xref: /linux/net/tipc/socket.c (revision 1f2367a39f17bd553a75e179a747f9b257bc9478)
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 
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 = 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 	sk_stop_timer(sk, &sk->sk_timer);
586 	tipc_sk_remove(tsk);
587 
588 	sock_orphan(sk);
589 	/* Reject any messages that accumulated in backlog queue */
590 	release_sock(sk);
591 	tipc_dest_list_purge(&tsk->cong_links);
592 	tsk->cong_link_cnt = 0;
593 	call_rcu(&tsk->rcu, tipc_sk_callback);
594 	sock->sk = NULL;
595 
596 	return 0;
597 }
598 
599 /**
600  * tipc_bind - associate or disassocate TIPC name(s) with a socket
601  * @sock: socket structure
602  * @uaddr: socket address describing name(s) and desired operation
603  * @uaddr_len: size of socket address data structure
604  *
605  * Name and name sequence binding is indicated using a positive scope value;
606  * a negative scope value unbinds the specified name.  Specifying no name
607  * (i.e. a socket address length of 0) unbinds all names from the socket.
608  *
609  * Returns 0 on success, errno otherwise
610  *
611  * NOTE: This routine doesn't need to take the socket lock since it doesn't
612  *       access any non-constant socket information.
613  */
614 static int tipc_bind(struct socket *sock, struct sockaddr *uaddr,
615 		     int uaddr_len)
616 {
617 	struct sock *sk = sock->sk;
618 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
619 	struct tipc_sock *tsk = tipc_sk(sk);
620 	int res = -EINVAL;
621 
622 	lock_sock(sk);
623 	if (unlikely(!uaddr_len)) {
624 		res = tipc_sk_withdraw(tsk, 0, NULL);
625 		goto exit;
626 	}
627 	if (tsk->group) {
628 		res = -EACCES;
629 		goto exit;
630 	}
631 	if (uaddr_len < sizeof(struct sockaddr_tipc)) {
632 		res = -EINVAL;
633 		goto exit;
634 	}
635 	if (addr->family != AF_TIPC) {
636 		res = -EAFNOSUPPORT;
637 		goto exit;
638 	}
639 
640 	if (addr->addrtype == TIPC_ADDR_NAME)
641 		addr->addr.nameseq.upper = addr->addr.nameseq.lower;
642 	else if (addr->addrtype != TIPC_ADDR_NAMESEQ) {
643 		res = -EAFNOSUPPORT;
644 		goto exit;
645 	}
646 
647 	if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) &&
648 	    (addr->addr.nameseq.type != TIPC_TOP_SRV) &&
649 	    (addr->addr.nameseq.type != TIPC_CFG_SRV)) {
650 		res = -EACCES;
651 		goto exit;
652 	}
653 
654 	res = (addr->scope >= 0) ?
655 		tipc_sk_publish(tsk, addr->scope, &addr->addr.nameseq) :
656 		tipc_sk_withdraw(tsk, -addr->scope, &addr->addr.nameseq);
657 exit:
658 	release_sock(sk);
659 	return res;
660 }
661 
662 /**
663  * tipc_getname - get port ID of socket or peer socket
664  * @sock: socket structure
665  * @uaddr: area for returned socket address
666  * @uaddr_len: area for returned length of socket address
667  * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
668  *
669  * Returns 0 on success, errno otherwise
670  *
671  * NOTE: This routine doesn't need to take the socket lock since it only
672  *       accesses socket information that is unchanging (or which changes in
673  *       a completely predictable manner).
674  */
675 static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
676 			int peer)
677 {
678 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
679 	struct sock *sk = sock->sk;
680 	struct tipc_sock *tsk = tipc_sk(sk);
681 
682 	memset(addr, 0, sizeof(*addr));
683 	if (peer) {
684 		if ((!tipc_sk_connected(sk)) &&
685 		    ((peer != 2) || (sk->sk_state != TIPC_DISCONNECTING)))
686 			return -ENOTCONN;
687 		addr->addr.id.ref = tsk_peer_port(tsk);
688 		addr->addr.id.node = tsk_peer_node(tsk);
689 	} else {
690 		addr->addr.id.ref = tsk->portid;
691 		addr->addr.id.node = tipc_own_addr(sock_net(sk));
692 	}
693 
694 	addr->addrtype = TIPC_ADDR_ID;
695 	addr->family = AF_TIPC;
696 	addr->scope = 0;
697 	addr->addr.name.domain = 0;
698 
699 	return sizeof(*addr);
700 }
701 
702 /**
703  * tipc_poll - read and possibly block on pollmask
704  * @file: file structure associated with the socket
705  * @sock: socket for which to calculate the poll bits
706  * @wait: ???
707  *
708  * Returns pollmask value
709  *
710  * COMMENTARY:
711  * It appears that the usual socket locking mechanisms are not useful here
712  * since the pollmask info is potentially out-of-date the moment this routine
713  * exits.  TCP and other protocols seem to rely on higher level poll routines
714  * to handle any preventable race conditions, so TIPC will do the same ...
715  *
716  * IMPORTANT: The fact that a read or write operation is indicated does NOT
717  * imply that the operation will succeed, merely that it should be performed
718  * and will not block.
719  */
720 static __poll_t tipc_poll(struct file *file, struct socket *sock,
721 			      poll_table *wait)
722 {
723 	struct sock *sk = sock->sk;
724 	struct tipc_sock *tsk = tipc_sk(sk);
725 	__poll_t revents = 0;
726 
727 	sock_poll_wait(file, sock, wait);
728 	trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " ");
729 
730 	if (sk->sk_shutdown & RCV_SHUTDOWN)
731 		revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
732 	if (sk->sk_shutdown == SHUTDOWN_MASK)
733 		revents |= EPOLLHUP;
734 
735 	switch (sk->sk_state) {
736 	case TIPC_ESTABLISHED:
737 	case TIPC_CONNECTING:
738 		if (!tsk->cong_link_cnt && !tsk_conn_cong(tsk))
739 			revents |= EPOLLOUT;
740 		/* fall through */
741 	case TIPC_LISTEN:
742 		if (!skb_queue_empty(&sk->sk_receive_queue))
743 			revents |= EPOLLIN | EPOLLRDNORM;
744 		break;
745 	case TIPC_OPEN:
746 		if (tsk->group_is_open && !tsk->cong_link_cnt)
747 			revents |= EPOLLOUT;
748 		if (!tipc_sk_type_connectionless(sk))
749 			break;
750 		if (skb_queue_empty(&sk->sk_receive_queue))
751 			break;
752 		revents |= EPOLLIN | EPOLLRDNORM;
753 		break;
754 	case TIPC_DISCONNECTING:
755 		revents = EPOLLIN | EPOLLRDNORM | EPOLLHUP;
756 		break;
757 	}
758 	return revents;
759 }
760 
761 /**
762  * tipc_sendmcast - send multicast message
763  * @sock: socket structure
764  * @seq: destination address
765  * @msg: message to send
766  * @dlen: length of data to send
767  * @timeout: timeout to wait for wakeup
768  *
769  * Called from function tipc_sendmsg(), which has done all sanity checks
770  * Returns the number of bytes sent on success, or errno
771  */
772 static int tipc_sendmcast(struct  socket *sock, struct tipc_name_seq *seq,
773 			  struct msghdr *msg, size_t dlen, long timeout)
774 {
775 	struct sock *sk = sock->sk;
776 	struct tipc_sock *tsk = tipc_sk(sk);
777 	struct tipc_msg *hdr = &tsk->phdr;
778 	struct net *net = sock_net(sk);
779 	int mtu = tipc_bcast_get_mtu(net);
780 	struct tipc_mc_method *method = &tsk->mc_method;
781 	struct sk_buff_head pkts;
782 	struct tipc_nlist dsts;
783 	int rc;
784 
785 	if (tsk->group)
786 		return -EACCES;
787 
788 	/* Block or return if any destination link is congested */
789 	rc = tipc_wait_for_cond(sock, &timeout, !tsk->cong_link_cnt);
790 	if (unlikely(rc))
791 		return rc;
792 
793 	/* Lookup destination nodes */
794 	tipc_nlist_init(&dsts, tipc_own_addr(net));
795 	tipc_nametbl_lookup_dst_nodes(net, seq->type, seq->lower,
796 				      seq->upper, &dsts);
797 	if (!dsts.local && !dsts.remote)
798 		return -EHOSTUNREACH;
799 
800 	/* Build message header */
801 	msg_set_type(hdr, TIPC_MCAST_MSG);
802 	msg_set_hdr_sz(hdr, MCAST_H_SIZE);
803 	msg_set_lookup_scope(hdr, TIPC_CLUSTER_SCOPE);
804 	msg_set_destport(hdr, 0);
805 	msg_set_destnode(hdr, 0);
806 	msg_set_nametype(hdr, seq->type);
807 	msg_set_namelower(hdr, seq->lower);
808 	msg_set_nameupper(hdr, seq->upper);
809 
