xref: /linux/net/tipc/socket.c (revision 54c9ee39923c0ffa27b99a1b7321e0d07408c97b)
1 /*
2  * net/tipc/socket.c: TIPC socket API
3  *
4  * Copyright (c) 2001-2007, 2012-2015, 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 "core.h"
39 #include "name_table.h"
40 #include "node.h"
41 #include "link.h"
42 #include "name_distr.h"
43 #include "socket.h"
44 #include "bcast.h"
45 
46 #define SS_LISTENING		-1	/* socket is listening */
47 #define SS_READY		-2	/* socket is connectionless */
48 
49 #define CONN_TIMEOUT_DEFAULT	8000	/* default connect timeout = 8s */
50 #define CONN_PROBING_INTERVAL	msecs_to_jiffies(3600000)  /* [ms] => 1 h */
51 #define TIPC_FWD_MSG		1
52 #define TIPC_CONN_OK		0
53 #define TIPC_CONN_PROBING	1
54 #define TIPC_MAX_PORT		0xffffffff
55 #define TIPC_MIN_PORT		1
56 
57 /**
58  * struct tipc_sock - TIPC socket structure
59  * @sk: socket - interacts with 'port' and with user via the socket API
60  * @connected: non-zero if port is currently connected to a peer port
61  * @conn_type: TIPC type used when connection was established
62  * @conn_instance: TIPC instance used when connection was established
63  * @published: non-zero if port has one or more associated names
64  * @max_pkt: maximum packet size "hint" used when building messages sent by port
65  * @portid: unique port identity in TIPC socket hash table
66  * @phdr: preformatted message header used when sending messages
67  * @port_list: adjacent ports in TIPC's global list of ports
68  * @publications: list of publications for port
69  * @pub_count: total # of publications port has made during its lifetime
70  * @probing_state:
71  * @probing_intv:
72  * @conn_timeout: the time we can wait for an unresponded setup request
73  * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue
74  * @link_cong: non-zero if owner must sleep because of link congestion
75  * @sent_unacked: # messages sent by socket, and not yet acked by peer
76  * @rcv_unacked: # messages read by user, but not yet acked back to peer
77  * @remote: 'connected' peer for dgram/rdm
78  * @node: hash table node
79  * @rcu: rcu struct for tipc_sock
80  */
81 struct tipc_sock {
82 	struct sock sk;
83 	int connected;
84 	u32 conn_type;
85 	u32 conn_instance;
86 	int published;
87 	u32 max_pkt;
88 	u32 portid;
89 	struct tipc_msg phdr;
90 	struct list_head sock_list;
91 	struct list_head publications;
92 	u32 pub_count;
93 	u32 probing_state;
94 	unsigned long probing_intv;
95 	uint conn_timeout;
96 	atomic_t dupl_rcvcnt;
97 	bool link_cong;
98 	uint sent_unacked;
99 	uint rcv_unacked;
100 	struct sockaddr_tipc remote;
101 	struct rhash_head node;
102 	struct rcu_head rcu;
103 };
104 
105 static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *skb);
106 static void tipc_data_ready(struct sock *sk);
107 static void tipc_write_space(struct sock *sk);
108 static int tipc_release(struct socket *sock);
109 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags);
110 static int tipc_wait_for_sndmsg(struct socket *sock, long *timeo_p);
111 static void tipc_sk_timeout(unsigned long data);
112 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
113 			   struct tipc_name_seq const *seq);
114 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
115 			    struct tipc_name_seq const *seq);
116 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid);
117 static int tipc_sk_insert(struct tipc_sock *tsk);
118 static void tipc_sk_remove(struct tipc_sock *tsk);
119 static int __tipc_send_stream(struct socket *sock, struct msghdr *m,
120 			      size_t dsz);
121 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dsz);
122 
123 static const struct proto_ops packet_ops;
124 static const struct proto_ops stream_ops;
125 static const struct proto_ops msg_ops;
126 static struct proto tipc_proto;
127 
128 static const struct nla_policy tipc_nl_sock_policy[TIPC_NLA_SOCK_MAX + 1] = {
129 	[TIPC_NLA_SOCK_UNSPEC]		= { .type = NLA_UNSPEC },
130 	[TIPC_NLA_SOCK_ADDR]		= { .type = NLA_U32 },
131 	[TIPC_NLA_SOCK_REF]		= { .type = NLA_U32 },
132 	[TIPC_NLA_SOCK_CON]		= { .type = NLA_NESTED },
133 	[TIPC_NLA_SOCK_HAS_PUBL]	= { .type = NLA_FLAG }
134 };
135 
136 static const struct rhashtable_params tsk_rht_params;
137 
138 /*
139  * Revised TIPC socket locking policy:
140  *
141  * Most socket operations take the standard socket lock when they start
142  * and hold it until they finish (or until they need to sleep).  Acquiring
143  * this lock grants the owner exclusive access to the fields of the socket
144  * data structures, with the exception of the backlog queue.  A few socket
145  * operations can be done without taking the socket lock because they only
146  * read socket information that never changes during the life of the socket.
147  *
148  * Socket operations may acquire the lock for the associated TIPC port if they
149  * need to perform an operation on the port.  If any routine needs to acquire
150  * both the socket lock and the port lock it must take the socket lock first
151  * to avoid the risk of deadlock.
152  *
153  * The dispatcher handling incoming messages cannot grab the socket lock in
154  * the standard fashion, since invoked it runs at the BH level and cannot block.
155  * Instead, it checks to see if the socket lock is currently owned by someone,
156  * and either handles the message itself or adds it to the socket's backlog
157  * queue; in the latter case the queued message is processed once the process
158  * owning the socket lock releases it.
159  *
160  * NOTE: Releasing the socket lock while an operation is sleeping overcomes
161  * the problem of a blocked socket operation preventing any other operations
162  * from occurring.  However, applications must be careful if they have
163  * multiple threads trying to send (or receive) on the same socket, as these
164  * operations might interfere with each other.  For example, doing a connect
165  * and a receive at the same time might allow the receive to consume the
166  * ACK message meant for the connect.  While additional work could be done
167  * to try and overcome this, it doesn't seem to be worthwhile at the present.
168  *
169  * NOTE: Releasing the socket lock while an operation is sleeping also ensures
170  * that another operation that must be performed in a non-blocking manner is
171  * not delayed for very long because the lock has already been taken.
172  *
173  * NOTE: This code assumes that certain fields of a port/socket pair are
174  * constant over its lifetime; such fields can be examined without taking
175  * the socket lock and/or port lock, and do not need to be re-read even
176  * after resuming processing after waiting.  These fields include:
177  *   - socket type
178  *   - pointer to socket sk structure (aka tipc_sock structure)
179  *   - pointer to port structure
180  *   - port reference
181  */
182 
183 static u32 tsk_own_node(struct tipc_sock *tsk)
184 {
185 	return msg_prevnode(&tsk->phdr);
186 }
187 
188 static u32 tsk_peer_node(struct tipc_sock *tsk)
189 {
190 	return msg_destnode(&tsk->phdr);
191 }
192 
193 static u32 tsk_peer_port(struct tipc_sock *tsk)
194 {
195 	return msg_destport(&tsk->phdr);
196 }
197 
198 static  bool tsk_unreliable(struct tipc_sock *tsk)
199 {
200 	return msg_src_droppable(&tsk->phdr) != 0;
201 }
202 
203 static void tsk_set_unreliable(struct tipc_sock *tsk, bool unreliable)
204 {
205 	msg_set_src_droppable(&tsk->phdr, unreliable ? 1 : 0);
206 }
207 
208 static bool tsk_unreturnable(struct tipc_sock *tsk)
209 {
210 	return msg_dest_droppable(&tsk->phdr) != 0;
211 }
212 
213 static void tsk_set_unreturnable(struct tipc_sock *tsk, bool unreturnable)
214 {
215 	msg_set_dest_droppable(&tsk->phdr, unreturnable ? 1 : 0);
216 }
217 
218 static int tsk_importance(struct tipc_sock *tsk)
219 {
220 	return msg_importance(&tsk->phdr);
221 }
222 
223 static int tsk_set_importance(struct tipc_sock *tsk, int imp)
224 {
225 	if (imp > TIPC_CRITICAL_IMPORTANCE)
226 		return -EINVAL;
227 	msg_set_importance(&tsk->phdr, (u32)imp);
228 	return 0;
229 }
230 
231 static struct tipc_sock *tipc_sk(const struct sock *sk)
232 {
233 	return container_of(sk, struct tipc_sock, sk);
234 }
235 
236 static int tsk_conn_cong(struct tipc_sock *tsk)
237 {
238 	return tsk->sent_unacked >= TIPC_FLOWCTRL_WIN;
239 }
240 
241 /**
242  * tsk_advance_rx_queue - discard first buffer in socket receive queue
243  *
244  * Caller must hold socket lock
245  */
246 static void tsk_advance_rx_queue(struct sock *sk)
247 {
248 	kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
249 }
250 
251 /**
252  * tsk_rej_rx_queue - reject all buffers in socket receive queue
253  *
254  * Caller must hold socket lock
255  */
256 static void tsk_rej_rx_queue(struct sock *sk)
257 {
258 	struct sk_buff *skb;
259 	u32 dnode;
260 	u32 own_node = tsk_own_node(tipc_sk(sk));
261 
262 	while ((skb = __skb_dequeue(&sk->sk_receive_queue))) {
263 		if (tipc_msg_reverse(own_node, skb, &dnode, TIPC_ERR_NO_PORT))
264 			tipc_node_xmit_skb(sock_net(sk), skb, dnode, 0);
265 	}
266 }
267 
268 /* tsk_peer_msg - verify if message was sent by connected port's peer
269  *
270  * Handles cases where the node's network address has changed from
271  * the default of <0.0.0> to its configured setting.
272  */
273 static bool tsk_peer_msg(struct tipc_sock *tsk, struct tipc_msg *msg)
274 {
275 	struct tipc_net *tn = net_generic(sock_net(&tsk->sk), tipc_net_id);
276 	u32 peer_port = tsk_peer_port(tsk);
277 	u32 orig_node;
278 	u32 peer_node;
279 
280 	if (unlikely(!tsk->connected))
281 		return false;
282 
283 	if (unlikely(msg_origport(msg) != peer_port))
284 		return false;
285 
286 	orig_node = msg_orignode(msg);
287 	peer_node = tsk_peer_node(tsk);
288 
289 	if (likely(orig_node == peer_node))
290 		return true;
291 
292 	if (!orig_node && (peer_node == tn->own_addr))
293 		return true;
294 
295 	if (!peer_node && (orig_node == tn->own_addr))
296 		return true;
297 
298 	return false;
299 }
300 
301 /**
302  * tipc_sk_create - create a TIPC socket
303  * @net: network namespace (must be default network)
304  * @sock: pre-allocated socket structure
305  * @protocol: protocol indicator (must be 0)
306  * @kern: caused by kernel or by userspace?
307  *
308  * This routine creates additional data structures used by the TIPC socket,
309  * initializes them, and links them together.
310  *
311  * Returns 0 on success, errno otherwise
312  */
313 static int tipc_sk_create(struct net *net, struct socket *sock,
314 			  int protocol, int kern)
315 {
316 	struct tipc_net *tn;
317 	const struct proto_ops *ops;
318 	socket_state state;
319 	struct sock *sk;
320 	struct tipc_sock *tsk;
321 	struct tipc_msg *msg;
322 
323 	/* Validate arguments */
324 	if (unlikely(protocol != 0))
325 		return -EPROTONOSUPPORT;
326 
327 	switch (sock->type) {
328 	case SOCK_STREAM:
329 		ops = &stream_ops;
330 		state = SS_UNCONNECTED;
331 		break;
332 	case SOCK_SEQPACKET:
333 		ops = &packet_ops;
334 		state = SS_UNCONNECTED;
335 		break;
336 	case SOCK_DGRAM:
337 	case SOCK_RDM:
338 		ops = &msg_ops;
339 		state = SS_READY;
340 		break;
341 	default:
342 		return -EPROTOTYPE;
343 	}
344 
345 	/* Allocate socket's protocol area */
346 	sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto, kern);
347 	if (sk == NULL)
348 		return -ENOMEM;
349 
350 	tsk = tipc_sk(sk);
351 	tsk->max_pkt = MAX_PKT_DEFAULT;
352 	INIT_LIST_HEAD(&tsk->publications);
353 	msg = &tsk->phdr;
354 	tn = net_generic(sock_net(sk), tipc_net_id);
355 	tipc_msg_init(tn->own_addr, msg, TIPC_LOW_IMPORTANCE, TIPC_NAMED_MSG,
356 		      NAMED_H_SIZE, 0);
357 
358 	/* Finish initializing socket data structures */
359 	sock->ops = ops;
360 	sock->state = state;
361 	sock_init_data(sock, sk);
362 	if (tipc_sk_insert(tsk)) {
363 		pr_warn("Socket create failed; port numbrer exhausted\n");
364 		return -EINVAL;
365 	}
366 	msg_set_origport(msg, tsk->portid);
367 	setup_timer(&sk->sk_timer, tipc_sk_timeout, (unsigned long)tsk);
368 	sk->sk_backlog_rcv = tipc_backlog_rcv;
369 	sk->sk_rcvbuf = sysctl_tipc_rmem[1];
370 	sk->sk_data_ready = tipc_data_ready;
371 	sk->sk_write_space = tipc_write_space;
372 	tsk->conn_timeout = CONN_TIMEOUT_DEFAULT;
373 	tsk->sent_unacked = 0;
374 	atomic_set(&tsk->dupl_rcvcnt, 0);
375 
376 	if (sock->state == SS_READY) {
377 		tsk_set_unreturnable(tsk, true);
378 		if (sock->type == SOCK_DGRAM)
379 			tsk_set_unreliable(tsk, true);
380 	}
381 	return 0;
382 }
383 
384 static void tipc_sk_callback(struct rcu_head *head)
385 {
386 	struct tipc_sock *tsk = container_of(head, struct tipc_sock, rcu);
387 
388 	sock_put(&tsk->sk);
389 }
390 
391 /**
392  * tipc_release - destroy a TIPC socket
393  * @sock: socket to destroy
394  *
395  * This routine cleans up any messages that are still queued on the socket.
