next_in_no))
 839                 goto attr_msg_full;
 840         if (nla_put_u32(msg->skb, TIPC_NLA_LINK_TX, bcl->next_out_no))
 841                 goto attr_msg_full;
 842
 843         prop = nla_nest_start(msg->skb, TIPC_NLA_LINK_PROP);
 844         if (!prop)
 845                 goto attr_msg_full;
 846         if (nla_put_u32(msg->skb, TIPC_NLA_PROP_WIN, bcl->queue_limit[0]))
 847                 goto prop_msg_full;
 848         nla_nest_end(msg->skb, prop);
 849
 850         err = __tipc_nl_add_bc_link_stat(msg->skb, &bcl->stats);
 851         if (err)
 852                 goto attr_msg_full;
 853
 854         tipc_bclink_unlock();
 855         nla_nest_end(msg->skb, attrs);
 856         genlmsg_end(msg->skb, hdr);
 857
 858         return 0;
 859
 860 prop_msg_full:
 861         nla_nest_cancel(msg->skb, prop);
 862 attr_msg_full:
 863         nla_nest_cancel(msg->skb, attrs);
 864 msg_full:
 865         tipc_bclink_unlock();
 866         genlmsg_cancel(msg->skb, hdr);
 867
 868         return -EMSGSIZE;
 869 }
 870
 871 int tipc_bclink_stats(char *buf, const u32 buf_size)
 872 {
 873         int ret;
 874         struct tipc_stats *s;
 875
 876         if (!bcl)
 877                 return 0;
 878
 879         tipc_bclink_lock();
 880
 881         s = &bcl->stats;
 882
 883         ret = tipc_snprintf(buf, buf_size, "Link <%s>\n"
 884                             "  Window:%u packets\n",
 885                             bcl->name, bcl->queue_limit[0]);
 886         ret += tipc_snprintf(buf + ret, buf_size - ret,
 887                              "  RX packets:%u fragments:%u/%u bundles:%u/%u\n",
 888                              s->recv_info, s->recv_fragments,
 889                              s->recv_fragmented, s->recv_bundles,
 890                              s->recv_bundled);
 891         ret += tipc_snprintf(buf + ret, buf_size - ret,
 892                              "  TX packets:%u fragments:%u/%u bundles:%u/%u\n",
 893                              s->sent_info, s->sent_fragments,
 894                              s->sent_fragmented, s->sent_bundles,
 895                              s->sent_bundled);
 896         ret += tipc_snprintf(buf + ret, buf_size - ret,
 897                              "  RX naks:%u defs:%u dups:%u\n",
 898                              s->recv_nacks, s->deferred_recv, s->duplicates);
 899         ret += tipc_snprintf(buf + ret, buf_size - ret,
 900                              "  TX naks:%u acks:%u dups:%u\n",
 901                              s->sent_nacks, s->sent_acks, s->retransmitted);
 902         ret += tipc_snprintf(buf + ret, buf_size - ret,
 903                              "  Congestion link:%u  Send queue max:%u avg:%u\n",
 904                              s->link_congs, s->max_queue_sz,
 905                              s->queue_sz_counts ?
 906                              (s->accu_queue_sz / s->queue_sz_counts) : 0);
 907
 908         tipc_bclink_unlock();
 909         return ret;
 910 }
 911
 912 int tipc_bclink_reset_stats(void)
 913 {
 914         if (!bcl)
 915                 return -ENOPROTOOPT;
 916
 917         tipc_bclink_lock();
 918         memset(&bcl->stats, 0, sizeof(bcl->stats));
 919         tipc_bclink_unlock();
 920         return 0;
 921 }
 922
 923 int tipc_bclink_set_queue_limits(u32 limit)
 924 {
 925         if (!bcl)
 926                 return -ENOPROTOOPT;
 927         if ((limit < TIPC_MIN_LINK_WIN) || (limit > TIPC_MAX_LINK_WIN))
 928                 return -EINVAL;
 929
 930         tipc_bclink_lock();
 931         tipc_link_set_queue_limits(bcl, limit);
 932         tipc_bclink_unlock();
 933         return 0;
 934 }
 935
 936 int tipc_bclink_init(void)
 937 {
 938         bcbearer = kzalloc(sizeof(*bcbearer), GFP_ATOMIC);
 939         if (!bcbearer)
 940                 return -ENOMEM;
 941
 942         bclink = kzalloc(sizeof(*bclink), GFP_ATOMIC);
 943         if (!bclink) {
 944                 kfree(bcbearer);
 945                 return -ENOMEM;
 946         }
 947
 948         bcl = &bclink->link;
 949         bcbearer->bearer.media = &bcbearer->media;
 950         bcbearer->media.send_msg = tipc_bcbearer_send;
 951         sprintf(bcbearer->media.name, "tipc-broadcast");
 952
 953         spin_lock_init(&bclink->lock);
 954         __skb_queue_head_init(&bcl->outqueue);
 955         __skb_queue_head_init(&bcl->deferred_queue);
 956         skb_queue_head_init(&bcl->waiting_sks);