810 	/* Build message as chain of buffers */
811 	skb_queue_head_init(&pkts);
812 	rc = tipc_msg_build(hdr, msg, 0, dlen, mtu, &pkts);
813 
814 	/* Send message if build was successful */
815 	if (unlikely(rc == dlen)) {
816 		trace_tipc_sk_sendmcast(sk, skb_peek(&pkts),
817 					TIPC_DUMP_SK_SNDQ, " ");
818 		rc = tipc_mcast_xmit(net, &pkts, method, &dsts,
819 				     &tsk->cong_link_cnt);
820 	}
821 
822 	tipc_nlist_purge(&dsts);
823 
824 	return rc ? rc : dlen;
825 }
826 
827 /**
828  * tipc_send_group_msg - send a message to a member in the group
829  * @net: network namespace
830  * @m: message to send
831  * @mb: group member
832  * @dnode: destination node
833  * @dport: destination port
834  * @dlen: total length of message data
835  */
836 static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
837 			       struct msghdr *m, struct tipc_member *mb,
838 			       u32 dnode, u32 dport, int dlen)
839 {
840 	u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group);
841 	struct tipc_mc_method *method = &tsk->mc_method;
842 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
843 	struct tipc_msg *hdr = &tsk->phdr;
844 	struct sk_buff_head pkts;
845 	int mtu, rc;
846 
847 	/* Complete message header */
848 	msg_set_type(hdr, TIPC_GRP_UCAST_MSG);
849 	msg_set_hdr_sz(hdr, GROUP_H_SIZE);
850 	msg_set_destport(hdr, dport);
851 	msg_set_destnode(hdr, dnode);
852 	msg_set_grp_bc_seqno(hdr, bc_snd_nxt);
853 
854 	/* Build message as chain of buffers */
855 	skb_queue_head_init(&pkts);
856 	mtu = tipc_node_get_mtu(net, dnode, tsk->portid);
857 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
858 	if (unlikely(rc != dlen))
859 		return rc;
860 
861 	/* Send message */
862 	rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
863 	if (unlikely(rc == -ELINKCONG)) {
864 		tipc_dest_push(&tsk->cong_links, dnode, 0);
865 		tsk->cong_link_cnt++;
866 	}
867 
868 	/* Update send window */
869 	tipc_group_update_member(mb, blks);
870 
871 	/* A broadcast sent within next EXPIRE period must follow same path */
872 	method->rcast = true;
873 	method->mandatory = true;
874 	return dlen;
875 }
876 
877 /**
878  * tipc_send_group_unicast - send message to a member in the group
879  * @sock: socket structure
880  * @m: message to send
881  * @dlen: total length of message data
882  * @timeout: timeout to wait for wakeup
883  *
884  * Called from function tipc_sendmsg(), which has done all sanity checks
885  * Returns the number of bytes sent on success, or errno
886  */
887 static int tipc_send_group_unicast(struct socket *sock, struct msghdr *m,
888 				   int dlen, long timeout)
889 {
890 	struct sock *sk = sock->sk;
891 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
892 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
893 	struct tipc_sock *tsk = tipc_sk(sk);
894 	struct net *net = sock_net(sk);
895 	struct tipc_member *mb = NULL;
896 	u32 node, port;
897 	int rc;
898 
899 	node = dest->addr.id.node;
900 	port = dest->addr.id.ref;
901 	if (!port && !node)
902 		return -EHOSTUNREACH;
903 
904 	/* Block or return if destination link or member is congested */
905 	rc = tipc_wait_for_cond(sock, &timeout,
906 				!tipc_dest_find(&tsk->cong_links, node, 0) &&
907 				tsk->group &&
908 				!tipc_group_cong(tsk->group, node, port, blks,
909 						 &mb));
910 	if (unlikely(rc))
911 		return rc;
912 
913 	if (unlikely(!mb))
914 		return -EHOSTUNREACH;
915 
916 	rc = tipc_send_group_msg(net, tsk, m, mb, node, port, dlen);
917 
918 	return rc ? rc : dlen;
919 }
920 
921 /**
922  * tipc_send_group_anycast - send message to any member with given identity
923  * @sock: socket structure
924  * @m: message to send
925  * @dlen: total length of message data
926  * @timeout: timeout to wait for wakeup
927  *
928  * Called from function tipc_sendmsg(), which has done all sanity checks
929  * Returns the number of bytes sent on success, or errno
930  */
931 static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m,
932 				   int dlen, long timeout)
933 {
934 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
935 	struct sock *sk = sock->sk;
936 	struct tipc_sock *tsk = tipc_sk(sk);
937 	struct list_head *cong_links = &tsk->cong_links;
938 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
939 	struct tipc_msg *hdr = &tsk->phdr;
940 	struct tipc_member *first = NULL;
941 	struct tipc_member *mbr = NULL;
942 	struct net *net = sock_net(sk);
943 	u32 node, port, exclude;
944 	struct list_head dsts;
945 	u32 type, inst, scope;
946 	int lookups = 0;
947 	int dstcnt, rc;
948 	bool cong;
949 
950 	INIT_LIST_HEAD(&dsts);
951 
952 	type = msg_nametype(hdr);
953 	inst = dest->addr.name.name.instance;
954 	scope = msg_lookup_scope(hdr);
955 
956 	while (++lookups < 4) {
957 		exclude = tipc_group_exclude(tsk->group);
958 
959 		first = NULL;
960 
961 		/* Look for a non-congested destination member, if any */
962 		while (1) {
963 			if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
964 						 &dstcnt, exclude, false))
965 				return -EHOSTUNREACH;
966 			tipc_dest_pop(&dsts, &node, &port);
967 			cong = tipc_group_cong(tsk->group, node, port, blks,
968 					       &mbr);
969 			if (!cong)
970 				break;
971 			if (mbr == first)
972 				break;
973 			if (!first)
974 				first = mbr;
975 		}
976 
977 		/* Start over if destination was not in member list */
978 		if (unlikely(!mbr))
979 			continue;
980 
981 		if (likely(!cong && !tipc_dest_find(cong_links, node, 0)))
982 			break;
983 
984 		/* Block or return if destination link or member is congested */
985 		rc = tipc_wait_for_cond(sock, &timeout,
986 					!tipc_dest_find(cong_links, node, 0) &&
987 					tsk->group &&
988 					!tipc_group_cong(tsk->group, node, port,
989 							 blks, &mbr));
990 		if (unlikely(rc))
991 			return rc;
992 
993 		/* Send, unless destination disappeared while waiting */
994 		if (likely(mbr))
995 			break;
996 	}
997 
998 	if (unlikely(lookups >= 4))
999 		return -EHOSTUNREACH;
1000 
1001 	rc = tipc_send_group_msg(net, tsk, m, mbr, node, port, dlen);
1002 
1003 	return rc ? rc : dlen;
1004 }
1005 
1006 /**
1007  * tipc_send_group_bcast - send message to all members in communication group
1008  * @sk: socket structure
1009  * @m: message to send
1010  * @dlen: total length of message data
1011  * @timeout: timeout to wait for wakeup
1012  *
1013  * Called from function tipc_sendmsg(), which has done all sanity checks
1014  * Returns the number of bytes sent on success, or errno
1015  */
1016 static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
1017 				 int dlen, long timeout)
1018 {
1019 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1020 	struct sock *sk = sock->sk;
1021 	struct net *net = sock_net(sk);
1022 	struct tipc_sock *tsk = tipc_sk(sk);
1023 	struct tipc_nlist *dsts;
1024 	struct tipc_mc_method *method = &tsk->mc_method;
1025 	bool ack = method->mandatory && method->rcast;
1026 	int blks = tsk_blocks(MCAST_H_SIZE + dlen);
1027 	struct tipc_msg *hdr = &tsk->phdr;
1028 	int mtu = tipc_bcast_get_mtu(net);
1029 	struct sk_buff_head pkts;
1030 	int rc = -EHOSTUNREACH;
1031 
1032 	/* Block or return if any destination link or member is congested */
1033 	rc = tipc_wait_for_cond(sock, &timeout,
1034 				!tsk->cong_link_cnt && tsk->group &&
1035 				!tipc_group_bc_cong(tsk->group, blks));
1036 	if (unlikely(rc))
1037 		return rc;
1038 
1039 	dsts = tipc_group_dests(tsk->group);
1040 	if (!dsts->local && !dsts->remote)
1041 		return -EHOSTUNREACH;
1042 
1043 	/* Complete message header */
1044 	if (dest) {
1045 		msg_set_type(hdr, TIPC_GRP_MCAST_MSG);
