1 /* 2 * net/tipc/socket.c: TIPC socket API 3 * 4 * Copyright (c) 2001-2007, 2012 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 "core.h" 38 #include "port.h" 39 40 #include <linux/export.h> 41 #include <net/sock.h> 42 43 #define SS_LISTENING -1 /* socket is listening */ 44 #define SS_READY -2 /* socket is connectionless */ 45 46 #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */ 47 48 struct tipc_sock { 49 struct sock sk; 50 struct tipc_port *p; 51 struct tipc_portid peer_name; 52 unsigned int conn_timeout; 53 }; 54 55 #define tipc_sk(sk) ((struct tipc_sock *)(sk)) 56 #define tipc_sk_port(sk) (tipc_sk(sk)->p) 57 58 #define tipc_rx_ready(sock) (!skb_queue_empty(&sock->sk->sk_receive_queue) || \ 59 (sock->state == SS_DISCONNECTING)) 60 61 static int backlog_rcv(struct sock *sk, struct sk_buff *skb); 62 static u32 dispatch(struct tipc_port *tport, struct sk_buff *buf); 63 static void wakeupdispatch(struct tipc_port *tport); 64 static void tipc_data_ready(struct sock *sk, int len); 65 static void tipc_write_space(struct sock *sk); 66 static int release(struct socket *sock); 67 static int accept(struct socket *sock, struct socket *new_sock, int flags); 68 69 static const struct proto_ops packet_ops; 70 static const struct proto_ops stream_ops; 71 static const struct proto_ops msg_ops; 72 73 static struct proto tipc_proto; 74 static struct proto tipc_proto_kern; 75 76 static int sockets_enabled; 77 78 /* 79 * Revised TIPC socket locking policy: 80 * 81 * Most socket operations take the standard socket lock when they start 82 * and hold it until they finish (or until they need to sleep). Acquiring 83 * this lock grants the owner exclusive access to the fields of the socket 84 * data structures, with the exception of the backlog queue. A few socket 85 * operations can be done without taking the socket lock because they only 86 * read socket information that never changes during the life of the socket. 87 * 88 * Socket operations may acquire the lock for the associated TIPC port if they 89 * need to perform an operation on the port. If any routine needs to acquire 90 * both the socket lock and the port lock it must take the socket lock first 91 * to avoid the risk of deadlock. 92 * 93 * The dispatcher handling incoming messages cannot grab the socket lock in 94 * the standard fashion, since invoked it runs at the BH level and cannot block. 95 * Instead, it checks to see if the socket lock is currently owned by someone, 96 * and either handles the message itself or adds it to the socket's backlog 97 * queue; in the latter case the queued message is processed once the process 98 * owning the socket lock releases it. 99 * 100 * NOTE: Releasing the socket lock while an operation is sleeping overcomes 101 * the problem of a blocked socket operation preventing any other operations 102 * from occurring. However, applications must be careful if they have 103 * multiple threads trying to send (or receive) on the same socket, as these 104 * operations might interfere with each other. For example, doing a connect 105 * and a receive at the same time might allow the receive to consume the 106 * ACK message meant for the connect. While additional work could be done 107 * to try and overcome this, it doesn't seem to be worthwhile at the present. 108 * 109 * NOTE: Releasing the socket lock while an operation is sleeping also ensures 110 * that another operation that must be performed in a non-blocking manner is 111 * not delayed for very long because the lock has already been taken. 112 * 113 * NOTE: This code assumes that certain fields of a port/socket pair are 114 * constant over its lifetime; such fields can be examined without taking 115 * the socket lock and/or port lock, and do not need to be re-read even 116 * after resuming processing after waiting. These fields include: 117 * - socket type 118 * - pointer to socket sk structure (aka tipc_sock structure) 119 * - pointer to port structure 120 * - port reference 121 */ 122 123 /** 124 * advance_rx_queue - discard first buffer in socket receive queue 125 * 126 * Caller must hold socket lock 127 */ 128 static void advance_rx_queue(struct sock *sk) 129 { 130 kfree_skb(__skb_dequeue(&sk->sk_receive_queue)); 131 } 132 133 /** 134 * reject_rx_queue - reject all buffers in socket receive queue 135 * 136 * Caller must hold socket lock 137 */ 138 static void reject_rx_queue(struct sock *sk) 139 { 140 struct sk_buff *buf; 141 142 while ((buf = __skb_dequeue(&sk->sk_receive_queue))) 143 tipc_reject_msg(buf, TIPC_ERR_NO_PORT); 144 } 145 146 /** 147 * tipc_sk_create - create a TIPC socket 148 * @net: network namespace (must be default network) 149 * @sock: pre-allocated socket structure 150 * @protocol: protocol indicator (must be 0) 151 * @kern: caused by kernel or by userspace? 152 * 153 * This routine creates additional data structures used by the TIPC socket, 154 * initializes them, and links them together. 155 * 156 * Returns 0 on success, errno otherwise 157 */ 158 static int tipc_sk_create(struct net *net, struct socket *sock, int protocol, 159 int kern) 160 { 161 const struct proto_ops *ops; 162 socket_state state; 163 struct sock *sk; 164 struct tipc_port *tp_ptr; 165 166 /* Validate arguments */ 167 if (unlikely(protocol != 0)) 168 return -EPROTONOSUPPORT; 169 170 switch (sock->type) { 171 case SOCK_STREAM: 172 ops = &stream_ops; 173 state = SS_UNCONNECTED; 174 break; 175 case SOCK_SEQPACKET: 176 ops = &packet_ops; 177 state = SS_UNCONNECTED; 178 break; 179 case SOCK_DGRAM: 180 case SOCK_RDM: 181 ops = &msg_ops; 182 state = SS_READY; 183 break; 184 default: 185 return -EPROTOTYPE; 186 } 187 188 /* Allocate socket's protocol area */ 189 if (!kern) 190 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto); 191 else 192 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto_kern); 193 194 if (sk == NULL) 195 return -ENOMEM; 196 197 /* Allocate TIPC port for socket to use */ 198 tp_ptr = tipc_createport(sk, &dispatch, &wakeupdispatch, 199 TIPC_LOW_IMPORTANCE); 200 if (unlikely(!tp_ptr)) { 201 sk_free(sk); 202 return -ENOMEM; 203 } 204 205 /* Finish initializing socket data structures */ 206 sock->ops = ops; 207 sock->state = state; 208 209 sock_init_data(sock, sk); 210 sk->sk_backlog_rcv = backlog_rcv; 211 sk->sk_rcvbuf = sysctl_tipc_rmem[1]; 212 sk->sk_data_ready = tipc_data_ready; 213 sk->sk_write_space = tipc_write_space; 214 tipc_sk(sk)->p = tp_ptr; 215 tipc_sk(sk)->conn_timeout = CONN_TIMEOUT_DEFAULT; 216 217 spin_unlock_bh(tp_ptr->lock); 218 219 if (sock->state == SS_READY) { 220 tipc_set_portunreturnable(tp_ptr->ref, 1); 221 if (sock->type == SOCK_DGRAM) 222 tipc_set_portunreliable(tp_ptr->ref, 1); 223 } 224 225 return 0; 226 } 227 228 /** 229 * tipc_sock_create_local - create TIPC socket from inside TIPC module 230 * @type: socket type - SOCK_RDM or SOCK_SEQPACKET 231 * 232 * We cannot use sock_creat_kern here because it bumps module user count. 