1 /* 2 * net/tipc/socket.c: TIPC socket API 3 * 4 * Copyright (c) 2001-2007, 2012-2017, Ericsson AB 5 * Copyright (c) 2004-2008, 2010-2013, Wind River Systems 6 * All rights reserved. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions are met: 10 * 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the names of the copyright holders nor the names of its 17 * contributors may be used to endorse or promote products derived from 18 * this software without specific prior written permission. 19 * 20 * Alternatively, this software may be distributed under the terms of the 21 * GNU General Public License ("GPL") version 2 as published by the Free 22 * Software Foundation. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" 25 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE 28 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 29 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 30 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 31 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 32 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 33 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 34 * POSSIBILITY OF SUCH DAMAGE. 35 */ 36 37 #include <linux/rhashtable.h> 38 #include <linux/sched/signal.h> 39 40 #include "core.h" 41 #include "name_table.h" 42 #include "node.h" 43 #include "link.h" 44 #include "name_distr.h" 45 #include "socket.h" 46 #include "bcast.h" 47 #include "netlink.h" 48 #include "group.h" 49 #include "trace.h" 50 51 #define NAGLE_START_INIT 4 52 #define NAGLE_START_MAX 1024 53 #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */ 54 #define CONN_PROBING_INTV msecs_to_jiffies(3600000) /* [ms] => 1 h */ 55 #define TIPC_MAX_PORT 0xffffffff 56 #define TIPC_MIN_PORT 1 57 #define TIPC_ACK_RATE 4 /* ACK at 1/4 of rcv window size */ 58 59 enum { 60 TIPC_LISTEN = TCP_LISTEN, 61 TIPC_ESTABLISHED = TCP_ESTABLISHED, 62 TIPC_OPEN = TCP_CLOSE, 63 TIPC_DISCONNECTING = TCP_CLOSE_WAIT, 64 TIPC_CONNECTING = TCP_SYN_SENT, 65 }; 66 67 struct sockaddr_pair { 68 struct sockaddr_tipc sock; 69 struct sockaddr_tipc member; 70 }; 71 72 /** 73 * struct tipc_sock - TIPC socket structure 74 * @sk: socket - interacts with 'port' and with user via the socket API 75 * @conn_type: TIPC type used when connection was established 76 * @conn_instance: TIPC instance used when connection was established 77 * @published: non-zero if port has one or more associated names 78 * @max_pkt: maximum packet size "hint" used when building messages sent by port 79 * @maxnagle: maximum size of msg which can be subject to nagle 80 * @portid: unique port identity in TIPC socket hash table 81 * @phdr: preformatted message header used when sending messages 82 * #cong_links: list of congested links 83 * @publications: list of publications for port 84 * @blocking_link: address of the congested link we are currently sleeping on 85 * @pub_count: total # of publications port has made during its lifetime 86 * @conn_timeout: the time we can wait for an unresponded setup request 87 * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue 88 * @cong_link_cnt: number of congested links 89 * @snt_unacked: # messages sent by socket, and not yet acked by peer 90 * @rcv_unacked: # messages read by user, but not yet acked back to peer 91 * @peer: 'connected' peer for dgram/rdm 92 * @node: hash table node 93 * @mc_method: cookie for use between socket and broadcast layer 94 * @rcu: rcu struct for tipc_sock 95 */ 96 struct tipc_sock { 97 struct sock sk; 98 u32 conn_type; 99 u32 conn_instance; 100 int published; 101 u32 max_pkt; 102 u32 maxnagle; 103 u32 portid; 104 struct tipc_msg phdr; 105 struct list_head cong_links; 106 struct list_head publications; 107 u32 pub_count; 108 atomic_t dupl_rcvcnt; 109 u16 conn_timeout; 110 bool probe_unacked; 111 u16 cong_link_cnt; 112 u16 snt_unacked; 113 u16 snd_win; 114 u16 peer_caps; 115 u16 rcv_unacked; 116 u16 rcv_win; 117 struct sockaddr_tipc peer; 118 struct rhash_head node; 119 struct tipc_mc_method mc_method; 120 struct rcu_head rcu; 121 struct tipc_group *group; 122 u32 oneway; 123 u32 nagle_start; 124 u16 snd_backlog; 125 u16 msg_acc; 126 u16 pkt_cnt; 127 bool expect_ack; 128 bool nodelay; 129 bool group_is_open; 130 }; 131 132 static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb); 133 static void tipc_data_ready(struct sock *sk); 134 static void tipc_write_space(struct sock *sk); 135 static void tipc_sock_destruct(struct sock *sk); 136 static int tipc_release(struct socket *sock); 137 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags, 138 bool kern); 139 static void tipc_sk_timeout(struct timer_list *t); 140 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope, 141 struct tipc_name_seq const *seq); 142 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope, 143 struct tipc_name_seq const *seq); 144 static int tipc_sk_leave(struct tipc_sock *tsk); 145 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid); 146 static int tipc_sk_insert(struct tipc_sock *tsk); 147 static void tipc_sk_remove(struct tipc_sock *tsk); 148 static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz); 149 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dsz); 150 static void tipc_sk_push_backlog(struct tipc_sock *tsk, bool nagle_ack); 151 152 static const struct proto_ops packet_ops; 153 static const struct proto_ops stream_ops; 154 static const struct proto_ops msg_ops; 155 static struct proto tipc_proto; 156 static const struct rhashtable_params tsk_rht_params; 157 158 static u32 tsk_own_node(struct tipc_sock *tsk) 159 { 160 return msg_prevnode(&tsk->phdr); 161 } 162 163 static u32 tsk_peer_node(struct tipc_sock *tsk) 164 { 165 return msg_destnode(&tsk->phdr); 166 } 167 168 static u32 tsk_peer_port(struct tipc_sock *tsk) 169 { 170 return msg_destport(&tsk->phdr); 171 } 172 173 static bool tsk_unreliable(struct tipc_sock *tsk) 174 { 175 return msg_src_droppable(&tsk->phdr) != 0; 176 } 177 178 static void tsk_set_unreliable(struct tipc_sock *tsk, bool unreliable) 179 { 180 msg_set_src_droppable(&tsk->phdr, unreliable ? 1 : 0); 181 } 182 183 static bool tsk_unreturnable(struct tipc_sock *tsk) 184 { 185 return msg_dest_droppable(&tsk->phdr) != 0; 186 } 187 188 static void tsk_set_unreturnable(struct tipc_sock *tsk, bool unreturnable) 189 { 190 msg_set_dest_droppable(&tsk->phdr, unreturnable ? 1 : 0); 191 } 192 193 static int tsk_importance(struct tipc_sock *tsk) 194 { 195 return msg_importance(&tsk->phdr); 196 } 197 198 static struct tipc_sock *tipc_sk(const struct sock *sk) 199 { 200 return container_of(sk, struct tipc_sock, sk); 201 } 202 203 int tsk_set_importance(struct sock *sk, int imp) 204 { 205 if (imp > TIPC_CRITICAL_IMPORTANCE) 206 return -EINVAL; 207 msg_set_importance(&tipc_sk(sk)->phdr, (u32)imp); 208 return 0; 209 } 210 211 static bool tsk_conn_cong(struct tipc_sock *tsk) 212 { 213 return tsk->snt_unacked > tsk->snd_win; 214 } 215 216 static u16 tsk_blocks(int len) 217 { 218 return ((len / FLOWCTL_BLK_SZ) + 1); 219 } 220 221 /* tsk_blocks(): translate a buffer size in bytes to number of 222 * advertisable blocks, taking into account the ratio truesize(len)/len 223 * We can trust that this ratio is always < 4 for len >= FLOWCTL_BLK_SZ 224 */ 225 static u16 tsk_adv_blocks(int len) 226 { 227 return len / FLOWCTL_BLK_SZ / 4; 228 } 229 230 /* tsk_inc(): increment counter for sent or received data 231 * - If block based flow control is not supported by peer we 232 * fall back to message based ditto, incrementing the counter 233 */ 234 static u16 tsk_inc(struct tipc_sock *tsk, int msglen) 235 { 236 if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL)) 237 return ((msglen / FLOWCTL_BLK_SZ) + 1); 238 return 1; 239 } 240 241 /* tsk_set_nagle - enable/disable nagle property by manipulating maxnagle 242 */ 243 static void tsk_set_nagle(struct tipc_sock *tsk) 244 { 245 struct sock *sk = &tsk->sk; 246 247 tsk->maxnagle = 0; 248 if (sk->sk_type != SOCK_STREAM) 249 return; 250 if (tsk->nodelay) 251 return; 252 if (!(tsk->peer_caps & TIPC_NAGLE)) 253 return; 254 /* Limit node local buffer size to avoid receive queue overflow */ 255 if (tsk->max_pkt == MAX_MSG_SIZE) 256 tsk->maxnagle = 1500; 257 else 258 tsk->maxnagle = tsk->max_pkt; 259 } 260 261 /** 262 * tsk_advance_rx_queue - discard first buffer in socket receive queue 263 * 264 * Caller must hold socket lock 265 */ 266 static void tsk_advance_rx_queue(struct sock *sk) 267 { 268 trace_tipc_sk_advance_rx(sk, NULL, TIPC_DUMP_SK_RCVQ, " "); 269 kfree_skb(__skb_dequeue(&sk->sk_receive_queue)); 270 } 271 272 /* tipc_sk_respond() : send response message back to sender 273 */ 274 static void tipc_sk_respond(struct sock *sk, struct sk_buff *skb, int err) 275 { 276 u32 selector; 277 u32 dnode; 278 u32 onode = tipc_own_addr(sock_net(sk)); 279 280 if (!tipc_msg_reverse(onode, &skb, err)) 281 return; 282 283 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE, "@sk_respond!"); 284 dnode = msg_destnode(buf_msg(skb)); 285 selector = msg_origport(buf_msg(skb)); 286 tipc_node_xmit_skb(sock_net(sk), skb, dnode, selector); 287 } 288 289 /** 290 * tsk_rej_rx_queue - reject all buffers in socket receive queue 291 * 292 * Caller must hold socket lock 293 */ 294 static void tsk_rej_rx_queue(struct sock *sk, int error) 295 { 296 struct sk_buff *skb; 297 298 while ((skb = __skb_dequeue(&sk->sk_receive_queue))) 299 tipc_sk_respond(sk, skb, error); 300 } 301 302 static bool tipc_sk_connected(struct sock *sk) 303 { 304 return sk->sk_state == TIPC_ESTABLISHED; 305 } 306 307 /* tipc_sk_type_connectionless - check if the socket is datagram socket 308 * @sk: socket 309 * 310 * Returns true if connection less, false otherwise 311 */ 312 static bool tipc_sk_type_connectionless(struct sock *sk) 313 { 314 return sk->sk_type == SOCK_RDM || sk->sk_type == SOCK_DGRAM; 315 } 316 317 /* tsk_peer_msg - verify if message was sent by connected port's peer 318 * 319 * Handles cases where the node's network address has changed from 320 * the default of <0.0.0> to its configured setting. 321 */ 322 static bool tsk_peer_msg(struct tipc_sock *tsk, struct tipc_msg *msg) 323 { 324 struct sock *sk = &tsk->sk; 325 u32 self = tipc_own_addr(sock_net(sk)); 326 u32 peer_port = tsk_peer_port(tsk); 327 u32 orig_node, peer_node; 328 329 if (unlikely(!tipc_sk_connected(sk))) 330 return false; 331 332 if (unlikely(msg_origport(msg) != peer_port)) 333 return false; 334 335 orig_node = msg_orignode(msg); 336 peer_node = tsk_peer_node(tsk); 337 338 if (likely(orig_node == peer_node)) 339 return true; 340 341 if (!orig_node && peer_node == self) 342 return true; 343 344 if (!peer_node && orig_node == self) 345 return true; 346 347 return false; 348 } 349 350 /* tipc_set_sk_state - set the sk_state of the socket 351 * @sk: socket 352 * 353 * Caller must hold socket lock 354 * 355 * Returns 0 on success, errno otherwise 356 */ 357 static int tipc_set_sk_state(struct sock *sk, int state) 358 { 359 int oldsk_state = sk->sk_state; 360 int res = -EINVAL; 361 362 switch (state) { 363 case TIPC_OPEN: 364 res = 0; 365 break; 366 case TIPC_LISTEN: 367 case TIPC_CONNECTING: 368 if (oldsk_state == TIPC_OPEN) 369 res = 0; 370 break; 371 case TIPC_ESTABLISHED: 372 if (oldsk_state == TIPC_CONNECTING || 373 oldsk_state == TIPC_OPEN) 374 res = 0; 375 break; 376 case TIPC_DISCONNECTING: 377 if (oldsk_state == TIPC_CONNECTING || 378 oldsk_state == TIPC_ESTABLISHED) 379 res = 0; 380 break; 381 } 382 383 if (!res) 384 sk->sk_state = state; 385 386 return res; 387 } 388 389 static int tipc_sk_sock_err(struct socket *sock, long *timeout) 390 { 391 struct sock *sk = sock->sk; 392 int err = sock_error(sk); 393 int typ = sock->type; 394 395 if (err) 396 return err; 397 if (typ == SOCK_STREAM || typ == SOCK_SEQPACKET) { 398 if (sk->sk_state == TIPC_DISCONNECTING) 399 return -EPIPE; 400 else if (!tipc_sk_connected(sk)) 401 return -ENOTCONN; 402 } 403 if (!*timeout) 404 return -EAGAIN; 405 if (signal_pending(current)) 406 return sock_intr_errno(*timeout); 407 408 return 0; 409 } 410 411 #define tipc_wait_for_cond(sock_, timeo_, condition_) \ 412 ({ \ 413 DEFINE_WAIT_FUNC(wait_, woken_wake_function); \ 414 struct sock *sk_; \ 415 int rc_; \ 416 \ 417 while ((rc_ = !(condition_))) { \ 418 /* coupled with smp_wmb() in tipc_sk_proto_rcv() */ \ 419 smp_rmb(); \ 420 sk_ = (sock_)->sk; \ 421 rc_ = tipc_sk_sock_err((sock_), timeo_); \ 422 if (rc_) \ 423 break; \ 424 add_wait_queue(sk_sleep(sk_), &wait_); \ 425 release_sock(sk_); \ 426 *(timeo_) = wait_woken(&wait_, TASK_INTERRUPTIBLE, *(timeo_)); \ 427 sched_annotate_sleep(); \ 428 lock_sock(sk_); \ 429 remove_wait_queue(sk_sleep(sk_), &wait_); \ 430 } \ 431 rc_; \ 432 }) 433 434 /** 435 * tipc_sk_create - create a TIPC socket 436 * @net: network namespace (must be default network) 437 * @sock: pre-allocated socket structure 438 * @protocol: protocol indicator (must be 0) 439 * @kern: caused by kernel or by userspace? 440 * 441 * This routine creates additional data structures used by the TIPC socket, 442 * initializes them, and links them together. 