1046 		msg_set_nameinst(hdr, dest->addr.name.name.instance);
1047 	} else {
1048 		msg_set_type(hdr, TIPC_GRP_BCAST_MSG);
1049 		msg_set_nameinst(hdr, 0);
1050 	}
1051 	msg_set_hdr_sz(hdr, GROUP_H_SIZE);
1052 	msg_set_destport(hdr, 0);
1053 	msg_set_destnode(hdr, 0);
1054 	msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(tsk->group));
1055 
1056 	/* Avoid getting stuck with repeated forced replicasts */
1057 	msg_set_grp_bc_ack_req(hdr, ack);
1058 
1059 	/* Build message as chain of buffers */
1060 	skb_queue_head_init(&pkts);
1061 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1062 	if (unlikely(rc != dlen))
1063 		return rc;
1064 
1065 	/* Send message */
1066 	rc = tipc_mcast_xmit(net, &pkts, method, dsts, &tsk->cong_link_cnt);
1067 	if (unlikely(rc))
1068 		return rc;
1069 
1070 	/* Update broadcast sequence number and send windows */
1071 	tipc_group_update_bc_members(tsk->group, blks, ack);
1072 
1073 	/* Broadcast link is now free to choose method for next broadcast */
1074 	method->mandatory = false;
1075 	method->expires = jiffies;
1076 
1077 	return dlen;
1078 }
1079 
1080 /**
1081  * tipc_send_group_mcast - send message to all members with given identity
1082  * @sock: socket structure
1083  * @m: message to send
1084  * @dlen: total length of message data
1085  * @timeout: timeout to wait for wakeup
1086  *
1087  * Called from function tipc_sendmsg(), which has done all sanity checks
1088  * Returns the number of bytes sent on success, or errno
1089  */
1090 static int tipc_send_group_mcast(struct socket *sock, struct msghdr *m,
1091 				 int dlen, long timeout)
1092 {
1093 	struct sock *sk = sock->sk;
1094 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1095 	struct tipc_sock *tsk = tipc_sk(sk);
1096 	struct tipc_group *grp = tsk->group;
1097 	struct tipc_msg *hdr = &tsk->phdr;
1098 	struct net *net = sock_net(sk);
1099 	u32 type, inst, scope, exclude;
1100 	struct list_head dsts;
1101 	u32 dstcnt;
1102 
1103 	INIT_LIST_HEAD(&dsts);
1104 
1105 	type = msg_nametype(hdr);
1106 	inst = dest->addr.name.name.instance;
1107 	scope = msg_lookup_scope(hdr);
1108 	exclude = tipc_group_exclude(grp);
1109 
1110 	if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
1111 				 &dstcnt, exclude, true))
1112 		return -EHOSTUNREACH;
1113 
1114 	if (dstcnt == 1) {
1115 		tipc_dest_pop(&dsts, &dest->addr.id.node, &dest->addr.id.ref);
1116 		return tipc_send_group_unicast(sock, m, dlen, timeout);
1117 	}
1118 
1119 	tipc_dest_list_purge(&dsts);
1120 	return tipc_send_group_bcast(sock, m, dlen, timeout);
1121 }
1122 
1123 /**
1124  * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
1125  * @arrvq: queue with arriving messages, to be cloned after destination lookup
1126  * @inputq: queue with cloned messages, delivered to socket after dest lookup
1127  *
1128  * Multi-threaded: parallel calls with reference to same queues may occur
1129  */
1130 void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
1131 		       struct sk_buff_head *inputq)
1132 {
1133 	u32 self = tipc_own_addr(net);
1134 	u32 type, lower, upper, scope;
1135 	struct sk_buff *skb, *_skb;
1136 	u32 portid, onode;
1137 	struct sk_buff_head tmpq;
1138 	struct list_head dports;
1139 	struct tipc_msg *hdr;
1140 	int user, mtyp, hlen;
1141 	bool exact;
1142 
1143 	__skb_queue_head_init(&tmpq);
1144 	INIT_LIST_HEAD(&dports);
1145 
1146 	skb = tipc_skb_peek(arrvq, &inputq->lock);
1147 	for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) {
1148 		hdr = buf_msg(skb);
1149 		user = msg_user(hdr);
1150 		mtyp = msg_type(hdr);
1151 		hlen = skb_headroom(skb) + msg_hdr_sz(hdr);
1152 		onode = msg_orignode(hdr);
1153 		type = msg_nametype(hdr);
1154 
1155 		if (mtyp == TIPC_GRP_UCAST_MSG || user == GROUP_PROTOCOL) {
1156 			spin_lock_bh(&inputq->lock);
1157 			if (skb_peek(arrvq) == skb) {
1158 				__skb_dequeue(arrvq);
1159 				__skb_queue_tail(inputq, skb);
1160 			}
1161 			kfree_skb(skb);
1162 			spin_unlock_bh(&inputq->lock);
1163 			continue;
1164 		}
1165 
1166 		/* Group messages require exact scope match */
1167 		if (msg_in_group(hdr)) {
1168 			lower = 0;
1169 			upper = ~0;
1170 			scope = msg_lookup_scope(hdr);
1171 			exact = true;
1172 		} else {
1173 			/* TIPC_NODE_SCOPE means "any scope" in this context */
1174 			if (onode == self)
1175 				scope = TIPC_NODE_SCOPE;
1176 			else
1177 				scope = TIPC_CLUSTER_SCOPE;
1178 			exact = false;
1179 			lower = msg_namelower(hdr);
1180 			upper = msg_nameupper(hdr);
1181 		}
1182 
1183 		/* Create destination port list: */
1184 		tipc_nametbl_mc_lookup(net, type, lower, upper,
1185 				       scope, exact, &dports);
1186 
1187 		/* Clone message per destination */
1188 		while (tipc_dest_pop(&dports, NULL, &portid)) {
1189 			_skb = __pskb_copy(skb, hlen, GFP_ATOMIC);
1190 			if (_skb) {
1191 				msg_set_destport(buf_msg(_skb), portid);
1192 				__skb_queue_tail(&tmpq, _skb);
1193 				continue;
1194 			}
1195 			pr_warn("Failed to clone mcast rcv buffer\n");
1196 		}
1197 		/* Append to inputq if not already done by other thread */
1198 		spin_lock_bh(&inputq->lock);
1199 		if (skb_peek(arrvq) == skb) {
1200 			skb_queue_splice_tail_init(&tmpq, inputq);
1201 			kfree_skb(__skb_dequeue(arrvq));
1202 		}
1203 		spin_unlock_bh(&inputq->lock);
1204 		__skb_queue_purge(&tmpq);
1205 		kfree_skb(skb);
1206 	}
1207 	tipc_sk_rcv(net, inputq);
1208 }
1209 
1210 /**
1211  * tipc_sk_conn_proto_rcv - receive a connection mng protocol message
1212  * @tsk: receiving socket
1213  * @skb: pointer to message buffer.
1214  */
1215 static void tipc_sk_conn_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb,
1216 				   struct sk_buff_head *inputq,
1217 				   struct sk_buff_head *xmitq)
1218 {
1219 	struct tipc_msg *hdr = buf_msg(skb);
1220 	u32 onode = tsk_own_node(tsk);
1221 	struct sock *sk = &tsk->sk;
1222 	int mtyp = msg_type(hdr);
1223 	bool conn_cong;
1224 
1225 	/* Ignore if connection cannot be validated: */
1226 	if (!tsk_peer_msg(tsk, hdr)) {
1227 		trace_tipc_sk_drop_msg(sk, skb, TIPC_DUMP_NONE, "@proto_rcv!");
1228 		goto exit;
1229 	}
1230 
1231 	if (unlikely(msg_errcode(hdr))) {
1232 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
1233 		tipc_node_remove_conn(sock_net(sk), tsk_peer_node(tsk),
1234 				      tsk_peer_port(tsk));
1235 		sk->sk_state_change(sk);
1236 
1237 		/* State change is ignored if socket already awake,
1238 		 * - convert msg to abort msg and add to inqueue
1239 		 */
1240 		msg_set_user(hdr, TIPC_CRITICAL_IMPORTANCE);
1241 		msg_set_type(hdr, TIPC_CONN_MSG);
1242 		msg_set_size(hdr, BASIC_H_SIZE);
1243 		msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1244 		__skb_queue_tail(inputq, skb);
1245 		return;
1246 	}
1247 
1248 	tsk->probe_unacked = false;
1249 
1250 	if (mtyp == CONN_PROBE) {
1251 		msg_set_type(hdr, CONN_PROBE_REPLY);
1252 		if (tipc_msg_reverse(onode, &skb, TIPC_OK))
1253 			__skb_queue_tail(xmitq, skb);
1254 		return;
1255 	} else if (mtyp == CONN_ACK) {
1256 		conn_cong = tsk_conn_cong(tsk);
1257 		tsk->snt_unacked -= msg_conn_ack(hdr);
1258 		if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1259 			tsk->snd_win = msg_adv_win(hdr);
1260 		if (conn_cong)
1261 			sk->sk_write_space(sk);
1262 	} else if (mtyp != CONN_PROBE_REPLY) {
1263 		pr_warn("Received unknown CONN_PROTO msg\n");
1264 	}
1265 exit:
1266 	kfree_skb(skb);
1267 }
1268 
1269 /**
1270  * tipc_sendmsg - send message in connectionless manner
1271  * @sock: socket structure
1272  * @m: message to send
1273  * @dsz: amount of user data to be sent
1274  *
1275  * Message must have an destination specified explicitly.