233 * Since socket owner and creator is the same module we must make sure 234 * that module count remains zero for module local sockets, otherwise 235 * we cannot do rmmod. 236 * 237 * Returns 0 on success, errno otherwise 238 */ 239 int tipc_sock_create_local(int type, struct socket **res) 240 { 241 int rc; 242 struct sock *sk; 243 244 rc = sock_create_lite(AF_TIPC, type, 0, res); 245 if (rc < 0) { 246 pr_err("Failed to create kernel socket\n"); 247 return rc; 248 } 249 tipc_sk_create(&init_net, *res, 0, 1); 250 251 sk = (*res)->sk; 252 253 return 0; 254 } 255 256 /** 257 * tipc_sock_release_local - release socket created by tipc_sock_create_local 258 * @sock: the socket to be released. 259 * 260 * Module reference count is not incremented when such sockets are created, 261 * so we must keep it from being decremented when they are released. 262 */ 263 void tipc_sock_release_local(struct socket *sock) 264 { 265 release(sock); 266 sock->ops = NULL; 267 sock_release(sock); 268 } 269 270 /** 271 * tipc_sock_accept_local - accept a connection on a socket created 272 * with tipc_sock_create_local. Use this function to avoid that 273 * module reference count is inadvertently incremented. 274 * 275 * @sock: the accepting socket 276 * @newsock: reference to the new socket to be created 277 * @flags: socket flags 278 */ 279 280 int tipc_sock_accept_local(struct socket *sock, struct socket **newsock, 281 int flags) 282 { 283 struct sock *sk = sock->sk; 284 int ret; 285 286 ret = sock_create_lite(sk->sk_family, sk->sk_type, 287 sk->sk_protocol, newsock); 288 if (ret < 0) 289 return ret; 290 291 ret = accept(sock, *newsock, flags); 292 if (ret < 0) { 293 sock_release(*newsock); 294 return ret; 295 } 296 (*newsock)->ops = sock->ops; 297 return ret; 298 } 299 300 /** 301 * release - destroy a TIPC socket 302 * @sock: socket to destroy 303 * 304 * This routine cleans up any messages that are still queued on the socket. 305 * For DGRAM and RDM socket types, all queued messages are rejected. 306 * For SEQPACKET and STREAM socket types, the first message is rejected 307 * and any others are discarded. (If the first message on a STREAM socket 308 * is partially-read, it is discarded and the next one is rejected instead.) 309 * 310 * NOTE: Rejected messages are not necessarily returned to the sender! They 311 * are returned or discarded according to the "destination droppable" setting 312 * specified for the message by the sender. 313 * 314 * Returns 0 on success, errno otherwise 315 */ 316 static int release(struct socket *sock) 317 { 318 struct sock *sk = sock->sk; 319 struct tipc_port *tport; 320 struct sk_buff *buf; 321 int res; 322 323 /* 324 * Exit if socket isn't fully initialized (occurs when a failed accept() 325 * releases a pre-allocated child socket that was never used) 326 */ 327 if (sk == NULL) 328 return 0; 329 330 tport = tipc_sk_port(sk); 331 lock_sock(sk); 332 333 /* 334 * Reject all unreceived messages, except on an active connection 335 * (which disconnects locally & sends a 'FIN+' to peer) 336 */ 337 while (sock->state != SS_DISCONNECTING) { 338 buf = __skb_dequeue(&sk->sk_receive_queue); 339 if (buf == NULL) 340 break; 341 if (TIPC_SKB_CB(buf)->handle != 0) 342 kfree_skb(buf); 343 else { 344 if ((sock->state == SS_CONNECTING) || 345 (sock->state == SS_CONNECTED)) { 346 sock->state = SS_DISCONNECTING; 347 tipc_disconnect(tport->ref); 348 } 349 tipc_reject_msg(buf, TIPC_ERR_NO_PORT); 350 } 351 } 352 353 /* 354 * Delete TIPC port; this ensures no more messages are queued 355 * (also disconnects an active connection & sends a 'FIN-' to peer) 356 */ 357 res = tipc_deleteport(tport->ref); 358 359 /* Discard any remaining (connection-based) messages in receive queue */ 360 __skb_queue_purge(&sk->sk_receive_queue); 361 362 /* Reject any messages that accumulated in backlog queue */ 363 sock->state = SS_DISCONNECTING; 364 release_sock(sk); 365 366 sock_put(sk); 367 sock->sk = NULL; 368 369 return res; 370 } 371 372 /** 373 * bind - associate or disassocate TIPC name(s) with a socket 374 * @sock: socket structure 375 * @uaddr: socket address describing name(s) and desired operation 376 * @uaddr_len: size of socket address data structure 377 * 378 * Name and name sequence binding is indicated using a positive scope value; 379 * a negative scope value unbinds the specified name. Specifying no name 380 * (i.e. a socket address length of 0) unbinds all names from the socket. 381 * 382 * Returns 0 on success, errno otherwise 383 * 384 * NOTE: This routine doesn't need to take the socket lock since it doesn't 385 * access any non-constant socket information. 386 */ 387 static int bind(struct socket *sock, struct sockaddr *uaddr, int uaddr_len) 388 { 389 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr; 390 u32 portref = tipc_sk_port(sock->sk)->ref; 391 392 if (unlikely(!uaddr_len)) 393 return tipc_withdraw(portref, 0, NULL); 394 395 if (uaddr_len < sizeof(struct sockaddr_tipc)) 396 return -EINVAL; 397 if (addr->family != AF_TIPC) 398 return -EAFNOSUPPORT; 399 400 if (addr->addrtype == TIPC_ADDR_NAME) 401 addr->addr.nameseq.upper = addr->addr.nameseq.lower; 402 else if (addr->addrtype != TIPC_ADDR_NAMESEQ) 403 return -EAFNOSUPPORT; 404 405 if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) && 406 (addr->addr.nameseq.type != TIPC_TOP_SRV) && 407 (addr->addr.nameseq.type != TIPC_CFG_SRV)) 408 return -EACCES; 409 410 return (addr->scope > 0) ? 411 tipc_publish(portref, addr->scope, &addr->addr.nameseq) : 412 tipc_withdraw(portref, -addr->scope, &addr->addr.nameseq); 413 } 414 415 /** 416 * get_name - get port ID of socket or peer socket 417 * @sock: socket structure 418 * @uaddr: area for returned socket address 419 * @uaddr_len: area for returned length of socket address 420 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID 421 * 422 * Returns 0 on success, errno otherwise 423 * 424 * NOTE: This routine doesn't need to take the socket lock since it only 425 * accesses socket information that is unchanging (or which changes in 426 * a completely predictable manner). 427 */ 428 static int get_name(struct socket *sock, struct sockaddr *uaddr, 429 int *uaddr_len, int peer) 430 { 431 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr; 432 struct tipc_sock *tsock = tipc_sk(sock->sk); 433 434 memset(addr, 0, sizeof(*addr)); 435 if (peer) { 436 if ((sock->state != SS_CONNECTED) && 437 ((peer != 2) || (sock->state != SS_DISCONNECTING))) 438 return -ENOTCONN; 439 addr->addr.id.ref = tsock->peer_name.ref; 440 addr->addr.id.node = tsock->peer_name.node; 441 } else { 442 addr->addr.id.ref = tsock->p->ref; 443 addr->addr.id.node = tipc_own_addr; 444 } 445 446 *uaddr_len = sizeof(*addr); 447 addr->addrtype = TIPC_ADDR_ID; 448 addr->family = AF_TIPC; 449 addr->scope = 0; 450 addr->addr.name.domain = 0; 451 452 return 0; 453 } 454 455 /** 456 * poll - read and possibly block on pollmask 457 * @file: file structure associated with the socket 458 * @sock: socket for which to calculate the poll bits 459 * @wait: ??? 