net/tipc/socket.c

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

DEFINITIONS