443 * 444 * Returns 0 on success, errno otherwise 445 */ 446 static int tipc_sk_create(struct net *net, struct socket *sock, 447 int protocol, int kern) 448 { 449 const struct proto_ops *ops; 450 struct sock *sk; 451 struct tipc_sock *tsk; 452 struct tipc_msg *msg; 453 454 /* Validate arguments */ 455 if (unlikely(protocol != 0)) 456 return -EPROTONOSUPPORT; 457 458 switch (sock->type) { 459 case SOCK_STREAM: 460 ops = &stream_ops; 461 break; 462 case SOCK_SEQPACKET: 463 ops = &packet_ops; 464 break; 465 case SOCK_DGRAM: 466 case SOCK_RDM: 467 ops = &msg_ops; 468 break; 469 default: 470 return -EPROTOTYPE; 471 } 472 473 /* Allocate socket's protocol area */ 474 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto, kern); 475 if (sk == NULL) 476 return -ENOMEM; 477 478 tsk = tipc_sk(sk); 479 tsk->max_pkt = MAX_PKT_DEFAULT; 480 tsk->maxnagle = 0; 481 tsk->nagle_start = NAGLE_START_INIT; 482 INIT_LIST_HEAD(&tsk->publications); 483 INIT_LIST_HEAD(&tsk->cong_links); 484 msg = &tsk->phdr; 485 486 /* Finish initializing socket data structures */ 487 sock->ops = ops; 488 sock_init_data(sock, sk); 489 tipc_set_sk_state(sk, TIPC_OPEN); 490 if (tipc_sk_insert(tsk)) { 491 pr_warn("Socket create failed; port number exhausted\n"); 492 return -EINVAL; 493 } 494 495 /* Ensure tsk is visible before we read own_addr. */ 496 smp_mb(); 497 498 tipc_msg_init(tipc_own_addr(net), msg, TIPC_LOW_IMPORTANCE, 499 TIPC_NAMED_MSG, NAMED_H_SIZE, 0); 500 501 msg_set_origport(msg, tsk->portid); 502 timer_setup(&sk->sk_timer, tipc_sk_timeout, 0); 503 sk->sk_shutdown = 0; 504 sk->sk_backlog_rcv = tipc_sk_backlog_rcv; 505 sk->sk_rcvbuf = sysctl_tipc_rmem[1]; 506 sk->sk_data_ready = tipc_data_ready; 507 sk->sk_write_space = tipc_write_space; 508 sk->sk_destruct = tipc_sock_destruct; 509 tsk->conn_timeout = CONN_TIMEOUT_DEFAULT; 510 tsk->group_is_open = true; 511 atomic_set(&tsk->dupl_rcvcnt, 0); 512 513 /* Start out with safe limits until we receive an advertised window */ 514 tsk->snd_win = tsk_adv_blocks(RCVBUF_MIN); 515 tsk->rcv_win = tsk->snd_win; 516 517 if (tipc_sk_type_connectionless(sk)) { 518 tsk_set_unreturnable(tsk, true); 519 if (sock->type == SOCK_DGRAM) 520 tsk_set_unreliable(tsk, true); 521 } 522 __skb_queue_head_init(&tsk->mc_method.deferredq); 523 trace_tipc_sk_create(sk, NULL, TIPC_DUMP_NONE, " "); 524 return 0; 525 } 526 527 static void tipc_sk_callback(struct rcu_head *head) 528 { 529 struct tipc_sock *tsk = container_of(head, struct tipc_sock, rcu); 530 531 sock_put(&tsk->sk); 532 } 533 534 /* Caller should hold socket lock for the socket. */ 535 static void __tipc_shutdown(struct socket *sock, int error) 536 { 537 struct sock *sk = sock->sk; 538 struct tipc_sock *tsk = tipc_sk(sk); 539 struct net *net = sock_net(sk); 540 long timeout = msecs_to_jiffies(CONN_TIMEOUT_DEFAULT); 541 u32 dnode = tsk_peer_node(tsk); 542 struct sk_buff *skb; 543 544 /* Avoid that hi-prio shutdown msgs bypass msgs in link wakeup queue */ 545 tipc_wait_for_cond(sock, &timeout, (!tsk->cong_link_cnt && 546 !tsk_conn_cong(tsk))); 547 548 /* Push out delayed messages if in Nagle mode */ 549 tipc_sk_push_backlog(tsk, false); 550 /* Remove pending SYN */ 551 __skb_queue_purge(&sk->sk_write_queue); 552 553 /* Remove partially received buffer if any */ 554 skb = skb_peek(&sk->sk_receive_queue); 555 if (skb && TIPC_SKB_CB(skb)->bytes_read) { 556 __skb_unlink(skb, &sk->sk_receive_queue); 557 kfree_skb(skb); 558 } 559 560 /* Reject all unreceived messages if connectionless */ 561 if (tipc_sk_type_connectionless(sk)) { 562 tsk_rej_rx_queue(sk, error); 563 return; 564 } 565 566 switch (sk->sk_state) { 567 case TIPC_CONNECTING: 568 case TIPC_ESTABLISHED: 569 tipc_set_sk_state(sk, TIPC_DISCONNECTING); 570 tipc_node_remove_conn(net, dnode, tsk->portid); 571 /* Send a FIN+/- to its peer */ 572 skb = __skb_dequeue(&sk->sk_receive_queue); 573 if (skb) { 574 __skb_queue_purge(&sk->sk_receive_queue); 575 tipc_sk_respond(sk, skb, error); 576 break; 577 } 578 skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE, 579 TIPC_CONN_MSG, SHORT_H_SIZE, 0, dnode, 580 tsk_own_node(tsk), tsk_peer_port(tsk), 581 tsk->portid, error); 582 if (skb) 583 tipc_node_xmit_skb(net, skb, dnode, tsk->portid); 584 break; 585 case TIPC_LISTEN: 586 /* Reject all SYN messages */ 587 tsk_rej_rx_queue(sk, error); 588 break; 589 default: 590 __skb_queue_purge(&sk->sk_receive_queue); 591 break; 592 } 593 } 594 595 /** 596 * tipc_release - destroy a TIPC socket 597 * @sock: socket to destroy 598 * 599 * This routine cleans up any messages that are still queued on the socket. 600 * For DGRAM and RDM socket types, all queued messages are rejected. 601 * For SEQPACKET and STREAM socket types, the first message is rejected 602 * and any others are discarded. (If the first message on a STREAM socket 603 * is partially-read, it is discarded and the next one is rejected instead.) 604 * 605 * NOTE: Rejected messages are not necessarily returned to the sender! They 606 * are returned or discarded according to the "destination droppable" setting 607 * specified for the message by the sender. 608 * 609 * Returns 0 on success, errno otherwise 610 */ 611 static int tipc_release(struct socket *sock) 612 { 613 struct sock *sk = sock->sk; 614 struct tipc_sock *tsk; 615 616 /* 617 * Exit if socket isn't fully initialized (occurs when a failed accept() 618 * releases a pre-allocated child socket that was never used) 619 */ 620 if (sk == NULL) 621 return 0; 622 623 tsk = tipc_sk(sk); 624 lock_sock(sk); 625 626 trace_tipc_sk_release(sk, NULL, TIPC_DUMP_ALL, " "); 627 __tipc_shutdown(sock, TIPC_ERR_NO_PORT); 628 sk->sk_shutdown = SHUTDOWN_MASK; 629 tipc_sk_leave(tsk); 630 tipc_sk_withdraw(tsk, 0, NULL); 631 __skb_queue_purge(&tsk->mc_method.deferredq); 632 sk_stop_timer(sk, &sk->sk_timer); 633 tipc_sk_remove(tsk); 634 635 sock_orphan(sk); 636 /* Reject any messages that accumulated in backlog queue */ 637 release_sock(sk); 638 tipc_dest_list_purge(&tsk->cong_links); 639 tsk->cong_link_cnt = 0; 640 call_rcu(&tsk->rcu, tipc_sk_callback); 641 sock->sk = NULL; 642 643 return 0; 644 } 645 646 /** 647 * tipc_bind - associate or disassocate TIPC name(s) with a socket 648 * @sock: socket structure 649 * @uaddr: socket address describing name(s) and desired operation 650 * @uaddr_len: size of socket address data structure 651 * 652 * Name and name sequence binding is indicated using a positive scope value; 653 * a negative scope value unbinds the specified name. Specifying no name 654 * (i.e. a socket address length of 0) unbinds all names from the socket. 655 * 656 * Returns 0 on success, errno otherwise 657 * 658 * NOTE: This routine doesn't need to take the socket lock since it doesn't 659 * access any non-constant socket information. 660 */ 661 static int tipc_bind(struct socket *sock, struct sockaddr *uaddr, 662 int uaddr_len) 663 { 664 struct sock *sk = sock->sk; 665 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr; 666 struct tipc_sock *tsk = tipc_sk(sk); 667 int res = -EINVAL; 668 669 lock_sock(sk); 670 if (unlikely(!uaddr_len)) { 671 res = tipc_sk_withdraw(tsk, 0, NULL); 672 goto exit; 673 } 674 if (tsk->group) { 675 res = -EACCES; 676 goto exit; 677 } 678 if (uaddr_len < sizeof(struct sockaddr_tipc)) { 679 res = -EINVAL; 680 goto exit; 681 } 682 if (addr->family != AF_TIPC) { 683 res = -EAFNOSUPPORT; 684 goto exit; 685 } 686 687 if (addr->addrtype == TIPC_ADDR_NAME) 688 addr->addr.nameseq.upper = addr->addr.nameseq.lower; 689 else if (addr->addrtype != TIPC_ADDR_NAMESEQ) { 690 res = -EAFNOSUPPORT; 691 goto exit; 692 } 693 694 if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) && 695 (addr->addr.nameseq.type != TIPC_TOP_SRV) && 696 (addr->addr.nameseq.type != TIPC_CFG_SRV)) { 697 res = -EACCES; 698 goto exit; 699 } 700 701 res = (addr->scope >= 0) ? 702 tipc_sk_publish(tsk, addr->scope, &addr->addr.nameseq) : 703 tipc_sk_withdraw(tsk, -addr->scope, &addr->addr.nameseq); 704 exit: 705 release_sock(sk); 706 return res; 707 } 708 709 /** 710 * tipc_getname - get port ID of socket or peer socket 711 * @sock: socket structure 712 * @uaddr: area for returned socket address 713 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID 714 * 715 * Returns 0 on success, errno otherwise 716 * 717 * NOTE: This routine doesn't need to take the socket lock since it only 718 * accesses socket information that is unchanging (or which changes in 719 * a completely predictable manner). 720 */ 721 static int tipc_getname(struct socket *sock, struct sockaddr *uaddr, 722 int peer) 723 { 724 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr; 725 struct sock *sk = sock->sk; 726 struct tipc_sock *tsk = tipc_sk(sk); 727 728 memset(addr, 0, sizeof(*addr)); 729 if (peer) { 730 if ((!tipc_sk_connected(sk)) && 731 ((peer != 2) || (sk->sk_state != TIPC_DISCONNECTING))) 732 return -ENOTCONN; 733 addr->addr.id.ref = tsk_peer_port(tsk); 734 addr->addr.id.node = tsk_peer_node(tsk); 735 } else { 736 addr->addr.id.ref = tsk->portid; 737 addr->addr.id.node = tipc_own_addr(sock_net(sk)); 738 } 739 740 addr->addrtype = TIPC_ADDR_ID; 741 addr->family = AF_TIPC; 742 addr->scope = 0; 743 addr->addr.name.domain = 0; 744 745 return sizeof(*addr); 746 } 747 748 /** 749 * tipc_poll - read and possibly block on pollmask 750 * @file: file structure associated with the socket 751 * @sock: socket for which to calculate the poll bits 752 * @wait: ??? 753 * 754 * Returns pollmask value 755 * 756 * COMMENTARY: 757 * It appears that the usual socket locking mechanisms are not useful here 758 * since the pollmask info is potentially out-of-date the moment this routine 759 * exits. TCP and other protocols seem to rely on higher level poll routines 760 * to handle any preventable race conditions, so TIPC will do the same ... 761 * 762 * IMPORTANT: The fact that a read or write operation is indicated does NOT 763 * imply that the operation will succeed, merely that it should be performed 764 * and will not block. 765 */ 766 static __poll_t tipc_poll(struct file *file, struct socket *sock, 767 poll_table *wait) 768 { 769 struct sock *sk = sock->sk; 770 struct tipc_sock *tsk = tipc_sk(sk); 771 __poll_t revents = 0; 772 773 sock_poll_wait(file, sock, wait); 774 trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " "); 775 776 if (sk->sk_shutdown & RCV_SHUTDOWN) 777 revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM; 778 if (sk->sk_shutdown == SHUTDOWN_MASK) 779 revents |= EPOLLHUP; 780 781 switch (sk->sk_state) { 782 case TIPC_ESTABLISHED: 783 if (!tsk->cong_link_cnt && !tsk_conn_cong(tsk)) 784 revents |= EPOLLOUT; 785 fallthrough; 786 case TIPC_LISTEN: 787 case TIPC_CONNECTING: 788 if (!skb_queue_empty_lockless(&sk->sk_receive_queue)) 789 revents |= EPOLLIN | EPOLLRDNORM; 790 break; 791 case TIPC_OPEN: 792 if (tsk->group_is_open && !tsk->cong_link_cnt) 793 revents |= EPOLLOUT; 794 if (!tipc_sk_type_connectionless(sk)) 795 break; 796 if (skb_queue_empty_lockless(&sk->sk_receive_queue)) 797 break; 798 revents |= EPOLLIN | EPOLLRDNORM; 799 break; 800 case TIPC_DISCONNECTING: 801 revents = EPOLLIN | EPOLLRDNORM | EPOLLHUP; 802 break; 803 } 804 return revents; 805 } 806 807 /** 808 * tipc_sendmcast - send multicast message 809 * @sock: socket structure 810 * @seq: destination address 811 * @msg: message to send 812 * @dlen: length of data to send 813 * @timeout: timeout to wait for wakeup 814 * 815 * Called from function tipc_sendmsg(), which has done all sanity checks 816 * Returns the number of bytes sent on success, or errno 817 */ 818 static int tipc_sendmcast(struct socket *sock, struct tipc_name_seq *seq, 819 struct msghdr *msg, size_t dlen, long timeout) 820 { 821 struct sock *sk = sock->sk; 822 struct tipc_sock *tsk = tipc_sk(sk); 823 struct tipc_msg *hdr = &tsk->phdr; 824 struct net *net = sock_net(sk); 825 int mtu = tipc_bcast_get_mtu(net); 826 struct tipc_mc_method *method = &tsk->mc_method; 827 struct sk_buff_head pkts; 828 struct tipc_nlist dsts; 829 int rc; 830 831 if (tsk->group) 832 return -EACCES; 833 834 /* Block or return if any destination link is congested */ 835 rc = tipc_wait_for_cond(sock, &timeout, !tsk->cong_link_cnt); 836 if (unlikely(rc)) 837 return rc; 838 839 /* Lookup destination nodes */ 840 tipc_nlist_init(&dsts, tipc_own_addr(net)); 841 tipc_nametbl_lookup_dst_nodes(net, seq->type, seq->lower, 842 seq->upper, &dsts); 843 if (!dsts.local && !dsts.remote) 844 return -EHOSTUNREACH; 845 846 /* Build message header */ 847 msg_set_type(hdr, TIPC_MCAST_MSG); 848 msg_set_hdr_sz(hdr, MCAST_H_SIZE); 849 msg_set_lookup_scope(hdr, TIPC_CLUSTER_SCOPE); 850 msg_set_destport(hdr, 0); 851 msg_set_destnode(hdr, 0); 852 msg_set_nametype(hdr, seq->type); 853 msg_set_namelower(hdr, seq->lower); 854 msg_set_nameupper(hdr, seq->upper); 855 856 /* Build message as chain of buffers */ 857 __skb_queue_head_init(&pkts); 858 rc = tipc_msg_build(hdr, msg, 0, dlen, mtu, &pkts); 859 860 /* Send message if build was successful */ 861 if (unlikely(rc == dlen)) { 862 trace_tipc_sk_sendmcast(sk, skb_peek(&pkts), 863 TIPC_DUMP_SK_SNDQ, " "); 864 rc = tipc_mcast_xmit(net, &pkts, method, &dsts, 865 &tsk->cong_link_cnt); 866 } 867 868 tipc_nlist_purge(&dsts); 869 870 return rc ? rc : dlen; 871 } 872 873 /** 874 * tipc_send_group_msg - send a message to a member in the group 875 * @net: network namespace 876 * @m: message to send 877 * @mb: group member 878 * @dnode: destination node 879 * @dport: destination port 880 * @dlen: total length of message data 881 */ 882 static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk, 883 struct msghdr *m, struct tipc_member *mb, 884 u32 dnode, u32 dport, int dlen) 885 { 886 u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group); 887 struct tipc_mc_method *method = &tsk->mc_method; 888 int blks = tsk_blocks(GROUP_H_SIZE + dlen); 889 struct tipc_msg *hdr = &tsk->phdr; 890 struct sk_buff_head pkts; 891 int mtu, rc; 892 893 /* Complete message header */ 894 msg_set_type(hdr, TIPC_GRP_UCAST_MSG); 895 msg_set_hdr_sz(hdr, GROUP_H_SIZE); 896 msg_set_destport(hdr, dport); 897 msg_set_destnode(hdr, dnode); 898 msg_set_grp_bc_seqno(hdr, bc_snd_nxt); 899 900 /* Build message as chain of buffers */ 901 __skb_queue_head_init(&pkts); 902 mtu = tipc_node_get_mtu(net, dnode, tsk->portid, false); 903 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts); 904 if (unlikely(rc != dlen)) 905 return rc; 906 907 /* Send message */ 908 rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid); 909 if (unlikely(rc == -ELINKCONG)) { 910 tipc_dest_push(&tsk->cong_links, dnode, 0); 911 tsk->cong_link_cnt++; 912 } 913 914 /* Update send window */ 915 tipc_group_update_member(mb, blks); 916 917 /* A broadcast sent within next EXPIRE period must follow same path */ 918 method->rcast = true; 919 method->mandatory = true; 920 return dlen; 921 } 922 923 /** 924 * tipc_send_group_unicast - send message to a member in the group 925 * @sock: socket structure 926 * @m: message to send 927 * @dlen: total length of message data 928 * @timeout: timeout to wait for wakeup 929 * 930 * Called from function tipc_sendmsg(), which has done all sanity checks 931 * Returns the number of bytes sent on success, or errno 932 */ 933 static int tipc_send_group_unicast(struct socket *sock, struct msghdr *m, 934 int dlen, long timeout) 935 { 936 struct sock *sk = sock->sk; 937 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name); 938 int blks = tsk_blocks(GROUP_H_SIZE + dlen); 939 struct tipc_sock *tsk = tipc_sk(sk); 940 struct net *net = sock_net(sk); 941 struct tipc_member *mb = NULL; 942 u32 node, port; 943 int rc; 944 945 node = dest->addr.id.node; 946 port = dest->addr.id.ref; 947 if (!port && !node) 948 return -EHOSTUNREACH; 949 950 /* Block or return if destination link or member is congested */ 951 rc = tipc_wait_for_cond(sock, &timeout, 952 !tipc_dest_find(&tsk->cong_links, node, 0) && 953 tsk->group && 954 !tipc_group_cong(tsk->group, node, port, blks, 955 &mb)); 956 if (unlikely(rc)) 957 return rc; 958 959 if (unlikely(!mb)) 960 return -EHOSTUNREACH; 961 962 rc = tipc_send_group_msg(net, tsk, m, mb, node, port, dlen); 963 964 return rc ? rc : dlen; 965 } 966 967 /** 968 * tipc_send_group_anycast - send message to any member with given identity 969 * @sock: socket structure 970 * @m: message to send 971 * @dlen: total length