460 * 461 * Returns pollmask value 462 * 463 * COMMENTARY: 464 * It appears that the usual socket locking mechanisms are not useful here 465 * since the pollmask info is potentially out-of-date the moment this routine 466 * exits. TCP and other protocols seem to rely on higher level poll routines 467 * to handle any preventable race conditions, so TIPC will do the same ... 468 * 469 * TIPC sets the returned events as follows: 470 * 471 * socket state flags set 472 * ------------ --------- 473 * unconnected no read flags 474 * POLLOUT if port is not congested 475 * 476 * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue 477 * no write flags 478 * 479 * connected POLLIN/POLLRDNORM if data in rx queue 480 * POLLOUT if port is not congested 481 * 482 * disconnecting POLLIN/POLLRDNORM/POLLHUP 483 * no write flags 484 * 485 * listening POLLIN if SYN in rx queue 486 * no write flags 487 * 488 * ready POLLIN/POLLRDNORM if data in rx queue 489 * [connectionless] POLLOUT (since port cannot be congested) 490 * 491 * IMPORTANT: The fact that a read or write operation is indicated does NOT 492 * imply that the operation will succeed, merely that it should be performed 493 * and will not block. 494 */ 495 static unsigned int poll(struct file *file, struct socket *sock, 496 poll_table *wait) 497 { 498 struct sock *sk = sock->sk; 499 u32 mask = 0; 500 501 sock_poll_wait(file, sk_sleep(sk), wait); 502 503 switch ((int)sock->state) { 504 case SS_UNCONNECTED: 505 if (!tipc_sk_port(sk)->congested) 506 mask |= POLLOUT; 507 break; 508 case SS_READY: 509 case SS_CONNECTED: 510 if (!tipc_sk_port(sk)->congested) 511 mask |= POLLOUT; 512 /* fall thru' */ 513 case SS_CONNECTING: 514 case SS_LISTENING: 515 if (!skb_queue_empty(&sk->sk_receive_queue)) 516 mask |= (POLLIN | POLLRDNORM); 517 break; 518 case SS_DISCONNECTING: 519 mask = (POLLIN | POLLRDNORM | POLLHUP); 520 break; 521 } 522 523 return mask; 524 } 525 526 /** 527 * dest_name_check - verify user is permitted to send to specified port name 528 * @dest: destination address 529 * @m: descriptor for message to be sent 530 * 531 * Prevents restricted configuration commands from being issued by 532 * unauthorized users. 533 * 534 * Returns 0 if permission is granted, otherwise errno 535 */ 536 static int dest_name_check(struct sockaddr_tipc *dest, struct msghdr *m) 537 { 538 struct tipc_cfg_msg_hdr hdr; 539 540 if (likely(dest->addr.name.name.type >= TIPC_RESERVED_TYPES)) 541 return 0; 542 if (likely(dest->addr.name.name.type == TIPC_TOP_SRV)) 543 return 0; 544 if (likely(dest->addr.name.name.type != TIPC_CFG_SRV)) 545 return -EACCES; 546 547 if (!m->msg_iovlen || (m->msg_iov[0].iov_len < sizeof(hdr))) 548 return -EMSGSIZE; 549 if (copy_from_user(&hdr, m->msg_iov[0].iov_base, sizeof(hdr))) 550 return -EFAULT; 551 if ((ntohs(hdr.tcm_type) & 0xC000) && (!capable(CAP_NET_ADMIN))) 552 return -EACCES; 553 554 return 0; 555 } 556 557 /** 558 * send_msg - send message in connectionless manner 559 * @iocb: if NULL, indicates that socket lock is already held 560 * @sock: socket structure 561 * @m: message to send 562 * @total_len: length of message 563 * 564 * Message must have an destination specified explicitly. 565 * Used for SOCK_RDM and SOCK_DGRAM messages, 566 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections. 567 * (Note: 'SYN+' is prohibited on SOCK_STREAM.) 568 * 569 * Returns the number of bytes sent on success, or errno otherwise 570 */ 571 static int send_msg(struct kiocb *iocb, struct socket *sock, 572 struct msghdr *m, size_t total_len) 573 { 574 struct sock *sk = sock->sk; 575 struct tipc_port *tport = tipc_sk_port(sk); 576 struct sockaddr_tipc *dest = (struct sockaddr_tipc *)m->msg_name; 577 int needs_conn; 578 long timeout_val; 579 int res = -EINVAL; 580 581 if (unlikely(!dest)) 582 return -EDESTADDRREQ; 583 if (unlikely((m->msg_namelen < sizeof(*dest)) || 584 (dest->family != AF_TIPC))) 585 return -EINVAL; 586 if (total_len > TIPC_MAX_USER_MSG_SIZE) 587 return -EMSGSIZE; 588 589 if (iocb) 590 lock_sock(sk); 591 592 needs_conn = (sock->state != SS_READY); 593 if (unlikely(needs_conn)) { 594 if (sock->state == SS_LISTENING) { 595 res = -EPIPE; 596 goto exit; 597 } 598 if (sock->state != SS_UNCONNECTED) { 599 res = -EISCONN; 600 goto exit; 601 } 602 if (tport->published) { 603 res = -EOPNOTSUPP; 604 goto exit; 605 } 606 if (dest->addrtype == TIPC_ADDR_NAME) { 607 tport->conn_type = dest->addr.name.name.type; 608 tport->conn_instance = dest->addr.name.name.instance; 609 } 610 611 /* Abort any pending connection attempts (very unlikely) */ 612 reject_rx_queue(sk); 613 } 614 615 timeout_val = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT); 616 617 do { 618 if (dest->addrtype == TIPC_ADDR_NAME) { 619 res = dest_name_check(dest, m); 620 if (res) 621 break; 622 res = tipc_send2name(tport->ref, 623 &dest->addr.name.name, 624 dest->addr.name.domain, 625 m->msg_iovlen, 626 m->msg_iov, 627 total_len); 628 } else if (dest->addrtype == TIPC_ADDR_ID) { 629 res = tipc_send2port(tport->ref, 630 &dest->addr.id, 631 m->msg_iovlen, 632 m->msg_iov, 633 total_len); 634 } else if (dest->addrtype == TIPC_ADDR_MCAST) { 635 if (needs_conn) { 636 res = -EOPNOTSUPP; 637 break; 638 } 639 res = dest_name_check(dest, m); 640 if (res) 641 break; 642 res = tipc_multicast(tport->ref, 643 &dest->addr.nameseq, 644 m->msg_iovlen, 645 m->msg_iov, 646 total_len); 647 } 648 if (likely(res != -ELINKCONG)) { 649 if (needs_conn && (res >= 0)) 650 sock->state = SS_CONNECTING; 651 break; 652 } 653 if (timeout_val <= 0L) { 654 res = timeout_val ? timeout_val : -EWOULDBLOCK; 655 break; 656 } 657 release_sock(sk); 658 timeout_val = wait_event_interruptible_timeout(*sk_sleep(sk), 659 !tport->congested, timeout_val); 660 lock_sock(sk); 661 } while (1); 662 663 exit: 664 if (iocb) 665 release_sock(sk); 666 return res; 667 } 668 669 /** 670 * send_packet - send a connection-oriented message 671 * @iocb: if NULL, indicates that socket lock is already held 672 * @sock: socket structure 673 * @m: message to send 674 * @total_len: length of message 675 * 676 * Used for SOCK_SEQPACKET messages and SOCK_STREAM data. 677 * 678 * Returns the number of bytes sent on success, or errno otherwise 679 */ 680 static int send_packet(struct kiocb *iocb, struct socket *sock, 681 struct msghdr *m, size_t total_len) 682 { 683 struct sock *sk = sock->sk; 684 struct tipc_port *tport = tipc_sk_port(sk); 685 struct sockaddr_tipc *dest = (struct sockaddr_tipc *)m->msg_name; 686 long