396  * For DGRAM and RDM socket types, all queued messages are rejected.
397  * For SEQPACKET and STREAM socket types, the first message is rejected
398  * and any others are discarded.  (If the first message on a STREAM socket
399  * is partially-read, it is discarded and the next one is rejected instead.)
400  *
401  * NOTE: Rejected messages are not necessarily returned to the sender!  They
402  * are returned or discarded according to the "destination droppable" setting
403  * specified for the message by the sender.
404  *
405  * Returns 0 on success, errno otherwise
406  */
407 static int tipc_release(struct socket *sock)
408 {
409 	struct sock *sk = sock->sk;
410 	struct net *net;
411 	struct tipc_sock *tsk;
412 	struct sk_buff *skb;
413 	u32 dnode;
414 
415 	/*
416 	 * Exit if socket isn't fully initialized (occurs when a failed accept()
417 	 * releases a pre-allocated child socket that was never used)
418 	 */
419 	if (sk == NULL)
420 		return 0;
421 
422 	net = sock_net(sk);
423 	tsk = tipc_sk(sk);
424 	lock_sock(sk);
425 
426 	/*
427 	 * Reject all unreceived messages, except on an active connection
428 	 * (which disconnects locally & sends a 'FIN+' to peer)
429 	 */
430 	dnode = tsk_peer_node(tsk);
431 	while (sock->state != SS_DISCONNECTING) {
432 		skb = __skb_dequeue(&sk->sk_receive_queue);
433 		if (skb == NULL)
434 			break;
435 		if (TIPC_SKB_CB(skb)->handle != NULL)
436 			kfree_skb(skb);
437 		else {
438 			if ((sock->state == SS_CONNECTING) ||
439 			    (sock->state == SS_CONNECTED)) {
440 				sock->state = SS_DISCONNECTING;
441 				tsk->connected = 0;
442 				tipc_node_remove_conn(net, dnode, tsk->portid);
443 			}
444 			if (tipc_msg_reverse(tsk_own_node(tsk), skb, &dnode,
445 					     TIPC_ERR_NO_PORT))
446 				tipc_node_xmit_skb(net, skb, dnode, 0);
447 		}
448 	}
449 
450 	tipc_sk_withdraw(tsk, 0, NULL);
451 	sk_stop_timer(sk, &sk->sk_timer);
452 	tipc_sk_remove(tsk);
453 	if (tsk->connected) {
454 		skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE,
455 				      TIPC_CONN_MSG, SHORT_H_SIZE, 0, dnode,
456 				      tsk_own_node(tsk), tsk_peer_port(tsk),
457 				      tsk->portid, TIPC_ERR_NO_PORT);
458 		if (skb)
459 			tipc_node_xmit_skb(net, skb, dnode, tsk->portid);
460 		tipc_node_remove_conn(net, dnode, tsk->portid);
461 	}
462 
463 	/* Discard any remaining (connection-based) messages in receive queue */
464 	__skb_queue_purge(&sk->sk_receive_queue);
465 
466 	/* Reject any messages that accumulated in backlog queue */
467 	sock->state = SS_DISCONNECTING;
468 	release_sock(sk);
469 
470 	call_rcu(&tsk->rcu, tipc_sk_callback);
471 	sock->sk = NULL;
472 
473 	return 0;
474 }
475 
476 /**
477  * tipc_bind - associate or disassocate TIPC name(s) with a socket
478  * @sock: socket structure
479  * @uaddr: socket address describing name(s) and desired operation
480  * @uaddr_len: size of socket address data structure
481  *
482  * Name and name sequence binding is indicated using a positive scope value;
483  * a negative scope value unbinds the specified name.  Specifying no name
484  * (i.e. a socket address length of 0) unbinds all names from the socket.
485  *
486  * Returns 0 on success, errno otherwise
487  *
488  * NOTE: This routine doesn't need to take the socket lock since it doesn't
489  *       access any non-constant socket information.
490  */
491 static int tipc_bind(struct socket *sock, struct sockaddr *uaddr,
492 		     int uaddr_len)
493 {
494 	struct sock *sk = sock->sk;
495 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
496 	struct tipc_sock *tsk = tipc_sk(sk);
497 	int res = -EINVAL;
498 
499 	lock_sock(sk);
500 	if (unlikely(!uaddr_len)) {
501 		res = tipc_sk_withdraw(tsk, 0, NULL);
502 		goto exit;
503 	}
504 
505 	if (uaddr_len < sizeof(struct sockaddr_tipc)) {
506 		res = -EINVAL;
507 		goto exit;
508 	}
509 	if (addr->family != AF_TIPC) {
510 		res = -EAFNOSUPPORT;
511 		goto exit;
512 	}
513 
514 	if (addr->addrtype == TIPC_ADDR_NAME)
515 		addr->addr.nameseq.upper = addr->addr.nameseq.lower;
516 	else if (addr->addrtype != TIPC_ADDR_NAMESEQ) {
517 		res = -EAFNOSUPPORT;
518 		goto exit;
519 	}
520 
521 	if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) &&
522 	    (addr->addr.nameseq.type != TIPC_TOP_SRV) &&
523 	    (addr->addr.nameseq.type != TIPC_CFG_SRV)) {
524 		res = -EACCES;
525 		goto exit;
526 	}
527 
528 	res = (addr->scope > 0) ?
529 		tipc_sk_publish(tsk, addr->scope, &addr->addr.nameseq) :
530 		tipc_sk_withdraw(tsk, -addr->scope, &addr->addr.nameseq);
531 exit:
532 	release_sock(sk);
533 	return res;
534 }
535 
536 /**
537  * tipc_getname - get port ID of socket or peer socket
538  * @sock: socket structure
539  * @uaddr: area for returned socket address
540  * @uaddr_len: area for returned length of socket address
541  * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
542  *
543  * Returns 0 on success, errno otherwise
544  *
545  * NOTE: This routine doesn't need to take the socket lock since it only
546  *       accesses socket information that is unchanging (or which changes in
547  *       a completely predictable manner).
548  */
549 static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
550 			int *uaddr_len, int peer)
551 {
552 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
553 	struct tipc_sock *tsk = tipc_sk(sock->sk);
554 	struct tipc_net *tn = net_generic(sock_net(sock->sk), tipc_net_id);
555 
556 	memset(addr, 0, sizeof(*addr));
557 	if (peer) {
558 		if ((sock->state != SS_CONNECTED) &&
559 			((peer != 2) || (sock->state != SS_DISCONNECTING)))
560 			return -ENOTCONN;
561 		addr->addr.id.ref = tsk_peer_port(tsk);
562 		addr->addr.id.node = tsk_peer_node(tsk);
563 	} else {
564 		addr->addr.id.ref = tsk->portid;
565 		addr->addr.id.node = tn->own_addr;
566 	}
567 
568 	*uaddr_len = sizeof(*addr);
569 	addr->addrtype = TIPC_ADDR_ID;
570 	addr->family = AF_TIPC;
571 	addr->scope = 0;
572 	addr->addr.name.domain = 0;
573 
574 	return 0;
575 }
576 
577 /**
578  * tipc_poll - read and possibly block on pollmask
579  * @file: file structure associated with the socket
580  * @sock: socket for which to calculate the poll bits
581  * @wait: ???
582  *
583  * Returns pollmask value
584  *
585  * COMMENTARY:
586  * It appears that the usual socket locking mechanisms are not useful here
587  * since the pollmask info is potentially out-of-date the moment this routine
588  * exits.  TCP and other protocols seem to rely on higher level poll routines
589  * to handle any preventable race conditions, so TIPC will do the same ...
590  *
591  * TIPC sets the returned events as follows:
592  *
593  * socket state		flags set
594  * ------------		---------
595  * unconnected		no read flags
596  *			POLLOUT if port is not congested
597  *
598  * connecting		POLLIN/POLLRDNORM if ACK/NACK in rx queue
599  *			no write flags
600  *
601  * connected		POLLIN/POLLRDNORM if data in rx queue
602  *			POLLOUT if port is not congested
603  *
604  * disconnecting	POLLIN/POLLRDNORM/POLLHUP
605  *			no write flags
606  *
607  * listening		POLLIN if SYN in rx queue
608  *			no write flags
609  *
610  * ready		POLLIN/POLLRDNORM if data in rx queue
611  * [connectionless]	POLLOUT (since port cannot be congested)
612  *
613  * IMPORTANT: The fact that a read or write operation is indicated does NOT
614  * imply that the operation will succeed, merely that it should be performed
615  * and will not block.
616  */
617 static unsigned int tipc_poll(struct file *file, struct socket *sock,
618 			      poll_table *wait)
619 {
620 	struct sock *sk = sock->sk;
621 	struct tipc_sock *tsk = tipc_sk(sk);
622 	u32 mask = 0;
623 
624 	sock_poll_wait(file, sk_sleep(sk), wait);
625 
626 	switch ((int)sock->state) {
627 	case SS_UNCONNECTED:
628 		if (!tsk->link_cong)
629 			mask |= POLLOUT;
630 		break;
631 	case SS_READY:
632 	case SS_CONNECTED:
633 		if (!tsk->link_cong && !tsk_conn_cong(tsk))
634 			mask |= POLLOUT;
635 		/* fall thru' */
636 	case SS_CONNECTING:
637 	case SS_LISTENING:
638 		if (!skb_queue_empty(&sk->sk_receive_queue))
639 			mask |= (POLLIN | POLLRDNORM);
640 		break;
641 	case SS_DISCONNECTING:
642 		mask = (POLLIN | POLLRDNORM | POLLHUP);
643 		break;
644 	}
645 
646 	return mask;
647 }
648 
649 /**
650  * tipc_sendmcast - send multicast message
651  * @sock: socket structure
652  * @seq: destination address
653  * @msg: message to send
654  * @dsz: total length of message data
655  * @timeo: timeout to wait for wakeup
656  *
657  * Called from function tipc_sendmsg(), which has done all sanity checks
658  * Returns the number of bytes sent on success, or errno
659  */
660 static int tipc_sendmcast(struct  socket *sock, struct tipc_name_seq *seq,
661 			  struct msghdr *msg, size_t dsz, long timeo)
662 {
663 	struct sock *sk = sock->sk;
664 	struct tipc_sock *tsk = tipc_sk(sk);
665 	struct net *net = sock_net(sk);
666 	struct tipc_msg *mhdr = &tsk->phdr;
667 	struct sk_buff_head *pktchain = &sk->sk_write_queue;
668 	struct iov_iter save = msg->msg_iter;
669 	uint mtu;
670 	int rc;
671 
672 	msg_set_type(mhdr, TIPC_MCAST_MSG);
673 	msg_set_lookup_scope(mhdr, TIPC_CLUSTER_SCOPE);
674 	msg_set_destport(mhdr, 0);
675 	msg_set_destnode(mhdr, 0);
676 	msg_set_nametype(mhdr, seq->type);
677 	msg_set_namelower(mhdr, seq->lower);
678 	msg_set_nameupper(mhdr, seq->upper);
679 	msg_set_hdr_sz(mhdr, MCAST_H_SIZE);
680 
681 new_mtu:
682 	mtu = tipc_bclink_get_mtu();
683 	rc = tipc_msg_build(mhdr, msg, 0, dsz, mtu, pktchain);
684 	if (unlikely(rc < 0))
685 		return rc;
686 
687 	do {
688 		rc = tipc_bclink_xmit(net, pktchain);
689 		if (likely(!rc))
690 			return dsz;
691 
692 		if (rc == -ELINKCONG) {
693 			tsk->link_cong = 1;
694 			rc = tipc_wait_for_sndmsg(sock, &timeo);
695 			if (!rc)
696 				continue;
697 		}
698 		__skb_queue_purge(pktchain);
699 		if (rc == -EMSGSIZE) {
700 			msg->msg_iter = save;
701 			goto new_mtu;
702 		}
703 		break;
704 	} while (1);
705 	return rc;
706 }
707 
708 /**
709  * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
710  * @arrvq: queue with arriving messages, to be cloned after destination lookup
711  * @inputq: queue with cloned messages, delivered to socket after dest lookup
712  *
713  * Multi-threaded: parallel calls with reference to same queues may occur
714  */
715 void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
716 		       struct sk_buff_head *inputq)
717 {
718 	struct tipc_msg *msg;
719 	struct tipc_plist dports;
720 	u32 portid;
721 	u32 scope = TIPC_CLUSTER_SCOPE;
722 	struct sk_buff_head tmpq;
723 	uint hsz;
724 	struct sk_buff *skb, *_skb;
725 
726 	__skb_queue_head_init(&tmpq);
727 	tipc_plist_init(&dports);
728 
729 	skb = tipc_skb_peek(arrvq, &inputq->lock);
730 	for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) {
731 		msg = buf_msg(skb);
732 		hsz = skb_headroom(skb) + msg_hdr_sz(msg);
733 
734 		if (in_own_node(net, msg_orignode(msg)))
735 			scope = TIPC_NODE_SCOPE;
736 
737 		/* Create destination port list and message clones: */
738 		tipc_nametbl_mc_translate(net,
739 					  msg_nametype(msg), msg_namelower(msg),
740 					  msg_nameupper(msg), scope, &dports);
741 		portid = tipc_plist_pop(&dports);
742 		for (; portid; portid = tipc_plist_pop(&dports)) {
743 			_skb = __pskb_copy(skb, hsz, GFP_ATOMIC);
744 			if (_skb) {
745 				msg_set_destport(buf_msg(_skb), portid);
746 				__skb_queue_tail(&tmpq, _skb);
747 				continue;
748 			}
749 			pr_warn("Failed to clone mcast rcv buffer\n");
750 		}
751 		/* Append to inputq if not already done by other thread */
752 		spin_lock_bh(&inputq->lock);
753 		if (skb_peek(arrvq) == skb) {
754 			skb_queue_splice_tail_init(&tmpq, inputq);
755 			kfree_skb(__skb_dequeue(arrvq));
756 		}
757 		spin_unlock_bh(&inputq->lock);
758 		__skb_queue_purge(&tmpq);
759 		kfree_skb(skb);
760 	}
761 	tipc_sk_rcv(net, inputq);
762 }
763 
764 /**
765  * tipc_sk_proto_rcv - receive a connection mng protocol message
766  * @tsk: receiving socket
767  * @skb: pointer to message buffer. Set to NULL if buffer is consumed.