of message data 972 * @timeout: timeout to wait for wakeup 973 * 974 * Called from function tipc_sendmsg(), which has done all sanity checks 975 * Returns the number of bytes sent on success, or errno 976 */ 977 static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m, 978 int dlen, long timeout) 979 { 980 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name); 981 struct sock *sk = sock->sk; 982 struct tipc_sock *tsk = tipc_sk(sk); 983 struct list_head *cong_links = &tsk->cong_links; 984 int blks = tsk_blocks(GROUP_H_SIZE + dlen); 985 struct tipc_msg *hdr = &tsk->phdr; 986 struct tipc_member *first = NULL; 987 struct tipc_member *mbr = NULL; 988 struct net *net = sock_net(sk); 989 u32 node, port, exclude; 990 struct list_head dsts; 991 u32 type, inst, scope; 992 int lookups = 0; 993 int dstcnt, rc; 994 bool cong; 995 996 INIT_LIST_HEAD(&dsts); 997 998 type = msg_nametype(hdr); 999 inst = dest->addr.name.name.instance; 1000 scope = msg_lookup_scope(hdr); 1001 1002 while (++lookups < 4) { 1003 exclude = tipc_group_exclude(tsk->group); 1004 1005 first = NULL; 1006 1007 /* Look for a non-congested destination member, if any */ 1008 while (1) { 1009 if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts, 1010 &dstcnt, exclude, false)) 1011 return -EHOSTUNREACH; 1012 tipc_dest_pop(&dsts, &node, &port); 1013 cong = tipc_group_cong(tsk->group, node, port, blks, 1014 &mbr); 1015 if (!cong) 1016 break; 1017 if (mbr == first) 1018 break; 1019 if (!first) 1020 first = mbr; 1021 } 1022 1023 /* Start over if destination was not in member list */ 1024 if (unlikely(!mbr)) 1025 continue; 1026 1027 if (likely(!cong && !tipc_dest_find(cong_links, node, 0))) 1028 break; 1029 1030 /* Block or return if destination link or member is congested */ 1031 rc = tipc_wait_for_cond(sock, &timeout, 1032 !tipc_dest_find(cong_links, node, 0) && 1033 tsk->group && 1034 !tipc_group_cong(tsk->group, node, port, 1035 blks, &mbr)); 1036 if (unlikely(rc)) 1037 return rc; 1038 1039 /* Send, unless destination disappeared while waiting */ 1040 if (likely(mbr)) 1041 break; 1042 } 1043 1044 if (unlikely(lookups >= 4)) 1045 return -EHOSTUNREACH; 1046 1047 rc = tipc_send_group_msg(net, tsk, m, mbr, node, port, dlen); 1048 1049 return rc ? rc : dlen; 1050 } 1051 1052 /** 1053 * tipc_send_group_bcast - send message to all members in communication group 1054 * @sock: socket structure 1055 * @m: message to send 1056 * @dlen: total length of message data 1057 * @timeout: timeout to wait for wakeup 1058 * 1059 * Called from function tipc_sendmsg(), which has done all sanity checks 1060 * Returns the number of bytes sent on success, or errno 1061 */ 1062 static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m, 1063 int dlen, long timeout) 1064 { 1065 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name); 1066 struct sock *sk = sock->sk; 1067 struct net *net = sock_net(sk); 1068 struct tipc_sock *tsk = tipc_sk(sk); 1069 struct tipc_nlist *dsts; 1070 struct tipc_mc_method *method = &tsk->mc_method; 1071 bool ack = method->mandatory && method->rcast; 1072 int blks = tsk_blocks(MCAST_H_SIZE + dlen); 1073 struct tipc_msg *hdr = &tsk->phdr; 1074 int mtu = tipc_bcast_get_mtu(net); 1075 struct sk_buff_head pkts; 1076 int rc = -EHOSTUNREACH; 1077 1078 /* Block or return if any destination link or member is congested */ 1079 rc = tipc_wait_for_cond(sock, &timeout, 1080 !tsk->cong_link_cnt && tsk->group && 1081 !tipc_group_bc_cong(tsk->group, blks)); 1082 if (unlikely(rc)) 1083 return rc; 1084 1085 dsts = tipc_group_dests(tsk->group); 1086 if (!dsts->local && !dsts->remote) 1087 return -EHOSTUNREACH; 1088 1089 /* Complete message header */ 1090 if (dest) { 1091 msg_set_type(hdr, TIPC_GRP_MCAST_MSG); 1092 msg_set_nameinst(hdr, dest->addr.name.name.instance); 1093 } else { 1094 msg_set_type(hdr, TIPC_GRP_BCAST_MSG); 1095 msg_set_nameinst(hdr, 0); 1096 } 1097 msg_set_hdr_sz(hdr, GROUP_H_SIZE); 1098 msg_set_destport(hdr, 0); 1099 msg_set_destnode(hdr, 0); 1100 msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(tsk->group)); 1101 1102 /* Avoid getting stuck with repeated forced replicasts */ 1103 msg_set_grp_bc_ack_req(hdr, ack); 1104 1105 /* Build message as chain of buffers */ 1106 __skb_queue_head_init(&pkts); 1107 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts); 1108 if (unlikely(rc != dlen)) 1109 return rc; 1110 1111 /* Send message */ 1112 rc = tipc_mcast_xmit(net, &pkts, method, dsts, &tsk->cong_link_cnt); 1113 if (unlikely(rc)) 1114 return rc; 1115 1116 /* Update broadcast sequence number and send windows */ 1117 tipc_group_update_bc_members(tsk->group, blks, ack); 1118 1119 /* Broadcast link is now free to choose method for next broadcast */ 1120 method->mandatory = false; 1121 method->expires = jiffies; 1122 1123 return dlen; 1124 } 1125 1126 /** 1127 * tipc_send_group_mcast - send message to all members with given identity 1128 * @sock: socket structure 1129 * @m: message to send 1130 * @dlen: total length of message data 1131 * @timeout: timeout to wait for wakeup 1132 * 1133 * Called from function tipc_sendmsg(), which has done all sanity checks 1134 * Returns the number of bytes sent on success, or errno 1135 */ 1136 static int tipc_send_group_mcast(struct socket *sock, struct msghdr *m, 1137 int dlen, long timeout) 1138 { 1139 struct sock *sk = sock->sk; 1140 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name); 1141 struct tipc_sock *tsk = tipc_sk(sk); 1142 struct tipc_group *grp = tsk->group; 1143 struct tipc_msg *hdr = &tsk->phdr; 1144 struct net *net = sock_net(sk); 1145 u32 type, inst, scope, exclude; 1146 struct list_head dsts; 1147 u32 dstcnt; 1148 1149 INIT_LIST_HEAD(&dsts); 1150 1151 type = msg_nametype(hdr); 1152 inst = dest->addr.name.name.instance; 1153 scope = msg_lookup_scope(hdr); 1154 exclude = tipc_group_exclude(grp); 1155 1156 if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts, 1157 &dstcnt, exclude, true)) 1158 return -EHOSTUNREACH; 1159 1160 if (dstcnt == 1) { 1161 tipc_dest_pop(&dsts, &dest->addr.id.node, &dest->addr.id.ref); 1162 return tipc_send_group_unicast(sock, m, dlen, timeout); 1163 } 1164 1165 tipc_dest_list_purge(&dsts); 1166 return tipc_send_group_bcast(sock, m, dlen, timeout); 1167 } 1168 1169 /** 1170 * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets 1171 * @arrvq: queue with arriving messages, to be cloned after destination lookup 1172 * @inputq: queue with cloned messages, delivered to socket after dest lookup 1173 * 1174 * Multi-threaded: parallel calls with reference to same queues may occur 1175 */ 1176 void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq, 1177 struct sk_buff_head *inputq) 1178 { 1179 u32 self = tipc_own_addr(net); 1180 u32 type, lower, upper, scope; 1181 struct sk_buff *skb, *_skb; 1182 u32 portid, onode; 1183 struct sk_buff_head tmpq; 1184 struct list_head dports; 1185 struct tipc_msg *hdr; 1186 int user, mtyp, hlen; 1187 bool exact; 1188 1189 __skb_queue_head_init(&tmpq); 1190 INIT_LIST_HEAD(&dports); 1191 1192 skb = tipc_skb_peek(arrvq, &inputq->lock); 1193 for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) { 1194 hdr = buf_msg(skb); 1195 user = msg_user(hdr); 1196 mtyp = msg_type(hdr); 1197 hlen = skb_headroom(skb) + msg_hdr_sz(hdr); 1198 onode = msg_orignode(hdr); 1199 type = msg_nametype(hdr); 1200 1201 if (mtyp == TIPC_GRP_UCAST_MSG || user == GROUP_PROTOCOL) { 1202 spin_lock_bh(&inputq->lock); 1203 if (skb_peek(arrvq) == skb) { 1204 __skb_dequeue(arrvq); 1205 __skb_queue_tail(inputq, skb); 1206 } 1207 kfree_skb(skb); 1208 spin_unlock_bh(&inputq->lock); 1209 continue; 1210 } 1211 1212 /* Group messages require exact scope match */ 1213 if (msg_in_group(hdr)) { 1214 lower = 0; 1215 upper = ~0; 1216 scope = msg_lookup_scope(hdr); 1217 exact = true; 1218 } else { 1219 /* TIPC_NODE_SCOPE means "any scope" in this context */ 1220 if (onode == self) 1221 scope = TIPC_NODE_SCOPE; 1222 else 1223 scope = TIPC_CLUSTER_SCOPE; 1224 exact = false; 1225 lower = msg_namelower(hdr); 1226 upper = msg_nameupper(hdr); 1227 } 1228 1229 /* Create destination port list: */ 1230 tipc_nametbl_mc_lookup(net, type, lower, upper, 1231 scope, exact, &dports); 1232 1233 /* Clone message per destination */ 1234 while (tipc_dest_pop(&dports, NULL, &portid)) { 1235 _skb = __pskb_copy(skb, hlen, GFP_ATOMIC); 1236 if (_skb) { 1237 msg_set_destport(buf_msg(_skb), portid); 1238 __skb_queue_tail(&tmpq, _skb); 1239 continue; 1240 } 1241 pr_warn("Failed to clone mcast rcv buffer\n"); 1242 } 1243 /* Append to inputq if not already done by other thread */ 1244 spin_lock_bh(&inputq->lock); 1245 if (skb_peek(arrvq) == skb) { 1246 skb_queue_splice_tail_init(&tmpq, inputq); 1247 kfree_skb(__skb_dequeue(arrvq)); 1248 } 1249 spin_unlock_bh(&inputq->lock); 1250 __skb_queue_purge(&tmpq); 1251 kfree_skb(skb); 1252 } 1253 tipc_sk_rcv(net, inputq); 1254 } 1255 1256 /* tipc_sk_push_backlog(): send accumulated buffers in socket write queue 1257 * when socket is in Nagle mode 1258 */ 1259 static void tipc_sk_push_backlog(struct tipc_sock *tsk, bool nagle_ack) 1260 { 1261 struct sk_buff_head *txq = &tsk->sk.sk_write_queue; 1262 struct sk_buff *skb = skb_peek_tail(txq); 1263 struct net *net = sock_net(&tsk->sk); 1264 u32 dnode = tsk_peer_node(tsk); 1265 int rc; 1266 1267 if (nagle_ack) { 1268 tsk->pkt_cnt += skb_queue_len(txq); 1269 if (!tsk->pkt_cnt || tsk->msg_acc / tsk->pkt_cnt < 2) { 1270 tsk->oneway = 0; 1271 if (tsk->nagle_start < NAGLE_START_MAX) 1272 tsk->nagle_start *= 2; 1273 tsk->expect_ack = false; 1274 pr_debug("tsk %10u: bad nagle %u -> %u, next start %u!\n", 1275 tsk->portid, tsk->msg_acc, tsk->pkt_cnt, 1276 tsk->nagle_start); 1277 } else { 1278 tsk->nagle_start = NAGLE_START_INIT; 1279 if (skb) { 1280 msg_set_ack_required(buf_msg(skb)); 1281 tsk->expect_ack = true; 1282 } else { 1283 tsk->expect_ack = false; 1284 } 1285 } 1286 tsk->msg_acc = 0; 1287 tsk->pkt_cnt = 0; 1288 } 1289 1290 if (!skb || tsk->cong_link_cnt) 1291 return; 1292 1293 /* Do not send SYN again after congestion */ 1294 if (msg_is_syn(buf_msg(skb))) 1295 return; 1296 1297 if (tsk->msg_acc) 1298 tsk->pkt_cnt += skb_queue_len(txq); 1299 tsk->snt_unacked += tsk->snd_backlog; 1300 tsk->snd_backlog = 0; 1301 rc = tipc_node_xmit(net, txq, dnode, tsk->portid); 1302 if (rc == -ELINKCONG) 1303 tsk->cong_link_cnt = 1; 1304 } 1305 1306 /** 1307 * tipc_sk_conn_proto_rcv - receive a connection mng protocol message 1308 * @tsk: receiving socket 1309 * @skb: pointer to message buffer. 1310 */ 1311 static void tipc_sk_conn_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb, 1312 struct sk_buff_head *inputq, 1313 struct sk_buff_head *xmitq) 1314 { 1315 struct tipc_msg *hdr = buf_msg(skb); 1316 u32 onode = tsk_own_node(tsk); 1317 struct sock *sk = &tsk->sk; 1318 int mtyp = msg_type(hdr); 1319 bool was_cong; 1320 1321 /* Ignore if connection cannot be validated: */ 1322 if (!tsk_peer_msg(tsk, hdr)) { 1323 trace_tipc_sk_drop_msg(sk, skb, TIPC_DUMP_NONE, "@proto_rcv!"); 1324 goto exit; 1325 } 1326 1327 if (unlikely(msg_errcode(hdr))) { 1328 tipc_set_sk_state(sk, TIPC_DISCONNECTING); 1329 tipc_node_remove_conn(sock_net(sk), tsk_peer_node(tsk), 1330 tsk_peer_port(tsk)); 1331 sk->sk_state_change(sk); 1332 1333 /* State change is ignored if socket already awake, 1334 * - convert msg to abort msg and add to inqueue 1335 */ 1336 msg_set_user(hdr, TIPC_CRITICAL_IMPORTANCE); 1337 msg_set_type(hdr, TIPC_CONN_MSG); 1338 msg_set_size(hdr, BASIC_H_SIZE); 1339 msg_set_hdr_sz(hdr, BASIC_H_SIZE); 1340 __skb_queue_tail(inputq, skb); 1341 return; 1342 } 1343 1344 tsk->probe_unacked = false; 1345 1346 if (mtyp == CONN_PROBE) { 1347 msg_set_type(hdr, CONN_PROBE_REPLY); 1348 if (tipc_msg_reverse(onode, &skb, TIPC_OK)) 1349 __skb_queue_tail(xmitq, skb); 1350 return; 1351 } else if (mtyp == CONN_ACK) { 1352 was_cong = tsk_conn_cong(tsk); 1353 tipc_sk_push_backlog(tsk, msg_nagle_ack(hdr)); 1354 tsk->snt_unacked -= msg_conn_ack(hdr); 1355 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) 1356 tsk->snd_win = msg_adv_win(hdr); 1357 if (was_cong && !tsk_conn_cong(tsk)) 1358 sk->sk_write_space(sk); 1359 } else if (mtyp != CONN_PROBE_REPLY) { 1360 pr_warn("Received unknown CONN_PROTO msg\n"); 1361 } 1362 exit: 1363 kfree_skb(skb); 1364 } 1365 1366 /** 1367 * tipc_sendmsg - send message in connectionless manner 1368 * @sock: socket structure 1369 * @m: message to send 1370 * @dsz: amount of user data to be sent 1371 * 1372 * Message must have an destination specified explicitly. 1373 * Used for SOCK_RDM and SOCK_DGRAM messages, 1374 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections. 1375 * (Note: 'SYN+' is prohibited on SOCK_STREAM.) 