timeout_val; 687 int res; 688 689 /* Handle implied connection establishment */ 690 if (unlikely(dest)) 691 return send_msg(iocb, sock, m, total_len); 692 693 if (total_len > TIPC_MAX_USER_MSG_SIZE) 694 return -EMSGSIZE; 695 696 if (iocb) 697 lock_sock(sk); 698 699 timeout_val = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT); 700 701 do { 702 if (unlikely(sock->state != SS_CONNECTED)) { 703 if (sock->state == SS_DISCONNECTING) 704 res = -EPIPE; 705 else 706 res = -ENOTCONN; 707 break; 708 } 709 710 res = tipc_send(tport->ref, m->msg_iovlen, m->msg_iov, 711 total_len); 712 if (likely(res != -ELINKCONG)) 713 break; 714 if (timeout_val <= 0L) { 715 res = timeout_val ? timeout_val : -EWOULDBLOCK; 716 break; 717 } 718 release_sock(sk); 719 timeout_val = wait_event_interruptible_timeout(*sk_sleep(sk), 720 (!tport->congested || !tport->connected), timeout_val); 721 lock_sock(sk); 722 } while (1); 723 724 if (iocb) 725 release_sock(sk); 726 return res; 727 } 728 729 /** 730 * send_stream - send stream-oriented data 731 * @iocb: (unused) 732 * @sock: socket structure 733 * @m: data to send 734 * @total_len: total length of data to be sent 735 * 736 * Used for SOCK_STREAM data. 737 * 738 * Returns the number of bytes sent on success (or partial success), 739 * or errno if no data sent 740 */ 741 static int send_stream(struct kiocb *iocb, struct socket *sock, 742 struct msghdr *m, size_t total_len) 743 { 744 struct sock *sk = sock->sk; 745 struct tipc_port *tport = tipc_sk_port(sk); 746 struct msghdr my_msg; 747 struct iovec my_iov; 748 struct iovec *curr_iov; 749 int curr_iovlen; 750 char __user *curr_start; 751 u32 hdr_size; 752 int curr_left; 753 int bytes_to_send; 754 int bytes_sent; 755 int res; 756 757 lock_sock(sk); 758 759 /* Handle special cases where there is no connection */ 760 if (unlikely(sock->state != SS_CONNECTED)) { 761 if (sock->state == SS_UNCONNECTED) { 762 res = send_packet(NULL, sock, m, total_len); 763 goto exit; 764 } else if (sock->state == SS_DISCONNECTING) { 765 res = -EPIPE; 766 goto exit; 767 } else { 768 res = -ENOTCONN; 769 goto exit; 770 } 771 } 772 773 if (unlikely(m->msg_name)) { 774 res = -EISCONN; 775 goto exit; 776 } 777 778 if (total_len > (unsigned int)INT_MAX) { 779 res = -EMSGSIZE; 780 goto exit; 781 } 782 783 /* 784 * Send each iovec entry using one or more messages 785 * 786 * Note: This algorithm is good for the most likely case 787 * (i.e. one large iovec entry), but could be improved to pass sets 788 * of small iovec entries into send_packet(). 789 */ 790 curr_iov = m->msg_iov; 791 curr_iovlen = m->msg_iovlen; 792 my_msg.msg_iov = &my_iov; 793 my_msg.msg_iovlen = 1; 794 my_msg.msg_flags = m->msg_flags; 795 my_msg.msg_name = NULL; 796 bytes_sent = 0; 797 798 hdr_size = msg_hdr_sz(&tport->phdr); 799 800 while (curr_iovlen--) { 801 curr_start = curr_iov->iov_base; 802 curr_left = curr_iov->iov_len; 803 804 while (curr_left) { 805 bytes_to_send = tport->max_pkt - hdr_size; 806 if (bytes_to_send > TIPC_MAX_USER_MSG_SIZE) 807 bytes_to_send = TIPC_MAX_USER_MSG_SIZE; 808 if (curr_left < bytes_to_send) 809 bytes_to_send = curr_left; 810 my_iov.iov_base = curr_start; 811 my_iov.iov_len = bytes_to_send; 812 res = send_packet(NULL, sock, &my_msg, bytes_to_send); 813 if (res < 0) { 814 if (bytes_sent) 815 res = bytes_sent; 816 goto exit; 817 } 818 curr_left -= bytes_to_send; 819 curr_start += bytes_to_send; 820 bytes_sent += bytes_to_send; 821 } 822 823 curr_iov++; 824 } 825 res = bytes_sent; 826 exit: 827 release_sock(sk); 828 return res; 829 } 830 831 /** 832 * auto_connect - complete connection setup to a remote port 833 * @sock: socket structure 834 * @msg: peer's response message 835 * 836 * Returns 0 on success, errno otherwise 837 */ 838 static int auto_connect(struct socket *sock, struct tipc_msg *msg) 839 { 840 struct tipc_sock *tsock = tipc_sk(sock->sk); 841 struct tipc_port *p_ptr; 842 843 tsock->peer_name.ref = msg_origport(msg); 844 tsock->peer_name.node = msg_orignode(msg); 845 p_ptr = tipc_port_deref(tsock->p->ref); 846 if (!p_ptr) 847 return -EINVAL; 848 849 __tipc_connect(tsock->p->ref, p_ptr, &tsock->peer_name); 850 851 if (msg_importance(msg) > TIPC_CRITICAL_IMPORTANCE) 852 return -EINVAL; 853 msg_set_importance(&p_ptr->phdr, (u32)msg_importance(msg)); 854 sock->state = SS_CONNECTED; 855 return 0; 856 } 857 858 /** 859 * set_orig_addr - capture sender's address for received message 860 * @m: descriptor for message info 861 * @msg: received message header 862 * 863 * Note: Address is not captured if not requested by receiver. 864 */ 865 static void set_orig_addr(struct msghdr *m, struct tipc_msg *msg) 866 { 867 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)m->msg_name; 868 869 if (addr) { 870 addr->family = AF_TIPC; 871 addr->addrtype = TIPC_ADDR_ID; 872 memset(&addr->addr, 0, sizeof(addr->addr)); 873 addr->addr.id.ref = msg_origport(msg); 874 addr->addr.id.node = msg_orignode(msg); 875 addr->addr.name.domain = 0; /* could leave uninitialized */ 876 addr->scope = 0; /* could leave uninitialized */ 877 m->msg_namelen = sizeof(struct sockaddr_tipc); 878 } 879 } 880 881 /** 882 * anc_data_recv - optionally capture ancillary data for received message 883 * @m: descriptor for message info 884 * @msg: received message header 885 * @tport: TIPC port associated with message 886 * 887 * Note: Ancillary data is not captured if not requested by receiver. 888 * 889 * Returns 0 if successful, otherwise errno 890 */ 891 static int anc_data_recv(struct msghdr *m, struct tipc_msg *msg, 892 struct tipc_port *tport) 893 { 894 u32 anc_data[3]; 895 u32 err; 896 u32 dest_type; 897 int has_name; 898 int res; 899 900 if (likely(m->msg_controllen == 0)) 901 return 0; 902 903 /* Optionally capture errored message object(s) */ 904 err = msg ? msg_errcode(msg) : 0; 905 if (unlikely(err)) { 906 anc_data[0] = err; 907 anc_data[1] = msg_data_sz(msg); 908 res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data); 909 if (res) 910 return res; 911 if (anc_data[1]) { 912 res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1], 913 msg_data(msg)); 914 if (res) 915 return res; 916 } 917 } 918 919 /* Optionally capture message destination object */ 920 dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG; 921 switch (dest_type) { 922 case TIPC_NAMED_MSG: 923 has_name = 1; 924 anc_data[0] = msg_nametype(msg); 925 anc_data[1] = msg_namelower(msg); 926 anc_data[2] = msg_namelower(msg); 927 break; 928 case TIPC_MCAST_MSG: 929 has_name = 1; 930 anc_data[0] = msg_nametype(msg); 931 anc_data[1] = msg_namelower(msg); 932 anc_data[2] = msg_nameupper(msg); 933 break; 934 case TIPC_CONN_MSG: 935 has_name = (tport->conn_type != 0); 936 anc_data[0] = tport->conn_type; 937 anc_data[1] = tport->conn_instance; 938 anc_data[2] = tport->conn_instance; 939 break; 940 default: 