768  */
769 static void tipc_sk_proto_rcv(struct tipc_sock *tsk, struct sk_buff **skb)
770 {
771 	struct tipc_msg *msg = buf_msg(*skb);
772 	int conn_cong;
773 	u32 dnode;
774 	u32 own_node = tsk_own_node(tsk);
775 	/* Ignore if connection cannot be validated: */
776 	if (!tsk_peer_msg(tsk, msg))
777 		goto exit;
778 
779 	tsk->probing_state = TIPC_CONN_OK;
780 
781 	if (msg_type(msg) == CONN_ACK) {
782 		conn_cong = tsk_conn_cong(tsk);
783 		tsk->sent_unacked -= msg_msgcnt(msg);
784 		if (conn_cong)
785 			tsk->sk.sk_write_space(&tsk->sk);
786 	} else if (msg_type(msg) == CONN_PROBE) {
787 		if (tipc_msg_reverse(own_node, *skb, &dnode, TIPC_OK)) {
788 			msg_set_type(msg, CONN_PROBE_REPLY);
789 			return;
790 		}
791 	}
792 	/* Do nothing if msg_type() == CONN_PROBE_REPLY */
793 exit:
794 	kfree_skb(*skb);
795 	*skb = NULL;
796 }
797 
798 static int tipc_wait_for_sndmsg(struct socket *sock, long *timeo_p)
799 {
800 	struct sock *sk = sock->sk;
801 	struct tipc_sock *tsk = tipc_sk(sk);
802 	DEFINE_WAIT(wait);
803 	int done;
804 
805 	do {
806 		int err = sock_error(sk);
807 		if (err)
808 			return err;
809 		if (sock->state == SS_DISCONNECTING)
810 			return -EPIPE;
811 		if (!*timeo_p)
812 			return -EAGAIN;
813 		if (signal_pending(current))
814 			return sock_intr_errno(*timeo_p);
815 
816 		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
817 		done = sk_wait_event(sk, timeo_p, !tsk->link_cong);
818 		finish_wait(sk_sleep(sk), &wait);
819 	} while (!done);
820 	return 0;
821 }
822 
823 /**
824  * tipc_sendmsg - send message in connectionless manner
825  * @sock: socket structure
826  * @m: message to send
827  * @dsz: amount of user data to be sent
828  *
829  * Message must have an destination specified explicitly.
830  * Used for SOCK_RDM and SOCK_DGRAM messages,
831  * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
832  * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
833  *
834  * Returns the number of bytes sent on success, or errno otherwise
835  */
836 static int tipc_sendmsg(struct socket *sock,
837 			struct msghdr *m, size_t dsz)
838 {
839 	struct sock *sk = sock->sk;
840 	int ret;
841 
842 	lock_sock(sk);
843 	ret = __tipc_sendmsg(sock, m, dsz);
844 	release_sock(sk);
845 
846 	return ret;
847 }
848 
849 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dsz)
850 {
851 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
852 	struct sock *sk = sock->sk;
853 	struct tipc_sock *tsk = tipc_sk(sk);
854 	struct net *net = sock_net(sk);
855 	struct tipc_msg *mhdr = &tsk->phdr;
856 	u32 dnode, dport;
857 	struct sk_buff_head *pktchain = &sk->sk_write_queue;
858 	struct sk_buff *skb;
859 	struct tipc_name_seq *seq;
860 	struct iov_iter save;
861 	u32 mtu;
862 	long timeo;
863 	int rc;
864 
865 	if (dsz > TIPC_MAX_USER_MSG_SIZE)
866 		return -EMSGSIZE;
867 	if (unlikely(!dest)) {
868 		if (tsk->connected && sock->state == SS_READY)
869 			dest = &tsk->remote;
870 		else
871 			return -EDESTADDRREQ;
872 	} else if (unlikely(m->msg_namelen < sizeof(*dest)) ||
873 		   dest->family != AF_TIPC) {
874 		return -EINVAL;
875 	}
876 	if (unlikely(sock->state != SS_READY)) {
877 		if (sock->state == SS_LISTENING)
878 			return -EPIPE;
879 		if (sock->state != SS_UNCONNECTED)
880 			return -EISCONN;
881 		if (tsk->published)
882 			return -EOPNOTSUPP;
883 		if (dest->addrtype == TIPC_ADDR_NAME) {
884 			tsk->conn_type = dest->addr.name.name.type;
885 			tsk->conn_instance = dest->addr.name.name.instance;
886 		}
887 	}
888 	seq = &dest->addr.nameseq;
889 	timeo = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
890 
891 	if (dest->addrtype == TIPC_ADDR_MCAST) {
892 		return tipc_sendmcast(sock, seq, m, dsz, timeo);
893 	} else if (dest->addrtype == TIPC_ADDR_NAME) {
894 		u32 type = dest->addr.name.name.type;
895 		u32 inst = dest->addr.name.name.instance;
896 		u32 domain = dest->addr.name.domain;
897 
898 		dnode = domain;
899 		msg_set_type(mhdr, TIPC_NAMED_MSG);
900 		msg_set_hdr_sz(mhdr, NAMED_H_SIZE);
901 		msg_set_nametype(mhdr, type);
902 		msg_set_nameinst(mhdr, inst);
903 		msg_set_lookup_scope(mhdr, tipc_addr_scope(domain));
904 		dport = tipc_nametbl_translate(net, type, inst, &dnode);
905 		msg_set_destnode(mhdr, dnode);
906 		msg_set_destport(mhdr, dport);
907 		if (unlikely(!dport && !dnode))
908 			return -EHOSTUNREACH;
909 	} else if (dest->addrtype == TIPC_ADDR_ID) {
910 		dnode = dest->addr.id.node;
911 		msg_set_type(mhdr, TIPC_DIRECT_MSG);
912 		msg_set_lookup_scope(mhdr, 0);
913 		msg_set_destnode(mhdr, dnode);
914 		msg_set_destport(mhdr, dest->addr.id.ref);
915 		msg_set_hdr_sz(mhdr, BASIC_H_SIZE);
916 	}
917 
918 	save = m->msg_iter;
919 new_mtu:
920 	mtu = tipc_node_get_mtu(net, dnode, tsk->portid);
921 	rc = tipc_msg_build(mhdr, m, 0, dsz, mtu, pktchain);
922 	if (rc < 0)
923 		return rc;
924 
925 	do {
926 		skb = skb_peek(pktchain);
927 		TIPC_SKB_CB(skb)->wakeup_pending = tsk->link_cong;
928 		rc = tipc_node_xmit(net, pktchain, dnode, tsk->portid);
929 		if (likely(!rc)) {
930 			if (sock->state != SS_READY)
931 				sock->state = SS_CONNECTING;
932 			return dsz;
933 		}
934 		if (rc == -ELINKCONG) {
935 			tsk->link_cong = 1;
936 			rc = tipc_wait_for_sndmsg(sock, &timeo);
937 			if (!rc)
938 				continue;
939 		}
940 		__skb_queue_purge(pktchain);
941 		if (rc == -EMSGSIZE) {
942 			m->msg_iter = save;
943 			goto new_mtu;
944 		}
945 		break;
946 	} while (1);
947 
948 	return rc;
949 }
950 
951 static int tipc_wait_for_sndpkt(struct socket *sock, long *timeo_p)
952 {
953 	struct sock *sk = sock->sk;
954 	struct tipc_sock *tsk = tipc_sk(sk);
955 	DEFINE_WAIT(wait);
956 	int done;
957 
958 	do {
959 		int err = sock_error(sk);
960 		if (err)
961 			return err;
962 		if (sock->state == SS_DISCONNECTING)
963 			return -EPIPE;
964 		else if (sock->state != SS_CONNECTED)
965 			return -ENOTCONN;
966 		if (!*timeo_p)
967 			return -EAGAIN;
968 		if (signal_pending(current))
969 			return sock_intr_errno(*timeo_p);
970 
971 		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
972 		done = sk_wait_event(sk, timeo_p,
973 				     (!tsk->link_cong &&
974 				      !tsk_conn_cong(tsk)) ||
975 				     !tsk->connected);
976 		finish_wait(sk_sleep(sk), &wait);
977 	} while (!done);
978 	return 0;
979 }
980 
981 /**
982  * tipc_send_stream - send stream-oriented data
983  * @sock: socket structure
984  * @m: data to send
985  * @dsz: total length of data to be transmitted
986  *
987  * Used for SOCK_STREAM data.
988  *
989  * Returns the number of bytes sent on success (or partial success),
990  * or errno if no data sent
991  */
992 static int tipc_send_stream(struct socket *sock, struct msghdr *m, size_t dsz)
993 {
994 	struct sock *sk = sock->sk;
995 	int ret;
996 
997 	lock_sock(sk);
998 	ret = __tipc_send_stream(sock, m, dsz);
999 	release_sock(sk);
1000 
1001 	return ret;
1002 }
1003 
1004 static int __tipc_send_stream(struct socket *sock, struct msghdr *m, size_t dsz)
1005 {
1006 	struct sock *sk = sock->sk;
1007 	struct net *net = sock_net(sk);
1008 	struct tipc_sock *tsk = tipc_sk(sk);
1009 	struct tipc_msg *mhdr = &tsk->phdr;
1010 	struct sk_buff_head *pktchain = &sk->sk_write_queue;
1011 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1012 	u32 portid = tsk->portid;
1013 	int rc = -EINVAL;
1014 	long timeo;
1015 	u32 dnode;
1016 	uint mtu, send, sent = 0;
1017 	struct iov_iter save;
1018 
1019 	/* Handle implied connection establishment */
1020 	if (unlikely(dest)) {
1021 		rc = __tipc_sendmsg(sock, m, dsz);
1022 		if (dsz && (dsz == rc))
1023 			tsk->sent_unacked = 1;
1024 		return rc;
1025 	}
1026 	if (dsz > (uint)INT_MAX)
1027 		return -EMSGSIZE;
1028 
1029 	if (unlikely(sock->state != SS_CONNECTED)) {
1030 		if (sock->state == SS_DISCONNECTING)
1031 			return -EPIPE;
1032 		else
1033 			return -ENOTCONN;
1034 	}
1035 
1036 	timeo = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1037 	dnode = tsk_peer_node(tsk);
1038 
1039 next:
1040 	save = m->msg_iter;
1041 	mtu = tsk->max_pkt;
1042 	send = min_t(uint, dsz - sent, TIPC_MAX_USER_MSG_SIZE);
1043 	rc = tipc_msg_build(mhdr, m, sent, send, mtu, pktchain);
1044 	if (unlikely(rc < 0))
1045 		return rc;
1046 	do {
1047 		if (likely(!tsk_conn_cong(tsk))) {
1048 			rc = tipc_node_xmit(net, pktchain, dnode, portid);
1049 			if (likely(!rc)) {
1050 				tsk->sent_unacked++;
1051 				sent += send;
1052 				if (sent == dsz)
1053 					return dsz;
1054 				goto next;
1055 			}
1056 			if (rc == -EMSGSIZE) {
1057 				__skb_queue_purge(pktchain);
1058 				tsk->max_pkt = tipc_node_get_mtu(net, dnode,
1059 								 portid);
1060 				m->msg_iter = save;
1061 				goto next;
1062 			}
1063 			if (rc != -ELINKCONG)
1064 				break;
1065 
1066 			tsk->link_cong = 1;
1067 		}
1068 		rc = tipc_wait_for_sndpkt(sock, &timeo);
1069 	} while (!rc);
1070 
1071 	__skb_queue_purge(pktchain);
1072 	return sent ? sent : rc;
1073 }
1074 
1075 /**
1076  * tipc_send_packet - send a connection-oriented message
1077  * @sock: socket structure
1078  * @m: message to send
1079  * @dsz: length of data to be transmitted
1080  *
1081  * Used for SOCK_SEQPACKET messages.