1276  * Used for SOCK_RDM and SOCK_DGRAM messages,
1277  * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
1278  * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
1279  *
1280  * Returns the number of bytes sent on success, or errno otherwise
1281  */
1282 static int tipc_sendmsg(struct socket *sock,
1283 			struct msghdr *m, size_t dsz)
1284 {
1285 	struct sock *sk = sock->sk;
1286 	int ret;
1287 
1288 	lock_sock(sk);
1289 	ret = __tipc_sendmsg(sock, m, dsz);
1290 	release_sock(sk);
1291 
1292 	return ret;
1293 }
1294 
1295 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)
1296 {
1297 	struct sock *sk = sock->sk;
1298 	struct net *net = sock_net(sk);
1299 	struct tipc_sock *tsk = tipc_sk(sk);
1300 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1301 	long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1302 	struct list_head *clinks = &tsk->cong_links;
1303 	bool syn = !tipc_sk_type_connectionless(sk);
1304 	struct tipc_group *grp = tsk->group;
1305 	struct tipc_msg *hdr = &tsk->phdr;
1306 	struct tipc_name_seq *seq;
1307 	struct sk_buff_head pkts;
1308 	u32 dport, dnode = 0;
1309 	u32 type, inst;
1310 	int mtu, rc;
1311 
1312 	if (unlikely(dlen > TIPC_MAX_USER_MSG_SIZE))
1313 		return -EMSGSIZE;
1314 
1315 	if (likely(dest)) {
1316 		if (unlikely(m->msg_namelen < sizeof(*dest)))
1317 			return -EINVAL;
1318 		if (unlikely(dest->family != AF_TIPC))
1319 			return -EINVAL;
1320 	}
1321 
1322 	if (grp) {
1323 		if (!dest)
1324 			return tipc_send_group_bcast(sock, m, dlen, timeout);
1325 		if (dest->addrtype == TIPC_ADDR_NAME)
1326 			return tipc_send_group_anycast(sock, m, dlen, timeout);
1327 		if (dest->addrtype == TIPC_ADDR_ID)
1328 			return tipc_send_group_unicast(sock, m, dlen, timeout);
1329 		if (dest->addrtype == TIPC_ADDR_MCAST)
1330 			return tipc_send_group_mcast(sock, m, dlen, timeout);
1331 		return -EINVAL;
1332 	}
1333 
1334 	if (unlikely(!dest)) {
1335 		dest = &tsk->peer;
1336 		if (!syn && dest->family != AF_TIPC)
1337 			return -EDESTADDRREQ;
1338 	}
1339 
1340 	if (unlikely(syn)) {
1341 		if (sk->sk_state == TIPC_LISTEN)
1342 			return -EPIPE;
1343 		if (sk->sk_state != TIPC_OPEN)
1344 			return -EISCONN;
1345 		if (tsk->published)
1346 			return -EOPNOTSUPP;
1347 		if (dest->addrtype == TIPC_ADDR_NAME) {
1348 			tsk->conn_type = dest->addr.name.name.type;
1349 			tsk->conn_instance = dest->addr.name.name.instance;
1350 		}
1351 		msg_set_syn(hdr, 1);
1352 	}
1353 
1354 	seq = &dest->addr.nameseq;
1355 	if (dest->addrtype == TIPC_ADDR_MCAST)
1356 		return tipc_sendmcast(sock, seq, m, dlen, timeout);
1357 
1358 	if (dest->addrtype == TIPC_ADDR_NAME) {
1359 		type = dest->addr.name.name.type;
1360 		inst = dest->addr.name.name.instance;
1361 		dnode = dest->addr.name.domain;
1362 		msg_set_type(hdr, TIPC_NAMED_MSG);
1363 		msg_set_hdr_sz(hdr, NAMED_H_SIZE);
1364 		msg_set_nametype(hdr, type);
1365 		msg_set_nameinst(hdr, inst);
1366 		msg_set_lookup_scope(hdr, tipc_node2scope(dnode));
1367 		dport = tipc_nametbl_translate(net, type, inst, &dnode);
1368 		msg_set_destnode(hdr, dnode);
1369 		msg_set_destport(hdr, dport);
1370 		if (unlikely(!dport && !dnode))
1371 			return -EHOSTUNREACH;
1372 	} else if (dest->addrtype == TIPC_ADDR_ID) {
1373 		dnode = dest->addr.id.node;
1374 		msg_set_type(hdr, TIPC_DIRECT_MSG);
1375 		msg_set_lookup_scope(hdr, 0);
1376 		msg_set_destnode(hdr, dnode);
1377 		msg_set_destport(hdr, dest->addr.id.ref);
1378 		msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1379 	} else {
1380 		return -EINVAL;
1381 	}
1382 
1383 	/* Block or return if destination link is congested */
1384 	rc = tipc_wait_for_cond(sock, &timeout,
1385 				!tipc_dest_find(clinks, dnode, 0));
1386 	if (unlikely(rc))
1387 		return rc;
1388 
1389 	skb_queue_head_init(&pkts);
1390 	mtu = tipc_node_get_mtu(net, dnode, tsk->portid);
1391 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1392 	if (unlikely(rc != dlen))
1393 		return rc;
1394 	if (unlikely(syn && !tipc_msg_skb_clone(&pkts, &sk->sk_write_queue)))
1395 		return -ENOMEM;
1396 
1397 	trace_tipc_sk_sendmsg(sk, skb_peek(&pkts), TIPC_DUMP_SK_SNDQ, " ");
1398 	rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
1399 	if (unlikely(rc == -ELINKCONG)) {
1400 		tipc_dest_push(clinks, dnode, 0);
1401 		tsk->cong_link_cnt++;
1402 		rc = 0;
1403 	}
1404 
1405 	if (unlikely(syn && !rc))
1406 		tipc_set_sk_state(sk, TIPC_CONNECTING);
1407 
1408 	return rc ? rc : dlen;
1409 }
1410 
1411 /**
1412  * tipc_sendstream - send stream-oriented data
1413  * @sock: socket structure
1414  * @m: data to send
1415  * @dsz: total length of data to be transmitted
1416  *
1417  * Used for SOCK_STREAM data.
1418  *
1419  * Returns the number of bytes sent on success (or partial success),
1420  * or errno if no data sent
1421  */
1422 static int tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz)
1423 {
1424 	struct sock *sk = sock->sk;
1425 	int ret;
1426 
1427 	lock_sock(sk);
1428 	ret = __tipc_sendstream(sock, m, dsz);
1429 	release_sock(sk);
1430 
1431 	return ret;
1432 }
1433 
1434 static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dlen)
1435 {
1436 	struct sock *sk = sock->sk;
1437 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1438 	long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1439 	struct tipc_sock *tsk = tipc_sk(sk);
1440 	struct tipc_msg *hdr = &tsk->phdr;
1441 	struct net *net = sock_net(sk);
1442 	struct sk_buff_head pkts;
1443 	u32 dnode = tsk_peer_node(tsk);
1444 	int send, sent = 0;
1445 	int rc = 0;
1446 
1447 	skb_queue_head_init(&pkts);
1448 
1449 	if (unlikely(dlen > INT_MAX))
1450 		return -EMSGSIZE;
1451 
1452 	/* Handle implicit connection setup */
1453 	if (unlikely(dest)) {
1454 		rc = __tipc_sendmsg(sock, m, dlen);
1455 		if (dlen && dlen == rc) {
1456 			tsk->peer_caps = tipc_node_get_capabilities(net, dnode);
1457 			tsk->snt_unacked = tsk_inc(tsk, dlen + msg_hdr_sz(hdr));
1458 		}
1459 		return rc;
1460 	}
1461 
1462 	do {
1463 		rc = tipc_wait_for_cond(sock, &timeout,
1464 					(!tsk->cong_link_cnt &&
1465 					 !tsk_conn_cong(tsk) &&
1466 					 tipc_sk_connected(sk)));
1467 		if (unlikely(rc))
1468 			break;
1469 
1470 		send = min_t(size_t, dlen - sent, TIPC_MAX_USER_MSG_SIZE);
1471 		rc = tipc_msg_build(hdr, m, sent, send, tsk->max_pkt, &pkts);
1472 		if (unlikely(rc != send))
1473 			break;
1474 
1475 		trace_tipc_sk_sendstream(sk, skb_peek(&pkts),
1476 					 TIPC_DUMP_SK_SNDQ, " ");
1477 		rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
1478 		if (unlikely(rc == -ELINKCONG)) {
1479 			tsk->cong_link_cnt = 1;
1480 			rc = 0;
1481 		}
1482 		if (likely(!rc)) {
1483 			tsk->snt_unacked += tsk_inc(tsk, send + MIN_H_SIZE);
1484 			sent += send;
1485 		}
1486 	} while (sent < dlen && !rc);
1487 
1488 	return sent ? sent : rc;
1489 }
1490 
1491 /**
1492  * tipc_send_packet - send a connection-oriented message
1493  * @sock: socket structure
1494  * @m: message to send
1495  * @dsz: length of data to be transmitted
1496  *
1497  * Used for SOCK_SEQPACKET messages.
1498  *
1499  * Returns the number of bytes sent on success, or errno otherwise
1500  */
1501 static int tipc_send_packet(struct socket *sock, struct msghdr *m, size_t dsz)
1502 {
1503 	if (dsz > TIPC_MAX_USER_MSG_SIZE)
1504 		return -EMSGSIZE;
1505 
1506 	return tipc_sendstream(sock, m, dsz);
1507 }
1508 
1509 /* tipc_sk_finish_conn - complete the setup of a connection
1510  */
1511 static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port,
1512 				u32 peer_node)
1513 {
1514 	struct sock *sk = &tsk->sk;
1515 	struct net *net = sock_net(sk);
1516 	struct tipc_msg *msg = &tsk->phdr;
1517 
1518 	msg_set_syn(msg, 0);
1519 	msg_set_destnode(msg, peer_node);
1520 	msg_set_destport(msg, peer_port);
1521 	msg_set_type(msg, TIPC_CONN_MSG);
1522 	msg_set_lookup_scope(msg, 0);
1523 	msg_set_hdr_sz(msg, SHORT_H_SIZE);
1524 
1525 	sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
1526 	tipc_set_sk_state(sk, TIPC_ESTABLISHED);
1527 	tipc_node_add_conn(net, peer_node, tsk->portid, peer_port);
1528 	tsk->max_pkt = tipc_node_get_mtu(net, peer_node, tsk->portid);
1529 	tsk->peer_caps = tipc_node_get_capabilities(net, peer_node);
1530 	__skb_queue_purge(&sk->sk_write_queue);
1531 	if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1532 		return;
1533 
1534 	/* Fall back to message based flow control */
1535 	tsk->rcv_win = FLOWCTL_MSG_WIN;
1536 	tsk->snd_win = FLOWCTL_MSG_WIN;
1537 }
1538 
1539 /**
1540  * tipc_sk_set_orig_addr - capture sender's address for received message
1541  * @m: descriptor for message info
1542  * @hdr: received message header
1543  *
1544  * Note: Address is not captured if not requested by receiver.