1376 * 1377 * Returns the number of bytes sent on success, or errno otherwise 1378 */ 1379 static int tipc_sendmsg(struct socket *sock, 1380 struct msghdr *m, size_t dsz) 1381 { 1382 struct sock *sk = sock->sk; 1383 int ret; 1384 1385 lock_sock(sk); 1386 ret = __tipc_sendmsg(sock, m, dsz); 1387 release_sock(sk); 1388 1389 return ret; 1390 } 1391 1392 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen) 1393 { 1394 struct sock *sk = sock->sk; 1395 struct net *net = sock_net(sk); 1396 struct tipc_sock *tsk = tipc_sk(sk); 1397 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name); 1398 long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT); 1399 struct list_head *clinks = &tsk->cong_links; 1400 bool syn = !tipc_sk_type_connectionless(sk); 1401 struct tipc_group *grp = tsk->group; 1402 struct tipc_msg *hdr = &tsk->phdr; 1403 struct tipc_name_seq *seq; 1404 struct sk_buff_head pkts; 1405 u32 dport = 0, dnode = 0; 1406 u32 type = 0, inst = 0; 1407 int mtu, rc; 1408 1409 if (unlikely(dlen > TIPC_MAX_USER_MSG_SIZE)) 1410 return -EMSGSIZE; 1411 1412 if (likely(dest)) { 1413 if (unlikely(m->msg_namelen < sizeof(*dest))) 1414 return -EINVAL; 1415 if (unlikely(dest->family != AF_TIPC)) 1416 return -EINVAL; 1417 } 1418 1419 if (grp) { 1420 if (!dest) 1421 return tipc_send_group_bcast(sock, m, dlen, timeout); 1422 if (dest->addrtype == TIPC_ADDR_NAME) 1423 return tipc_send_group_anycast(sock, m, dlen, timeout); 1424 if (dest->addrtype == TIPC_ADDR_ID) 1425 return tipc_send_group_unicast(sock, m, dlen, timeout); 1426 if (dest->addrtype == TIPC_ADDR_MCAST) 1427 return tipc_send_group_mcast(sock, m, dlen, timeout); 1428 return -EINVAL; 1429 } 1430 1431 if (unlikely(!dest)) { 1432 dest = &tsk->peer; 1433 if (!syn && dest->family != AF_TIPC) 1434 return -EDESTADDRREQ; 1435 } 1436 1437 if (unlikely(syn)) { 1438 if (sk->sk_state == TIPC_LISTEN) 1439 return -EPIPE; 1440 if (sk->sk_state != TIPC_OPEN) 1441 return -EISCONN; 1442 if (tsk->published) 1443 return -EOPNOTSUPP; 1444 if (dest->addrtype == TIPC_ADDR_NAME) { 1445 tsk->conn_type = dest->addr.name.name.type; 1446 tsk->conn_instance = dest->addr.name.name.instance; 1447 } 1448 msg_set_syn(hdr, 1); 1449 } 1450 1451 seq = &dest->addr.nameseq; 1452 if (dest->addrtype == TIPC_ADDR_MCAST) 1453 return tipc_sendmcast(sock, seq, m, dlen, timeout); 1454 1455 if (dest->addrtype == TIPC_ADDR_NAME) { 1456 type = dest->addr.name.name.type; 1457 inst = dest->addr.name.name.instance; 1458 dnode = dest->addr.name.domain; 1459 dport = tipc_nametbl_translate(net, type, inst, &dnode); 1460 if (unlikely(!dport && !dnode)) 1461 return -EHOSTUNREACH; 1462 } else if (dest->addrtype == TIPC_ADDR_ID) { 1463 dnode = dest->addr.id.node; 1464 } else { 1465 return -EINVAL; 1466 } 1467 1468 /* Block or return if destination link is congested */ 1469 rc = tipc_wait_for_cond(sock, &timeout, 1470 !tipc_dest_find(clinks, dnode, 0)); 1471 if (unlikely(rc)) 1472 return rc; 1473 1474 if (dest->addrtype == TIPC_ADDR_NAME) { 1475 msg_set_type(hdr, TIPC_NAMED_MSG); 1476 msg_set_hdr_sz(hdr, NAMED_H_SIZE); 1477 msg_set_nametype(hdr, type); 1478 msg_set_nameinst(hdr, inst); 1479 msg_set_lookup_scope(hdr, tipc_node2scope(dnode)); 1480 msg_set_destnode(hdr, dnode); 1481 msg_set_destport(hdr, dport); 1482 } else { /* TIPC_ADDR_ID */ 1483 msg_set_type(hdr, TIPC_DIRECT_MSG); 1484 msg_set_lookup_scope(hdr, 0); 1485 msg_set_destnode(hdr, dnode); 1486 msg_set_destport(hdr, dest->addr.id.ref); 1487 msg_set_hdr_sz(hdr, BASIC_H_SIZE); 1488 } 1489 1490 __skb_queue_head_init(&pkts); 1491 mtu = tipc_node_get_mtu(net, dnode, tsk->portid, true); 1492 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts); 1493 if (unlikely(rc != dlen)) 1494 return rc; 1495 if (unlikely(syn && !tipc_msg_skb_clone(&pkts, &sk->sk_write_queue))) { 1496 __skb_queue_purge(&pkts); 1497 return -ENOMEM; 1498 } 1499 1500 trace_tipc_sk_sendmsg(sk, skb_peek(&pkts), TIPC_DUMP_SK_SNDQ, " "); 1501 rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid); 1502 if (unlikely(rc == -ELINKCONG)) { 1503 tipc_dest_push(clinks, dnode, 0); 1504 tsk->cong_link_cnt++; 1505 rc = 0; 1506 } 1507 1508 if (unlikely(syn && !rc)) 1509 tipc_set_sk_state(sk, TIPC_CONNECTING); 1510 1511 return rc ? rc : dlen; 1512 } 1513 1514 /** 1515 * tipc_sendstream - send stream-oriented data 1516 * @sock: socket structure 1517 * @m: data to send 1518 * @dsz: total length of data to be transmitted 1519 * 1520 * Used for SOCK_STREAM data. 1521 * 1522 * Returns the number of bytes sent on success (or partial success), 1523 * or errno if no data sent 1524 */ 1525 static int tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz) 1526 { 1527 struct sock *sk = sock->sk; 1528 int ret; 1529 1530 lock_sock(sk); 1531 ret = __tipc_sendstream(sock, m, dsz); 1532 release_sock(sk); 1533 1534 return ret; 1535 } 1536 1537 static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dlen) 1538 { 1539 struct sock *sk = sock->sk; 1540 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name); 1541 long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT); 1542 struct sk_buff_head *txq = &sk->sk_write_queue; 1543 struct tipc_sock *tsk = tipc_sk(sk); 1544 struct tipc_msg *hdr = &tsk->phdr; 1545 struct net *net = sock_net(sk); 1546 struct sk_buff *skb; 1547 u32 dnode = tsk_peer_node(tsk); 1548 int maxnagle = tsk->maxnagle; 1549 int maxpkt = tsk->max_pkt; 1550 int send, sent = 0; 1551 int blocks, rc = 0; 1552 1553 if (unlikely(dlen > INT_MAX)) 1554 return -EMSGSIZE; 1555 1556 /* Handle implicit connection setup */ 1557 if (unlikely(dest)) { 1558 rc = __tipc_sendmsg(sock, m, dlen); 1559 if (dlen && dlen == rc) { 1560 tsk->peer_caps = tipc_node_get_capabilities(net, dnode); 1561 tsk->snt_unacked = tsk_inc(tsk, dlen + msg_hdr_sz(hdr)); 1562 } 1563 return rc; 1564 } 1565 1566 do { 1567 rc = tipc_wait_for_cond(sock, &timeout, 1568 (!tsk->cong_link_cnt && 1569 !tsk_conn_cong(tsk) && 1570 tipc_sk_connected(sk))); 1571 if (unlikely(rc)) 1572 break; 1573 send = min_t(size_t, dlen - sent, TIPC_MAX_USER_MSG_SIZE); 1574 blocks = tsk->snd_backlog; 1575 if (tsk->oneway++ >= tsk->nagle_start && maxnagle && 1576 send <= maxnagle) { 1577 rc = tipc_msg_append(hdr, m, send, maxnagle, txq); 1578 if (unlikely(rc < 0)) 1579 break; 1580 blocks += rc; 1581 tsk->msg_acc++; 1582 if (blocks <= 64 && tsk->expect_ack) { 1583 tsk->snd_backlog = blocks; 1584 sent += send; 1585 break; 1586 } else if (blocks > 64) { 1587 tsk->pkt_cnt += skb_queue_len(txq); 1588 } else { 1589 skb = skb_peek_tail(txq); 1590 if (skb) { 1591 msg_set_ack_required(buf_msg(skb)); 1592 tsk->expect_ack = true; 1593 } else { 1594 tsk->expect_ack = false; 1595 } 1596 tsk->msg_acc = 0; 1597 tsk->pkt_cnt = 0; 1598 } 1599 } else { 1600 rc = tipc_msg_build(hdr, m, sent, send, maxpkt, txq); 1601 if (unlikely(rc != send)) 1602 break; 1603 blocks += tsk_inc(tsk, send + MIN_H_SIZE); 1604 } 1605 trace_tipc_sk_sendstream(sk, skb_peek(txq), 1606 TIPC_DUMP_SK_SNDQ, " "); 1607 rc = tipc_node_xmit(net, txq, dnode, tsk->portid); 1608 if (unlikely(rc == -ELINKCONG)) { 1609 tsk->cong_link_cnt = 1; 1610 rc = 0; 1611 } 1612 if (likely(!rc)) { 1613 tsk->snt_unacked += blocks; 1614 tsk->snd_backlog = 0; 1615 sent += send; 1616 } 1617 } while (sent < dlen && !rc); 1618 1619 return sent ? sent : rc; 1620 } 1621 1622 /** 1623 * tipc_send_packet - send a connection-oriented message 1624 * @sock: socket structure 1625 * @m: message to send 1626 * @dsz: length of data to be transmitted 1627 * 1628 * Used for SOCK_SEQPACKET messages. 1629 * 1630 * Returns the number of bytes sent on success, or errno otherwise 1631 */ 1632 static int tipc_send_packet(struct socket *sock, struct msghdr *m, size_t dsz) 1633 { 1634 if (dsz > TIPC_MAX_USER_MSG_SIZE) 1635 return -EMSGSIZE; 1636 1637 return tipc_sendstream(sock, m, dsz); 1638 } 1639 1640 /* tipc_sk_finish_conn - complete the setup of a connection 1641 */ 1642 static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port, 1643 u32 peer_node) 1644 { 1645 struct sock *sk = &tsk->sk; 1646 struct net *net = sock_net(sk); 1647 struct tipc_msg *msg = &tsk->phdr; 1648 1649 msg_set_syn(msg, 0); 1650 msg_set_destnode(msg, peer_node); 1651 msg_set_destport(msg, peer_port); 1652 msg_set_type(msg, TIPC_CONN_MSG); 1653 msg_set_lookup_scope(msg, 0); 1654 msg_set_hdr_sz(msg, SHORT_H_SIZE); 1655 1656 sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV); 1657 tipc_set_sk_state(sk, TIPC_ESTABLISHED); 1658 tipc_node_add_conn(net, peer_node, tsk->portid, peer_port); 1659 tsk->max_pkt = tipc_node_get_mtu(net, peer_node, tsk->portid, true); 1660 tsk->peer_caps = tipc_node_get_capabilities(net, peer_node); 1661 tsk_set_nagle(tsk); 1662 __skb_queue_purge(&sk->sk_write_queue); 1663 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) 1664 return; 1665 1666 /* Fall back to message based flow control */ 1667 tsk->rcv_win = FLOWCTL_MSG_WIN; 1668 tsk->snd_win = FLOWCTL_MSG_WIN; 1669 } 1670 1671 /** 1672 * tipc_sk_set_orig_addr - capture sender's address for received message 1673 * @m: descriptor for message info 1674 * @skb: received message 1675 * 1676 * Note: Address is not captured if not requested by receiver. 1677 */ 1678 static void tipc_sk_set_orig_addr(struct msghdr *m, struct sk_buff *skb) 1679 { 1680 DECLARE_SOCKADDR(struct sockaddr_pair *, srcaddr, m->msg_name); 1681 struct tipc_msg *hdr = buf_msg(skb); 1682 1683 if (!srcaddr) 1684 return; 1685 1686 srcaddr->sock.family = AF_TIPC; 1687 srcaddr->sock.addrtype = TIPC_ADDR_ID; 1688 srcaddr->sock.scope = 0; 1689 srcaddr->sock.addr.id.ref = msg_origport(hdr); 1690 srcaddr->sock.addr.id.node = msg_orignode(hdr); 1691 srcaddr->sock.addr.name.domain = 0; 1692 m->msg_namelen = sizeof(struct sockaddr_tipc); 1693 1694 if (!msg_in_group(hdr)) 1695 return; 1696 1697 /* Group message users may also want to know sending member's id */ 1698 srcaddr->member.family = AF_TIPC; 1699 srcaddr->member.addrtype = TIPC_ADDR_NAME; 1700 srcaddr->member.scope = 0; 1701 srcaddr->member.addr.name.name.type = msg_nametype(hdr); 1702 srcaddr->member.addr.name.name.instance = TIPC_SKB_CB(skb)->orig_member; 1703 srcaddr->member.addr.name.domain = 0; 1704 m->msg_namelen = sizeof(*srcaddr); 1705 } 1706 1707 /** 1708 * tipc_sk_anc_data_recv - optionally capture ancillary data for received message 1709 * @m: descriptor for message info 1710 * @skb: received message buffer 1711 * @tsk: TIPC port associated with message 1712 * 1713 * Note: Ancillary data is not captured if not requested by receiver. 1714 * 1715 * Returns 0 if successful, otherwise errno 1716 */ 1717 static int tipc_sk_anc_data_recv(struct msghdr *m, struct sk_buff *skb, 1718 struct tipc_sock *tsk) 1719 { 1720 struct tipc_msg *msg; 1721 u32 anc_data[3]; 1722 u32 err; 1723 u32 dest_type; 1724 int has_name; 1725 int res; 1726 1727 if (likely(m->msg_controllen == 0)) 1728 return 0; 1729 msg = buf_msg(skb); 1730 1731 /* Optionally capture errored message object(s) */ 1732 err = msg ? msg_errcode(msg) : 0; 1733 if (unlikely(err)) { 1734 anc_data[0] = err; 1735 anc_data[1] = msg_data_sz(msg); 1736 res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data); 1737 if (res) 1738 return res; 1739 if (anc_data[1]) { 1740 if (skb_linearize(skb)) 1741 return -ENOMEM; 1742 msg = buf_msg(skb); 1743 res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1], 1744 msg_data(msg)); 1745 if (res) 1746 return res; 1747 } 1748 } 1749 1750 /* Optionally capture message destination object */ 1751 dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG; 1752 switch (dest_type) { 1753 case TIPC_NAMED_MSG: 1754 has_name = 1; 1755 anc_data[0] = msg_nametype(msg); 1756 anc_data[1] = msg_namelower(msg); 1757 anc_data[2] = msg_namelower(msg); 1758 break; 1759 case TIPC_MCAST_MSG: 1760 has_name = 1; 1761 anc_data[0] = msg_nametype(msg); 1762 anc_data[1] = msg_namelower(msg); 1763 anc_data[2] = msg_nameupper(msg); 1764 break; 1765 case TIPC_CONN_MSG: 1766 has_name = (tsk->conn_type != 0); 1767 anc_data[0] = tsk->conn_type; 1768 anc_data[1] = tsk->conn_instance; 1769 anc_data[2] = tsk->conn_instance; 1770 break; 1771 default: 1772 has_name = 0; 1773 } 1774 if (has_name) { 1775 res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data); 1776 if (res) 1777 return res; 1778 } 1779 1780 return 0; 1781 } 1782 1783 static struct sk_buff *tipc_sk_build_ack(struct tipc_sock *tsk) 1784 { 1785 struct sock *sk = &tsk->sk; 1786 struct sk_buff *skb = NULL; 1787 struct tipc_msg *msg; 1788 u32 peer_port = tsk_peer_port(tsk); 1789 u32 dnode = tsk_peer_node(tsk); 1790 1791 if (!tipc_sk_connected(sk)) 1792 return NULL; 1793 skb = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0, 1794 dnode, tsk_own_node(tsk), peer_port, 1795 tsk->portid, TIPC_OK); 1796 if (!skb) 1797 return NULL; 1798 msg = buf_msg(skb); 1799 msg_set_conn_ack(msg, tsk->rcv_unacked); 1800 tsk->rcv_unacked = 0; 1801 1802 /* Adjust to and advertize the correct window limit */ 1803 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) { 1804 tsk->rcv_win = tsk_adv_blocks(tsk->sk.sk_rcvbuf); 1805 msg_set_adv_win(msg, tsk->rcv_win); 1806 } 1807 return skb; 1808 } 1809 1810 static void tipc_sk_send_ack(struct tipc_sock *tsk) 1811 { 1812 struct sk_buff *skb; 1813 1814 skb = tipc_sk_build_ack(tsk); 1815 if (!skb) 1816 return; 1817 1818 tipc_node_xmit_skb(sock_net(&tsk->sk), skb, tsk_peer_node(tsk), 1819 msg_link_selector(buf_msg(skb))); 1820 } 1821 1822 static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop) 1823 { 1824 struct sock *sk = sock->sk; 1825 DEFINE_WAIT_FUNC(wait, woken_wake_function); 1826 long timeo = *timeop; 1827 int err = sock_error(sk); 1828 1829 if (err) 1830 return err; 1831 1832 for (;;) { 1833 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) { 1834 if (sk->sk_shutdown & RCV_SHUTDOWN) { 1835 err = -ENOTCONN; 1836 break; 1837 } 1838 add_wait_queue(sk_sleep(sk), &wait); 1839 release_sock(sk); 1840 timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo); 1841 sched_annotate_sleep(); 1842 lock_sock(sk); 1843 remove_wait_queue(sk_sleep(sk), &wait); 1844 } 1845 err = 0; 1846 if (!skb_queue_empty(&sk->sk_receive_queue)) 1847 break; 1848 err = -EAGAIN; 1849 if (!timeo) 1850 break; 1851 err = sock_intr_errno(timeo); 1852 if (signal_pending(current)) 1853 break; 1854 1855 err = sock_error(sk); 1856 if (err) 1857 break; 1858 } 1859 *timeop = timeo; 1860 return err; 1861 } 1862 1863 /** 1864 * tipc_recvmsg - receive packet-oriented message 1865 * @m: descriptor for message info 1866 * @buflen: length of user buffer area 1867 * @flags: receive flags 1868 * 1869 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages. 1870 * If the complete message doesn't fit in user area, truncate it. 1871 * 1872 * Returns size of returned message data, errno otherwise 1873 */ 1874 static int tipc_recvmsg(struct socket *sock, struct msghdr *m, 1875 size_t buflen, int flags) 1876 { 1877 struct sock *sk = sock->sk; 1878 bool connected = !tipc_sk_type_connectionless(sk); 1879 struct tipc_sock *tsk = tipc_sk(sk); 1880 int rc, err, hlen, dlen, copy; 1881 struct sk_buff_head xmitq; 1882 struct tipc_msg *hdr; 1883 struct sk_buff *skb; 1884 bool grp_evt; 1885 long timeout; 1886 1887 /* Catch invalid receive requests */ 1888 if (unlikely(!buflen)) 1889 return -EINVAL; 1890 1891 lock_sock(sk); 1892 if (unlikely(connected && sk->sk_state == TIPC_OPEN)) { 1893 rc = -ENOTCONN; 1894 goto exit; 1895 } 1896 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT); 1897 1898 /* Step rcv queue to first msg with data or error; wait if necessary */ 1899 do { 1900 rc = tipc_wait_for_rcvmsg(sock, &timeout); 1901 if (unlikely(rc)) 1902 goto exit; 1903 skb = skb_peek(&sk->sk_receive_queue); 1904 hdr = buf_msg(skb); 1905 dlen = msg_data_sz(hdr); 1906 hlen = msg_hdr_sz(hdr); 1907 err = msg_errcode(hdr); 1908 grp_evt = msg_is_grp_evt(hdr); 1909 if (likely(dlen || err)) 1910 break; 1911 tsk_advance_rx_queue(sk); 1912 } while (1); 1913 1914 /* Collect msg meta data, including error code and rejected data */ 1915 tipc_sk_set_orig_addr(m, skb); 1916 rc = tipc_sk_anc_data_recv(m, skb, tsk); 1917 if (unlikely(rc)) 1918 goto exit; 1919 hdr = buf_msg(skb); 1920 1921 /* Capture data if non-error msg, otherwise just set return value */ 1922 if (likely(!err)) { 1923 copy = min_t(int, dlen, buflen); 1924 if (unlikely(copy != dlen)) 1925 m->msg_flags |= MSG_TRUNC; 1926 rc = skb_copy_datagram_msg(skb, hlen, m, copy); 1927 } else { 1928 copy = 0; 1929 rc = 0; 1930 if (err != TIPC_CONN_SHUTDOWN && connected && !m->msg_control) 1931 rc = -ECONNRESET; 1932 } 1933 if (unlikely(rc)) 1934 goto exit; 1935 1936 /* Mark message as group event if applicable */ 1937 if (unlikely(grp_evt)) { 1938 if (msg_grp_evt(hdr) == TIPC_WITHDRAWN) 1939 m->msg_flags |= MSG_EOR; 1940 m->msg_flags |= MSG_OOB; 1941 copy = 0; 1942 } 1943 1944 /* Caption of data or error code/rejected data was successful */ 1945 if (unlikely(flags & MSG_PEEK)) 1946 goto exit; 1947 1948 /* Send group flow control advertisement when applicable */ 1949 if (tsk->group && msg_in_group(hdr) && !grp_evt) { 1950 __skb_queue_head_init(&xmitq); 1951 tipc_group_update_rcv_win(tsk->group, tsk_blocks(hlen + dlen), 1952 msg_orignode(hdr), msg_origport(hdr), 1953 &xmitq); 1954 tipc_node_distr_xmit(sock_net(sk), &xmitq); 1955 } 1956 1957 tsk_advance_rx_queue(sk); 1958 1959 if (likely(!connected)) 1960 goto exit; 1961 1962 /* Send connection flow control advertisement when applicable */ 1963 tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen); 1964 if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE) 1965 tipc_sk_send_ack(tsk); 1966 exit: 1967 release_sock(sk); 1968 return rc ? rc : copy; 1969 } 1970 1971 /** 1972 * tipc_recvstream - receive stream-oriented data 1973 * @m: descriptor for message info 1974 * @buflen: total size of user buffer area 1975 * @flags: receive flags 1976 * 1977 * Used for SOCK_STREAM messages only. If not enough data is available 1978 * will optionally wait for more; never truncates data. 1979 * 1980 * Returns size of returned message data, errno otherwise 1981 */ 1982 static int tipc_recvstream(struct socket *sock, struct msghdr *m, 1983 size_t buflen, int flags) 1984 { 1985 struct sock *sk = sock->sk; 1986 struct tipc_sock *tsk = tipc_sk(sk); 1987 struct sk_buff *skb; 1988 struct tipc_msg *hdr; 1989 struct tipc_skb_cb *skb_cb; 1990 bool peek = flags & MSG_PEEK; 1991 int offset, required, copy, copied = 0; 1992 int hlen, dlen, err, rc; 1993 long timeout; 1994 1995 /* Catch invalid receive attempts */ 1996 if (unlikely(!buflen)) 1997 return -EINVAL; 1998 1999 lock_sock(sk); 2000 2001 if (unlikely(sk->sk_state == TIPC_OPEN)) { 2002 rc = -ENOTCONN; 2003 goto exit; 2004 } 2005 required = sock_rcvlowat(sk, flags & MSG_WAITALL, buflen); 2006 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT); 2007 2008 do { 2009 /* Look at first msg in receive queue; wait if necessary */ 2010 rc = tipc_wait_for_rcvmsg(sock, &timeout); 2011 if (unlikely(rc)) 2012 break; 2013 skb = skb_peek(&sk->sk_receive_queue); 2014 skb_cb = TIPC_SKB_CB(skb); 2015 hdr = buf_msg(skb); 2016 dlen = msg_data_sz(hdr); 2017 hlen = msg_hdr_sz(hdr); 2018 err = msg_errcode(hdr); 2019 2020 /* Discard any empty non-errored (SYN-) message */ 2021 if (unlikely(!dlen && !err)) { 2022 tsk_advance_rx_queue(sk); 2023 continue; 2024 } 2025 2026 /* Collect msg meta data, incl. error code and rejected data */ 2027 if (!copied) { 2028 tipc_sk_set_orig_addr(m, skb); 2029 rc = tipc_sk_anc_data_recv(m, skb, tsk); 2030 if (rc) 2031 break; 2032 hdr = buf_msg(skb); 2033 } 2034 2035 /* Copy data if msg ok, otherwise return error/partial data */ 2036 if (likely(!err)) { 2037 offset = skb_cb->bytes_read; 2038 copy = min_t(int, dlen - offset, buflen - copied); 2039 rc = skb_copy_datagram_msg(skb, hlen + offset, m, copy); 2040 if (unlikely(rc)) 2041 break; 2042 copied += copy; 2043 offset += copy; 2044 if (unlikely(offset < dlen)) { 2045 if (!peek) 2046 skb_cb->bytes_read = offset; 2047 break; 2048 } 2049 } else { 2050 rc = 0; 2051 if ((err != TIPC_CONN_SHUTDOWN) && !m->msg_control) 2052 rc = -ECONNRESET; 2053 if (copied || rc) 2054 break; 2055 } 2056 2057 if (unlikely(peek)) 2058 break; 2059 2060 tsk_advance_rx_queue(sk); 2061 2062 /* Send connection flow control advertisement when applicable */ 2063 tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen); 2064 if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE) 2065 tipc_sk_send_ack(tsk); 2066 2067 /* Exit if all requested data or FIN/error received */ 2068 if (copied == buflen || err) 2069 break; 2070 2071 } while (!skb_queue_empty(&sk->sk_receive_queue) || copied < required); 2072 exit: 2073 release_sock(sk); 2074 return copied ? copied : rc; 2075 } 2076 2077 /** 2078 * tipc_write_space - wake up thread if port congestion is released 2079 * @sk: socket 2080 */ 2081 static void tipc_write_space(struct sock *sk) 2082 { 2083 struct socket_wq *wq; 2084 2085 rcu_read_lock(); 2086 wq = rcu_dereference(sk->sk_wq); 2087 if (skwq_has_sleeper(wq)) 2088 wake_up_interruptible_sync_poll(&wq->wait, EPOLLOUT | 2089 EPOLLWRNORM | EPOLLWRBAND); 2090 rcu_read_unlock(); 2091 } 2092 2093 /** 2094 * tipc_data_ready - wake up threads to indicate messages have been received 2095 * @sk: socket 2096 */ 2097 static void tipc_data_ready(struct sock *sk) 2098 { 2099 struct socket_wq *wq; 2100 2101 rcu_read_lock(); 2102 wq = rcu_dereference(sk->sk_wq); 2103 if (skwq_has_sleeper(wq)) 2104 wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN | 2105 EPOLLRDNORM | EPOLLRDBAND); 2106 rcu_read_unlock(); 2107 } 2108 2109 static void tipc_sock_destruct(struct sock *sk) 2110 { 2111 __skb_queue_purge(&sk->sk_receive_queue); 2112 } 2113 2114 static void tipc_sk_proto_rcv(struct sock *sk, 2115 struct sk_buff_head *inputq, 2116 struct sk_buff_head *xmitq) 2117 { 2118 struct sk_buff *skb = __skb_dequeue(inputq); 2119 struct tipc_sock *tsk = tipc_sk(sk); 2120 struct tipc_msg *hdr = buf_msg(skb); 2121 struct tipc_group *grp = tsk->group; 2122 bool wakeup = false; 2123 2124 switch (msg_user(hdr)) { 2125 case CONN_MANAGER: 2126 tipc_sk_conn_proto_rcv(tsk, skb, inputq, xmitq); 2127 return; 2128 case SOCK_WAKEUP: 2129 tipc_dest_del(&tsk->cong_links, msg_orignode(hdr), 0); 2130 /* coupled with smp_rmb() in tipc_wait_for_cond() */ 2131 smp_wmb(); 2132 tsk->cong_link_cnt--; 2133 wakeup = true; 2134 tipc_sk_push_backlog(tsk, false); 2135 break; 2136 case GROUP_PROTOCOL: 2137 tipc_group_proto_rcv(grp, &wakeup, hdr, inputq, xmitq); 2138 break; 2139 case TOP_SRV: 2140 tipc_group_member_evt(tsk->group, &wakeup, &sk->sk_rcvbuf, 2141 hdr, inputq, xmitq); 2142 break; 2143 default: 2144 break; 2145 } 2146 2147 if (wakeup) 2148 sk->sk_write_space(sk); 2149 2150 kfree_skb(skb); 2151 } 2152 2153 /** 2154 * tipc_sk_filter_connect - check incoming message for a connection-based socket 2155 * @tsk: TIPC socket 2156 * @skb: pointer to message buffer. 2157 * @xmitq: for Nagle ACK if any 2158 * Returns true if message should be added to receive queue, false otherwise 2159 */ 2160 static bool tipc_sk_filter_connect(struct tipc_sock *tsk, struct sk_buff *skb, 2161 struct sk_buff_head *xmitq) 2162 { 2163 struct sock *sk = &tsk->sk; 2164 struct net *net = sock_net(sk); 2165 struct tipc_msg *hdr = buf_msg(skb); 2166 bool con_msg = msg_connected(hdr); 2167 u32 pport = tsk_peer_port(tsk); 2168 u32 pnode = tsk_peer_node(tsk); 2169 u32 oport = msg_origport(hdr); 2170 u32 onode = msg_orignode(hdr); 2171 int err = msg_errcode(hdr); 2172 unsigned long delay; 2173 2174 if (unlikely(msg_mcast(hdr))) 2175 return false; 2176 tsk->oneway = 0; 2177 2178 switch (sk->sk_state) { 2179 case TIPC_CONNECTING: 2180 /* Setup ACK */ 2181 if (likely(con_msg)) { 2182 if (err) 2183 break; 2184 tipc_sk_finish_conn(tsk, oport, onode); 2185 msg_set_importance(&tsk->phdr, msg_importance(hdr)); 2186 /* ACK+ message with data is added to receive queue */ 2187 if (msg_data_sz(hdr)) 2188 return true; 2189 /* Empty ACK-, - wake up sleeping connect() and drop */ 2190 sk->sk_state_change(sk); 2191 msg_set_dest_droppable(hdr, 1); 2192 return false; 2193 } 2194 /* Ignore connectionless message if not from listening socket */ 2195 if (oport != pport || onode != pnode) 2196 return false; 2197 2198 /* Rejected SYN */ 2199 if (err != TIPC_ERR_OVERLOAD) 2200 break; 2201 2202 /* Prepare for new setup attempt if we have a SYN clone */ 2203 if (skb_queue_empty(&sk->sk_write_queue)) 2204 break; 2205 get_random_bytes(&delay, 2); 2206 delay %= (tsk->conn_timeout / 4); 2207 delay = msecs_to_jiffies(delay + 100); 2208 sk_reset_timer(sk, &sk->sk_timer, jiffies + delay); 2209 return false; 2210 case TIPC_OPEN: 2211 case TIPC_DISCONNECTING: 2212 return false; 2213 case TIPC_LISTEN: 2214 /* Accept only SYN message */ 2215 if (!msg_is_syn(hdr) && 2216 tipc_node_get_capabilities(net, onode) & TIPC_SYN_BIT) 2217 return false; 2218 if (!con_msg && !err) 2219 return true; 2220 return false; 2221 case TIPC_ESTABLISHED: 2222 if (!skb_queue_empty(&sk->sk_write_queue)) 2223 tipc_sk_push_backlog(tsk, false); 2224 /* Accept only connection-based messages sent by peer */ 2225 if (likely(con_msg && !err && pport == oport && 2226 pnode == onode)) { 2227 if (msg_ack_required(hdr)) { 2228 struct sk_buff *skb; 2229 2230 skb = tipc_sk_build_ack(tsk); 2231 if (skb) { 2232 msg_set_nagle_ack(buf_msg(skb)); 2233 __skb_queue_tail(xmitq, skb); 2234 } 2235 } 2236 return true; 2237 } 2238 if (!tsk_peer_msg(tsk, hdr)) 2239 return false; 2240 if (!err) 2241 return true; 2242 tipc_set_sk_state(sk, TIPC_DISCONNECTING); 2243 tipc_node_remove_conn(net, pnode, tsk->portid); 2244 sk->sk_state_change(sk); 2245 return true; 2246 default: 2247 pr_err("Unknown sk_state %u\n", sk->sk_state); 2248 } 2249 /* Abort connection setup attempt */ 2250 tipc_set_sk_state(sk, TIPC_DISCONNECTING); 2251 sk->sk_err = ECONNREFUSED; 2252 sk->sk_state_change(sk); 2253 return true; 2254 } 2255 2256 /** 2257 * rcvbuf_limit - get proper overload limit of socket receive queue 2258 * @sk: socket 2259 * @skb: message 2260 * 2261 * For connection oriented messages, irrespective of importance, 2262 * default queue limit is 2 MB. 2263 * 2264 * For connectionless messages, queue limits are based on message 2265 * importance as follows: 2266 * 2267 * TIPC_LOW_IMPORTANCE (2 MB) 2268 * TIPC_MEDIUM_IMPORTANCE (4 MB) 2269 * TIPC_HIGH_IMPORTANCE (8 MB) 2270 * TIPC_CRITICAL_IMPORTANCE (16 MB) 2271 * 2272 * Returns overload limit according to corresponding message importance 2273 */ 2274 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *skb) 2275 { 2276 struct tipc_sock *tsk = tipc_sk(sk); 2277 struct tipc_msg *hdr = buf_msg(skb); 2278 2279 if (unlikely(msg_in_group(hdr))) 2280 return READ_ONCE(sk->sk_rcvbuf); 2281 2282 if (unlikely(!msg_connected(hdr))) 2283 return READ_ONCE(sk->sk_rcvbuf) << msg_importance(hdr); 2284 2285 if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL)) 2286 return READ_ONCE(sk->sk_rcvbuf); 2287 2288 return FLOWCTL_MSG_LIM; 2289 } 2290 2291 /** 2292 * tipc_sk_filter_rcv - validate incoming message 2293 * @sk: socket 2294 * @skb: pointer to message. 2295 * 2296 * Enqueues message on receive queue if acceptable; optionally handles 2297 * disconnect indication for a connected socket. 2298 * 2299 * Called with socket lock already taken 2300 * 2301 */ 2302 static void tipc_sk_filter_rcv(struct sock *sk, struct sk_buff *skb, 2303 struct sk_buff_head *xmitq) 2304 { 2305 bool sk_conn = !tipc_sk_type_connectionless(sk); 2306 struct tipc_sock *tsk = tipc_sk(sk); 2307 struct tipc_group *grp = tsk->group; 2308 struct tipc_msg *hdr = buf_msg(skb); 2309 struct net *net = sock_net(sk); 2310 struct sk_buff_head inputq; 2311 int mtyp = msg_type(hdr); 2312 int limit, err = TIPC_OK; 2313 2314 trace_tipc_sk_filter_rcv(sk, skb, TIPC_DUMP_ALL, " "); 2315 TIPC_SKB_CB(skb)->bytes_read = 0; 2316 __skb_queue_head_init(&inputq); 2317 __skb_queue_tail(&inputq, skb); 2318 2319 if (unlikely(!msg_isdata(hdr))) 2320 tipc_sk_proto_rcv(sk, &inputq, xmitq); 2321 2322 if (unlikely(grp)) 2323 tipc_group_filter_msg(grp, &inputq, xmitq); 2324 2325 if (unlikely(!grp) && mtyp == TIPC_MCAST_MSG) 2326 tipc_mcast_filter_msg(net, &tsk->mc_method.deferredq, &inputq); 2327 2328 /* Validate and add to receive buffer if there is space */ 2329 while ((skb = __skb_dequeue(&inputq))) { 2330 hdr = buf_msg(skb); 2331 limit = rcvbuf_limit(sk, skb); 2332 if ((sk_conn && !tipc_sk_filter_connect(tsk, skb, xmitq)) || 2333 (!sk_conn && msg_connected(hdr)) || 2334 (!grp && msg_in_group(hdr))) 2335 err = TIPC_ERR_NO_PORT; 2336 else if (sk_rmem_alloc_get(sk) + skb->truesize >= limit) { 2337 trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL, 2338 "err_overload2!"); 2339 atomic_inc(&sk->sk_drops); 2340 err = TIPC_ERR_OVERLOAD; 2341 } 2342 2343 if (unlikely(err)) { 2344 if (tipc_msg_reverse(tipc_own_addr(net), &skb, err)) { 2345 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE, 2346 "@filter_rcv!"); 2347 __skb_queue_tail(xmitq, skb); 2348 } 2349 err = TIPC_OK; 2350 continue; 2351 } 2352 __skb_queue_tail(&sk->sk_receive_queue, skb); 2353 skb_set_owner_r(skb, sk); 2354 trace_tipc_sk_overlimit2(sk, skb, TIPC_DUMP_ALL, 2355 "rcvq >90% allocated!"); 2356 sk->sk_data_ready(sk); 2357 } 2358 } 2359 2360 /** 2361 * tipc_sk_backlog_rcv - handle incoming message from backlog queue 2362 * @sk: socket 2363 * @skb: message 2364 * 2365 * Caller must hold socket lock 2366 */ 2367 static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb) 2368 { 2369 unsigned int before = sk_rmem_alloc_get(sk); 2370 struct sk_buff_head xmitq; 2371 unsigned int added; 2372 2373 __skb_queue_head_init(&xmitq); 2374 2375 tipc_sk_filter_rcv(sk, skb, &xmitq); 2376 added = sk_rmem_alloc_get(sk) - before; 2377 atomic_add(added, &tipc_sk(sk)->dupl_rcvcnt); 2378 2379 /* Send pending response/rejected messages, if any */ 2380 tipc_node_distr_xmit(sock_net(sk), &xmitq); 2381 return 0; 2382 } 2383 2384 /** 2385 * tipc_sk_enqueue - extract all buffers with destination 'dport' from 2386 * inputq and try adding them to socket or backlog queue 2387 * @inputq: list of incoming buffers with potentially different destinations 2388 * @sk: socket where the buffers should be enqueued 2389 * @dport: port number for the socket 2390 * 2391 * Caller must hold socket lock 2392 */ 2393 static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk, 2394 u32 dport, struct sk_buff_head *xmitq) 2395 { 2396 unsigned long time_limit = jiffies + 2; 2397 struct sk_buff *skb; 2398 unsigned int lim; 2399 atomic_t *dcnt; 2400 u32 onode; 2401 2402 while (skb_queue_len(inputq)) { 2403 if (unlikely(time_after_eq(jiffies, time_limit))) 2404 return; 2405 2406 skb = tipc_skb_dequeue(inputq, dport); 2407 if (unlikely(!skb)) 2408 return; 2409 2410 /* Add message directly to receive queue if possible */ 2411 if (!sock_owned_by_user(sk)) { 2412 tipc_sk_filter_rcv(sk, skb, xmitq); 2413 continue; 2414 } 2415 2416 /* Try backlog, compensating for double-counted bytes */ 2417 dcnt = &tipc_sk(sk)->dupl_rcvcnt; 2418 if (!sk->sk_backlog.len) 2419 atomic_set(dcnt, 0); 2420 lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt); 2421 if (likely(!sk_add_backlog(sk, skb, lim))) { 2422 trace_tipc_sk_overlimit1(sk, skb, TIPC_DUMP_ALL, 2423 "bklg & rcvq >90% allocated!"); 2424 continue; 2425 } 2426 2427 trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL, "err_overload!"); 2428 /* Overload => reject message back to sender */ 2429 onode = tipc_own_addr(sock_net(sk)); 2430 atomic_inc(&sk->sk_drops); 2431 if (tipc_msg_reverse(onode, &skb, TIPC_ERR_OVERLOAD)) { 2432 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_ALL, 2433 "@sk_enqueue!"); 2434 __skb_queue_tail(xmitq, skb); 2435 } 2436 break; 2437 } 2438 } 2439 2440 /** 2441 * tipc_sk_rcv - handle a chain of incoming buffers 2442 * @inputq: buffer list containing the buffers 2443 * Consumes all buffers in list until inputq is empty 2444 * Note: may be called in multiple threads referring to the same queue 2445 */ 2446 void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq) 2447 { 2448 struct sk_buff_head xmitq; 2449 u32 dnode, dport = 0; 2450 int err; 2451 struct tipc_sock *tsk; 2452 struct sock *sk; 2453 struct sk_buff *skb; 2454 2455 __skb_queue_head_init(&xmitq); 2456 while (skb_queue_len(inputq)) { 2457 dport = tipc_skb_peek_port(inputq, dport); 2458 tsk = tipc_sk_lookup(net, dport); 2459 2460 if (likely(tsk)) { 2461 sk = &tsk->sk; 2462 if (likely(spin_trylock_bh(&sk->sk_lock.slock))) { 2463 tipc_sk_enqueue(inputq, sk, dport, &xmitq); 2464 spin_unlock_bh(&sk->sk_lock.slock); 2465 } 2466 /* Send pending response/rejected messages, if any */ 2467 tipc_node_distr_xmit(sock_net(sk), &xmitq); 2468 sock_put(sk); 2469 continue; 2470 } 2471 /* No destination socket => dequeue skb if still there */ 2472 skb = tipc_skb_dequeue(inputq, dport); 2473 if (!skb) 2474 return; 2475 2476 /* Try secondary lookup if unresolved named message */ 2477 err = TIPC_ERR_NO_PORT; 2478 if (tipc_msg_lookup_dest(net, skb, &err)) 2479 goto xmit; 2480 2481 /* Prepare for message rejection */ 2482 if (!tipc_msg_reverse(tipc_own_addr(net), &skb, err)) 2483 continue; 2484 2485 trace_tipc_sk_rej_msg(NULL, skb, TIPC_DUMP_NONE, "@sk_rcv!"); 2486 xmit: 2487 dnode = msg_destnode(buf_msg(skb)); 2488 tipc_node_xmit_skb(net, skb, dnode, dport); 2489 } 2490 } 2491 2492 static int tipc_wait_for_connect(struct socket *sock, long *timeo_p) 2493 { 2494 DEFINE_WAIT_FUNC(wait, woken_wake_function); 2495 struct sock *sk = sock->sk; 2496 int done; 2497 2498 do { 2499 int err = sock_error(sk); 2500 if (err) 2501 return err; 2502 if (!