1082  *
1083  * Returns the number of bytes sent on success, or errno otherwise
1084  */
1085 static int tipc_send_packet(struct socket *sock, struct msghdr *m, size_t dsz)
1086 {
1087 	if (dsz > TIPC_MAX_USER_MSG_SIZE)
1088 		return -EMSGSIZE;
1089 
1090 	return tipc_send_stream(sock, m, dsz);
1091 }
1092 
1093 /* tipc_sk_finish_conn - complete the setup of a connection
1094  */
1095 static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port,
1096 				u32 peer_node)
1097 {
1098 	struct sock *sk = &tsk->sk;
1099 	struct net *net = sock_net(sk);
1100 	struct tipc_msg *msg = &tsk->phdr;
1101 
1102 	msg_set_destnode(msg, peer_node);
1103 	msg_set_destport(msg, peer_port);
1104 	msg_set_type(msg, TIPC_CONN_MSG);
1105 	msg_set_lookup_scope(msg, 0);
1106 	msg_set_hdr_sz(msg, SHORT_H_SIZE);
1107 
1108 	tsk->probing_intv = CONN_PROBING_INTERVAL;
1109 	tsk->probing_state = TIPC_CONN_OK;
1110 	tsk->connected = 1;
1111 	sk_reset_timer(sk, &sk->sk_timer, jiffies + tsk->probing_intv);
1112 	tipc_node_add_conn(net, peer_node, tsk->portid, peer_port);
1113 	tsk->max_pkt = tipc_node_get_mtu(net, peer_node, tsk->portid);
1114 }
1115 
1116 /**
1117  * set_orig_addr - capture sender's address for received message
1118  * @m: descriptor for message info
1119  * @msg: received message header
1120  *
1121  * Note: Address is not captured if not requested by receiver.
1122  */
1123 static void set_orig_addr(struct msghdr *m, struct tipc_msg *msg)
1124 {
1125 	DECLARE_SOCKADDR(struct sockaddr_tipc *, addr, m->msg_name);
1126 
1127 	if (addr) {
1128 		addr->family = AF_TIPC;
1129 		addr->addrtype = TIPC_ADDR_ID;
1130 		memset(&addr->addr, 0, sizeof(addr->addr));
1131 		addr->addr.id.ref = msg_origport(msg);
1132 		addr->addr.id.node = msg_orignode(msg);
1133 		addr->addr.name.domain = 0;	/* could leave uninitialized */
1134 		addr->scope = 0;		/* could leave uninitialized */
1135 		m->msg_namelen = sizeof(struct sockaddr_tipc);
1136 	}
1137 }
1138 
1139 /**
1140  * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
1141  * @m: descriptor for message info
1142  * @msg: received message header
1143  * @tsk: TIPC port associated with message
1144  *
1145  * Note: Ancillary data is not captured if not requested by receiver.
1146  *
1147  * Returns 0 if successful, otherwise errno
1148  */
1149 static int tipc_sk_anc_data_recv(struct msghdr *m, struct tipc_msg *msg,
1150 				 struct tipc_sock *tsk)
1151 {
1152 	u32 anc_data[3];
1153 	u32 err;
1154 	u32 dest_type;
1155 	int has_name;
1156 	int res;
1157 
1158 	if (likely(m->msg_controllen == 0))
1159 		return 0;
1160 
1161 	/* Optionally capture errored message object(s) */
1162 	err = msg ? msg_errcode(msg) : 0;
1163 	if (unlikely(err)) {
1164 		anc_data[0] = err;
1165 		anc_data[1] = msg_data_sz(msg);
1166 		res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
1167 		if (res)
1168 			return res;
1169 		if (anc_data[1]) {
1170 			res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
1171 				       msg_data(msg));
1172 			if (res)
1173 				return res;
1174 		}
1175 	}
1176 
1177 	/* Optionally capture message destination object */
1178 	dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
1179 	switch (dest_type) {
1180 	case TIPC_NAMED_MSG:
1181 		has_name = 1;
1182 		anc_data[0] = msg_nametype(msg);
1183 		anc_data[1] = msg_namelower(msg);
1184 		anc_data[2] = msg_namelower(msg);
1185 		break;
1186 	case TIPC_MCAST_MSG:
1187 		has_name = 1;
1188 		anc_data[0] = msg_nametype(msg);
1189 		anc_data[1] = msg_namelower(msg);
1190 		anc_data[2] = msg_nameupper(msg);
1191 		break;
1192 	case TIPC_CONN_MSG:
1193 		has_name = (tsk->conn_type != 0);
1194 		anc_data[0] = tsk->conn_type;
1195 		anc_data[1] = tsk->conn_instance;
1196 		anc_data[2] = tsk->conn_instance;
1197 		break;
1198 	default:
1199 		has_name = 0;
1200 	}
1201 	if (has_name) {
1202 		res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
1203 		if (res)
1204 			return res;
1205 	}
1206 
1207 	return 0;
1208 }
1209 
1210 static void tipc_sk_send_ack(struct tipc_sock *tsk, uint ack)
1211 {
1212 	struct net *net = sock_net(&tsk->sk);
1213 	struct sk_buff *skb = NULL;
1214 	struct tipc_msg *msg;
1215 	u32 peer_port = tsk_peer_port(tsk);
1216 	u32 dnode = tsk_peer_node(tsk);
1217 
1218 	if (!tsk->connected)
1219 		return;
1220 	skb = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0,
1221 			      dnode, tsk_own_node(tsk), peer_port,
1222 			      tsk->portid, TIPC_OK);
1223 	if (!skb)
1224 		return;
1225 	msg = buf_msg(skb);
1226 	msg_set_msgcnt(msg, ack);
1227 	tipc_node_xmit_skb(net, skb, dnode, msg_link_selector(msg));
1228 }
1229 
1230 static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop)
1231 {
1232 	struct sock *sk = sock->sk;
1233 	DEFINE_WAIT(wait);
1234 	long timeo = *timeop;
1235 	int err;
1236 
1237 	for (;;) {
1238 		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1239 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1240 			if (sock->state == SS_DISCONNECTING) {
1241 				err = -ENOTCONN;
1242 				break;
1243 			}
1244 			release_sock(sk);
1245 			timeo = schedule_timeout(timeo);
1246 			lock_sock(sk);
1247 		}
1248 		err = 0;
1249 		if (!skb_queue_empty(&sk->sk_receive_queue))
1250 			break;
1251 		err = -EAGAIN;
1252 		if (!timeo)
1253 			break;
1254 		err = sock_intr_errno(timeo);
1255 		if (signal_pending(current))
1256 			break;
1257 	}
1258 	finish_wait(sk_sleep(sk), &wait);
1259 	*timeop = timeo;
1260 	return err;
1261 }
1262 
1263 /**
1264  * tipc_recvmsg - receive packet-oriented message
1265  * @m: descriptor for message info
1266  * @buf_len: total size of user buffer area
1267  * @flags: receive flags
1268  *
1269  * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1270  * If the complete message doesn't fit in user area, truncate it.
1271  *
1272  * Returns size of returned message data, errno otherwise
1273  */
1274 static int tipc_recvmsg(struct socket *sock, struct msghdr *m, size_t buf_len,
1275 			int flags)
1276 {
1277 	struct sock *sk = sock->sk;
1278 	struct tipc_sock *tsk = tipc_sk(sk);
1279 	struct sk_buff *buf;
1280 	struct tipc_msg *msg;
1281 	long timeo;
1282 	unsigned int sz;
1283 	u32 err;
1284 	int res;
1285 
1286 	/* Catch invalid receive requests */
1287 	if (unlikely(!buf_len))
1288 		return -EINVAL;
1289 
1290 	lock_sock(sk);
1291 
1292 	if (unlikely(sock->state == SS_UNCONNECTED)) {
1293 		res = -ENOTCONN;
1294 		goto exit;
1295 	}
1296 
1297 	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1298 restart:
1299 
1300 	/* Look for a message in receive queue; wait if necessary */
1301 	res = tipc_wait_for_rcvmsg(sock, &timeo);
1302 	if (res)
1303 		goto exit;
1304 
1305 	/* Look at first message in receive queue */
1306 	buf = skb_peek(&sk->sk_receive_queue);
1307 	msg = buf_msg(buf);
1308 	sz = msg_data_sz(msg);
1309 	err = msg_errcode(msg);
1310 
1311 	/* Discard an empty non-errored message & try again */
1312 	if ((!sz) && (!err)) {
1313 		tsk_advance_rx_queue(sk);
1314 		goto restart;
1315 	}
1316 
1317 	/* Capture sender's address (optional) */
1318 	set_orig_addr(m, msg);
1319 
1320 	/* Capture ancillary data (optional) */
1321 	res = tipc_sk_anc_data_recv(m, msg, tsk);
1322 	if (res)
1323 		goto exit;
1324 
1325 	/* Capture message data (if valid) & compute return value (always) */
1326 	if (!err) {
1327 		if (unlikely(buf_len < sz)) {
1328 			sz = buf_len;
1329 			m->msg_flags |= MSG_TRUNC;
1330 		}
1331 		res = skb_copy_datagram_msg(buf, msg_hdr_sz(msg), m, sz);
1332 		if (res)
1333 			goto exit;
1334 		res = sz;
1335 	} else {
1336 		if ((sock->state == SS_READY) ||
1337 		    ((err == TIPC_CONN_SHUTDOWN) || m->msg_control))
1338 			res = 0;
1339 		else
1340 			res = -ECONNRESET;
1341 	}
1342 
1343 	/* Consume received message (optional) */
1344 	if (likely(!(flags & MSG_PEEK))) {
1345 		if ((sock->state != SS_READY) &&
1346 		    (++tsk->rcv_unacked >= TIPC_CONNACK_INTV)) {
1347 			tipc_sk_send_ack(tsk, tsk->rcv_unacked);
1348 			tsk->rcv_unacked = 0;
1349 		}
1350 		tsk_advance_rx_queue(sk);
1351 	}
1352 exit:
1353 	release_sock(sk);
1354 	return res;
1355 }
1356 
1357 /**
1358  * tipc_recv_stream - receive stream-oriented data
1359  * @m: descriptor for message info
1360  * @buf_len: total size of user buffer area
1361  * @flags: receive flags
1362  *
1363  * Used for SOCK_STREAM messages only.  If not enough data is available
1364  * will optionally wait for more; never truncates data.