1545  */
1546 static void tipc_sk_set_orig_addr(struct msghdr *m, struct sk_buff *skb)
1547 {
1548 	DECLARE_SOCKADDR(struct sockaddr_pair *, srcaddr, m->msg_name);
1549 	struct tipc_msg *hdr = buf_msg(skb);
1550 
1551 	if (!srcaddr)
1552 		return;
1553 
1554 	srcaddr->sock.family = AF_TIPC;
1555 	srcaddr->sock.addrtype = TIPC_ADDR_ID;
1556 	srcaddr->sock.scope = 0;
1557 	srcaddr->sock.addr.id.ref = msg_origport(hdr);
1558 	srcaddr->sock.addr.id.node = msg_orignode(hdr);
1559 	srcaddr->sock.addr.name.domain = 0;
1560 	m->msg_namelen = sizeof(struct sockaddr_tipc);
1561 
1562 	if (!msg_in_group(hdr))
1563 		return;
1564 
1565 	/* Group message users may also want to know sending member's id */
1566 	srcaddr->member.family = AF_TIPC;
1567 	srcaddr->member.addrtype = TIPC_ADDR_NAME;
1568 	srcaddr->member.scope = 0;
1569 	srcaddr->member.addr.name.name.type = msg_nametype(hdr);
1570 	srcaddr->member.addr.name.name.instance = TIPC_SKB_CB(skb)->orig_member;
1571 	srcaddr->member.addr.name.domain = 0;
1572 	m->msg_namelen = sizeof(*srcaddr);
1573 }
1574 
1575 /**
1576  * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
1577  * @m: descriptor for message info
1578  * @skb: received message buffer
1579  * @tsk: TIPC port associated with message
1580  *
1581  * Note: Ancillary data is not captured if not requested by receiver.
1582  *
1583  * Returns 0 if successful, otherwise errno
1584  */
1585 static int tipc_sk_anc_data_recv(struct msghdr *m, struct sk_buff *skb,
1586 				 struct tipc_sock *tsk)
1587 {
1588 	struct tipc_msg *msg;
1589 	u32 anc_data[3];
1590 	u32 err;
1591 	u32 dest_type;
1592 	int has_name;
1593 	int res;
1594 
1595 	if (likely(m->msg_controllen == 0))
1596 		return 0;
1597 	msg = buf_msg(skb);
1598 
1599 	/* Optionally capture errored message object(s) */
1600 	err = msg ? msg_errcode(msg) : 0;
1601 	if (unlikely(err)) {
1602 		anc_data[0] = err;
1603 		anc_data[1] = msg_data_sz(msg);
1604 		res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
1605 		if (res)
1606 			return res;
1607 		if (anc_data[1]) {
1608 			if (skb_linearize(skb))
1609 				return -ENOMEM;
1610 			msg = buf_msg(skb);
1611 			res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
1612 				       msg_data(msg));
1613 			if (res)
1614 				return res;
1615 		}
1616 	}
1617 
1618 	/* Optionally capture message destination object */
1619 	dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
1620 	switch (dest_type) {
1621 	case TIPC_NAMED_MSG:
1622 		has_name = 1;
1623 		anc_data[0] = msg_nametype(msg);
1624 		anc_data[1] = msg_namelower(msg);
1625 		anc_data[2] = msg_namelower(msg);
1626 		break;
1627 	case TIPC_MCAST_MSG:
1628 		has_name = 1;
1629 		anc_data[0] = msg_nametype(msg);
1630 		anc_data[1] = msg_namelower(msg);
1631 		anc_data[2] = msg_nameupper(msg);
1632 		break;
1633 	case TIPC_CONN_MSG:
1634 		has_name = (tsk->conn_type != 0);
1635 		anc_data[0] = tsk->conn_type;
1636 		anc_data[1] = tsk->conn_instance;
1637 		anc_data[2] = tsk->conn_instance;
1638 		break;
1639 	default:
1640 		has_name = 0;
1641 	}
1642 	if (has_name) {
1643 		res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
1644 		if (res)
1645 			return res;
1646 	}
1647 
1648 	return 0;
1649 }
1650 
1651 static void tipc_sk_send_ack(struct tipc_sock *tsk)
1652 {
1653 	struct sock *sk = &tsk->sk;
1654 	struct net *net = sock_net(sk);
1655 	struct sk_buff *skb = NULL;
1656 	struct tipc_msg *msg;
1657 	u32 peer_port = tsk_peer_port(tsk);
1658 	u32 dnode = tsk_peer_node(tsk);
1659 
1660 	if (!tipc_sk_connected(sk))
1661 		return;
1662 	skb = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0,
1663 			      dnode, tsk_own_node(tsk), peer_port,
1664 			      tsk->portid, TIPC_OK);
1665 	if (!skb)
1666 		return;
1667 	msg = buf_msg(skb);
1668 	msg_set_conn_ack(msg, tsk->rcv_unacked);
1669 	tsk->rcv_unacked = 0;
1670 
1671 	/* Adjust to and advertize the correct window limit */
1672 	if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) {
1673 		tsk->rcv_win = tsk_adv_blocks(tsk->sk.sk_rcvbuf);
1674 		msg_set_adv_win(msg, tsk->rcv_win);
1675 	}
1676 	tipc_node_xmit_skb(net, skb, dnode, msg_link_selector(msg));
1677 }
1678 
1679 static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop)
1680 {
1681 	struct sock *sk = sock->sk;
1682 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
1683 	long timeo = *timeop;
1684 	int err = sock_error(sk);
1685 
1686 	if (err)
1687 		return err;
1688 
1689 	for (;;) {
1690 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1691 			if (sk->sk_shutdown & RCV_SHUTDOWN) {
1692 				err = -ENOTCONN;
1693 				break;
1694 			}
1695 			add_wait_queue(sk_sleep(sk), &wait);
1696 			release_sock(sk);
1697 			timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
1698 			sched_annotate_sleep();
1699 			lock_sock(sk);
1700 			remove_wait_queue(sk_sleep(sk), &wait);
1701 		}
1702 		err = 0;
1703 		if (!skb_queue_empty(&sk->sk_receive_queue))
1704 			break;
1705 		err = -EAGAIN;
1706 		if (!timeo)
1707 			break;
1708 		err = sock_intr_errno(timeo);
1709 		if (signal_pending(current))
1710 			break;
1711 
1712 		err = sock_error(sk);
1713 		if (err)
1714 			break;
1715 	}
1716 	*timeop = timeo;
1717 	return err;
1718 }
1719 
1720 /**
1721  * tipc_recvmsg - receive packet-oriented message
1722  * @m: descriptor for message info
1723  * @buflen: length of user buffer area
1724  * @flags: receive flags
1725  *
1726  * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1727  * If the complete message doesn't fit in user area, truncate it.
1728  *
1729  * Returns size of returned message data, errno otherwise
1730  */
1731 static int tipc_recvmsg(struct socket *sock, struct msghdr *m,
1732 			size_t buflen,	int flags)
1733 {
1734 	struct sock *sk = sock->sk;
1735 	bool connected = !tipc_sk_type_connectionless(sk);
1736 	struct tipc_sock *tsk = tipc_sk(sk);
1737 	int rc, err, hlen, dlen, copy;
1738 	struct sk_buff_head xmitq;
1739 	struct tipc_msg *hdr;
1740 	struct sk_buff *skb;
1741 	bool grp_evt;
1742 	long timeout;
1743 
1744 	/* Catch invalid receive requests */
1745 	if (unlikely(!buflen))
1746 		return -EINVAL;
1747 
1748 	lock_sock(sk);
1749 	if (unlikely(connected && sk->sk_state == TIPC_OPEN)) {
1750 		rc = -ENOTCONN;
1751 		goto exit;
1752 	}
1753 	timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1754 
1755 	/* Step rcv queue to first msg with data or error; wait if necessary */
1756 	do {
1757 		rc = tipc_wait_for_rcvmsg(sock, &timeout);
1758 		if (unlikely(rc))
1759 			goto exit;
1760 		skb = skb_peek(&sk->sk_receive_queue);
1761 		hdr = buf_msg(skb);
1762 		dlen = msg_data_sz(hdr);
1763 		hlen = msg_hdr_sz(hdr);
1764 		err = msg_errcode(hdr);
1765 		grp_evt = msg_is_grp_evt(hdr);
1766 		if (likely(dlen || err))
1767 			break;
1768 		tsk_advance_rx_queue(sk);
1769 	} while (1);
1770 
1771 	/* Collect msg meta data, including error code and rejected data */
1772 	tipc_sk_set_orig_addr(m, skb);
1773 	rc = tipc_sk_anc_data_recv(m, skb, tsk);
1774 	if (unlikely(rc))
1775 		goto exit;
1776 	hdr = buf_msg(skb);
1777 
1778 	/* Capture data if non-error msg, otherwise just set return value */
1779 	if (likely(!err)) {
1780 		copy = min_t(int, dlen, buflen);
1781 		if (unlikely(copy != dlen))
1782 			m->msg_flags |= MSG_TRUNC;
1783 		rc = skb_copy_datagram_msg(skb, hlen, m, copy);
1784 	} else {
1785 		copy = 0;
1786 		rc = 0;
1787 		if (err != TIPC_CONN_SHUTDOWN && connected && !m->msg_control)
1788 			rc = -ECONNRESET;
1789 	}
1790 	if (unlikely(rc))
1791 		goto exit;
1792 
1793 	/* Mark message as group event if applicable */
1794 	if (unlikely(grp_evt)) {
1795 		if (msg_grp_evt(hdr) == TIPC_WITHDRAWN)
1796 			m->msg_flags |= MSG_EOR;
1797 		m->msg_flags |= MSG_OOB;
1798 		copy = 0;
1799 	}
1800 
1801 	/* Caption of data or error code/rejected data was successful */
1802 	if (unlikely(flags & MSG_PEEK))
1803 		goto exit;
1804 
1805 	/* Send group flow control advertisement when applicable */
1806 	if (tsk->group && msg_in_group(hdr) && !grp_evt) {
1807 		skb_queue_head_init(&xmitq);
1808 		tipc_group_update_rcv_win(tsk->group, tsk_blocks(hlen + dlen),
1809 					  msg_orignode(hdr), msg_origport(hdr),
1810 					  &xmitq);
1811 		tipc_node_distr_xmit(sock_net(sk), &xmitq);
1812 	}
1813 
1814 	tsk_advance_rx_queue(sk);
1815 
1816 	if (likely(!connected))
1817 		goto exit;
1818 
1819 	/* Send connection flow control advertisement when applicable */
1820 	tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
1821 	if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
1822 		tipc_sk_send_ack(tsk);
1823 exit:
1824 	release_sock(sk);
1825 	return rc ? rc : copy;
1826 }
1827 
1828 /**
1829  * tipc_recvstream - receive stream-oriented data
1830  * @m: descriptor for message info
1831  * @buflen: total size of user buffer area
1832  * @flags: receive flags
1833  *
1834  * Used for SOCK_STREAM messages only.  If not enough data is available
1835  * will optionally wait for more; never truncates data.