*timeo_p) 2503 return -ETIMEDOUT; 2504 if (signal_pending(current)) 2505 return sock_intr_errno(*timeo_p); 2506 if (sk->sk_state == TIPC_DISCONNECTING) 2507 break; 2508 2509 add_wait_queue(sk_sleep(sk), &wait); 2510 done = sk_wait_event(sk, timeo_p, tipc_sk_connected(sk), 2511 &wait); 2512 remove_wait_queue(sk_sleep(sk), &wait); 2513 } while (!done); 2514 return 0; 2515 } 2516 2517 static bool tipc_sockaddr_is_sane(struct sockaddr_tipc *addr) 2518 { 2519 if (addr->family != AF_TIPC) 2520 return false; 2521 if (addr->addrtype == TIPC_SERVICE_RANGE) 2522 return (addr->addr.nameseq.lower <= addr->addr.nameseq.upper); 2523 return (addr->addrtype == TIPC_SERVICE_ADDR || 2524 addr->addrtype == TIPC_SOCKET_ADDR); 2525 } 2526 2527 /** 2528 * tipc_connect - establish a connection to another TIPC port 2529 * @sock: socket structure 2530 * @dest: socket address for destination port 2531 * @destlen: size of socket address data structure 2532 * @flags: file-related flags associated with socket 2533 * 2534 * Returns 0 on success, errno otherwise 2535 */ 2536 static int tipc_connect(struct socket *sock, struct sockaddr *dest, 2537 int destlen, int flags) 2538 { 2539 struct sock *sk = sock->sk; 2540 struct tipc_sock *tsk = tipc_sk(sk); 2541 struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest; 2542 struct msghdr m = {NULL,}; 2543 long timeout = (flags & O_NONBLOCK) ? 0 : tsk->conn_timeout; 2544 int previous; 2545 int res = 0; 2546 2547 if (destlen != sizeof(struct sockaddr_tipc)) 2548 return -EINVAL; 2549 2550 lock_sock(sk); 2551 2552 if (tsk->group) { 2553 res = -EINVAL; 2554 goto exit; 2555 } 2556 2557 if (dst->family == AF_UNSPEC) { 2558 memset(&tsk->peer, 0, sizeof(struct sockaddr_tipc)); 2559 if (!tipc_sk_type_connectionless(sk)) 2560 res = -EINVAL; 2561 goto exit; 2562 } 2563 if (!tipc_sockaddr_is_sane(dst)) { 2564 res = -EINVAL; 2565 goto exit; 2566 } 2567 /* DGRAM/RDM connect(), just save the destaddr */ 2568 if (tipc_sk_type_connectionless(sk)) { 2569 memcpy(&tsk->peer, dest, destlen); 2570 goto exit; 2571 } else if (dst->addrtype == TIPC_SERVICE_RANGE) { 2572 res = -EINVAL; 2573 goto exit; 2574 } 2575 2576 previous = sk->sk_state; 2577 2578 switch (sk->sk_state) { 2579 case TIPC_OPEN: 2580 /* Send a 'SYN-' to destination */ 2581 m.msg_name = dest; 2582 m.msg_namelen = destlen; 2583 2584 /* If connect is in non-blocking case, set MSG_DONTWAIT to 2585 * indicate send_msg() is never blocked. 2586 */ 2587 if (!timeout) 2588 m.msg_flags = MSG_DONTWAIT; 2589 2590 res = __tipc_sendmsg(sock, &m, 0); 2591 if ((res < 0) && (res != -EWOULDBLOCK)) 2592 goto exit; 2593 2594 /* Just entered TIPC_CONNECTING state; the only 2595 * difference is that return value in non-blocking 2596 * case is EINPROGRESS, rather than EALREADY. 2597 */ 2598 res = -EINPROGRESS; 2599 fallthrough; 2600 case TIPC_CONNECTING: 2601 if (!timeout) { 2602 if (previous == TIPC_CONNECTING) 2603 res = -EALREADY; 2604 goto exit; 2605 } 2606 timeout = msecs_to_jiffies(timeout); 2607 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */ 2608 res = tipc_wait_for_connect(sock, &timeout); 2609 break; 2610 case TIPC_ESTABLISHED: 2611 res = -EISCONN; 2612 break; 2613 default: 2614 res = -EINVAL; 2615 } 2616 2617 exit: 2618 release_sock(sk); 2619 return res; 2620 } 2621 2622 /** 2623 * tipc_listen - allow socket to listen for incoming connections 2624 * @sock: socket structure 2625 * @len: (unused) 2626 * 2627 * Returns 0 on success, errno otherwise 2628 */ 2629 static int tipc_listen(struct socket *sock, int len) 2630 { 2631 struct sock *sk = sock->sk; 2632 int res; 2633 2634 lock_sock(sk); 2635 res = tipc_set_sk_state(sk, TIPC_LISTEN); 2636 release_sock(sk); 2637 2638 return res; 2639 } 2640 2641 static int tipc_wait_for_accept(struct socket *sock, long timeo) 2642 { 2643 struct sock *sk = sock->sk; 2644 DEFINE_WAIT(wait); 2645 int err; 2646 2647 /* True wake-one mechanism for incoming connections: only 2648 * one process gets woken up, not the 'whole herd'. 2649 * Since we do not 'race & poll' for established sockets 2650 * anymore, the common case will execute the loop only once. 2651 */ 2652 for (;;) { 2653 prepare_to_wait_exclusive(sk_sleep(sk), &wait, 2654 TASK_INTERRUPTIBLE); 2655 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) { 2656 release_sock(sk); 2657 timeo = schedule_timeout(timeo); 2658 lock_sock(sk); 2659 } 2660 err = 0; 2661 if (!skb_queue_empty(&sk->sk_receive_queue)) 2662 break; 2663 err = -EAGAIN; 2664 if (!timeo) 2665 break; 2666 err = sock_intr_errno(timeo); 2667 if (signal_pending(current)) 2668 break; 2669 } 2670 finish_wait(sk_sleep(sk), &wait); 2671 return err; 2672 } 2673 2674 /** 2675 * tipc_accept - wait for connection request 2676 * @sock: listening socket 2677 * @new_sock: new socket that is to be connected 2678 * @flags: file-related flags associated with socket 2679 * 2680 * Returns 0 on success, errno otherwise 2681 */ 2682 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags, 2683 bool kern) 2684 { 2685 struct sock *new_sk, *sk = sock->sk; 2686 struct sk_buff *buf; 2687 struct tipc_sock *new_tsock; 2688 struct tipc_msg *msg; 2689 long timeo; 2690 int res; 2691 2692 lock_sock(sk); 2693 2694 if (sk->sk_state != TIPC_LISTEN) { 2695 res = -EINVAL; 2696 goto exit; 2697 } 2698 timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK); 2699 res = tipc_wait_for_accept(sock, timeo); 2700 if (res) 2701 goto exit; 2702 2703 buf = skb_peek(&sk->sk_receive_queue); 2704 2705 res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, kern); 2706 if (res) 2707 goto exit; 2708 security_sk_clone(sock->sk, new_sock->sk); 2709 2710 new_sk = new_sock->sk; 2711 new_tsock = tipc_sk(new_sk); 2712 msg = buf_msg(buf); 2713 2714 /* we lock on new_sk; but lockdep sees the lock on sk */ 2715 lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING); 2716 2717 /* 2718 * Reject any stray messages received by new socket 2719 * before the socket lock was taken (very, very unlikely) 2720 */ 2721 tsk_rej_rx_queue(new_sk, TIPC_ERR_NO_PORT); 2722 2723 /* Connect new socket to it's peer */ 2724 tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg)); 2725 2726 tsk_set_importance(new_sk, msg_importance(msg)); 2727 if (msg_named(msg)) { 2728 new_tsock->conn_type = msg_nametype(msg); 2729 new_tsock->conn_instance = msg_nameinst(msg); 2730 } 2731 2732 /* 2733 * Respond to 'SYN-' by discarding it & returning 'ACK'-. 2734 * Respond to 'SYN+' by queuing it on new socket. 2735 */ 2736 if (!msg_data_sz(msg)) { 2737 struct msghdr m = {NULL,}; 2738 2739 tsk_advance_rx_queue(sk); 2740 __tipc_sendstream(new_sock, &m, 0); 2741 } else { 2742 __skb_dequeue(&sk->sk_receive_queue); 2743 __skb_queue_head(&new_sk->sk_receive_queue, buf); 2744 skb_set_owner_r(buf, new_sk); 2745 } 2746 release_sock(new_sk); 2747 exit: 2748 release_sock(sk); 2749 return res; 2750 } 2751 2752 /** 2753 * tipc_shutdown - shutdown socket connection 2754 * @sock: socket structure 2755 * @how: direction to close (must be SHUT_RDWR) 2756 * 2757 * Terminates connection (if necessary), then purges socket's receive queue. 2758 * 2759 * Returns 0 on success, errno otherwise 2760 */ 2761 static int tipc_shutdown(struct socket *sock, int how) 2762 { 2763 struct sock *sk = sock->sk; 2764 int res; 2765 2766 if (how != SHUT_RDWR) 2767 return -EINVAL; 2768 2769 lock_sock(sk); 2770 2771 trace_tipc_sk_shutdown(sk, NULL, TIPC_DUMP_ALL, " "); 2772 __tipc_shutdown(sock, TIPC_CONN_SHUTDOWN); 2773 if (tipc_sk_type_connectionless(sk)) 2774 sk->sk_shutdown = SHUTDOWN_MASK; 2775 else 2776 sk->sk_shutdown = SEND_SHUTDOWN; 2777 2778 if (sk->sk_state == TIPC_DISCONNECTING) { 2779 /* Discard any unreceived messages */ 2780 __skb_queue_purge(&sk->sk_receive_queue); 2781 2782 res = 0; 2783 } else { 2784 res = -ENOTCONN; 2785 } 2786 /* Wake up anyone sleeping in poll. */ 2787 sk->sk_state_change(sk); 2788 2789 release_sock(sk); 2790 return res; 2791 } 2792 2793 static void tipc_sk_check_probing_state(struct sock *sk, 2794 struct sk_buff_head *list) 2795 { 2796 struct tipc_sock *tsk = tipc_sk(sk); 2797 u32 pnode = tsk_peer_node(tsk); 2798 u32 pport = tsk_peer_port(tsk); 2799 u32 self = tsk_own_node(tsk); 2800 u32 oport = tsk->portid; 2801 struct sk_buff *skb; 2802 2803 if (tsk->probe_unacked) { 2804 tipc_set_sk_state(sk, TIPC_DISCONNECTING); 2805 sk->sk_err = ECONNABORTED; 2806 tipc_node_remove_conn(sock_net(sk), pnode, pport); 2807 sk->sk_state_change(sk); 2808 return; 2809 } 2810 /* Prepare new probe */ 2811 skb = tipc_msg_create(CONN_MANAGER, CONN_PROBE, INT_H_SIZE, 0, 2812 pnode, self, pport, oport, TIPC_OK); 2813 if (skb) 2814 __skb_queue_tail(list, skb); 2815 tsk->probe_unacked = true; 2816 sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV); 2817 } 2818 2819 static void tipc_sk_retry_connect(struct sock *sk, struct sk_buff_head *list) 2820 { 2821 struct tipc_sock *tsk = tipc_sk(sk); 2822 2823 /* Try again later if dest link is congested */ 2824 if (tsk->cong_link_cnt) { 2825 sk_reset_timer(sk, &sk->sk_timer, msecs_to_jiffies(100)); 2826 return; 2827 } 2828 /* Prepare SYN for retransmit */ 2829 tipc_msg_skb_clone(&sk->sk_write_queue, list); 2830 } 2831 2832 static void tipc_sk_timeout(struct timer_list *t) 2833 { 2834 struct sock *sk = from_timer(sk, t, sk_timer); 2835 struct tipc_sock *tsk = tipc_sk(sk); 2836 u32 pnode = tsk_peer_node(tsk); 2837 struct sk_buff_head list; 2838 int rc = 0; 2839 2840 __skb_queue_head_init(&list); 2841 bh_lock_sock(sk); 2842 2843 /* Try again later if socket is busy */ 2844 if (sock_owned_by_user(sk)) { 2845 sk_reset_timer(sk, &sk->sk_timer, jiffies + HZ / 20); 2846 bh_unlock_sock(sk); 2847 sock_put(sk); 2848 return; 2849 } 2850 2851 if (sk->sk_state == TIPC_ESTABLISHED) 2852 tipc_sk_check_probing_state(sk, &list); 2853 else if (sk->sk_state == TIPC_CONNECTING) 2854 tipc_sk_retry_connect(sk, &list); 2855 2856 bh_unlock_sock(sk); 2857 2858 if (!skb_queue_empty(&list)) 2859 rc = tipc_node_xmit(sock_net(sk), &list, pnode, tsk->portid); 2860 2861 /* SYN messages may cause link congestion */ 2862 if (rc == -ELINKCONG) { 2863 tipc_dest_push(&tsk->cong_links, pnode, 0); 2864 tsk->cong_link_cnt = 1; 2865 } 2866 sock_put(sk); 2867 } 2868 2869 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope, 2870 struct tipc_name_seq const *seq) 2871 { 2872 struct sock *sk = &tsk->sk; 2873 struct net *net = sock_net(sk); 2874 struct publication *publ; 2875 u32 key; 2876 2877 if (scope != TIPC_NODE_SCOPE) 2878 scope = TIPC_CLUSTER_SCOPE; 2879 2880 if (tipc_sk_connected(sk)) 2881 return -EINVAL; 2882 key = tsk->portid + tsk->pub_count + 1; 2883 if (key == tsk->portid) 2884 return -EADDRINUSE; 2885 2886 publ = tipc_nametbl_publish(net, seq->type, seq->lower, seq->upper, 2887 scope, tsk->portid, key); 2888 if (unlikely(!publ)) 2889 return -EINVAL; 2890 2891 list_add(&publ->binding_sock, &tsk->publications); 2892 tsk->pub_count++; 2893 tsk->published = 1; 2894 return 0; 2895 } 2896 2897 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope, 2898 struct tipc_name_seq const *seq) 2899 { 2900 struct net *net = sock_net(&tsk->sk); 2901 struct publication *publ; 2902 struct publication *safe; 2903 int rc = -EINVAL; 2904 2905 if (scope != TIPC_NODE_SCOPE) 2906 scope = TIPC_CLUSTER_SCOPE; 2907 2908 list_for_each_entry_safe(publ, safe, &tsk->publications, binding_sock) { 2909 if (seq) { 2910 if (publ->scope != scope) 2911 continue; 2912 if (publ->type != seq->type) 2913 continue; 2914 if (publ->lower != seq->lower) 2915 continue; 2916 if (publ->upper != seq->upper) 2917 break; 2918 tipc_nametbl_withdraw(net, publ->type, publ->lower, 2919 publ->upper, publ->key); 2920 rc = 0; 2921 break; 2922 } 2923 tipc_nametbl_withdraw(net, publ->type, publ->lower, 2924 publ->upper, publ->key); 2925 rc = 0; 2926 } 2927 if (list_empty(&tsk->publications)) 2928 tsk->published = 0; 2929 return rc; 2930 } 2931 2932 /* tipc_sk_reinit: set non-zero address in all existing sockets 2933 * when we go from standalone to network mode. 2934 */ 2935 void tipc_sk_reinit(struct net *net) 2936 { 2937 struct tipc_net *tn = net_generic(net, tipc_net_id); 2938 struct rhashtable_iter iter; 2939 struct tipc_sock *tsk; 2940 struct tipc_msg *msg; 2941 2942 rhashtable_walk_enter(&tn->sk_rht, &iter); 2943 2944 do { 2945 rhashtable_walk_start(&iter); 2946 2947 while ((tsk = rhashtable_walk_next(&iter)) && !IS_ERR(tsk)) { 2948 sock_hold(&tsk->sk); 2949 rhashtable_walk_stop(&iter); 2950 lock_sock(&tsk->sk); 2951 msg = &tsk->phdr; 2952 msg_set_prevnode(msg, tipc_own_addr(net)); 2953 msg_set_orignode(msg, tipc_own_addr(net)); 2954 release_sock(&tsk->sk); 