1365  *
1366  * Returns size of returned message data, errno otherwise
1367  */
1368 static int tipc_recv_stream(struct socket *sock, struct msghdr *m,
1369 			    size_t buf_len, int flags)
1370 {
1371 	struct sock *sk = sock->sk;
1372 	struct tipc_sock *tsk = tipc_sk(sk);
1373 	struct sk_buff *buf;
1374 	struct tipc_msg *msg;
1375 	long timeo;
1376 	unsigned int sz;
1377 	int sz_to_copy, target, needed;
1378 	int sz_copied = 0;
1379 	u32 err;
1380 	int res = 0;
1381 
1382 	/* Catch invalid receive attempts */
1383 	if (unlikely(!buf_len))
1384 		return -EINVAL;
1385 
1386 	lock_sock(sk);
1387 
1388 	if (unlikely(sock->state == SS_UNCONNECTED)) {
1389 		res = -ENOTCONN;
1390 		goto exit;
1391 	}
1392 
1393 	target = sock_rcvlowat(sk, flags & MSG_WAITALL, buf_len);
1394 	timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1395 
1396 restart:
1397 	/* Look for a message in receive queue; wait if necessary */
1398 	res = tipc_wait_for_rcvmsg(sock, &timeo);
1399 	if (res)
1400 		goto exit;
1401 
1402 	/* Look at first message in receive queue */
1403 	buf = skb_peek(&sk->sk_receive_queue);
1404 	msg = buf_msg(buf);
1405 	sz = msg_data_sz(msg);
1406 	err = msg_errcode(msg);
1407 
1408 	/* Discard an empty non-errored message & try again */
1409 	if ((!sz) && (!err)) {
1410 		tsk_advance_rx_queue(sk);
1411 		goto restart;
1412 	}
1413 
1414 	/* Optionally capture sender's address & ancillary data of first msg */
1415 	if (sz_copied == 0) {
1416 		set_orig_addr(m, msg);
1417 		res = tipc_sk_anc_data_recv(m, msg, tsk);
1418 		if (res)
1419 			goto exit;
1420 	}
1421 
1422 	/* Capture message data (if valid) & compute return value (always) */
1423 	if (!err) {
1424 		u32 offset = (u32)(unsigned long)(TIPC_SKB_CB(buf)->handle);
1425 
1426 		sz -= offset;
1427 		needed = (buf_len - sz_copied);
1428 		sz_to_copy = (sz <= needed) ? sz : needed;
1429 
1430 		res = skb_copy_datagram_msg(buf, msg_hdr_sz(msg) + offset,
1431 					    m, sz_to_copy);
1432 		if (res)
1433 			goto exit;
1434 
1435 		sz_copied += sz_to_copy;
1436 
1437 		if (sz_to_copy < sz) {
1438 			if (!(flags & MSG_PEEK))
1439 				TIPC_SKB_CB(buf)->handle =
1440 				(void *)(unsigned long)(offset + sz_to_copy);
1441 			goto exit;
1442 		}
1443 	} else {
1444 		if (sz_copied != 0)
1445 			goto exit; /* can't add error msg to valid data */
1446 
1447 		if ((err == TIPC_CONN_SHUTDOWN) || m->msg_control)
1448 			res = 0;
1449 		else
1450 			res = -ECONNRESET;
1451 	}
1452 
1453 	/* Consume received message (optional) */
1454 	if (likely(!(flags & MSG_PEEK))) {
1455 		if (unlikely(++tsk->rcv_unacked >= TIPC_CONNACK_INTV)) {
1456 			tipc_sk_send_ack(tsk, tsk->rcv_unacked);
1457 			tsk->rcv_unacked = 0;
1458 		}
1459 		tsk_advance_rx_queue(sk);
1460 	}
1461 
1462 	/* Loop around if more data is required */
1463 	if ((sz_copied < buf_len) &&	/* didn't get all requested data */
1464 	    (!skb_queue_empty(&sk->sk_receive_queue) ||
1465 	    (sz_copied < target)) &&	/* and more is ready or required */
1466 	    (!(flags & MSG_PEEK)) &&	/* and aren't just peeking at data */
1467 	    (!err))			/* and haven't reached a FIN */
1468 		goto restart;
1469 
1470 exit:
1471 	release_sock(sk);
1472 	return sz_copied ? sz_copied : res;
1473 }
1474 
1475 /**
1476  * tipc_write_space - wake up thread if port congestion is released
1477  * @sk: socket
1478  */
1479 static void tipc_write_space(struct sock *sk)
1480 {
1481 	struct socket_wq *wq;
1482 
1483 	rcu_read_lock();
1484 	wq = rcu_dereference(sk->sk_wq);
1485 	if (wq_has_sleeper(wq))
1486 		wake_up_interruptible_sync_poll(&wq->wait, POLLOUT |
1487 						POLLWRNORM | POLLWRBAND);
1488 	rcu_read_unlock();
1489 }
1490 
1491 /**
1492  * tipc_data_ready - wake up threads to indicate messages have been received
1493  * @sk: socket
1494  * @len: the length of messages
1495  */
1496 static void tipc_data_ready(struct sock *sk)
1497 {
1498 	struct socket_wq *wq;
1499 
1500 	rcu_read_lock();
1501 	wq = rcu_dereference(sk->sk_wq);
1502 	if (wq_has_sleeper(wq))
1503 		wake_up_interruptible_sync_poll(&wq->wait, POLLIN |
1504 						POLLRDNORM | POLLRDBAND);
1505 	rcu_read_unlock();
1506 }
1507 
1508 /**
1509  * filter_connect - Handle all incoming messages for a connection-based socket
1510  * @tsk: TIPC socket
1511  * @skb: pointer to message buffer. Set to NULL if buffer is consumed
1512  *
1513  * Returns 0 (TIPC_OK) if everything ok, -TIPC_ERR_NO_PORT otherwise
1514  */
1515 static int filter_connect(struct tipc_sock *tsk, struct sk_buff **skb)
1516 {
1517 	struct sock *sk = &tsk->sk;
1518 	struct net *net = sock_net(sk);
1519 	struct socket *sock = sk->sk_socket;
1520 	struct tipc_msg *msg = buf_msg(*skb);
1521 	int retval = -TIPC_ERR_NO_PORT;
1522 
1523 	if (msg_mcast(msg))
1524 		return retval;
1525 
1526 	switch ((int)sock->state) {
1527 	case SS_CONNECTED:
1528 		/* Accept only connection-based messages sent by peer */
1529 		if (tsk_peer_msg(tsk, msg)) {
1530 			if (unlikely(msg_errcode(msg))) {
1531 				sock->state = SS_DISCONNECTING;
1532 				tsk->connected = 0;
1533 				/* let timer expire on it's own */
1534 				tipc_node_remove_conn(net, tsk_peer_node(tsk),
1535 						      tsk->portid);
1536 			}
1537 			retval = TIPC_OK;
1538 		}
1539 		break;
1540 	case SS_CONNECTING:
1541 		/* Accept only ACK or NACK message */
1542 
1543 		if (unlikely(!msg_connected(msg)))
1544 			break;
1545 
1546 		if (unlikely(msg_errcode(msg))) {
1547 			sock->state = SS_DISCONNECTING;
1548 			sk->sk_err = ECONNREFUSED;
1549 			retval = TIPC_OK;
1550 			break;
1551 		}
1552 
1553 		if (unlikely(msg_importance(msg) > TIPC_CRITICAL_IMPORTANCE)) {
1554 			sock->state = SS_DISCONNECTING;
1555 			sk->sk_err = EINVAL;
1556 			retval = TIPC_OK;
1557 			break;
1558 		}
1559 
1560 		tipc_sk_finish_conn(tsk, msg_origport(msg), msg_orignode(msg));
1561 		msg_set_importance(&tsk->phdr, msg_importance(msg));
1562 		sock->state = SS_CONNECTED;
1563 
1564 		/* If an incoming message is an 'ACK-', it should be
1565 		 * discarded here because it doesn't contain useful
1566 		 * data. In addition, we should try to wake up
1567 		 * connect() routine if sleeping.
1568 		 */
1569 		if (msg_data_sz(msg) == 0) {
1570 			kfree_skb(*skb);
1571 			*skb = NULL;
1572 			if (waitqueue_active(sk_sleep(sk)))
1573 				wake_up_interruptible(sk_sleep(sk));
1574 		}
1575 		retval = TIPC_OK;
1576 		break;
1577 	case SS_LISTENING:
1578 	case SS_UNCONNECTED:
1579 		/* Accept only SYN message */
1580 		if (!msg_connected(msg) && !(msg_errcode(msg)))
1581 			retval = TIPC_OK;
1582 		break;
1583 	case SS_DISCONNECTING:
1584 		break;
1585 	default:
1586 		pr_err("Unknown socket state %u\n", sock->state);
1587 	}
1588 	return retval;
1589 }
1590 
1591 /**
1592  * rcvbuf_limit - get proper overload limit of socket receive queue
1593  * @sk: socket
1594  * @buf: message
1595  *
1596  * For all connection oriented messages, irrespective of importance,
1597  * the default overload value (i.e. 67MB) is set as limit.
1598  *
1599  * For all connectionless messages, by default new queue limits are
1600  * as belows:
1601  *
1602  * TIPC_LOW_IMPORTANCE       (4 MB)
1603  * TIPC_MEDIUM_IMPORTANCE    (8 MB)
1604  * TIPC_HIGH_IMPORTANCE      (16 MB)
1605  * TIPC_CRITICAL_IMPORTANCE  (32 MB)
1606  *
1607  * Returns overload limit according to corresponding message importance
1608  */
1609 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *buf)
1610 {
1611 	struct tipc_msg *msg = buf_msg(buf);
1612 
1613 	if (msg_connected(msg))
1614 		return sysctl_tipc_rmem[2];
1615 
1616 	return sk->sk_rcvbuf >> TIPC_CRITICAL_IMPORTANCE <<
1617 		msg_importance(msg);
1618 }
1619 
1620 /**
1621  * filter_rcv - validate incoming message
1622  * @sk: socket
1623  * @skb: pointer to message. Set to NULL if buffer is consumed.
1624  *
1625  * Enqueues message on receive queue if acceptable; optionally handles
1626  * disconnect indication for a connected socket.
1627  *
1628  * Called with socket lock already taken
1629  *
1630  * Returns 0 (TIPC_OK) if message was ok, -TIPC error code if rejected
1631  */
1632 static int filter_rcv(struct sock *sk, struct sk_buff **skb)
1633 {
1634 	struct socket *sock = sk->sk_socket;
1635 	struct tipc_sock *tsk = tipc_sk(sk);
1636 	struct tipc_msg *msg = buf_msg(*skb);
1637 	unsigned int limit = rcvbuf_limit(sk, *skb);
1638 	int rc = TIPC_OK;
1639 
1640 	if (unlikely(msg_user(msg) == CONN_MANAGER)) {
1641 		tipc_sk_proto_rcv(tsk, skb);
1642 		return TIPC_OK;
1643 	}
1644 
1645 	if (unlikely(msg_user(msg) == SOCK_WAKEUP)) {
1646 		kfree_skb(*skb);
1647 		tsk->link_cong = 0;
1648 		sk->sk_write_space(sk);
1649 		*skb = NULL;
1650 		return TIPC_OK;
1651 	}
1652 
1653 	/* Reject message if it is wrong sort of message for socket */
1654 	if (msg_type(msg) > TIPC_DIRECT_MSG)
1655 		return -TIPC_ERR_NO_PORT;
1656 
1657 	if (sock->state == SS_READY) {
1658 		if (msg_connected(msg))
1659 			return -TIPC_ERR_NO_PORT;
1660 	} else {
1661 		rc = filter_connect(tsk, skb);
1662 		if (rc != TIPC_OK || !*skb)
1663 			return rc;
1664 	}
1665 
1666 	/* Reject message if there isn't room to queue it */
1667 	if (sk_rmem_alloc_get(sk) + (*skb)->truesize >= limit)
1668 		return -TIPC_ERR_OVERLOAD;
1669 
1670 	/* Enqueue message */
1671 	TIPC_SKB_CB(*skb)->handle = NULL;
1672 	__skb_queue_tail(&sk->sk_receive_queue, *skb);
1673 	skb_set_owner_r(*skb, sk);
1674 
1675 	sk->sk_data_ready(sk);
1676 	*skb = NULL;
1677 	return TIPC_OK;
1678 }
1679 
1680 /**
1681  * tipc_backlog_rcv - handle incoming message from backlog queue
1682  * @sk: socket
1683  * @skb: message
1684  *
1685  * Caller must hold socket lock
1686  *
1687  * Returns 0
1688  */
1689 static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *skb)
1690 {
1691 	int err;
1692 	atomic_t *dcnt;
1693 	u32 dnode;
1694 	struct tipc_sock *tsk = tipc_sk(sk);
1695 	struct net *net = sock_net(sk);
1696 	uint truesize = skb->truesize;
1697 
1698 	err = filter_rcv(sk, &skb);
1699 	if (likely(!skb)) {
1700 		dcnt = &tsk->dupl_rcvcnt;
1701 		if (atomic_read(dcnt) < TIPC_CONN_OVERLOAD_LIMIT)
1702 			atomic_add(truesize, dcnt);
1703 		return 0;
1704 	}
1705 	if (!err || tipc_msg_reverse(tsk_own_node(tsk), skb, &dnode, -err))
1706 		tipc_node_xmit_skb(net, skb, dnode, tsk->portid);
1707 	return 0;
1708 }
1709 
1710 /**
1711  * tipc_sk_enqueue - extract all buffers with destination 'dport' from
1712  *                   inputq and try adding them to socket or backlog queue