1836  *
1837  * Returns size of returned message data, errno otherwise
1838  */
1839 static int tipc_recvstream(struct socket *sock, struct msghdr *m,
1840 			   size_t buflen, int flags)
1841 {
1842 	struct sock *sk = sock->sk;
1843 	struct tipc_sock *tsk = tipc_sk(sk);
1844 	struct sk_buff *skb;
1845 	struct tipc_msg *hdr;
1846 	struct tipc_skb_cb *skb_cb;
1847 	bool peek = flags & MSG_PEEK;
1848 	int offset, required, copy, copied = 0;
1849 	int hlen, dlen, err, rc;
1850 	long timeout;
1851 
1852 	/* Catch invalid receive attempts */
1853 	if (unlikely(!buflen))
1854 		return -EINVAL;
1855 
1856 	lock_sock(sk);
1857 
1858 	if (unlikely(sk->sk_state == TIPC_OPEN)) {
1859 		rc = -ENOTCONN;
1860 		goto exit;
1861 	}
1862 	required = sock_rcvlowat(sk, flags & MSG_WAITALL, buflen);
1863 	timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1864 
1865 	do {
1866 		/* Look at first msg in receive queue; wait if necessary */
1867 		rc = tipc_wait_for_rcvmsg(sock, &timeout);
1868 		if (unlikely(rc))
1869 			break;
1870 		skb = skb_peek(&sk->sk_receive_queue);
1871 		skb_cb = TIPC_SKB_CB(skb);
1872 		hdr = buf_msg(skb);
1873 		dlen = msg_data_sz(hdr);
1874 		hlen = msg_hdr_sz(hdr);
1875 		err = msg_errcode(hdr);
1876 
1877 		/* Discard any empty non-errored (SYN-) message */
1878 		if (unlikely(!dlen && !err)) {
1879 			tsk_advance_rx_queue(sk);
1880 			continue;
1881 		}
1882 
1883 		/* Collect msg meta data, incl. error code and rejected data */
1884 		if (!copied) {
1885 			tipc_sk_set_orig_addr(m, skb);
1886 			rc = tipc_sk_anc_data_recv(m, skb, tsk);
1887 			if (rc)
1888 				break;
1889 			hdr = buf_msg(skb);
1890 		}
1891 
1892 		/* Copy data if msg ok, otherwise return error/partial data */
1893 		if (likely(!err)) {
1894 			offset = skb_cb->bytes_read;
1895 			copy = min_t(int, dlen - offset, buflen - copied);
1896 			rc = skb_copy_datagram_msg(skb, hlen + offset, m, copy);
1897 			if (unlikely(rc))
1898 				break;
1899 			copied += copy;
1900 			offset += copy;
1901 			if (unlikely(offset < dlen)) {
1902 				if (!peek)
1903 					skb_cb->bytes_read = offset;
1904 				break;
1905 			}
1906 		} else {
1907 			rc = 0;
1908 			if ((err != TIPC_CONN_SHUTDOWN) && !m->msg_control)
1909 				rc = -ECONNRESET;
1910 			if (copied || rc)
1911 				break;
1912 		}
1913 
1914 		if (unlikely(peek))
1915 			break;
1916 
1917 		tsk_advance_rx_queue(sk);
1918 
1919 		/* Send connection flow control advertisement when applicable */
1920 		tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
1921 		if (unlikely(tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE))
1922 			tipc_sk_send_ack(tsk);
1923 
1924 		/* Exit if all requested data or FIN/error received */
1925 		if (copied == buflen || err)
1926 			break;
1927 
1928 	} while (!skb_queue_empty(&sk->sk_receive_queue) || copied < required);
1929 exit:
1930 	release_sock(sk);
1931 	return copied ? copied : rc;
1932 }
1933 
1934 /**
1935  * tipc_write_space - wake up thread if port congestion is released
1936  * @sk: socket
1937  */
1938 static void tipc_write_space(struct sock *sk)
1939 {
1940 	struct socket_wq *wq;
1941 
1942 	rcu_read_lock();
1943 	wq = rcu_dereference(sk->sk_wq);
1944 	if (skwq_has_sleeper(wq))
1945 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLOUT |
1946 						EPOLLWRNORM | EPOLLWRBAND);
1947 	rcu_read_unlock();
1948 }
1949 
1950 /**
1951  * tipc_data_ready - wake up threads to indicate messages have been received
1952  * @sk: socket
1953  * @len: the length of messages
1954  */
1955 static void tipc_data_ready(struct sock *sk)
1956 {
1957 	struct socket_wq *wq;
1958 
1959 	rcu_read_lock();
1960 	wq = rcu_dereference(sk->sk_wq);
1961 	if (skwq_has_sleeper(wq))
1962 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN |
1963 						EPOLLRDNORM | EPOLLRDBAND);
1964 	rcu_read_unlock();
1965 }
1966 
1967 static void tipc_sock_destruct(struct sock *sk)
1968 {
1969 	__skb_queue_purge(&sk->sk_receive_queue);
1970 }
1971 
1972 static void tipc_sk_proto_rcv(struct sock *sk,
1973 			      struct sk_buff_head *inputq,
1974 			      struct sk_buff_head *xmitq)
1975 {
1976 	struct sk_buff *skb = __skb_dequeue(inputq);
1977 	struct tipc_sock *tsk = tipc_sk(sk);
1978 	struct tipc_msg *hdr = buf_msg(skb);
1979 	struct tipc_group *grp = tsk->group;
1980 	bool wakeup = false;
1981 
1982 	switch (msg_user(hdr)) {
1983 	case CONN_MANAGER:
1984 		tipc_sk_conn_proto_rcv(tsk, skb, inputq, xmitq);
1985 		return;
1986 	case SOCK_WAKEUP:
1987 		tipc_dest_del(&tsk->cong_links, msg_orignode(hdr), 0);
1988 		/* coupled with smp_rmb() in tipc_wait_for_cond() */
1989 		smp_wmb();
1990 		tsk->cong_link_cnt--;
1991 		wakeup = true;
1992 		break;
1993 	case GROUP_PROTOCOL:
1994 		tipc_group_proto_rcv(grp, &wakeup, hdr, inputq, xmitq);
1995 		break;
1996 	case TOP_SRV:
1997 		tipc_group_member_evt(tsk->group, &wakeup, &sk->sk_rcvbuf,
1998 				      hdr, inputq, xmitq);
1999 		break;
2000 	default:
2001 		break;
2002 	}
2003 
2004 	if (wakeup)
2005 		sk->sk_write_space(sk);
2006 
2007 	kfree_skb(skb);
2008 }
2009 
2010 /**
2011  * tipc_sk_filter_connect - check incoming message for a connection-based socket
2012  * @tsk: TIPC socket
2013  * @skb: pointer to message buffer.