941 has_name = 0; 942 } 943 if (has_name) { 944 res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data); 945 if (res) 946 return res; 947 } 948 949 return 0; 950 } 951 952 /** 953 * recv_msg - receive packet-oriented message 954 * @iocb: (unused) 955 * @m: descriptor for message info 956 * @buf_len: total size of user buffer area 957 * @flags: receive flags 958 * 959 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages. 960 * If the complete message doesn't fit in user area, truncate it. 961 * 962 * Returns size of returned message data, errno otherwise 963 */ 964 static int recv_msg(struct kiocb *iocb, struct socket *sock, 965 struct msghdr *m, size_t buf_len, int flags) 966 { 967 struct sock *sk = sock->sk; 968 struct tipc_port *tport = tipc_sk_port(sk); 969 struct sk_buff *buf; 970 struct tipc_msg *msg; 971 long timeout; 972 unsigned int sz; 973 u32 err; 974 int res; 975 976 /* Catch invalid receive requests */ 977 if (unlikely(!buf_len)) 978 return -EINVAL; 979 980 lock_sock(sk); 981 982 if (unlikely(sock->state == SS_UNCONNECTED)) { 983 res = -ENOTCONN; 984 goto exit; 985 } 986 987 /* will be updated in set_orig_addr() if needed */ 988 m->msg_namelen = 0; 989 990 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT); 991 restart: 992 993 /* Look for a message in receive queue; wait if necessary */ 994 while (skb_queue_empty(&sk->sk_receive_queue)) { 995 if (sock->state == SS_DISCONNECTING) { 996 res = -ENOTCONN; 997 goto exit; 998 } 999 if (timeout <= 0L) { 1000 res = timeout ? timeout : -EWOULDBLOCK; 1001 goto exit; 1002 } 1003 release_sock(sk); 1004 timeout = wait_event_interruptible_timeout(*sk_sleep(sk), 1005 tipc_rx_ready(sock), 1006 timeout); 1007 lock_sock(sk); 1008 } 1009 1010 /* Look at first message in receive queue */ 1011 buf = skb_peek(&sk->sk_receive_queue); 1012 msg = buf_msg(buf); 1013 sz = msg_data_sz(msg); 1014 err = msg_errcode(msg); 1015 1016 /* Discard an empty non-errored message & try again */ 1017 if ((!sz) && (!err)) { 1018 advance_rx_queue(sk); 1019 goto restart; 1020 } 1021 1022 /* Capture sender's address (optional) */ 1023 set_orig_addr(m, msg); 1024 1025 /* Capture ancillary data (optional) */ 1026 res = anc_data_recv(m, msg, tport); 1027 if (res) 1028 goto exit; 1029 1030 /* Capture message data (if valid) & compute return value (always) */ 1031 if (!err) { 1032 if (unlikely(buf_len < sz)) { 1033 sz = buf_len; 1034 m->msg_flags |= MSG_TRUNC; 1035 } 1036 res = skb_copy_datagram_iovec(buf, msg_hdr_sz(msg), 1037 m->msg_iov, sz); 1038 if (res) 1039 goto exit; 1040 res = sz; 1041 } else { 1042 if ((sock->state == SS_READY) || 1043 ((err == TIPC_CONN_SHUTDOWN) || m->msg_control)) 1044 res = 0; 1045 else 1046 res = -ECONNRESET; 1047 } 1048 1049 /* Consume received message (optional) */ 1050 if (likely(!(flags & MSG_PEEK))) { 1051 if ((sock->state != SS_READY) && 1052 (++tport->conn_unacked >= TIPC_FLOW_CONTROL_WIN)) 1053 tipc_acknowledge(tport->ref, tport->conn_unacked); 1054 advance_rx_queue(sk); 1055 } 1056 exit: 1057 release_sock(sk); 1058 return res; 1059 } 1060 1061 /** 1062 * recv_stream - receive stream-oriented data 1063 * @iocb: (unused) 1064 * @m: descriptor for message info 1065 * @buf_len: total size of user buffer area 1066 * @flags: receive flags 1067 * 1068 * Used for SOCK_STREAM messages only. If not enough data is available 1069 * will optionally wait for more; never truncates data. 1070 * 1071 * Returns size of returned message data, errno otherwise 1072 */ 1073 static int recv_stream(struct kiocb *iocb, struct socket *sock, 1074 struct msghdr *m, size_t buf_len, int flags) 1075 { 1076 struct sock *sk = sock->sk; 1077 struct tipc_port *tport = tipc_sk_port(sk); 1078 struct sk_buff *buf; 1079 struct tipc_msg *msg; 1080 long timeout; 1081 unsigned int sz; 1082 int sz_to_copy, target, needed; 1083 int sz_copied = 0; 1084 u32 err; 1085 int res = 0; 1086 1087 /* Catch invalid receive attempts */ 1088 if (unlikely(!buf_len)) 1089 return -EINVAL; 1090 1091 lock_sock(sk); 1092 1093 if (unlikely((sock->state == SS_UNCONNECTED))) { 1094 res = -ENOTCONN; 1095 goto exit; 1096 } 1097 1098 /* will be updated in set_orig_addr() if needed */ 1099 m->msg_namelen = 0; 1100 1101 target = sock_rcvlowat(sk, flags & MSG_WAITALL, buf_len); 1102 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT); 1103 1104 restart: 1105 /* Look for a message in receive queue; wait if necessary */ 1106 while (skb_queue_empty(&sk->sk_receive_queue)) { 1107 if (sock->state == SS_DISCONNECTING) { 1108 res = -ENOTCONN; 1109 goto exit; 1110 } 1111 if (timeout <= 0L) { 1112 res = timeout ? timeout : -EWOULDBLOCK; 1113 goto exit; 1114 } 1115 release_sock(sk); 1116 timeout = wait_event_interruptible_timeout(*sk_sleep(sk), 1117 tipc_rx_ready(sock), 1118 timeout); 1119 lock_sock(sk); 1120 } 1121 1122 /* Look at first message in receive queue */ 1123 buf = skb_peek(&sk->sk_receive_queue); 1124 msg = buf_msg(buf); 1125 sz = msg_data_sz(msg); 1126 err = msg_errcode(msg); 1127 1128 /* Discard an empty non-errored message & try again */ 1129 if ((!sz) && (!err)) { 1130 advance_rx_queue(sk); 1131 goto restart; 1132 } 1133 1134 /* Optionally capture sender's address & ancillary data of first msg */ 1135 if (sz_copied == 0) { 1136 set_orig_addr(m, msg); 1137 res = anc_data_recv(m, msg, tport); 1138 if (res) 1139 goto exit; 1140 } 1141 1142 /* Capture message data (if valid) & compute return value (always) */ 1143 if (!err) { 1144 u32 offset = (u32)(unsigned long)(TIPC_SKB_CB(buf)->handle); 1145 1146 sz -= offset; 1147 needed = (buf_len - sz_copied); 1148 sz_to_copy = (sz <= needed) ? sz : needed; 1149 1150 res = skb_copy_datagram_iovec(buf, msg_hdr_sz(msg) + offset, 1151 m->msg_iov, sz_to_copy); 1152 if (res) 1153 goto exit; 1154 1155 sz_copied += sz_to_copy; 1156 1157 if (sz_to_copy < sz) { 1158 if (!(flags & MSG_PEEK)) 1159 TIPC_SKB_CB(buf)->handle = 1160 (void *)(unsigned long)(offset + sz_to_copy); 1161 goto exit; 1162 } 1163 } else { 1164 if (sz_copied != 0) 1165 goto exit; /* can't add error msg to valid data */ 1166 1167 if ((err == TIPC_CONN_SHUTDOWN) || m->msg_control) 1168 res = 0; 1169 else 1170 res = -ECONNRESET; 1171 } 1172 1173 /* Consume received message (optional) */ 1174 if (likely(!(flags & MSG_PEEK))) { 1175 if (unlikely(++tport->conn_unacked >= TIPC_FLOW_CONTROL_WIN)) 1176 tipc_acknowledge(tport->ref, tport->conn_unacked); 1177 advance_rx_queue(sk); 1178 } 1179 1180 /* Loop around if more data is required */ 1181 if ((sz_copied < buf_len) && /* didn't get all requested data */ 1182 (!skb_queue_empty(&sk->sk_receive_queue) || 1183 (sz_copied < target)) && /* and more is ready or required */ 1184 (!