1713  * @inputq: list of incoming buffers with potentially different destinations
1714  * @sk: socket where the buffers should be enqueued
1715  * @dport: port number for the socket
1716  * @_skb: returned buffer to be forwarded or rejected, if applicable
1717  *
1718  * Caller must hold socket lock
1719  *
1720  * Returns TIPC_OK if all buffers enqueued, otherwise -TIPC_ERR_OVERLOAD
1721  * or -TIPC_ERR_NO_PORT
1722  */
1723 static int tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk,
1724 			   u32 dport, struct sk_buff **_skb)
1725 {
1726 	unsigned int lim;
1727 	atomic_t *dcnt;
1728 	int err;
1729 	struct sk_buff *skb;
1730 	unsigned long time_limit = jiffies + 2;
1731 
1732 	while (skb_queue_len(inputq)) {
1733 		if (unlikely(time_after_eq(jiffies, time_limit)))
1734 			return TIPC_OK;
1735 		skb = tipc_skb_dequeue(inputq, dport);
1736 		if (unlikely(!skb))
1737 			return TIPC_OK;
1738 		if (!sock_owned_by_user(sk)) {
1739 			err = filter_rcv(sk, &skb);
1740 			if (likely(!skb))
1741 				continue;
1742 			*_skb = skb;
1743 			return err;
1744 		}
1745 		dcnt = &tipc_sk(sk)->dupl_rcvcnt;
1746 		if (sk->sk_backlog.len)
1747 			atomic_set(dcnt, 0);
1748 		lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
1749 		if (likely(!sk_add_backlog(sk, skb, lim)))
1750 			continue;
1751 		*_skb = skb;
1752 		return -TIPC_ERR_OVERLOAD;
1753 	}
1754 	return TIPC_OK;
1755 }
1756 
1757 /**
1758  * tipc_sk_rcv - handle a chain of incoming buffers
1759  * @inputq: buffer list containing the buffers
1760  * Consumes all buffers in list until inputq is empty
1761  * Note: may be called in multiple threads referring to the same queue
1762  * Returns 0 if last buffer was accepted, otherwise -EHOSTUNREACH
1763  * Only node local calls check the return value, sending single-buffer queues
1764  */
1765 int tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq)
1766 {
1767 	u32 dnode, dport = 0;
1768 	int err;
1769 	struct sk_buff *skb;
1770 	struct tipc_sock *tsk;
1771 	struct tipc_net *tn;
1772 	struct sock *sk;
1773 
1774 	while (skb_queue_len(inputq)) {
1775 		err = -TIPC_ERR_NO_PORT;
1776 		skb = NULL;
1777 		dport = tipc_skb_peek_port(inputq, dport);
1778 		tsk = tipc_sk_lookup(net, dport);
1779 		if (likely(tsk)) {
1780 			sk = &tsk->sk;
1781 			if (likely(spin_trylock_bh(&sk->sk_lock.slock))) {
1782 				err = tipc_sk_enqueue(inputq, sk, dport, &skb);
1783 				spin_unlock_bh(&sk->sk_lock.slock);
1784 				dport = 0;
1785 			}
1786 			sock_put(sk);
1787 		} else {
1788 			skb = tipc_skb_dequeue(inputq, dport);
1789 		}
1790 		if (likely(!skb))
1791 			continue;
1792 		if (tipc_msg_lookup_dest(net, skb, &dnode, &err))
1793 			goto xmit;
1794 		if (!err) {
1795 			dnode = msg_destnode(buf_msg(skb));
1796 			goto xmit;
1797 		}
1798 		tn = net_generic(net, tipc_net_id);
1799 		if (!tipc_msg_reverse(tn->own_addr, skb, &dnode, -err))
1800 			continue;
1801 xmit:
1802 		tipc_node_xmit_skb(net, skb, dnode, dport);
1803 	}
1804 	return err ? -EHOSTUNREACH : 0;
1805 }
1806 
1807 static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
1808 {
1809 	struct sock *sk = sock->sk;
1810 	DEFINE_WAIT(wait);
1811 	int done;
1812 
1813 	do {
1814 		int err = sock_error(sk);
1815 		if (err)
1816 			return err;
1817 		if (!*timeo_p)
1818 			return -ETIMEDOUT;
1819 		if (signal_pending(current))
1820 			return sock_intr_errno(*timeo_p);
1821 
1822 		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1823 		done = sk_wait_event(sk, timeo_p, sock->state != SS_CONNECTING);
1824 		finish_wait(sk_sleep(sk), &wait);
1825 	} while (!done);
1826 	return 0;
1827 }
1828 
1829 /**
1830  * tipc_connect - establish a connection to another TIPC port
1831  * @sock: socket structure
1832  * @dest: socket address for destination port
1833  * @destlen: size of socket address data structure
1834  * @flags: file-related flags associated with socket
1835  *
1836  * Returns 0 on success, errno otherwise
1837  */
1838 static int tipc_connect(struct socket *sock, struct sockaddr *dest,
1839 			int destlen, int flags)
1840 {
1841 	struct sock *sk = sock->sk;
1842 	struct tipc_sock *tsk = tipc_sk(sk);
1843 	struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
1844 	struct msghdr m = {NULL,};
1845 	long timeout = (flags & O_NONBLOCK) ? 0 : tsk->conn_timeout;
1846 	socket_state previous;
1847 	int res = 0;
1848 
1849 	lock_sock(sk);
1850 
1851 	/* DGRAM/RDM connect(), just save the destaddr */
1852 	if (sock->state == SS_READY) {
1853 		if (dst->family == AF_UNSPEC) {
1854 			memset(&tsk->remote, 0, sizeof(struct sockaddr_tipc));
1855 			tsk->connected = 0;
1856 		} else if (destlen != sizeof(struct sockaddr_tipc)) {
1857 			res = -EINVAL;
1858 		} else {
1859 			memcpy(&tsk->remote, dest, destlen);
1860 			tsk->connected = 1;
1861 		}
1862 		goto exit;
1863 	}
1864 
1865 	/*
1866 	 * Reject connection attempt using multicast address
1867 	 *
1868 	 * Note: send_msg() validates the rest of the address fields,
1869 	 *       so there's no need to do it here
1870 	 */
1871 	if (dst->addrtype == TIPC_ADDR_MCAST) {
1872 		res = -EINVAL;
1873 		goto exit;
1874 	}
1875 
1876 	previous = sock->state;
1877 	switch (sock->state) {
1878 	case SS_UNCONNECTED:
1879 		/* Send a 'SYN-' to destination */
1880 		m.msg_name = dest;
1881 		m.msg_namelen = destlen;
1882 
1883 		/* If connect is in non-blocking case, set MSG_DONTWAIT to
1884 		 * indicate send_msg() is never blocked.
1885 		 */
1886 		if (!timeout)
1887 			m.msg_flags = MSG_DONTWAIT;
1888 
1889 		res = __tipc_sendmsg(sock, &m, 0);
1890 		if ((res < 0) && (res != -EWOULDBLOCK))
1891 			goto exit;
1892 
1893 		/* Just entered SS_CONNECTING state; the only
1894 		 * difference is that return value in non-blocking
1895 		 * case is EINPROGRESS, rather than EALREADY.
1896 		 */
1897 		res = -EINPROGRESS;
1898 	case SS_CONNECTING:
1899 		if (previous == SS_CONNECTING)
1900 			res = -EALREADY;
1901 		if (!timeout)
1902 			goto exit;
1903 		timeout = msecs_to_jiffies(timeout);
1904 		/* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
1905 		res = tipc_wait_for_connect(sock, &timeout);
1906 		break;
1907 	case SS_CONNECTED:
1908 		res = -EISCONN;
1909 		break;
1910 	default:
1911 		res = -EINVAL;
1912 		break;
1913 	}
1914 exit:
1915 	release_sock(sk);
1916 	return res;
1917 }
1918 
1919 /**
1920  * tipc_listen - allow socket to listen for incoming connections
1921  * @sock: socket structure
1922  * @len: (unused)
1923  *
1924  * Returns 0 on success, errno otherwise
1925  */
1926 static int tipc_listen(struct socket *sock, int len)
1927 {
1928 	struct sock *sk = sock->sk;
1929 	int res;
1930 
1931 	lock_sock(sk);
1932 
1933 	if (sock->state != SS_UNCONNECTED)
1934 		res = -EINVAL;
1935 	else {
1936 		sock->state = SS_LISTENING;
1937 		res = 0;
1938 	}
1939 
1940 	release_sock(sk);
1941 	return res;
1942 }
1943 
1944 static int tipc_wait_for_accept(struct socket *sock, long timeo)
1945 {
1946 	struct sock *sk = sock->sk;
1947 	DEFINE_WAIT(wait);
1948 	int err;
1949 
1950 	/* True wake-one mechanism for incoming connections: only
1951 	 * one process gets woken up, not the 'whole herd'.
1952 	 * Since we do not 'race & poll' for established sockets
1953 	 * anymore, the common case will execute the loop only once.
1954 	*/
1955 	for (;;) {
1956 		prepare_to_wait_exclusive(sk_sleep(sk), &wait,
1957 					  TASK_INTERRUPTIBLE);
1958 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1959 			release_sock(sk);
1960 			timeo = schedule_timeout(timeo);
1961 			lock_sock(sk);
1962 		}
1963 		err = 0;
1964 		if (!skb_queue_empty(&sk->sk_receive_queue))
1965 			break;
1966 		err = -EINVAL;
1967 		if (sock->state != SS_LISTENING)
1968 			break;
1969 		err = -EAGAIN;
1970 		if (!timeo)
1971 			break;
1972 		err = sock_intr_errno(timeo);
1973 		if (signal_pending(current))
1974 			break;
1975 	}
1976 	finish_wait(sk_sleep(sk), &wait);
1977 	return err;
1978 }
1979 
1980 /**
1981  * tipc_accept - wait for connection request
1982  * @sock: listening socket
1983  * @newsock: new socket that is to be connected
1984  * @flags: file-related flags associated with socket
1985  *
1986  * Returns 0 on success, errno otherwise
1987  */
1988 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags)
1989 {
1990 	struct sock *new_sk, *sk = sock->sk;
1991 	struct sk_buff *buf;
1992 	struct tipc_sock *new_tsock;
1993 	struct tipc_msg *msg;
1994 	long timeo;
1995 	int res;
1996 
1997 	lock_sock(sk);
1998 
1999 	if (sock->state != SS_LISTENING) {
2000 		res = -EINVAL;
2001 		goto exit;
2002 	}
2003 	timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
2004 	res = tipc_wait_for_accept(sock, timeo);
2005 	if (res)
2006 		goto exit;
2007 
2008 	buf = skb_peek(&sk->sk_receive_queue);
2009 
2010 	res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, 1);
2011 	if (res)
2012 		goto exit;
2013 	security_sk_clone(sock->sk, new_sock->sk);
2014 
2015 	new_sk = new_sock->sk;
2016 	new_tsock = tipc_sk(new_sk);
2017 	msg = buf_msg(buf);
2018 
2019 	/* we lock on new_sk; but lockdep sees the lock on sk */
2020 	lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
2021 
2022 	/*
2023 	 * Reject any stray messages received by new socket
2024 	 * before the socket lock was taken (very, very unlikely)
2025 	 */
2026 	tsk_rej_rx_queue(new_sk);
2027 
2028 	/* Connect new socket to it's peer */
2029 	tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg));
2030 	new_sock->state = SS_CONNECTED;
2031 
2032 	tsk_set_importance(new_tsock, msg_importance(msg));
2033 	if (msg_named(msg)) {
2034 		new_tsock->conn_type = msg_nametype(msg);
2035 		new_tsock->conn_instance = msg_nameinst(msg);
2036 	}
2037 
2038 	/*
2039 	 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2040 	 * Respond to 'SYN+' by queuing it on new socket.
2041 	 */
2042 	if (!msg_data_sz(msg)) {
2043 		struct msghdr m = {NULL,};
2044 
2045 		tsk_advance_rx_queue(sk);
2046 		__tipc_send_stream(new_sock, &m, 0);
2047 	} else {
2048 		__skb_dequeue(&sk->sk_receive_queue);
2049 		__skb_queue_head(&new_sk->sk_receive_queue, buf);
2050 		skb_set_owner_r(buf, new_sk);
2051 	}
2052 	release_sock(new_sk);
2053 exit:
2054 	release_sock(sk);
2055 	return res;
2056 }
2057 
2058 /**
2059  * tipc_shutdown - shutdown socket connection
2060  * @sock: socket structure
2061  * @how: direction to close (must be SHUT_RDWR)
2062  *
2063  * Terminates connection (if necessary), then purges socket's receive queue.