2014  * Returns true if message should be added to receive queue, false otherwise
2015  */
2016 static bool tipc_sk_filter_connect(struct tipc_sock *tsk, struct sk_buff *skb)
2017 {
2018 	struct sock *sk = &tsk->sk;
2019 	struct net *net = sock_net(sk);
2020 	struct tipc_msg *hdr = buf_msg(skb);
2021 	bool con_msg = msg_connected(hdr);
2022 	u32 pport = tsk_peer_port(tsk);
2023 	u32 pnode = tsk_peer_node(tsk);
2024 	u32 oport = msg_origport(hdr);
2025 	u32 onode = msg_orignode(hdr);
2026 	int err = msg_errcode(hdr);
2027 	unsigned long delay;
2028 
2029 	if (unlikely(msg_mcast(hdr)))
2030 		return false;
2031 
2032 	switch (sk->sk_state) {
2033 	case TIPC_CONNECTING:
2034 		/* Setup ACK */
2035 		if (likely(con_msg)) {
2036 			if (err)
2037 				break;
2038 			tipc_sk_finish_conn(tsk, oport, onode);
2039 			msg_set_importance(&tsk->phdr, msg_importance(hdr));
2040 			/* ACK+ message with data is added to receive queue */
2041 			if (msg_data_sz(hdr))
2042 				return true;
2043 			/* Empty ACK-, - wake up sleeping connect() and drop */
2044 			sk->sk_data_ready(sk);
2045 			msg_set_dest_droppable(hdr, 1);
2046 			return false;
2047 		}
2048 		/* Ignore connectionless message if not from listening socket */
2049 		if (oport != pport || onode != pnode)
2050 			return false;
2051 
2052 		/* Rejected SYN */
2053 		if (err != TIPC_ERR_OVERLOAD)
2054 			break;
2055 
2056 		/* Prepare for new setup attempt if we have a SYN clone */
2057 		if (skb_queue_empty(&sk->sk_write_queue))
2058 			break;
2059 		get_random_bytes(&delay, 2);
2060 		delay %= (tsk->conn_timeout / 4);
2061 		delay = msecs_to_jiffies(delay + 100);
2062 		sk_reset_timer(sk, &sk->sk_timer, jiffies + delay);
2063 		return false;
2064 	case TIPC_OPEN:
2065 	case TIPC_DISCONNECTING:
2066 		return false;
2067 	case TIPC_LISTEN:
2068 		/* Accept only SYN message */
2069 		if (!msg_is_syn(hdr) &&
2070 		    tipc_node_get_capabilities(net, onode) & TIPC_SYN_BIT)
2071 			return false;
2072 		if (!con_msg && !err)
2073 			return true;
2074 		return false;
2075 	case TIPC_ESTABLISHED:
2076 		/* Accept only connection-based messages sent by peer */
2077 		if (likely(con_msg && !err && pport == oport && pnode == onode))
2078 			return true;
2079 		if (!tsk_peer_msg(tsk, hdr))
2080 			return false;
2081 		if (!err)
2082 			return true;
2083 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2084 		tipc_node_remove_conn(net, pnode, tsk->portid);
2085 		sk->sk_state_change(sk);
2086 		return true;
2087 	default:
2088 		pr_err("Unknown sk_state %u\n", sk->sk_state);
2089 	}
2090 	/* Abort connection setup attempt */
2091 	tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2092 	sk->sk_err = ECONNREFUSED;
2093 	sk->sk_state_change(sk);
2094 	return true;
2095 }
2096 
2097 /**
2098  * rcvbuf_limit - get proper overload limit of socket receive queue
2099  * @sk: socket
2100  * @skb: message
2101  *
2102  * For connection oriented messages, irrespective of importance,
2103  * default queue limit is 2 MB.
2104  *
2105  * For connectionless messages, queue limits are based on message
2106  * importance as follows:
2107  *
2108  * TIPC_LOW_IMPORTANCE       (2 MB)
2109  * TIPC_MEDIUM_IMPORTANCE    (4 MB)
2110  * TIPC_HIGH_IMPORTANCE      (8 MB)
2111  * TIPC_CRITICAL_IMPORTANCE  (16 MB)
2112  *
2113  * Returns overload limit according to corresponding message importance
2114  */
2115 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *skb)
2116 {
2117 	struct tipc_sock *tsk = tipc_sk(sk);
2118 	struct tipc_msg *hdr = buf_msg(skb);
2119 
2120 	if (unlikely(msg_in_group(hdr)))
2121 		return sk->sk_rcvbuf;
2122 
2123 	if (unlikely(!msg_connected(hdr)))
2124 		return sk->sk_rcvbuf << msg_importance(hdr);
2125 
2126 	if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
2127 		return sk->sk_rcvbuf;
2128 
2129 	return FLOWCTL_MSG_LIM;
2130 }
2131 
2132 /**
2133  * tipc_sk_filter_rcv - validate incoming message
2134  * @sk: socket
2135  * @skb: pointer to message.
2136  *
2137  * Enqueues message on receive queue if acceptable; optionally handles
2138  * disconnect indication for a connected socket.
2139  *
2140  * Called with socket lock already taken
2141  *
2142  */
2143 static void tipc_sk_filter_rcv(struct sock *sk, struct sk_buff *skb,
2144 			       struct sk_buff_head *xmitq)
2145 {
2146 	bool sk_conn = !tipc_sk_type_connectionless(sk);
2147 	struct tipc_sock *tsk = tipc_sk(sk);
2148 	struct tipc_group *grp = tsk->group;
2149 	struct tipc_msg *hdr = buf_msg(skb);
2150 	struct net *net = sock_net(sk);
2151 	struct sk_buff_head inputq;
2152 	int limit, err = TIPC_OK;
2153 
2154 	trace_tipc_sk_filter_rcv(sk, skb, TIPC_DUMP_ALL, " ");
2155 	TIPC_SKB_CB(skb)->bytes_read = 0;
2156 	__skb_queue_head_init(&inputq);
2157 	__skb_queue_tail(&inputq, skb);
2158 
2159 	if (unlikely(!msg_isdata(hdr)))
2160 		tipc_sk_proto_rcv(sk, &inputq, xmitq);
2161 
2162 	if (unlikely(grp))
2163 		tipc_group_filter_msg(grp, &inputq, xmitq);
2164 
2165 	/* Validate and add to receive buffer if there is space */
2166 	while ((skb = __skb_dequeue(&inputq))) {
2167 		hdr = buf_msg(skb);
2168 		limit = rcvbuf_limit(sk, skb);
2169 		if ((sk_conn && !tipc_sk_filter_connect(tsk, skb)) ||
2170 		    (!sk_conn && msg_connected(hdr)) ||
2171 		    (!grp && msg_in_group(hdr)))
2172 			err = TIPC_ERR_NO_PORT;
2173 		else if (sk_rmem_alloc_get(sk) + skb->truesize >= limit) {
2174 			trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL,
2175 					   "err_overload2!");
2176 			atomic_inc(&sk->sk_drops);
2177 			err = TIPC_ERR_OVERLOAD;
2178 		}
2179 
2180 		if (unlikely(err)) {
2181 			if (tipc_msg_reverse(tipc_own_addr(net), &skb, err)) {
2182 				trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE,
2183 						      "@filter_rcv!");
2184 				__skb_queue_tail(xmitq, skb);
2185 			}
2186 			err = TIPC_OK;
2187 			continue;
2188 		}
2189 		__skb_queue_tail(&sk->sk_receive_queue, skb);
2190 		skb_set_owner_r(skb, sk);
2191 		trace_tipc_sk_overlimit2(sk, skb, TIPC_DUMP_ALL,
2192 					 "rcvq >90% allocated!");
2193 		sk->sk_data_ready(sk);
2194 	}
2195 }
2196 
2197 /**
2198  * tipc_sk_backlog_rcv - handle incoming message from backlog queue
2199  * @sk: socket
2200  * @skb: message
2201  *
2202  * Caller must hold socket lock
2203  */
2204 static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb)
2205 {
2206 	unsigned int before = sk_rmem_alloc_get(sk);
2207 	struct sk_buff_head xmitq;
2208 	unsigned int added;
2209 
2210 	__skb_queue_head_init(&xmitq);
2211 
2212 	tipc_sk_filter_rcv(sk, skb, &xmitq);
2213 	added = sk_rmem_alloc_get(sk) - before;
2214 	atomic_add(added, &tipc_sk(sk)->dupl_rcvcnt);
2215 
2216 	/* Send pending response/rejected messages, if any */
2217 	tipc_node_distr_xmit(sock_net(sk), &xmitq);
2218 	return 0;
2219 }
2220 
2221 /**
2222  * tipc_sk_enqueue - extract all buffers with destination 'dport' from
2223  *                   inputq and try adding them to socket or backlog queue
2224  * @inputq: list of incoming buffers with potentially different destinations
2225  * @sk: socket where the buffers should be enqueued
2226  * @dport: port number for the socket
2227  *
2228  * Caller must hold socket lock
2229  */
2230 static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk,
2231 			    u32 dport, struct sk_buff_head *xmitq)
2232 {
2233 	unsigned long time_limit = jiffies + 2;
2234 	struct sk_buff *skb;
2235 	unsigned int lim;
2236 	atomic_t *dcnt;
2237 	u32 onode;
2238 
2239 	while (skb_queue_len(inputq)) {
2240 		if (unlikely(time_after_eq(jiffies, time_limit)))
2241 			return;
2242 
2243 		skb = tipc_skb_dequeue(inputq, dport);
2244 		if (unlikely(!skb))
2245 			return;
2246 
2247 		/* Add message directly to receive queue if possible */
2248 		if (!sock_owned_by_user(sk)) {
2249 			tipc_sk_filter_rcv(sk, skb, xmitq);
2250 			continue;
2251 		}
2252 
2253 		/* Try backlog, compensating for double-counted bytes */
2254 		dcnt = &tipc_sk(sk)->dupl_rcvcnt;
2255 		if (!sk->sk_backlog.len)
2256 			atomic_set(dcnt, 0);
2257 		lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
2258 		if (likely(!sk_add_backlog(sk, skb, lim))) {