(flags & MSG_PEEK)) && /* and aren't just peeking at data */ 1185 (!err)) /* and haven't reached a FIN */ 1186 goto restart; 1187 1188 exit: 1189 release_sock(sk); 1190 return sz_copied ? sz_copied : res; 1191 } 1192 1193 /** 1194 * tipc_write_space - wake up thread if port congestion is released 1195 * @sk: socket 1196 */ 1197 static void tipc_write_space(struct sock *sk) 1198 { 1199 struct socket_wq *wq; 1200 1201 rcu_read_lock(); 1202 wq = rcu_dereference(sk->sk_wq); 1203 if (wq_has_sleeper(wq)) 1204 wake_up_interruptible_sync_poll(&wq->wait, POLLOUT | 1205 POLLWRNORM | POLLWRBAND); 1206 rcu_read_unlock(); 1207 } 1208 1209 /** 1210 * tipc_data_ready - wake up threads to indicate messages have been received 1211 * @sk: socket 1212 * @len: the length of messages 1213 */ 1214 static void tipc_data_ready(struct sock *sk, int len) 1215 { 1216 struct socket_wq *wq; 1217 1218 rcu_read_lock(); 1219 wq = rcu_dereference(sk->sk_wq); 1220 if (wq_has_sleeper(wq)) 1221 wake_up_interruptible_sync_poll(&wq->wait, POLLIN | 1222 POLLRDNORM | POLLRDBAND); 1223 rcu_read_unlock(); 1224 } 1225 1226 /** 1227 * filter_connect - Handle all incoming messages for a connection-based socket 1228 * @tsock: TIPC socket 1229 * @msg: message 1230 * 1231 * Returns TIPC error status code and socket error status code 1232 * once it encounters some errors 1233 */ 1234 static u32 filter_connect(struct tipc_sock *tsock, struct sk_buff **buf) 1235 { 1236 struct socket *sock = tsock->sk.sk_socket; 1237 struct tipc_msg *msg = buf_msg(*buf); 1238 struct sock *sk = &tsock->sk; 1239 u32 retval = TIPC_ERR_NO_PORT; 1240 int res; 1241 1242 if (msg_mcast(msg)) 1243 return retval; 1244 1245 switch ((int)sock->state) { 1246 case SS_CONNECTED: 1247 /* Accept only connection-based messages sent by peer */ 1248 if (msg_connected(msg) && tipc_port_peer_msg(tsock->p, msg)) { 1249 if (unlikely(msg_errcode(msg))) { 1250 sock->state = SS_DISCONNECTING; 1251 __tipc_disconnect(tsock->p); 1252 } 1253 retval = TIPC_OK; 1254 } 1255 break; 1256 case SS_CONNECTING: 1257 /* Accept only ACK or NACK message */ 1258 if (unlikely(msg_errcode(msg))) { 1259 sock->state = SS_DISCONNECTING; 1260 sk->sk_err = ECONNREFUSED; 1261 retval = TIPC_OK; 1262 break; 1263 } 1264 1265 if (unlikely(!msg_connected(msg))) 1266 break; 1267 1268 res = auto_connect(sock, msg); 1269 if (res) { 1270 sock->state = SS_DISCONNECTING; 1271 sk->sk_err = -res; 1272 retval = TIPC_OK; 1273 break; 1274 } 1275 1276 /* If an incoming message is an 'ACK-', it should be 1277 * discarded here because it doesn't contain useful 1278 * data. In addition, we should try to wake up 1279 * connect() routine if sleeping. 1280 */ 1281 if (msg_data_sz(msg) == 0) { 1282 kfree_skb(*buf); 1283 *buf = NULL; 1284 if (waitqueue_active(sk_sleep(sk))) 1285 wake_up_interruptible(sk_sleep(sk)); 1286 } 1287 retval = TIPC_OK; 1288 break; 1289 case SS_LISTENING: 1290 case SS_UNCONNECTED: 1291 /* Accept only SYN message */ 1292 if (!msg_connected(msg) && !(msg_errcode(msg))) 1293 retval = TIPC_OK; 1294 break; 1295 case SS_DISCONNECTING: 1296 break; 1297 default: 1298 pr_err("Unknown socket state %u\n", sock->state); 1299 } 1300 return retval; 1301 } 1302 1303 /** 1304 * rcvbuf_limit - get proper overload limit of socket receive queue 1305 * @sk: socket 1306 * @buf: message 1307 * 1308 * For all connection oriented messages, irrespective of importance, 1309 * the default overload value (i.e. 67MB) is set as limit. 1310 * 1311 * For all connectionless messages, by default new queue limits are 1312 * as belows: 1313 * 1314 * TIPC_LOW_IMPORTANCE (4 MB) 1315 * TIPC_MEDIUM_IMPORTANCE (8 MB) 1316 * TIPC_HIGH_IMPORTANCE (16 MB) 1317 * TIPC_CRITICAL_IMPORTANCE (32 MB) 1318 * 1319 * Returns overload limit according to corresponding message importance 1320 */ 1321 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *buf) 1322 { 1323 struct tipc_msg *msg = buf_msg(buf); 1324 unsigned int limit; 1325 1326 if (msg_connected(msg)) 1327 limit = sysctl_tipc_rmem[2]; 1328 else 1329 limit = sk->sk_rcvbuf >> TIPC_CRITICAL_IMPORTANCE << 1330 msg_importance(msg); 1331 return limit; 1332 } 1333 1334 /** 1335 * filter_rcv - validate incoming message 1336 * @sk: socket 1337 * @buf: message 1338 * 1339 * Enqueues message on receive queue if acceptable; optionally handles 1340 * disconnect indication for a connected socket. 1341 * 1342 * Called with socket lock already taken; port lock may also be taken. 1343 * 1344 * Returns TIPC error status code (TIPC_OK if message is not to be rejected) 1345 */ 1346 static u32 filter_rcv(struct sock *sk, struct sk_buff *buf) 1347 { 1348 struct socket *sock = sk->sk_socket; 1349 struct tipc_msg *msg = buf_msg(buf); 1350 unsigned int limit = rcvbuf_limit(sk, buf); 1351 u32 res = TIPC_OK; 1352 1353 /* Reject message if it is wrong sort of message for socket */ 1354 if (msg_type(msg) > TIPC_DIRECT_MSG) 1355 return TIPC_ERR_NO_PORT; 1356 1357 if (sock->state == SS_READY) { 1358 if (msg_connected(msg)) 1359 return TIPC_ERR_NO_PORT; 1360 } else { 1361 res = filter_connect(tipc_sk(sk), &buf); 1362 if (res != TIPC_OK || buf == NULL) 1363 return res; 1364 } 1365 1366 /* Reject message if there isn't room to queue it */ 1367 if (sk_rmem_alloc_get(sk) + buf->truesize >= limit) 1368 return TIPC_ERR_OVERLOAD; 1369 1370 /* Enqueue message */ 1371 TIPC_SKB_CB(buf)->handle = 0; 1372 __skb_queue_tail(&sk->sk_receive_queue, buf); 1373 skb_set_owner_r(buf, sk); 1374 1375 sk->sk_data_ready(sk, 0); 1376 return TIPC_OK; 1377 } 1378 1379 /** 1380 * backlog_rcv - handle incoming message from backlog queue 1381 * @sk: socket 1382 * @buf: message 1383 * 1384 * Caller must hold socket lock, but not port lock. 1385 * 1386 * Returns 0 1387 */ 1388 static int backlog_rcv(struct sock *sk, struct sk_buff *buf) 1389 { 1390 u32 res; 1391 1392 res = filter_rcv(sk, buf); 1393 if (res) 1394 tipc_reject_msg(buf, res); 1395 return 0; 1396 } 1397 1398 /** 1399 * dispatch - handle incoming message 1400 * @tport: TIPC port that received message 1401 * @buf: message 1402 * 1403 * Called with port lock already taken. 1404 * 1405 * Returns TIPC error status code (TIPC_OK if message is not to be rejected) 1406 */ 1407 static u32 dispatch(struct tipc_port *tport, struct sk_buff *buf) 1408 { 1409 struct sock *sk = tport->sk; 1410 u32 res; 1411 1412 /* 1413 * Process message if socket is unlocked; otherwise add to backlog queue 1414 * 1415 * This code is based on sk_receive_skb(), but must be distinct from it 1416 * since a TIPC-specific filter/reject mechanism is utilized 1417 */ 1418 bh_lock_sock(sk); 1419 if (!sock_owned_by_user(sk)) { 1420 res = filter_rcv(sk, buf); 1421 } else { 1422 if (sk_add_backlog(sk, buf, rcvbuf_limit(sk, buf))) 1423 res = TIPC_ERR_OVERLOAD; 1424 else 1425 res = TIPC_OK; 1426 } 1427 bh_unlock_sock(sk); 1428 1429 return res; 1430 } 1431 1432 /** 1433 * wakeupdispatch - wake up port after congestion 1434 * @tport: port to wakeup 1435 * 1436 * Called with port lock already taken. 