2955 rhashtable_walk_start(&iter); 2956 sock_put(&tsk->sk); 2957 } 2958 2959 rhashtable_walk_stop(&iter); 2960 } while (tsk == ERR_PTR(-EAGAIN)); 2961 2962 rhashtable_walk_exit(&iter); 2963 } 2964 2965 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid) 2966 { 2967 struct tipc_net *tn = net_generic(net, tipc_net_id); 2968 struct tipc_sock *tsk; 2969 2970 rcu_read_lock(); 2971 tsk = rhashtable_lookup(&tn->sk_rht, &portid, tsk_rht_params); 2972 if (tsk) 2973 sock_hold(&tsk->sk); 2974 rcu_read_unlock(); 2975 2976 return tsk; 2977 } 2978 2979 static int tipc_sk_insert(struct tipc_sock *tsk) 2980 { 2981 struct sock *sk = &tsk->sk; 2982 struct net *net = sock_net(sk); 2983 struct tipc_net *tn = net_generic(net, tipc_net_id); 2984 u32 remaining = (TIPC_MAX_PORT - TIPC_MIN_PORT) + 1; 2985 u32 portid = prandom_u32() % remaining + TIPC_MIN_PORT; 2986 2987 while (remaining--) { 2988 portid++; 2989 if ((portid < TIPC_MIN_PORT) || (portid > TIPC_MAX_PORT)) 2990 portid = TIPC_MIN_PORT; 2991 tsk->portid = portid; 2992 sock_hold(&tsk->sk); 2993 if (!rhashtable_lookup_insert_fast(&tn->sk_rht, &tsk->node, 2994 tsk_rht_params)) 2995 return 0; 2996 sock_put(&tsk->sk); 2997 } 2998 2999 return -1; 3000 } 3001 3002 static void tipc_sk_remove(struct tipc_sock *tsk) 3003 { 3004 struct sock *sk = &tsk->sk; 3005 struct tipc_net *tn = net_generic(sock_net(sk), tipc_net_id); 3006 3007 if (!rhashtable_remove_fast(&tn->sk_rht, &tsk->node, tsk_rht_params)) { 3008 WARN_ON(refcount_read(&sk->sk_refcnt) == 1); 3009 __sock_put(sk); 3010 } 3011 } 3012 3013 static const struct rhashtable_params tsk_rht_params = { 3014 .nelem_hint = 192, 3015 .head_offset = offsetof(struct tipc_sock, node), 3016 .key_offset = offsetof(struct tipc_sock, portid), 3017 .key_len = sizeof(u32), /* portid */ 3018 .max_size = 1048576, 3019 .min_size = 256, 3020 .automatic_shrinking = true, 3021 }; 3022 3023 int tipc_sk_rht_init(struct net *net) 3024 { 3025 struct tipc_net *tn = net_generic(net, tipc_net_id); 3026 3027 return rhashtable_init(&tn->sk_rht, &tsk_rht_params); 3028 } 3029 3030 void tipc_sk_rht_destroy(struct net *net) 3031 { 3032 struct tipc_net *tn = net_generic(net, tipc_net_id); 3033 3034 /* Wait for socket readers to complete */ 3035 synchronize_net(); 3036 3037 rhashtable_destroy(&tn->sk_rht); 3038 } 3039 3040 static int tipc_sk_join(struct tipc_sock *tsk, struct tipc_group_req *mreq) 3041 { 3042 struct net *net = sock_net(&tsk->sk); 3043 struct tipc_group *grp = tsk->group; 3044 struct tipc_msg *hdr = &tsk->phdr; 3045 struct tipc_name_seq seq; 3046 int rc; 3047 3048 if (mreq->type < TIPC_RESERVED_TYPES) 3049 return -EACCES; 3050 if (mreq->scope > TIPC_NODE_SCOPE) 3051 return -EINVAL; 3052 if (grp) 3053 return -EACCES; 3054 grp = tipc_group_create(net, tsk->portid, mreq, &tsk->group_is_open); 3055 if (!grp) 3056 return -ENOMEM; 3057 tsk->group = grp; 3058 msg_set_lookup_scope(hdr, mreq->scope); 3059 msg_set_nametype(hdr, mreq->type); 3060 msg_set_dest_droppable(hdr, true); 3061 seq.type = mreq->type; 3062 seq.lower = mreq->instance; 3063 seq.upper = seq.lower; 3064 tipc_nametbl_build_group(net, grp, mreq->type, mreq->scope); 3065 rc = tipc_sk_publish(tsk, mreq->scope, &seq); 3066 if (rc) { 3067 tipc_group_delete(net, grp); 3068 tsk->group = NULL; 3069 return rc; 3070 } 3071 /* Eliminate any risk that a broadcast overtakes sent JOINs */ 3072 tsk->mc_method.rcast = true; 3073 tsk->mc_method.mandatory = true; 3074 tipc_group_join(net, grp, &tsk->sk.sk_rcvbuf); 3075 return rc; 3076 } 3077 3078 static int tipc_sk_leave(struct tipc_sock *tsk) 3079 { 3080 struct net *net = sock_net(&tsk->sk); 3081 struct tipc_group *grp = tsk->group; 3082 struct tipc_name_seq seq; 3083 int scope; 3084 3085 if (!grp) 3086 return -EINVAL; 3087 tipc_group_self(grp, &seq, &scope); 3088 tipc_group_delete(net, grp); 3089 tsk->group = NULL; 3090 tipc_sk_withdraw(tsk, scope, &seq); 3091 return 0; 3092 } 3093 3094 /** 3095 * tipc_setsockopt - set socket option 3096 * @sock: socket structure 3097 * @lvl: option level 3098 * @opt: option identifier 3099 * @ov: pointer to new option value 3100 * @ol: length of option value 3101 * 3102 * For stream sockets only, accepts and ignores all IPPROTO_TCP options 3103 * (to ease compatibility). 3104 * 3105 * Returns 0 on success, errno otherwise 3106 */ 3107 static int tipc_setsockopt(struct socket *sock, int lvl, int opt, 3108 sockptr_t ov, unsigned int ol) 3109 { 3110 struct sock *sk = sock->sk; 3111 struct tipc_sock *tsk = tipc_sk(sk); 3112 struct tipc_group_req mreq; 3113 u32 value = 0; 3114 int res = 0; 3115 3116 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM)) 3117 return 0; 3118 if (lvl != SOL_TIPC) 3119 return -ENOPROTOOPT; 3120 3121 switch (opt) { 3122 case TIPC_IMPORTANCE: 3123 case TIPC_SRC_DROPPABLE: 3124 case TIPC_DEST_DROPPABLE: 3125 case TIPC_CONN_TIMEOUT: 3126 case TIPC_NODELAY: 3127 if (ol < sizeof(value)) 3128 return -EINVAL; 3129 if (copy_from_sockptr(&value, ov, sizeof(u32))) 3130 return -EFAULT; 3131 break; 3132 case TIPC_GROUP_JOIN: 3133 if (ol < sizeof(mreq)) 3134 return -EINVAL; 3135 if (copy_from_sockptr(&mreq, ov, sizeof(mreq))) 3136 return -EFAULT; 3137 break; 3138 default: 3139 if (!sockptr_is_null(ov) || ol) 3140 return -EINVAL; 3141 } 3142 3143 lock_sock(sk); 3144 3145 switch (opt) { 3146 case TIPC_IMPORTANCE: 3147 res = tsk_set_importance(sk, value); 3148 break; 3149 case TIPC_SRC_DROPPABLE: 3150 if (sock->type != SOCK_STREAM) 3151 tsk_set_unreliable(tsk, value); 3152 else 3153 res = -ENOPROTOOPT; 3154 break; 3155 case TIPC_DEST_DROPPABLE: 3156 tsk_set_unreturnable(tsk, value); 3157 break; 3158 case TIPC_CONN_TIMEOUT: 3159 tipc_sk(sk)->conn_timeout = value; 3160 break; 3161 case TIPC_MCAST_BROADCAST: 3162 tsk->mc_method.rcast = false; 3163 tsk->mc_method.mandatory = true; 3164 break; 3165 case TIPC_MCAST_REPLICAST: 3166 tsk->mc_method.rcast = true; 3167 tsk->mc_method.mandatory = true; 3168 break; 3169 case TIPC_GROUP_JOIN: 3170 res = tipc_sk_join(tsk, &mreq); 3171 break; 3172 case TIPC_GROUP_LEAVE: 3173 res = tipc_sk_leave(tsk); 3174 break; 3175 case TIPC_NODELAY: 3176 tsk->nodelay = !!value; 3177 tsk_set_nagle(tsk); 3178 break; 3179 default: 3180 res = -EINVAL; 3181 } 3182 3183 release_sock(sk); 3184 3185 return res; 3186 } 3187 3188 /** 3189 * tipc_getsockopt - get socket option 3190 * @sock: socket structure 3191 * @lvl: option level 3192 * @opt: option identifier 3193 * @ov: receptacle for option value 3194 * @ol: receptacle for length of option value 3195 * 3196 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options 3197 * (to ease compatibility). 3198 * 3199 * Returns 0 on success, errno otherwise 3200 */ 3201 static int tipc_getsockopt(struct socket *sock, int lvl, int opt, 3202 char __user *ov, int __user *ol) 3203 { 3204 struct sock *sk = sock->sk; 3205 struct tipc_sock *tsk = tipc_sk(sk); 3206 struct tipc_name_seq seq; 3207 int len, scope; 3208 u32 value; 3209 int res; 3210 3211 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM)) 3212 return put_user(0, ol); 3213 if (lvl != SOL_TIPC) 3214 return -ENOPROTOOPT; 3215 res = get_user(len, ol); 3216 if (res) 3217 return res; 3218 3219 lock_sock(sk); 3220 3221 switch (opt) { 3222 case TIPC_IMPORTANCE: 3223 value = tsk_importance(tsk); 3224 break; 3225 case TIPC_SRC_DROPPABLE: 3226 value = tsk_unreliable(tsk); 3227 break; 3228 case TIPC_DEST_DROPPABLE: 3229 value = tsk_unreturnable(tsk); 3230 break; 3231 case TIPC_CONN_TIMEOUT: 3232 value = tsk->conn_timeout; 3233 /* no need to set "res", since already 0 at this point */ 3234 break; 3235 case TIPC_NODE_RECVQ_DEPTH: 3236 value = 0; /* was tipc_queue_size, now obsolete */ 3237 break; 3238 case TIPC_SOCK_RECVQ_DEPTH: 3239 value = skb_queue_len(&sk->sk_receive_queue); 3240 break; 3241 case TIPC_SOCK_RECVQ_USED: 3242 value = sk_rmem_alloc_get(sk); 3243 break; 3244 case TIPC_GROUP_JOIN: 3245 seq.type = 0; 3246 if (tsk->group) 3247 tipc_group_self(tsk->group, &seq, &scope); 3248 value = seq.type; 3249 break; 3250 default: 3251 res = -EINVAL; 3252 } 3253 3254 release_sock(sk); 3255 3256 if (res) 3257 return res; /* "get" failed */ 3258 3259 if (len < sizeof(value)) 3260 return -EINVAL; 3261 3262 if (copy_to_user(ov, &value, sizeof(value))) 3263 return -EFAULT; 3264 3265 return put_user(sizeof(value), ol); 3266 } 3267 3268 static int tipc_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg) 3269 { 3270 struct net *net = sock_net(sock->sk); 3271 struct tipc_sioc_nodeid_req nr = {0}; 3272 struct tipc_sioc_ln_req lnr; 3273 void __user *argp = (void __user *)arg; 3274 3275 switch (cmd) { 3276 case SIOCGETLINKNAME: 3277 if (copy_from_user(&lnr, argp, sizeof(lnr))) 3278 return -EFAULT; 3279 if (!tipc_node_get_linkname(net, 3280 lnr.bearer_id & 0xffff, lnr.peer, 3281 lnr.linkname, TIPC_MAX_LINK_NAME)) { 3282 if (copy_to_user(argp, &lnr, sizeof(lnr))) 3283 return -EFAULT; 3284 return 0; 3285 } 3286 return -EADDRNOTAVAIL; 3287 case SIOCGETNODEID: 3288 if (copy_from_user(&nr, argp, sizeof(nr))) 3289 return -EFAULT; 3290 if (!tipc_node_get_id(net, nr.peer, nr.node_id)) 3291 return -EADDRNOTAVAIL; 3292 if (copy_to_user(argp, &nr, sizeof(nr))) 3293 return -EFAULT; 3294 return 0; 3295 default: 3296 return -ENOIOCTLCMD; 3297 } 3298 } 3299 3300 static int tipc_socketpair(struct socket *sock1, struct socket *sock2) 3301 { 3302 struct tipc_sock *tsk2 = tipc_sk(sock2->sk); 3303 struct tipc_sock *tsk1 = tipc_sk(sock1->sk); 3304 u32 onode = tipc_own_addr(sock_net(sock1->sk)); 3305 3306 tsk1->peer.family = AF_TIPC; 3307 tsk1->peer.addrtype = TIPC_ADDR_ID; 3308 tsk1->peer.scope = TIPC_NODE_SCOPE; 3309 tsk1->peer.addr.id.ref = tsk2->portid; 3310 tsk1->peer.addr.id.node = onode; 3311 tsk2->peer.family = AF_TIPC; 3312 tsk2->peer.addrtype = TIPC_ADDR_ID; 3313 tsk2->peer.scope = TIPC_NODE_SCOPE; 3314 tsk2->peer.addr.id.ref = tsk1->portid; 3315 tsk2->peer.addr.id.node = onode; 3316 3317 tipc_sk_finish_conn(tsk1, tsk2->portid, onode); 3318 tipc_sk_finish_conn(tsk2, tsk1->portid, onode); 3319 return 0; 3320 } 3321 3322 /* Protocol switches for the various types of TIPC sockets */ 3323 3324 static const struct proto_ops msg_ops = { 3325 .owner = THIS_MODULE, 3326 .family = AF_TIPC, 3327 .release = tipc_release, 3328 .bind = tipc_bind, 3329 .connect = tipc_connect, 3330 .socketpair = tipc_socketpair, 3331 .accept = sock_no_accept, 3332 .getname = tipc_getname, 3333 .poll = tipc_poll, 3334 .ioctl = tipc_ioctl, 3335 .listen = sock_no_listen, 3336 .shutdown = tipc_shutdown, 3337 .setsockopt = tipc_setsockopt, 3338 .getsockopt = tipc_getsockopt, 3339 .sendmsg = tipc_sendmsg, 3340 .recvmsg = tipc_recvmsg, 3341 .mmap = sock_no_mmap, 3342 .sendpage = sock_no_sendpage 3343 }; 3344 3345 static const struct proto_ops packet_ops = { 3346 .owner = THIS_MODULE, 3347 .family = AF_TIPC, 3348 .release = tipc_release, 3349 .bind = tipc_bind, 3350 .connect = tipc_connect, 3351 .socketpair = tipc_socketpair, 3352 .accept = tipc_accept, 3353 .getname = tipc_getname, 3354 .poll = tipc_poll, 3355 .ioctl = tipc_ioctl, 3356 .listen = tipc_listen, 3357 .shutdown = tipc_shutdown, 3358 .setsockopt = tipc_setsockopt, 3359 .getsockopt = tipc_getsockopt, 3360 .sendmsg = tipc_send_packet, 3361 .recvmsg = tipc_recvmsg, 3362 .mmap = sock_no_mmap, 3363 .sendpage = sock_no_sendpage 3364 }; 3365 3366 static const struct proto_ops stream_ops = { 3367 .owner = THIS_MODULE, 3368 .family = AF_TIPC, 3369 .release = tipc_release, 3370 .bind = tipc_bind, 3371 .connect = tipc_connect, 3372 .socketpair = tipc_socketpair, 3373 .accept = tipc_accept, 3374 .getname = tipc_getname, 3375 .poll = tipc_poll, 3376 .ioctl = tipc_ioctl, 3377 .listen = tipc_listen, 3378 .shutdown = tipc_shutdown, 3379 .setsockopt = tipc_setsockopt, 3380 .getsockopt = tipc_getsockopt, 3381 .sendmsg = tipc_sendstream, 3382 .recvmsg = tipc_recvstream, 3383 .mmap = sock_no_mmap, 3384 .sendpage = sock_no_sendpage 3385 }; 3386 3387 static const struct net_proto_family tipc_family_ops = { 3388 .owner = THIS_MODULE, 3389 .family = AF_TIPC, 3390 .create = tipc_sk_create 3391 }; 3392 3393 static struct proto tipc_proto = { 3394 .name = "TIPC", 3395 .owner = THIS_MODULE, 3396 .obj_size = sizeof(struct tipc_sock), 3397 .sysctl_rmem = sysctl_tipc_rmem 3398 }; 3399 3400 /** 3401 * tipc_socket_init - initialize TIPC socket interface 3402 * 3403 * Returns 0 on success, errno otherwise 3404 */ 3405 int tipc_socket_init(void) 3406 { 3407 int res; 3408 3409 res = proto_register(&tipc_proto, 1); 3410 if (res) { 3411 pr_err("Failed to register TIPC protocol type\n"); 3412 goto out; 3413 } 3414 3415 res = sock_register(&tipc_family_ops); 3416 if (res) { 3417 pr_err("Failed to register TIPC socket type\n"); 3418 proto_unregister(&tipc_proto); 3419 goto out; 3420 } 3421 out: 3422 return res; 3423 } 3424 3425 /** 3426 * tipc_socket_stop - stop TIPC socket interface 3427 */ 3428 void tipc_socket_stop(void) 3429 { 3430 sock_unregister(tipc_family_ops.family); 3431 proto_unregister(&tipc_proto); 3432 } 3433 3434 /* Caller should hold socket lock for the passed tipc socket. */ 3435 static int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk) 3436 { 3437 u32 peer_node; 3438 u32 peer_port; 3439 struct nlattr *nest; 3440 3441 peer_node = tsk_peer_node(tsk); 3442 peer_port = tsk_peer_port(tsk); 3443 3444 nest = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_CON); 3445 if (!nest) 3446 return -EMSGSIZE; 3447 3448 if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node)) 3449 goto msg_full; 3450 if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port)) 3451 goto msg_full; 3452 3453 if (tsk->conn_type != 0) { 3454 if (nla_put_flag(skb, TIPC_NLA_CON_FLAG)) 3455 goto msg_full; 3456 if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, tsk->conn_type)) 3457 goto msg_full; 3458 if (nla_put_u32(skb, TIPC_NLA_CON_INST, tsk->conn_instance)) 3459 goto msg_full; 3460 } 3461 nla_nest_end(skb, nest); 3462 3463 return 0; 3464 3465 msg_full: 3466 nla_nest_cancel(skb, nest); 3467 3468 return -EMSGSIZE; 3469 } 3470 3471 static int __tipc_nl_add_sk_info(struct sk_buff *skb, struct tipc_sock 3472 *tsk) 3473 { 3474 struct net *net = sock_net(skb->sk); 3475 struct sock *sk = &tsk->sk; 3476 3477 if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->portid) || 3478 nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tipc_own_addr(net))) 3479 return -EMSGSIZE; 3480 3481 if (tipc_sk_connected(sk)) { 3482 if (__tipc_nl_add_sk_con(skb, tsk)) 3483 return -EMSGSIZE; 3484 } else if (!list_empty(&tsk->publications)) { 3485 if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL)) 3486 return -EMSGSIZE; 3487 } 3488 return 0; 3489 } 3490 3491 /* Caller should hold socket lock for the passed tipc socket. */ 