2064  *
2065  * Returns 0 on success, errno otherwise
2066  */
2067 static int tipc_shutdown(struct socket *sock, int how)
2068 {
2069 	struct sock *sk = sock->sk;
2070 	struct net *net = sock_net(sk);
2071 	struct tipc_sock *tsk = tipc_sk(sk);
2072 	struct sk_buff *skb;
2073 	u32 dnode;
2074 	int res;
2075 
2076 	if (how != SHUT_RDWR)
2077 		return -EINVAL;
2078 
2079 	lock_sock(sk);
2080 
2081 	switch (sock->state) {
2082 	case SS_CONNECTING:
2083 	case SS_CONNECTED:
2084 
2085 restart:
2086 		/* Disconnect and send a 'FIN+' or 'FIN-' message to peer */
2087 		skb = __skb_dequeue(&sk->sk_receive_queue);
2088 		if (skb) {
2089 			if (TIPC_SKB_CB(skb)->handle != NULL) {
2090 				kfree_skb(skb);
2091 				goto restart;
2092 			}
2093 			if (tipc_msg_reverse(tsk_own_node(tsk), skb, &dnode,
2094 					     TIPC_CONN_SHUTDOWN))
2095 				tipc_node_xmit_skb(net, skb, dnode,
2096 						   tsk->portid);
2097 		} else {
2098 			dnode = tsk_peer_node(tsk);
2099 
2100 			skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE,
2101 					      TIPC_CONN_MSG, SHORT_H_SIZE,
2102 					      0, dnode, tsk_own_node(tsk),
2103 					      tsk_peer_port(tsk),
2104 					      tsk->portid, TIPC_CONN_SHUTDOWN);
2105 			tipc_node_xmit_skb(net, skb, dnode, tsk->portid);
2106 		}
2107 		tsk->connected = 0;
2108 		sock->state = SS_DISCONNECTING;
2109 		tipc_node_remove_conn(net, dnode, tsk->portid);
2110 		/* fall through */
2111 
2112 	case SS_DISCONNECTING:
2113 
2114 		/* Discard any unreceived messages */
2115 		__skb_queue_purge(&sk->sk_receive_queue);
2116 
2117 		/* Wake up anyone sleeping in poll */
2118 		sk->sk_state_change(sk);
2119 		res = 0;
2120 		break;
2121 
2122 	default:
2123 		res = -ENOTCONN;
2124 	}
2125 
2126 	release_sock(sk);
2127 	return res;
2128 }
2129 
2130 static void tipc_sk_timeout(unsigned long data)
2131 {
2132 	struct tipc_sock *tsk = (struct tipc_sock *)data;
2133 	struct sock *sk = &tsk->sk;
2134 	struct sk_buff *skb = NULL;
2135 	u32 peer_port, peer_node;
2136 	u32 own_node = tsk_own_node(tsk);
2137 
2138 	bh_lock_sock(sk);
2139 	if (!tsk->connected) {
2140 		bh_unlock_sock(sk);
2141 		goto exit;
2142 	}
2143 	peer_port = tsk_peer_port(tsk);
2144 	peer_node = tsk_peer_node(tsk);
2145 
2146 	if (tsk->probing_state == TIPC_CONN_PROBING) {
2147 		if (!sock_owned_by_user(sk)) {
2148 			sk->sk_socket->state = SS_DISCONNECTING;
2149 			tsk->connected = 0;
2150 			tipc_node_remove_conn(sock_net(sk), tsk_peer_node(tsk),
2151 					      tsk_peer_port(tsk));
2152 			sk->sk_state_change(sk);
2153 		} else {
2154 			/* Try again later */
2155 			sk_reset_timer(sk, &sk->sk_timer, (HZ / 20));
2156 		}
2157 
2158 	} else {
2159 		skb = tipc_msg_create(CONN_MANAGER, CONN_PROBE,
2160 				      INT_H_SIZE, 0, peer_node, own_node,
2161 				      peer_port, tsk->portid, TIPC_OK);
2162 		tsk->probing_state = TIPC_CONN_PROBING;
2163 		sk_reset_timer(sk, &sk->sk_timer, jiffies + tsk->probing_intv);
2164 	}
2165 	bh_unlock_sock(sk);
2166 	if (skb)
2167 		tipc_node_xmit_skb(sock_net(sk), skb, peer_node, tsk->portid);
2168 exit:
2169 	sock_put(sk);
2170 }
2171 
2172 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
2173 			   struct tipc_name_seq const *seq)
2174 {
2175 	struct net *net = sock_net(&tsk->sk);
2176 	struct publication *publ;
2177 	u32 key;
2178 
2179 	if (tsk->connected)
2180 		return -EINVAL;
2181 	key = tsk->portid + tsk->pub_count + 1;
2182 	if (key == tsk->portid)
2183 		return -EADDRINUSE;
2184 
2185 	publ = tipc_nametbl_publish(net, seq->type, seq->lower, seq->upper,
2186 				    scope, tsk->portid, key);
2187 	if (unlikely(!publ))
2188 		return -EINVAL;
2189 
2190 	list_add(&publ->pport_list, &tsk->publications);
2191 	tsk->pub_count++;
2192 	tsk->published = 1;
2193 	return 0;
2194 }
2195 
2196 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
2197 			    struct tipc_name_seq const *seq)
2198 {
2199 	struct net *net = sock_net(&tsk->sk);
2200 	struct publication *publ;
2201 	struct publication *safe;
2202 	int rc = -EINVAL;
2203 
2204 	list_for_each_entry_safe(publ, safe, &tsk->publications, pport_list) {
2205 		if (seq) {
2206 			if (publ->scope != scope)
2207 				continue;
2208 			if (publ->type != seq->type)
2209 				continue;
2210 			if (publ->lower != seq->lower)
2211 				continue;
2212 			if (publ->upper != seq->upper)
2213 				break;
2214 			tipc_nametbl_withdraw(net, publ->type, publ->lower,
2215 					      publ->ref, publ->key);
2216 			rc = 0;
2217 			break;
2218 		}
2219 		tipc_nametbl_withdraw(net, publ->type, publ->lower,
2220 				      publ->ref, publ->key);
2221 		rc = 0;
2222 	}
2223 	if (list_empty(&tsk->publications))
2224 		tsk->published = 0;
2225 	return rc;
2226 }
2227 
2228 /* tipc_sk_reinit: set non-zero address in all existing sockets
2229  *                 when we go from standalone to network mode.
2230  */
2231 void tipc_sk_reinit(struct net *net)
2232 {
2233 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2234 	const struct bucket_table *tbl;
2235 	struct rhash_head *pos;
2236 	struct tipc_sock *tsk;
2237 	struct tipc_msg *msg;
2238 	int i;
2239 
2240 	rcu_read_lock();
2241 	tbl = rht_dereference_rcu((&tn->sk_rht)->tbl, &tn->sk_rht);
2242 	for (i = 0; i < tbl->size; i++) {
2243 		rht_for_each_entry_rcu(tsk, pos, tbl, i, node) {
2244 			spin_lock_bh(&tsk->sk.sk_lock.slock);
2245 			msg = &tsk->phdr;
2246 			msg_set_prevnode(msg, tn->own_addr);
2247 			msg_set_orignode(msg, tn->own_addr);
2248 			spin_unlock_bh(&tsk->sk.sk_lock.slock);
2249 		}
2250 	}
2251 	rcu_read_unlock();
2252 }
2253 
2254 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid)
2255 {
2256 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2257 	struct tipc_sock *tsk;
2258 
2259 	rcu_read_lock();
2260 	tsk = rhashtable_lookup_fast(&tn->sk_rht, &portid, tsk_rht_params);
2261 	if (tsk)
2262 		sock_hold(&tsk->sk);
2263 	rcu_read_unlock();
2264 
2265 	return tsk;
2266 }
2267 
2268 static int tipc_sk_insert(struct tipc_sock *tsk)
2269 {
2270 	struct sock *sk = &tsk->sk;
2271 	struct net *net = sock_net(sk);
2272 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2273 	u32 remaining = (TIPC_MAX_PORT - TIPC_MIN_PORT) + 1;
2274 	u32 portid = prandom_u32() % remaining + TIPC_MIN_PORT;
2275 
2276 	while (remaining--) {
2277 		portid++;
2278 		if ((portid < TIPC_MIN_PORT) || (portid > TIPC_MAX_PORT))
2279 			portid = TIPC_MIN_PORT;
2280 		tsk->portid = portid;
2281 		sock_hold(&tsk->sk);
2282 		if (!rhashtable_lookup_insert_fast(&tn->sk_rht, &tsk->node,
2283 						   tsk_rht_params))
2284 			return 0;
2285 		sock_put(&tsk->sk);
2286 	}
2287 
2288 	return -1;
2289 }
2290 
2291 static void tipc_sk_remove(struct tipc_sock *tsk)
2292 {
2293 	struct sock *sk = &tsk->sk;
2294 	struct tipc_net *tn = net_generic(sock_net(sk), tipc_net_id);
2295 
2296 	if (!rhashtable_remove_fast(&tn->sk_rht, &tsk->node, tsk_rht_params)) {
2297 		WARN_ON(atomic_read(&sk->sk_refcnt) == 1);
2298 		__sock_put(sk);
2299 	}
2300 }
2301 
2302 static const struct rhashtable_params tsk_rht_params = {
2303 	.nelem_hint = 192,
2304 	.head_offset = offsetof(struct tipc_sock, node),
2305 	.key_offset = offsetof(struct tipc_sock, portid),
2306 	.key_len = sizeof(u32), /* portid */
2307 	.max_size = 1048576,
2308 	.min_size = 256,
2309 	.automatic_shrinking = true,
2310 };
2311 
2312 int tipc_sk_rht_init(struct net *net)
2313 {
2314 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2315 
2316 	return rhashtable_init(&tn->sk_rht, &tsk_rht_params);
2317 }
2318 
2319 void tipc_sk_rht_destroy(struct net *net)
2320 {
2321 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2322 
2323 	/* Wait for socket readers to complete */
2324 	synchronize_net();
2325 
2326 	rhashtable_destroy(&tn->sk_rht);
2327 }
2328 
2329 /**
2330  * tipc_setsockopt - set socket option
2331  * @sock: socket structure
2332  * @lvl: option level
2333  * @opt: option identifier
2334  * @ov: pointer to new option value
2335  * @ol: length of option value
2336  *
2337  * For stream sockets only, accepts and ignores all IPPROTO_TCP options
2338  * (to ease compatibility).
2339  *
2340  * Returns 0 on success, errno otherwise
2341  */
2342 static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
2343 			   char __user *ov, unsigned int ol)
2344 {
2345 	struct sock *sk = sock->sk;
2346 	struct tipc_sock *tsk = tipc_sk(sk);
2347 	u32 value;
2348 	int res;
2349 
2350 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
2351 		return 0;
2352 	if (lvl != SOL_TIPC)
2353 		return -ENOPROTOOPT;
2354 	if (ol < sizeof(value))
2355 		return -EINVAL;
2356 	res = get_user(value, (u32 __user *)ov);
2357 	if (res)
2358 		return res;
2359 
2360 	lock_sock(sk);
2361 
2362 	switch (opt) {
2363 	case TIPC_IMPORTANCE:
2364 		res = tsk_set_importance(tsk, value);
2365 		break;
2366 	case TIPC_SRC_DROPPABLE:
2367 		if (sock->type != SOCK_STREAM)
2368 			tsk_set_unreliable(tsk, value);
2369 		else
2370 			res = -ENOPROTOOPT;
2371 		break;
2372 	case TIPC_DEST_DROPPABLE:
2373 		tsk_set_unreturnable(tsk, value);
2374 		break;
2375 	case TIPC_CONN_TIMEOUT:
2376 		tipc_sk(sk)->conn_timeout = value;
2377 		/* no need to set "res", since already 0 at this point */
2378 		break;
2379 	default:
2380 		res = -EINVAL;
2381 	}
2382 
2383 	release_sock(sk);
2384 
2385 	return res;
2386 }
2387 
2388 /**
2389  * tipc_getsockopt - get socket option
2390  * @sock: socket structure
2391  * @lvl: option level
2392  * @opt: option identifier
2393  * @ov: receptacle for option value
2394  * @ol: receptacle for length of option value
2395  *
2396  * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
2397  * (to ease compatibility).