1437 */ 1438 static void wakeupdispatch(struct tipc_port *tport) 1439 { 1440 struct sock *sk = tport->sk; 1441 1442 sk->sk_write_space(sk); 1443 } 1444 1445 /** 1446 * connect - establish a connection to another TIPC port 1447 * @sock: socket structure 1448 * @dest: socket address for destination port 1449 * @destlen: size of socket address data structure 1450 * @flags: file-related flags associated with socket 1451 * 1452 * Returns 0 on success, errno otherwise 1453 */ 1454 static int connect(struct socket *sock, struct sockaddr *dest, int destlen, 1455 int flags) 1456 { 1457 struct sock *sk = sock->sk; 1458 struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest; 1459 struct msghdr m = {NULL,}; 1460 unsigned int timeout; 1461 int res; 1462 1463 lock_sock(sk); 1464 1465 /* For now, TIPC does not allow use of connect() with DGRAM/RDM types */ 1466 if (sock->state == SS_READY) { 1467 res = -EOPNOTSUPP; 1468 goto exit; 1469 } 1470 1471 /* 1472 * Reject connection attempt using multicast address 1473 * 1474 * Note: send_msg() validates the rest of the address fields, 1475 * so there's no need to do it here 1476 */ 1477 if (dst->addrtype == TIPC_ADDR_MCAST) { 1478 res = -EINVAL; 1479 goto exit; 1480 } 1481 1482 timeout = (flags & O_NONBLOCK) ? 0 : tipc_sk(sk)->conn_timeout; 1483 1484 switch (sock->state) { 1485 case SS_UNCONNECTED: 1486 /* Send a 'SYN-' to destination */ 1487 m.msg_name = dest; 1488 m.msg_namelen = destlen; 1489 1490 /* If connect is in non-blocking case, set MSG_DONTWAIT to 1491 * indicate send_msg() is never blocked. 1492 */ 1493 if (!timeout) 1494 m.msg_flags = MSG_DONTWAIT; 1495 1496 res = send_msg(NULL, sock, &m, 0); 1497 if ((res < 0) && (res != -EWOULDBLOCK)) 1498 goto exit; 1499 1500 /* Just entered SS_CONNECTING state; the only 1501 * difference is that return value in non-blocking 1502 * case is EINPROGRESS, rather than EALREADY. 1503 */ 1504 res = -EINPROGRESS; 1505 break; 1506 case SS_CONNECTING: 1507 res = -EALREADY; 1508 break; 1509 case SS_CONNECTED: 1510 res = -EISCONN; 1511 break; 1512 default: 1513 res = -EINVAL; 1514 goto exit; 1515 } 1516 1517 if (sock->state == SS_CONNECTING) { 1518 if (!timeout) 1519 goto exit; 1520 1521 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */ 1522 release_sock(sk); 1523 res = wait_event_interruptible_timeout(*sk_sleep(sk), 1524 sock->state != SS_CONNECTING, 1525 timeout ? (long)msecs_to_jiffies(timeout) 1526 : MAX_SCHEDULE_TIMEOUT); 1527 lock_sock(sk); 1528 if (res <= 0) { 1529 if (res == 0) 1530 res = -ETIMEDOUT; 1531 else 1532 ; /* leave "res" unchanged */ 1533 goto exit; 1534 } 1535 } 1536 1537 if (unlikely(sock->state == SS_DISCONNECTING)) 1538 res = sock_error(sk); 1539 else 1540 res = 0; 1541 1542 exit: 1543 release_sock(sk); 1544 return res; 1545 } 1546 1547 /** 1548 * listen - allow socket to listen for incoming connections 1549 * @sock: socket structure 1550 * @len: (unused) 1551 * 1552 * Returns 0 on success, errno otherwise 1553 */ 1554 static int listen(struct socket *sock, int len) 1555 { 1556 struct sock *sk = sock->sk; 1557 int res; 1558 1559 lock_sock(sk); 1560 1561 if (sock->state != SS_UNCONNECTED) 1562 res = -EINVAL; 1563 else { 1564 sock->state = SS_LISTENING; 1565 res = 0; 1566 } 1567 1568 release_sock(sk); 1569 return res; 1570 } 1571 1572 /** 1573 * accept - wait for connection request 1574 * @sock: listening socket 1575 * @newsock: new socket that is to be connected 1576 * @flags: file-related flags associated with socket 1577 * 1578 * Returns 0 on success, errno otherwise 1579 */ 1580 static int accept(struct socket *sock, struct socket *new_sock, int flags) 1581 { 1582 struct sock *new_sk, *sk = sock->sk; 1583 struct sk_buff *buf; 1584 struct tipc_sock *new_tsock; 1585 struct tipc_port *new_tport; 1586 struct tipc_msg *msg; 1587 u32 new_ref; 1588 1589 int res; 1590 1591 lock_sock(sk); 1592 1593 if (sock->state != SS_LISTENING) { 1594 res = -EINVAL; 1595 goto exit; 1596 } 1597 1598 while (skb_queue_empty(&sk->sk_receive_queue)) { 1599 if (flags & O_NONBLOCK) { 1600 res = -EWOULDBLOCK; 1601 goto exit; 1602 } 1603 release_sock(sk); 1604 res = wait_event_interruptible(*sk_sleep(sk), 1605 (!skb_queue_empty(&sk->sk_receive_queue))); 1606 lock_sock(sk); 1607 if (res) 1608 goto exit; 1609 } 1610 1611 buf = skb_peek(&sk->sk_receive_queue); 1612 1613 res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, 1); 1614 if (res) 1615 goto exit; 1616 1617 new_sk = new_sock->sk; 1618 new_tsock = tipc_sk(new_sk); 1619 new_tport = new_tsock->p; 1620 new_ref = new_tport->ref; 1621 msg = buf_msg(buf); 1622 1623 /* we lock on new_sk; but lockdep sees the lock on sk */ 1624 lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING); 1625 1626 /* 1627 * Reject any stray messages received by new socket 1628 * before the socket lock was taken (very, very unlikely) 1629 */ 1630 reject_rx_queue(new_sk); 1631 1632 /* Connect new socket to it's peer */ 1633 new_tsock->peer_name.ref = msg_origport(msg); 1634 new_tsock->peer_name.node = msg_orignode(msg); 1635 tipc_connect(new_ref, &new_tsock->peer_name); 1636 new_sock->state = SS_CONNECTED; 1637 1638 tipc_set_portimportance(new_ref, msg_importance(msg)); 1639 if (msg_named(msg)) { 1640 new_tport->conn_type = msg_nametype(msg); 1641 new_tport->conn_instance = msg_nameinst(msg); 1642 } 1643 1644 /* 1645 * Respond to 'SYN-' by discarding it & returning 'ACK'-. 1646 * Respond to 'SYN+' by queuing it on new socket. 1647 */ 1648 if (!msg_data_sz(msg)) { 1649 struct msghdr m = {NULL,}; 1650 1651 advance_rx_queue(sk); 1652 send_packet(NULL, new_sock, &m, 0); 1653 } else { 1654 __skb_dequeue(&sk->sk_receive_queue); 1655 __skb_queue_head(&new_sk->sk_receive_queue, buf); 1656 skb_set_owner_r(buf, new_sk); 1657 } 1658 release_sock(new_sk); 1659 1660 exit: 1661 release_sock(sk); 1662 return res; 1663 } 1664 1665 /** 1666 * shutdown - shutdown socket connection 1667 * @sock: socket structure 1668 * @how: direction to close (must be SHUT_RDWR) 1669 * 1670 * Terminates connection (if necessary), then purges socket's receive queue. 1671 * 1672 * Returns 0 on success, errno otherwise 1673 */ 1674 static int shutdown(struct socket *sock, int how) 1675 { 1676 struct sock *sk = sock->sk; 1677 struct tipc_port *tport = tipc_sk_port(sk); 1678 struct sk_buff *buf; 1679 int res; 1680 1681 if (how != SHUT_RDWR) 1682 return -EINVAL; 1683 1684 lock_sock(sk); 1685 1686 switch (sock->state) { 1687 case SS_CONNECTING: 1688 case SS_CONNECTED: 1689 1690 restart: 1691 /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */ 1692 buf = __skb_dequeue(&sk->sk_receive_queue); 1693 if (buf) { 1694 if (TIPC_SKB_CB(buf)->handle != 0) { 1695 kfree_skb(buf); 1696 goto restart; 1697 } 1698 tipc_disconnect(tport->ref); 1699 tipc_reject_msg(buf, TIPC_CONN_SHUTDOWN); 1700 } else { 1701 tipc_shutdown(tport->ref); 1702 } 1703 1704 sock->state = SS_DISCONNECTING; 1705 1706 /* fall through */ 1707 1708 case SS_DISCONNECTING: 1709 1710 /* Discard any unreceived messages */ 1711 __skb_queue_purge(&sk->sk_receive_queue); 1712 1713 /* Wake up anyone sleeping in poll */ 1714 sk->sk_state_change(sk); 1715 res = 0; 1716 break; 1717 1718 default: 1719 res = -ENOTCONN; 1720 } 1721 1722 release_sock(sk); 1723 return res; 1724 } 1725 1726 /** 1727 * setsockopt - set socket option 1728 * @sock: socket structure 1729 * @lvl: option level 1730 * @opt: option identifier 1731 * @ov: pointer to new option value 1732 * @ol: length of option value 1733 * 1734 * For stream sockets only, accepts and ignores all IPPROTO_TCP options 1735 * (to ease compatibility). 