3492 static int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb, 3493 struct tipc_sock *tsk) 3494 { 3495 struct nlattr *attrs; 3496 void *hdr; 3497 3498 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq, 3499 &tipc_genl_family, NLM_F_MULTI, TIPC_NL_SOCK_GET); 3500 if (!hdr) 3501 goto msg_cancel; 3502 3503 attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK); 3504 if (!attrs) 3505 goto genlmsg_cancel; 3506 3507 if (__tipc_nl_add_sk_info(skb, tsk)) 3508 goto attr_msg_cancel; 3509 3510 nla_nest_end(skb, attrs); 3511 genlmsg_end(skb, hdr); 3512 3513 return 0; 3514 3515 attr_msg_cancel: 3516 nla_nest_cancel(skb, attrs); 3517 genlmsg_cancel: 3518 genlmsg_cancel(skb, hdr); 3519 msg_cancel: 3520 return -EMSGSIZE; 3521 } 3522 3523 int tipc_nl_sk_walk(struct sk_buff *skb, struct netlink_callback *cb, 3524 int (*skb_handler)(struct sk_buff *skb, 3525 struct netlink_callback *cb, 3526 struct tipc_sock *tsk)) 3527 { 3528 struct rhashtable_iter *iter = (void *)cb->args[4]; 3529 struct tipc_sock *tsk; 3530 int err; 3531 3532 rhashtable_walk_start(iter); 3533 while ((tsk = rhashtable_walk_next(iter)) != NULL) { 3534 if (IS_ERR(tsk)) { 3535 err = PTR_ERR(tsk); 3536 if (err == -EAGAIN) { 3537 err = 0; 3538 continue; 3539 } 3540 break; 3541 } 3542 3543 sock_hold(&tsk->sk); 3544 rhashtable_walk_stop(iter); 3545 lock_sock(&tsk->sk); 3546 err = skb_handler(skb, cb, tsk); 3547 if (err) { 3548 release_sock(&tsk->sk); 3549 sock_put(&tsk->sk); 3550 goto out; 3551 } 3552 release_sock(&tsk->sk); 3553 rhashtable_walk_start(iter); 3554 sock_put(&tsk->sk); 3555 } 3556 rhashtable_walk_stop(iter); 3557 out: 3558 return skb->len; 3559 } 3560 EXPORT_SYMBOL(tipc_nl_sk_walk); 3561 3562 int tipc_dump_start(struct netlink_callback *cb) 3563 { 3564 return __tipc_dump_start(cb, sock_net(cb->skb->sk)); 3565 } 3566 EXPORT_SYMBOL(tipc_dump_start); 3567 3568 int __tipc_dump_start(struct netlink_callback *cb, struct net *net) 3569 { 3570 /* tipc_nl_name_table_dump() uses cb->args[0...3]. */ 3571 struct rhashtable_iter *iter = (void *)cb->args[4]; 3572 struct tipc_net *tn = tipc_net(net); 3573 3574 if (!iter) { 3575 iter = kmalloc(sizeof(*iter), GFP_KERNEL); 3576 if (!iter) 3577 return -ENOMEM; 3578 3579 cb->args[4] = (long)iter; 3580 } 3581 3582 rhashtable_walk_enter(&tn->sk_rht, iter); 3583 return 0; 3584 } 3585 3586 int tipc_dump_done(struct netlink_callback *cb) 3587 { 3588 struct rhashtable_iter *hti = (void *)cb->args[4]; 3589 3590 rhashtable_walk_exit(hti); 3591 kfree(hti); 3592 return 0; 3593 } 3594 EXPORT_SYMBOL(tipc_dump_done); 3595 3596 int tipc_sk_fill_sock_diag(struct sk_buff *skb, struct netlink_callback *cb, 3597 struct tipc_sock *tsk, u32 sk_filter_state, 3598 u64 (*tipc_diag_gen_cookie)(struct sock *sk)) 3599 { 3600 struct sock *sk = &tsk->sk; 3601 struct nlattr *attrs; 3602 struct nlattr *stat; 3603 3604 /*filter response w.r.t sk_state*/ 3605 if (!(sk_filter_state & (1 << sk->sk_state))) 3606 return 0; 3607 3608 attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK); 3609 if (!attrs) 3610 goto msg_cancel; 3611 3612 if (__tipc_nl_add_sk_info(skb, tsk)) 3613 goto attr_msg_cancel; 3614 3615 if (nla_put_u32(skb, TIPC_NLA_SOCK_TYPE, (u32)sk->sk_type) || 3616 nla_put_u32(skb, TIPC_NLA_SOCK_TIPC_STATE, (u32)sk->sk_state) || 3617 nla_put_u32(skb, TIPC_NLA_SOCK_INO, sock_i_ino(sk)) || 3618 nla_put_u32(skb, TIPC_NLA_SOCK_UID, 3619 from_kuid_munged(sk_user_ns(NETLINK_CB(cb->skb).sk), 3620 sock_i_uid(sk))) || 3621 nla_put_u64_64bit(skb, TIPC_NLA_SOCK_COOKIE, 3622 tipc_diag_gen_cookie(sk), 3623 TIPC_NLA_SOCK_PAD)) 3624 goto attr_msg_cancel; 3625 3626 stat = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_STAT); 3627 if (!stat) 3628 goto attr_msg_cancel; 3629 3630 if (nla_put_u32(skb, TIPC_NLA_SOCK_STAT_RCVQ, 3631 skb_queue_len(&sk->sk_receive_queue)) || 3632 nla_put_u32(skb, TIPC_NLA_SOCK_STAT_SENDQ, 3633 skb_queue_len(&sk->sk_write_queue)) || 3634 nla_put_u32(skb, TIPC_NLA_SOCK_STAT_DROP, 3635 atomic_read(&sk->sk_drops))) 3636 goto stat_msg_cancel; 3637 3638 if (tsk->cong_link_cnt && 3639 nla_put_flag(skb, TIPC_NLA_SOCK_STAT_LINK_CONG)) 3640 goto stat_msg_cancel; 3641 3642 if (tsk_conn_cong(tsk) && 3643 nla_put_flag(skb, TIPC_NLA_SOCK_STAT_CONN_CONG)) 3644 goto stat_msg_cancel; 3645 3646 nla_nest_end(skb, stat); 3647 3648 if (tsk->group) 3649 if (tipc_group_fill_sock_diag(tsk->group, skb)) 3650 goto stat_msg_cancel; 3651 3652 nla_nest_end(skb, attrs); 3653 3654 return 0; 3655 3656 stat_msg_cancel: 3657 nla_nest_cancel(skb, stat); 3658 attr_msg_cancel: 3659 nla_nest_cancel(skb, attrs); 3660 msg_cancel: 3661 return -EMSGSIZE; 3662 } 3663 EXPORT_SYMBOL(tipc_sk_fill_sock_diag); 3664 3665 int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb) 3666 { 3667 return tipc_nl_sk_walk(skb, cb, __tipc_nl_add_sk); 3668 } 3669 3670 /* Caller should hold socket lock for the passed tipc socket. */ 3671 static int __tipc_nl_add_sk_publ(struct sk_buff *skb, 3672 struct netlink_callback *cb, 3673 struct publication *publ) 3674 { 3675 void *hdr; 3676 struct nlattr *attrs; 3677 3678 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq, 3679 &tipc_genl_family, NLM_F_MULTI, TIPC_NL_PUBL_GET); 3680 if (!hdr) 3681 goto msg_cancel; 3682 3683 attrs = nla_nest_start_noflag(skb, TIPC_NLA_PUBL); 3684 if (!attrs) 3685 goto genlmsg_cancel; 3686 3687 if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key)) 3688 goto attr_msg_cancel; 3689 if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->type)) 3690 goto attr_msg_cancel; 3691 if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->lower)) 3692 goto attr_msg_cancel; 3693 if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->upper)) 3694 goto attr_msg_cancel; 3695 3696 nla_nest_end(skb, attrs); 3697 genlmsg_end(skb, hdr); 3698 3699 return 0; 3700 3701 attr_msg_cancel: 3702 nla_nest_cancel(skb, attrs); 3703 genlmsg_cancel: 3704 genlmsg_cancel(skb, hdr); 3705 msg_cancel: 3706 return -EMSGSIZE; 3707 } 3708 3709 /* Caller should hold socket lock for the passed tipc socket. */ 3710 static int __tipc_nl_list_sk_publ(struct sk_buff *skb, 3711 struct netlink_callback *cb, 3712 struct tipc_sock *tsk, u32 *last_publ) 3713 { 3714 int err; 3715 struct publication *p; 3716 3717 if (*last_publ) { 3718 list_for_each_entry(p, &tsk->publications, binding_sock) { 3719 if (p->key == *last_publ) 3720 break; 3721 } 3722 if (p->key != *last_publ) { 3723 /* We never set seq or call nl_dump_check_consistent() 3724 * this means that setting prev_seq here will cause the 3725 * consistence check to fail in the netlink callback 3726 * handler. Resulting in the last NLMSG_DONE message 3727 * having the NLM_F_DUMP_INTR flag set. 3728 */ 3729 cb->prev_seq = 1; 3730 *last_publ = 0; 3731 return -EPIPE; 3732 } 3733 } else { 3734 p = list_first_entry(&tsk->publications, struct publication, 3735 binding_sock); 3736 } 3737 3738 list_for_each_entry_from(p, &tsk->publications, binding_sock) { 3739 err = __tipc_nl_add_sk_publ(skb, cb, p); 3740 if (err) { 3741 *last_publ = p->key; 3742 return err; 3743 } 3744 } 3745 *last_publ = 0; 3746 3747 return 0; 3748 } 3749 3750 int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb) 3751 { 3752 int err; 3753 u32 tsk_portid = cb->args[0]; 3754 u32 last_publ = cb->args[1]; 3755 u32 done = cb->args[2]; 3756 struct net *net = sock_net(skb->sk); 3757 struct tipc_sock *tsk; 3758 3759 if (!tsk_portid) { 3760 struct nlattr **attrs = genl_dumpit_info(cb)->attrs; 3761 struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1]; 3762 3763 if (!attrs[TIPC_NLA_SOCK]) 3764 return -EINVAL; 3765 3766 err = nla_parse_nested_deprecated(sock, TIPC_NLA_SOCK_MAX, 3767 attrs[TIPC_NLA_SOCK], 3768 tipc_nl_sock_policy, NULL); 3769 if (err) 3770 return err; 3771 3772 if (!sock[TIPC_NLA_SOCK_REF]) 3773 return -EINVAL; 3774 3775 tsk_portid = nla_get_u32(sock[TIPC_NLA_SOCK_REF]); 3776 } 3777 3778 if (done) 3779 return 0; 3780 3781 tsk = tipc_sk_lookup(net, tsk_portid); 3782 if (!tsk) 3783 return -EINVAL; 3784 3785 lock_sock(&tsk->sk); 3786 err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ); 3787 if (!err) 3788 done = 1; 3789 release_sock(&tsk->sk); 3790 sock_put(&tsk->sk); 3791 3792 cb->args[0] = tsk_portid; 3793 cb->args[1] = last_publ; 3794 cb->args[2] = done; 3795 3796 return skb->len; 3797 } 3798 3799 /** 3800 * tipc_sk_filtering - check if a socket should be traced 3801 * @sk: the socket to be examined 3802 * @sysctl_tipc_sk_filter[]: the socket tuple for filtering, 3803 * (portid, sock type, name type, name lower, name upper) 3804 * 3805 * Returns true if the socket meets the socket tuple data 3806 * (value 0 = 'any') or when there is no tuple set (all = 0), 3807 * otherwise false 3808 */ 3809 bool tipc_sk_filtering(struct sock *sk) 3810 { 3811 struct tipc_sock *tsk; 3812 struct publication *p; 3813 u32 _port, _sktype, _type, _lower, _upper; 3814 u32 type = 0, lower = 0, upper = 0; 3815 3816 if (!sk) 3817 return true; 3818 3819 tsk = tipc_sk(sk); 3820 3821 _port = sysctl_tipc_sk_filter[0]; 3822 _sktype = sysctl_tipc_sk_filter[1]; 3823 _type = sysctl_tipc_sk_filter[2]; 3824 _lower = sysctl_tipc_sk_filter[3]; 3825 _upper = sysctl_tipc_sk_filter[4]; 3826 3827 if (!_port && !_sktype && !_type && !_lower && !_upper) 3828 return true; 3829 3830 if (_port) 3831 return (_port == tsk->portid); 3832 3833 if (_sktype && _sktype != sk->sk_type) 3834 return false; 3835 3836 if (tsk->published) { 3837 p = list_first_entry_or_null(&tsk->publications, 3838 struct publication, binding_sock); 3839 if (p) { 3840 type = p->type; 3841 lower = p->lower; 3842 upper = p->upper; 3843 } 3844 } 3845 3846 if (!tipc_sk_type_connectionless(sk)) { 3847 type = tsk->conn_type; 3848 lower = tsk->conn_instance; 3849 upper = tsk->conn_instance; 3850 } 3851 3852 if ((_type && _type != type) || (_lower && _lower != lower) || 3853 (_upper && _upper != upper)) 3854 return false; 3855 3856 return true; 3857 } 3858 3859 u32 tipc_sock_get_portid(struct sock *sk) 3860 { 3861 return (sk) ? (tipc_sk(sk))->portid : 0; 3862 } 3863 3864 /** 3865 * tipc_sk_overlimit1 - check if socket rx queue is about to be overloaded, 3866 * both the rcv and backlog queues are considered 3867 * @sk: tipc sk to be checked 3868 * @skb: tipc msg to be checked 3869 * 3870 * Returns true if the socket rx queue allocation is > 90%, otherwise false 3871 */ 3872 3873 bool tipc_sk_overlimit1(struct sock *sk, struct sk_buff *skb) 3874 { 3875 atomic_t *dcnt = &tipc_sk(sk)->dupl_rcvcnt; 3876 unsigned int lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt); 3877 unsigned int qsize = sk->sk_backlog.len + sk_rmem_alloc_get(sk); 3878 3879 return (qsize > lim * 90 / 100); 3880 } 3881 3882 /** 3883 * tipc_sk_overlimit2 - check if socket rx queue is about to be overloaded, 3884 * only the rcv queue is considered 3885 * @sk: tipc sk to be checked 3886 * @skb: tipc msg to be checked 3887 * 3888 * Returns true if the socket rx queue allocation is > 90%, otherwise false 3889 */ 3890 3891 bool tipc_sk_overlimit2(struct sock *sk, struct sk_buff *skb) 3892 { 3893 unsigned int lim = rcvbuf_limit(sk, skb); 3894 unsigned int qsize = sk_rmem_alloc_get(sk); 3895 3896 return (qsize > lim * 90 / 100); 3897 } 3898 3899 /** 3900 * tipc_sk_dump - dump TIPC socket 3901 * @sk: tipc sk to be dumped 3902 * @dqueues: bitmask to decide if any socket queue to be dumped? 3903 * - TIPC_DUMP_NONE: don't dump socket queues 3904 * - TIPC_DUMP_SK_SNDQ: dump socket send queue 3905 * - TIPC_DUMP_SK_RCVQ: dump socket rcv queue 3906 * - TIPC_DUMP_SK_BKLGQ: dump socket backlog queue 3907 * - TIPC_DUMP_ALL: dump all the socket queues above 3908 * @buf: returned buffer of dump data in format 3909 */ 3910 int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf) 3911 { 3912 int i = 0; 3913 size_t sz = (dqueues) ? SK_LMAX : SK_LMIN; 3914 struct tipc_sock *tsk; 3915 struct publication *p; 3916 bool tsk_connected; 3917 3918 if (!sk) { 3919 i += scnprintf(buf, sz, "sk data: (null)\n"); 3920 return i; 3921 } 3922 3923 tsk = tipc_sk(sk); 3924 tsk_connected = !tipc_sk_type_connectionless(sk); 3925 3926 i += scnprintf(buf, sz, "sk data: %u", sk->sk_type); 3927 i += scnprintf(buf + i, sz - i, " %d", sk->sk_state); 3928 i += scnprintf(buf + i, sz - i, " %x", tsk_own_node(tsk)); 3929 i += scnprintf(buf + i, sz - i, " %u", tsk->portid); 3930 i += scnprintf(buf + i, sz - i, " | %u", tsk_connected); 3931 if (tsk_connected) { 3932 i += scnprintf(buf + i, sz - i, " %x", tsk_peer_node(tsk)); 3933 i += scnprintf(buf + i, sz - i, " %u", tsk_peer_port(tsk)); 3934 i += scnprintf(buf + i, sz - i, " %u", tsk->conn_type); 3935 i += scnprintf(buf + i, sz - i, " %u", tsk->conn_instance); 3936 } 3937 i += scnprintf(buf + i, sz - i, " | %u", tsk->published); 3938 if (tsk->published) { 3939 p = list_first_entry_or_null(&tsk->publications, 3940 struct publication, binding_sock); 3941 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->type : 0); 3942 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->lower : 0); 3943 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->upper : 0); 3944 } 3945 i += scnprintf(buf + i, sz - i, " | %u", tsk->snd_win); 3946 i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_win); 3947 i += scnprintf(buf + i, sz - i, " %u", tsk->max_pkt); 3948 i += scnprintf(buf + i, sz - i, " %x", tsk->peer_caps); 3949 i += scnprintf(buf + i, sz - i, " %u", tsk->cong_link_cnt); 3950 i += scnprintf(buf + i, sz - i, " %u", tsk->snt_unacked); 3951 i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_unacked); 3952 i += scnprintf(buf + i, sz - i, " %u", atomic_read(&tsk->dupl_rcvcnt)); 3953 i += scnprintf(buf + i, sz - i, " %u", sk->sk_shutdown); 3954 i += scnprintf(buf + i, sz - i, " | %d", sk_wmem_alloc_get(sk)); 3955 i += scnprintf(buf + i, sz - i, " %d", sk->sk_sndbuf); 3956 i += scnprintf(buf + i, sz - i, " | %d", sk_rmem_alloc_get(sk)); 3957 i += scnprintf(buf + i, sz - i, " %d", sk->sk_rcvbuf); 3958 i += scnprintf(buf + i, sz - i, " | %d\n", READ_ONCE(sk->sk_backlog.len)); 3959 3960 if (dqueues & TIPC_DUMP_SK_SNDQ) { 3961 i += scnprintf(buf + i, sz - i, "sk_write_queue: "); 3962 i += tipc_list_dump(&sk->sk_write_queue, false, buf + i); 3963 } 3964 3965 if (dqueues & TIPC_DUMP_SK_RCVQ) { 3966 i += scnprintf(buf + i, sz - i, "sk_receive_queue: "); 3967 i += tipc_list_dump(&sk->sk_receive_queue, false, buf + i); 3968 } 3969 3970 if (dqueues & TIPC_DUMP_SK_BKLGQ) { 3971 i += scnprintf(buf + i, sz - i, "sk_backlog:\n head "); 3972 i += tipc_skb_dump(sk->sk_backlog.head, false, buf + i); 3973 if (sk->sk_backlog.tail != sk->sk_backlog.head) { 3974 i += scnprintf(buf + i, sz - i, " tail "); 3975 i += tipc_skb_dump(sk->sk_backlog.tail, false, 3976 buf + i); 3977 } 3978 } 3979 3980 return i; 3981 } 3982