 957         bcl->next_out_no = 1;
 958         spin_lock_init(&bclink->node.lock);
 959         __skb_queue_head_init(&bclink->node.waiting_sks);
 960         bcl->owner = &bclink->node;
 961         bcl->max_pkt = MAX_PKT_DEFAULT_MCAST;
 962         tipc_link_set_queue_limits(bcl, BCLINK_WIN_DEFAULT);
 963         bcl->bearer_id = MAX_BEARERS;
 964         rcu_assign_pointer(bearer_list[MAX_BEARERS], &bcbearer->bearer);
 965         bcl->state = WORKING_WORKING;
 966         strlcpy(bcl->name, tipc_bclink_name, TIPC_MAX_LINK_NAME);
 967         return 0;
 968 }
 969
 970 void tipc_bclink_stop(void)
 971 {
 972         tipc_bclink_lock();
 973         tipc_link_purge_queues(bcl);
 974         tipc_bclink_unlock();
 975
 976         RCU_INIT_POINTER(bearer_list[BCBEARER], NULL);
 977         synchronize_net();
 978         kfree(bcbearer);
 979         kfree(bclink);
 980 }
 981
 982 /**
 983  * tipc_nmap_add - add a node to a node map
 984  */
 985 static void tipc_nmap_add(struct tipc_node_map *nm_ptr, u32 node)
 986 {
 987         int n = tipc_node(node);
 988         int w = n / WSIZE;
 989         u32 mask = (1 << (n % WSIZE));
 990
 991         if ((nm_ptr->map[w] & mask) == 0) {
 992                 nm_ptr->count++;
 993                 nm_ptr->map[w] |= mask;
 994         }
 995 }
 996
 997 /**
 998  * tipc_nmap_remove - remove a node from a node map
 999  */
1000 static void tipc_nmap_remove(struct tipc_node_map *nm_ptr, u32 node)
1001 {
1002         int n = tipc_node(node);
1003         int w = n / WSIZE;
1004         u32 mask = (1 << (n % WSIZE));
1005
1006         if ((nm_ptr->map[w] & mask) != 0) {
1007                 nm_ptr->map[w] &= ~mask;
1008                 nm_ptr->count--;
1009         }
1010 }
1011
1012 /**
1013  * tipc_nmap_diff - find differences between node maps
1014  * @nm_a: input node map A
1015  * @nm_b: input node map B
1016  * @nm_diff: output node map A-B (i.e. nodes of A that are not in B)
1017  */
1018 static void tipc_nmap_diff(struct tipc_node_map *nm_a,
1019                            struct tipc_node_map *nm_b,
1020                            struct tipc_node_map *nm_diff)
1021 {
1022         int stop = ARRAY_SIZE(nm_a->map);
1023         int w;
1024         int b;
1025         u32 map;
1026
1027         memset(nm_diff, 0, sizeof(*nm_diff));
1028         for (w = 0; w < stop; w++) {
1029                 map = nm_a->map[w] ^ (nm_a->map[w] & nm_b->map[w]);
1030                 nm_diff->map[w] = map;
1031                 if (map != 0) {
1032                         for (b = 0 ; b < WSIZE; b++) {
1033                                 if (map & (1 << b))
1034                                         nm_diff->count++;
1035                         }
1036                 }
1037         }
1038 }
1039
1040 /**
1041  * tipc_port_list_add - add a port to a port list, ensuring no duplicates
1042  */
1043 void tipc_port_list_add(struct tipc_port_list *pl_ptr, u32 port)
1044 {
1045         struct tipc_port_list *item = pl_ptr;
1046         int i;
1047         int item_sz = PLSIZE;
1048         int cnt = pl_ptr->count;
1049
1050         for (; ; cnt -= item_sz, item = item->next) {
1051                 if (cnt < PLSIZE)
1052                         item_sz = cnt;
1053                 for (i = 0; i < item_sz; i++)
1054                         if (item->ports[i] == port)
1055                                 return;
1056                 if (i < PLSIZE) {
1057                         item->ports[i] = port;
1058                         pl_ptr->count++;
1059                         return;
1060                 }
1061                 if (!item->next) {
1062                         item->next = kmalloc(sizeof(*item), GFP_ATOMIC);
1063                         if (!item->next) {
1064                                 pr_warn("Incomplete multicast delivery, no memory\n");
1065                                 return;
1066                         }
1067                         item->next->next = NULL;
1068                 }
1069         }
1070 }
1071
1072 /**
1073  * tipc_port_list_free - free dynamically created entries in port_list chain
1074  *
1075  */
1076 void tipc_port_list_free(struct tipc_port_list *pl_ptr)
1077 {
1078         struct tipc_port_list *item;
1079         struct tipc_port_list *next;
1080
1081         for (item = pl_ptr->next; item; item = next) {
1082                 next = item->next;
1083                 kfree(item);
1084         }
1085 }