1736 * 1737 * Returns 0 on success, errno otherwise 1738 */ 1739 static int setsockopt(struct socket *sock, int lvl, int opt, char __user *ov, 1740 unsigned int ol) 1741 { 1742 struct sock *sk = sock->sk; 1743 struct tipc_port *tport = tipc_sk_port(sk); 1744 u32 value; 1745 int res; 1746 1747 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM)) 1748 return 0; 1749 if (lvl != SOL_TIPC) 1750 return -ENOPROTOOPT; 1751 if (ol < sizeof(value)) 1752 return -EINVAL; 1753 res = get_user(value, (u32 __user *)ov); 1754 if (res) 1755 return res; 1756 1757 lock_sock(sk); 1758 1759 switch (opt) { 1760 case TIPC_IMPORTANCE: 1761 res = tipc_set_portimportance(tport->ref, value); 1762 break; 1763 case TIPC_SRC_DROPPABLE: 1764 if (sock->type != SOCK_STREAM) 1765 res = tipc_set_portunreliable(tport->ref, value); 1766 else 1767 res = -ENOPROTOOPT; 1768 break; 1769 case TIPC_DEST_DROPPABLE: 1770 res = tipc_set_portunreturnable(tport->ref, value); 1771 break; 1772 case TIPC_CONN_TIMEOUT: 1773 tipc_sk(sk)->conn_timeout = value; 1774 /* no need to set "res", since already 0 at this point */ 1775 break; 1776 default: 1777 res = -EINVAL; 1778 } 1779 1780 release_sock(sk); 1781 1782 return res; 1783 } 1784 1785 /** 1786 * getsockopt - get socket option 1787 * @sock: socket structure 1788 * @lvl: option level 1789 * @opt: option identifier 1790 * @ov: receptacle for option value 1791 * @ol: receptacle for length of option value 1792 * 1793 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options 1794 * (to ease compatibility). 1795 * 1796 * Returns 0 on success, errno otherwise 1797 */ 1798 static int getsockopt(struct socket *sock, int lvl, int opt, char __user *ov, 1799 int __user *ol) 1800 { 1801 struct sock *sk = sock->sk; 1802 struct tipc_port *tport = tipc_sk_port(sk); 1803 int len; 1804 u32 value; 1805 int res; 1806 1807 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM)) 1808 return put_user(0, ol); 1809 if (lvl != SOL_TIPC) 1810 return -ENOPROTOOPT; 1811 res = get_user(len, ol); 1812 if (res) 1813 return res; 1814 1815 lock_sock(sk); 1816 1817 switch (opt) { 1818 case TIPC_IMPORTANCE: 1819 res = tipc_portimportance(tport->ref, &value); 1820 break; 1821 case TIPC_SRC_DROPPABLE: 1822 res = tipc_portunreliable(tport->ref, &value); 1823 break; 1824 case TIPC_DEST_DROPPABLE: 1825 res = tipc_portunreturnable(tport->ref, &value); 1826 break; 1827 case TIPC_CONN_TIMEOUT: 1828 value = tipc_sk(sk)->conn_timeout; 1829 /* no need to set "res", since already 0 at this point */ 1830 break; 1831 case TIPC_NODE_RECVQ_DEPTH: 1832 value = 0; /* was tipc_queue_size, now obsolete */ 1833 break; 1834 case TIPC_SOCK_RECVQ_DEPTH: 1835 value = skb_queue_len(&sk->sk_receive_queue); 1836 break; 1837 default: 1838 res = -EINVAL; 1839 } 1840 1841 release_sock(sk); 1842 1843 if (res) 1844 return res; /* "get" failed */ 1845 1846 if (len < sizeof(value)) 1847 return -EINVAL; 1848 1849 if (copy_to_user(ov, &value, sizeof(value))) 1850 return -EFAULT; 1851 1852 return put_user(sizeof(value), ol); 1853 } 1854 1855 /* Protocol switches for the various types of TIPC sockets */ 1856 1857 static const struct proto_ops msg_ops = { 1858 .owner = THIS_MODULE, 1859 .family = AF_TIPC, 1860 .release = release, 1861 .bind = bind, 1862 .connect = connect, 1863 .socketpair = sock_no_socketpair, 1864 .accept = sock_no_accept, 1865 .getname = get_name, 1866 .poll = poll, 1867 .ioctl = sock_no_ioctl, 1868 .listen = sock_no_listen, 1869 .shutdown = shutdown, 1870 .setsockopt = setsockopt, 1871 .getsockopt = getsockopt, 1872 .sendmsg = send_msg, 1873 .recvmsg = recv_msg, 1874 .mmap = sock_no_mmap, 1875 .sendpage = sock_no_sendpage 1876 }; 1877 1878 static const struct proto_ops packet_ops = { 1879 .owner = THIS_MODULE, 1880 .family = AF_TIPC, 1881 .release = release, 1882 .bind = bind, 1883 .connect = connect, 1884 .socketpair = sock_no_socketpair, 1885 .accept = accept, 1886 .getname = get_name, 1887 .poll = poll, 1888 .ioctl = sock_no_ioctl, 1889 .listen = listen, 1890 .shutdown = shutdown, 1891 .setsockopt = setsockopt, 1892 .getsockopt = getsockopt, 1893 .sendmsg = send_packet, 1894 .recvmsg = recv_msg, 1895 .mmap = sock_no_mmap, 1896 .sendpage = sock_no_sendpage 1897 }; 1898 1899 static const struct proto_ops stream_ops = { 1900 .owner = THIS_MODULE, 1901 .family = AF_TIPC, 1902 .release = release, 1903 .bind = bind, 1904 .connect = connect, 1905 .socketpair = sock_no_socketpair, 1906 .accept = accept, 1907 .getname = get_name, 1908 .poll = poll, 1909 .ioctl = sock_no_ioctl, 1910 .listen = listen, 1911 .shutdown = shutdown, 1912 .setsockopt = setsockopt, 1913 .getsockopt = getsockopt, 1914 .sendmsg = send_stream, 1915 .recvmsg = recv_stream, 1916 .mmap = sock_no_mmap, 1917 .sendpage = sock_no_sendpage 1918 }; 1919 1920 static const struct net_proto_family tipc_family_ops = { 1921 .owner = THIS_MODULE, 1922 .family = AF_TIPC, 1923 .create = tipc_sk_create 1924 }; 1925 1926 static struct proto tipc_proto = { 1927 .name = "TIPC", 1928 .owner = THIS_MODULE, 1929 .obj_size = sizeof(struct tipc_sock), 1930 .sysctl_rmem = sysctl_tipc_rmem 1931 }; 1932 1933 static struct proto tipc_proto_kern = { 1934 .name = "TIPC", 1935 .obj_size = sizeof(struct tipc_sock), 1936 .sysctl_rmem = sysctl_tipc_rmem 1937 }; 1938 1939 /** 1940 * tipc_socket_init - initialize TIPC socket interface 1941 * 1942 * Returns 0 on success, errno otherwise 1943 */ 1944 int tipc_socket_init(void) 1945 { 1946 int res; 1947 1948 res = proto_register(&tipc_proto, 1); 1949 if (res) { 1950 pr_err("Failed to register TIPC protocol type\n"); 1951 goto out; 1952 } 1953 1954 res = sock_register(&tipc_family_ops); 1955 if (res) { 1956 pr_err("Failed to register TIPC socket type\n"); 1957 proto_unregister(&tipc_proto); 1958 goto out; 1959 } 1960 1961 sockets_enabled = 1; 1962 out: 1963 return res; 1964 } 1965 1966 /** 1967 * tipc_socket_stop - stop TIPC socket interface 1968 */ 1969 void tipc_socket_stop(void) 1970 { 1971 if (!sockets_enabled) 1972 return; 1973 1974 sockets_enabled = 0; 1975 sock_unregister(tipc_family_ops.family); 1976 proto_unregister(&tipc_proto); 1977 } 1978