2398  *
2399  * Returns 0 on success, errno otherwise
2400  */
2401 static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
2402 			   char __user *ov, int __user *ol)
2403 {
2404 	struct sock *sk = sock->sk;
2405 	struct tipc_sock *tsk = tipc_sk(sk);
2406 	int len;
2407 	u32 value;
2408 	int res;
2409 
2410 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
2411 		return put_user(0, ol);
2412 	if (lvl != SOL_TIPC)
2413 		return -ENOPROTOOPT;
2414 	res = get_user(len, ol);
2415 	if (res)
2416 		return res;
2417 
2418 	lock_sock(sk);
2419 
2420 	switch (opt) {
2421 	case TIPC_IMPORTANCE:
2422 		value = tsk_importance(tsk);
2423 		break;
2424 	case TIPC_SRC_DROPPABLE:
2425 		value = tsk_unreliable(tsk);
2426 		break;
2427 	case TIPC_DEST_DROPPABLE:
2428 		value = tsk_unreturnable(tsk);
2429 		break;
2430 	case TIPC_CONN_TIMEOUT:
2431 		value = tsk->conn_timeout;
2432 		/* no need to set "res", since already 0 at this point */
2433 		break;
2434 	case TIPC_NODE_RECVQ_DEPTH:
2435 		value = 0; /* was tipc_queue_size, now obsolete */
2436 		break;
2437 	case TIPC_SOCK_RECVQ_DEPTH:
2438 		value = skb_queue_len(&sk->sk_receive_queue);
2439 		break;
2440 	default:
2441 		res = -EINVAL;
2442 	}
2443 
2444 	release_sock(sk);
2445 
2446 	if (res)
2447 		return res;	/* "get" failed */
2448 
2449 	if (len < sizeof(value))
2450 		return -EINVAL;
2451 
2452 	if (copy_to_user(ov, &value, sizeof(value)))
2453 		return -EFAULT;
2454 
2455 	return put_user(sizeof(value), ol);
2456 }
2457 
2458 static int tipc_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
2459 {
2460 	struct sock *sk = sock->sk;
2461 	struct tipc_sioc_ln_req lnr;
2462 	void __user *argp = (void __user *)arg;
2463 
2464 	switch (cmd) {
2465 	case SIOCGETLINKNAME:
2466 		if (copy_from_user(&lnr, argp, sizeof(lnr)))
2467 			return -EFAULT;
2468 		if (!tipc_node_get_linkname(sock_net(sk),
2469 					    lnr.bearer_id & 0xffff, lnr.peer,
2470 					    lnr.linkname, TIPC_MAX_LINK_NAME)) {
2471 			if (copy_to_user(argp, &lnr, sizeof(lnr)))
2472 				return -EFAULT;
2473 			return 0;
2474 		}
2475 		return -EADDRNOTAVAIL;
2476 	default:
2477 		return -ENOIOCTLCMD;
2478 	}
2479 }
2480 
2481 /* Protocol switches for the various types of TIPC sockets */
2482 
2483 static const struct proto_ops msg_ops = {
2484 	.owner		= THIS_MODULE,
2485 	.family		= AF_TIPC,
2486 	.release	= tipc_release,
2487 	.bind		= tipc_bind,
2488 	.connect	= tipc_connect,
2489 	.socketpair	= sock_no_socketpair,
2490 	.accept		= sock_no_accept,
2491 	.getname	= tipc_getname,
2492 	.poll		= tipc_poll,
2493 	.ioctl		= tipc_ioctl,
2494 	.listen		= sock_no_listen,
2495 	.shutdown	= tipc_shutdown,
2496 	.setsockopt	= tipc_setsockopt,
2497 	.getsockopt	= tipc_getsockopt,
2498 	.sendmsg	= tipc_sendmsg,
2499 	.recvmsg	= tipc_recvmsg,
2500 	.mmap		= sock_no_mmap,
2501 	.sendpage	= sock_no_sendpage
2502 };
2503 
2504 static const struct proto_ops packet_ops = {
2505 	.owner		= THIS_MODULE,
2506 	.family		= AF_TIPC,
2507 	.release	= tipc_release,
2508 	.bind		= tipc_bind,
2509 	.connect	= tipc_connect,
2510 	.socketpair	= sock_no_socketpair,
2511 	.accept		= tipc_accept,
2512 	.getname	= tipc_getname,
2513 	.poll		= tipc_poll,
2514 	.ioctl		= tipc_ioctl,
2515 	.listen		= tipc_listen,
2516 	.shutdown	= tipc_shutdown,
2517 	.setsockopt	= tipc_setsockopt,
2518 	.getsockopt	= tipc_getsockopt,
2519 	.sendmsg	= tipc_send_packet,
2520 	.recvmsg	= tipc_recvmsg,
2521 	.mmap		= sock_no_mmap,
2522 	.sendpage	= sock_no_sendpage
2523 };
2524 
2525 static const struct proto_ops stream_ops = {
2526 	.owner		= THIS_MODULE,
2527 	.family		= AF_TIPC,
2528 	.release	= tipc_release,
2529 	.bind		= tipc_bind,
2530 	.connect	= tipc_connect,
2531 	.socketpair	= sock_no_socketpair,
2532 	.accept		= tipc_accept,
2533 	.getname	= tipc_getname,
2534 	.poll		= tipc_poll,
2535 	.ioctl		= tipc_ioctl,
2536 	.listen		= tipc_listen,
2537 	.shutdown	= tipc_shutdown,
2538 	.setsockopt	= tipc_setsockopt,
2539 	.getsockopt	= tipc_getsockopt,
2540 	.sendmsg	= tipc_send_stream,
2541 	.recvmsg	= tipc_recv_stream,
2542 	.mmap		= sock_no_mmap,
2543 	.sendpage	= sock_no_sendpage
2544 };
2545 
2546 static const struct net_proto_family tipc_family_ops = {
2547 	.owner		= THIS_MODULE,
2548 	.family		= AF_TIPC,
2549 	.create		= tipc_sk_create
2550 };
2551 
2552 static struct proto tipc_proto = {
2553 	.name		= "TIPC",
2554 	.owner		= THIS_MODULE,
2555 	.obj_size	= sizeof(struct tipc_sock),
2556 	.sysctl_rmem	= sysctl_tipc_rmem
2557 };
2558 
2559 /**
2560  * tipc_socket_init - initialize TIPC socket interface
2561  *
2562  * Returns 0 on success, errno otherwise
2563  */
2564 int tipc_socket_init(void)
2565 {
2566 	int res;
2567 
2568 	res = proto_register(&tipc_proto, 1);
2569 	if (res) {
2570 		pr_err("Failed to register TIPC protocol type\n");
2571 		goto out;
2572 	}
2573 
2574 	res = sock_register(&tipc_family_ops);
2575 	if (res) {
2576 		pr_err("Failed to register TIPC socket type\n");
2577 		proto_unregister(&tipc_proto);
2578 		goto out;
2579 	}
2580  out:
2581 	return res;
2582 }
2583 
2584 /**
2585  * tipc_socket_stop - stop TIPC socket interface
2586  */
2587 void tipc_socket_stop(void)
2588 {
2589 	sock_unregister(tipc_family_ops.family);
2590 	proto_unregister(&tipc_proto);
2591 }
2592 
2593 /* Caller should hold socket lock for the passed tipc socket. */
2594 static int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk)
2595 {
2596 	u32 peer_node;
2597 	u32 peer_port;
2598 	struct nlattr *nest;
2599 
2600 	peer_node = tsk_peer_node(tsk);
2601 	peer_port = tsk_peer_port(tsk);
2602 
2603 	nest = nla_nest_start(skb, TIPC_NLA_SOCK_CON);
2604 
2605 	if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node))
2606 		goto msg_full;
2607 	if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port))
2608 		goto msg_full;
2609 
2610 	if (tsk->conn_type != 0) {
2611 		if (nla_put_flag(skb, TIPC_NLA_CON_FLAG))
2612 			goto msg_full;
2613 		if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, tsk->conn_type))
2614 			goto msg_full;
2615 		if (nla_put_u32(skb, TIPC_NLA_CON_INST, tsk->conn_instance))
2616 			goto msg_full;
2617 	}
2618 	nla_nest_end(skb, nest);
2619 
2620 	return 0;
2621 
2622 msg_full:
2623 	nla_nest_cancel(skb, nest);
2624 
2625 	return -EMSGSIZE;
2626 }
2627 
2628 /* Caller should hold socket lock for the passed tipc socket. */
2629 static int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb,
2630 			    struct tipc_sock *tsk)
2631 {
2632 	int err;
2633 	void *hdr;
2634 	struct nlattr *attrs;
2635 	struct net *net = sock_net(skb->sk);
2636 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2637 
2638 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
2639 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_SOCK_GET);
2640 	if (!hdr)
2641 		goto msg_cancel;
2642 
2643 	attrs = nla_nest_start(skb, TIPC_NLA_SOCK);
2644 	if (!attrs)
2645 		goto genlmsg_cancel;
2646 	if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->portid))
2647 		goto attr_msg_cancel;
2648 	if (nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tn->own_addr))
2649 		goto attr_msg_cancel;
2650 
2651 	if (tsk->connected) {
2652 		err = __tipc_nl_add_sk_con(skb, tsk);
2653 		if (err)
2654 			goto attr_msg_cancel;
2655 	} else if (!list_empty(&tsk->publications)) {
2656 		if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL))
2657 			goto attr_msg_cancel;
2658 	}
2659 	nla_nest_end(skb, attrs);
2660 	genlmsg_end(skb, hdr);
2661 
2662 	return 0;
2663 
2664 attr_msg_cancel:
2665 	nla_nest_cancel(skb, attrs);
2666 genlmsg_cancel:
2667 	genlmsg_cancel(skb, hdr);
2668 msg_cancel:
2669 	return -EMSGSIZE;
2670 }
2671 
2672 int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb)
2673 {
2674 	int err;
2675 	struct tipc_sock *tsk;
2676 	const struct bucket_table *tbl;
2677 	struct rhash_head *pos;
2678 	struct net *net = sock_net(skb->sk);
2679 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2680 	u32 tbl_id = cb->args[0];
2681 	u32 prev_portid = cb->args[1];
2682 
2683 	rcu_read_lock();
2684 	tbl = rht_dereference_rcu((&tn->sk_rht)->tbl, &tn->sk_rht);
2685 	for (; tbl_id < tbl->size; tbl_id++) {
2686 		rht_for_each_entry_rcu(tsk, pos, tbl, tbl_id, node) {
2687 			spin_lock_bh(&tsk->sk.sk_lock.slock);
2688 			if (prev_portid && prev_portid != tsk->portid) {
2689 				spin_unlock_bh(&tsk->sk.sk_lock.slock);
2690 				continue;
2691 			}
2692 
2693 			err = __tipc_nl_add_sk(skb, cb, tsk);
2694 			if (err) {
2695 				prev_portid = tsk->portid;
2696 				spin_unlock_bh(&tsk->sk.sk_lock.slock);
2697 				goto out;
2698 			}
2699 			prev_portid = 0;
2700 			spin_unlock_bh(&tsk->sk.sk_lock.slock);
2701 		}
2702 	}
2703 out:
2704 	rcu_read_unlock();
2705 	cb->args[0] = tbl_id;
2706 	cb->args[1] = prev_portid;
2707 
2708 	return skb->len;
2709 }
2710 
2711 /* Caller should hold socket lock for the passed tipc socket. */
2712 static int __tipc_nl_add_sk_publ(struct sk_buff *skb,
2713 				 struct netlink_callback *cb,
2714 				 struct publication *publ)
2715 {
2716 	void *hdr;
2717 	struct nlattr *attrs;
2718 
2719 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
2720 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_PUBL_GET);
2721 	if (!hdr)
2722 		goto msg_cancel;
2723 
2724 	attrs = nla_nest_start(skb, TIPC_NLA_PUBL);
2725 	if (!attrs)
2726 		goto genlmsg_cancel;
2727 
2728 	if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key))
2729 		goto attr_msg_cancel;
2730 	if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->type))
2731 		goto attr_msg_cancel;
2732 	if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->lower))
2733 		goto attr_msg_cancel;
2734 	if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->upper))
2735 		goto attr_msg_cancel;
2736 
2737 	nla_nest_end(skb, attrs);
2738 	genlmsg_end(skb, hdr);
2739 
2740 	return 0;
2741 
2742 attr_msg_cancel:
2743 	nla_nest_cancel(skb, attrs);
2744 genlmsg_cancel:
2745 	genlmsg_cancel(skb, hdr);
2746 msg_cancel:
2747 	return -EMSGSIZE;
2748 }
2749 
2750 /* Caller should hold socket lock for the passed tipc socket. */
2751 static int __tipc_nl_list_sk_publ(struct sk_buff *skb,
2752 				  struct netlink_callback *cb,
2753 				  struct tipc_sock *tsk, u32 *last_publ)
2754 {
2755 	int err;
2756 	struct publication *p;
2757 
2758 	if (*last_publ) {
2759 		list_for_each_entry(p, &tsk->publications, pport_list) {
2760 			if (p->key == *last_publ)
2761 				break;
2762 		}
2763 		if (p->key != *last_publ) {
2764 			/* We never set seq or call nl_dump_check_consistent()
2765 			 * this means that setting prev_seq here will cause the
2766 			 * consistence check to fail in the netlink callback
2767 			 * handler. Resulting in the last NLMSG_DONE message
2768 			 * having the NLM_F_DUMP_INTR flag set.
2769 			 */
2770 			cb->prev_seq = 1;
2771 			*last_publ = 0;
2772 			return -EPIPE;
2773 		}
2774 	} else {
2775 		p = list_first_entry(&tsk->publications, struct publication,
2776 				     pport_list);
2777 	}
2778 
2779 	list_for_each_entry_from(p, &tsk->publications, pport_list) {
2780 		err = __tipc_nl_add_sk_publ(skb, cb, p);
2781 		if (err) {
2782 			*last_publ = p->key;
2783 			return err;
2784 		}
2785 	}
2786 	*last_publ = 0;
2787 
2788 	return 0;
2789 }
2790 
2791 int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)
2792 {
2793 	int err;
2794 	u32 tsk_portid = cb->args[0];
2795 	u32 last_publ = cb->args[1];
2796 	u32 done = cb->args[2];
2797 	struct net *net = sock_net(skb->sk);
2798 	struct tipc_sock *tsk;
2799 
2800 	if (!tsk_portid) {
2801 		struct nlattr **attrs;
2802 		struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1];
2803 
2804 		err = tipc_nlmsg_parse(cb->nlh, &attrs);
2805 		if (err)
2806 			return err;
2807 
2808 		err = nla_parse_nested(sock, TIPC_NLA_SOCK_MAX,
2809 				       attrs[TIPC_NLA_SOCK],
2810 				       tipc_nl_sock_policy);
2811 		if (err)
2812 			return err;
2813 
2814 		if (!sock[TIPC_NLA_SOCK_REF])
2815 			return -EINVAL;
2816 
2817 		tsk_portid = nla_get_u32(sock[TIPC_NLA_SOCK_REF]);
2818 	}
2819 
2820 	if (done)
2821 		return 0;
2822 
2823 	tsk = tipc_sk_lookup(net, tsk_portid);
2824 	if (!tsk)
2825 		return -EINVAL;
2826 
2827 	lock_sock(&tsk->sk);
2828 	err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ);
2829 	if (!err)
2830 		done = 1;
2831 	release_sock(&tsk->sk);
2832 	sock_put(&tsk->sk);
2833 
2834 	cb->args[0] = tsk_portid;
2835 	cb->args[1] = last_publ;
2836 	cb->args[2] = done;
2837 
2838 	return skb->len;
2839 }
2840