2259 			trace_tipc_sk_overlimit1(sk, skb, TIPC_DUMP_ALL,
2260 						 "bklg & rcvq >90% allocated!");
2261 			continue;
2262 		}
2263 
2264 		trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL, "err_overload!");
2265 		/* Overload => reject message back to sender */
2266 		onode = tipc_own_addr(sock_net(sk));
2267 		atomic_inc(&sk->sk_drops);
2268 		if (tipc_msg_reverse(onode, &skb, TIPC_ERR_OVERLOAD)) {
2269 			trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_ALL,
2270 					      "@sk_enqueue!");
2271 			__skb_queue_tail(xmitq, skb);
2272 		}
2273 		break;
2274 	}
2275 }
2276 
2277 /**
2278  * tipc_sk_rcv - handle a chain of incoming buffers
2279  * @inputq: buffer list containing the buffers
2280  * Consumes all buffers in list until inputq is empty
2281  * Note: may be called in multiple threads referring to the same queue
2282  */
2283 void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq)
2284 {
2285 	struct sk_buff_head xmitq;
2286 	u32 dnode, dport = 0;
2287 	int err;
2288 	struct tipc_sock *tsk;
2289 	struct sock *sk;
2290 	struct sk_buff *skb;
2291 
2292 	__skb_queue_head_init(&xmitq);
2293 	while (skb_queue_len(inputq)) {
2294 		dport = tipc_skb_peek_port(inputq, dport);
2295 		tsk = tipc_sk_lookup(net, dport);
2296 
2297 		if (likely(tsk)) {
2298 			sk = &tsk->sk;
2299 			if (likely(spin_trylock_bh(&sk->sk_lock.slock))) {
2300 				tipc_sk_enqueue(inputq, sk, dport, &xmitq);
2301 				spin_unlock_bh(&sk->sk_lock.slock);
2302 			}
2303 			/* Send pending response/rejected messages, if any */
2304 			tipc_node_distr_xmit(sock_net(sk), &xmitq);
2305 			sock_put(sk);
2306 			continue;
2307 		}
2308 		/* No destination socket => dequeue skb if still there */
2309 		skb = tipc_skb_dequeue(inputq, dport);
2310 		if (!skb)
2311 			return;
2312 
2313 		/* Try secondary lookup if unresolved named message */
2314 		err = TIPC_ERR_NO_PORT;
2315 		if (tipc_msg_lookup_dest(net, skb, &err))
2316 			goto xmit;
2317 
2318 		/* Prepare for message rejection */
2319 		if (!tipc_msg_reverse(tipc_own_addr(net), &skb, err))
2320 			continue;
2321 
2322 		trace_tipc_sk_rej_msg(NULL, skb, TIPC_DUMP_NONE, "@sk_rcv!");
2323 xmit:
2324 		dnode = msg_destnode(buf_msg(skb));
2325 		tipc_node_xmit_skb(net, skb, dnode, dport);
2326 	}
2327 }
2328 
2329 static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
2330 {
2331 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
2332 	struct sock *sk = sock->sk;
2333 	int done;
2334 
2335 	do {
2336 		int err = sock_error(sk);
2337 		if (err)
2338 			return err;
2339 		if (!*timeo_p)
2340 			return -ETIMEDOUT;
2341 		if (signal_pending(current))
2342 			return sock_intr_errno(*timeo_p);
2343 
2344 		add_wait_queue(sk_sleep(sk), &wait);
2345 		done = sk_wait_event(sk, timeo_p,
2346 				     sk->sk_state != TIPC_CONNECTING, &wait);
2347 		remove_wait_queue(sk_sleep(sk), &wait);
2348 	} while (!done);
2349 	return 0;
2350 }
2351 
2352 /**
2353  * tipc_connect - establish a connection to another TIPC port
2354  * @sock: socket structure
2355  * @dest: socket address for destination port
2356  * @destlen: size of socket address data structure
2357  * @flags: file-related flags associated with socket
2358  *
2359  * Returns 0 on success, errno otherwise
2360  */
2361 static int tipc_connect(struct socket *sock, struct sockaddr *dest,
2362 			int destlen, int flags)
2363 {
2364 	struct sock *sk = sock->sk;
2365 	struct tipc_sock *tsk = tipc_sk(sk);
2366 	struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
2367 	struct msghdr m = {NULL,};
2368 	long timeout = (flags & O_NONBLOCK) ? 0 : tsk->conn_timeout;
2369 	int previous;
2370 	int res = 0;
2371 
2372 	if (destlen != sizeof(struct sockaddr_tipc))
2373 		return -EINVAL;
2374 
2375 	lock_sock(sk);
2376 
2377 	if (tsk->group) {
2378 		res = -EINVAL;
2379 		goto exit;
2380 	}
2381 
2382 	if (dst->family == AF_UNSPEC) {
2383 		memset(&tsk->peer, 0, sizeof(struct sockaddr_tipc));
2384 		if (!tipc_sk_type_connectionless(sk))
2385 			res = -EINVAL;
2386 		goto exit;
2387 	} else if (dst->family != AF_TIPC) {
2388 		res = -EINVAL;
2389 	}
2390 	if (dst->addrtype != TIPC_ADDR_ID && dst->addrtype != TIPC_ADDR_NAME)
2391 		res = -EINVAL;
2392 	if (res)
2393 		goto exit;
2394 
2395 	/* DGRAM/RDM connect(), just save the destaddr */
2396 	if (tipc_sk_type_connectionless(sk)) {
2397 		memcpy(&tsk->peer, dest, destlen);
2398 		goto exit;
2399 	}
2400 
2401 	previous = sk->sk_state;
2402 
2403 	switch (sk->sk_state) {
2404 	case TIPC_OPEN:
2405 		/* Send a 'SYN-' to destination */
2406 		m.msg_name = dest;
2407 		m.msg_namelen = destlen;
2408 
2409 		/* If connect is in non-blocking case, set MSG_DONTWAIT to
2410 		 * indicate send_msg() is never blocked.
2411 		 */
2412 		if (!timeout)
2413 			m.msg_flags = MSG_DONTWAIT;
2414 
2415 		res = __tipc_sendmsg(sock, &m, 0);
2416 		if ((res < 0) && (res != -EWOULDBLOCK))
2417 			goto exit;
2418 
2419 		/* Just entered TIPC_CONNECTING state; the only
2420 		 * difference is that return value in non-blocking
2421 		 * case is EINPROGRESS, rather than EALREADY.
2422 		 */
2423 		res = -EINPROGRESS;
2424 		/* fall through */
2425 	case TIPC_CONNECTING:
2426 		if (!timeout) {
2427 			if (previous == TIPC_CONNECTING)
2428 				res = -EALREADY;
2429 			goto exit;
2430 		}
2431 		timeout = msecs_to_jiffies(timeout);
2432 		/* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
2433 		res = tipc_wait_for_connect(sock, &timeout);
2434 		break;
2435 	case TIPC_ESTABLISHED:
2436 		res = -EISCONN;
2437 		break;
2438 	default:
2439 		res = -EINVAL;
2440 	}
2441 
2442 exit:
2443 	release_sock(sk);
2444 	return res;
2445 }
2446 
2447 /**
2448  * tipc_listen - allow socket to listen for incoming connections
2449  * @sock: socket structure
2450  * @len: (unused)
2451  *
2452  * Returns 0 on success, errno otherwise
2453  */
2454 static int tipc_listen(struct socket *sock, int len)
2455 {
2456 	struct sock *sk = sock->sk;
2457 	int res;
2458 
2459 	lock_sock(sk);
2460 	res = tipc_set_sk_state(sk, TIPC_LISTEN);
2461 	release_sock(sk);
2462 
2463 	return res;
2464 }
2465 
2466 static int tipc_wait_for_accept(struct socket *sock, long timeo)
2467 {
2468 	struct sock *sk = sock->sk;
2469 	DEFINE_WAIT(wait);
2470 	int err;
2471 
2472 	/* True wake-one mechanism for incoming connections: only
2473 	 * one process gets woken up, not the 'whole herd'.
2474 	 * Since we do not 'race & poll' for established sockets
2475 	 * anymore, the common case will execute the loop only once.
2476 	*/
2477 	for (;;) {
2478 		prepare_to_wait_exclusive(sk_sleep(sk), &wait,
2479 					  TASK_INTERRUPTIBLE);
2480 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
2481 			release_sock(sk);
2482 			timeo = schedule_timeout(timeo);
2483 			lock_sock(sk);
2484 		}
2485 		err = 0;
2486 		if (!skb_queue_empty(&sk->sk_receive_queue))
2487 			break;
2488 		err = -EAGAIN;
2489 		if (!timeo)
2490 			break;
2491 		err = sock_intr_errno(timeo);
2492 		if (signal_pending(current))
2493 			break;
2494 	}
2495 	finish_wait(sk_sleep(sk), &wait);
2496 	return err;
2497 }
2498 
2499 /**
2500  * tipc_accept - wait for connection request
2501  * @sock: listening socket
2502  * @newsock: new socket that is to be connected
2503  * @flags: file-related flags associated with socket
2504  *
2505  * Returns 0 on success, errno otherwise
2506  */
2507 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags,
2508 		       bool kern)
2509 {
2510 	struct sock *new_sk, *sk = sock->sk;
2511 	struct sk_buff *buf;
2512 	struct tipc_sock *new_tsock;
2513 	struct tipc_msg *msg;
2514 	long timeo;
2515 	int res;
2516 
2517 	lock_sock(sk);
2518 
2519 	if (sk->sk_state != TIPC_LISTEN) {
2520 		res = -EINVAL;
2521 		goto exit;
2522 	}
2523 	timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
2524 	res = tipc_wait_for_accept(sock, timeo);
2525 	if (res)
2526 		goto exit;
2527 
2528 	buf = skb_peek(&sk->sk_receive_queue);
2529 
2530 	res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, kern);
2531 	if (res)
2532 		goto exit;
2533 	security_sk_clone(sock->sk, new_sock->sk);
2534 
2535 	new_sk = new_sock->sk;
2536 	new_tsock = tipc_sk(new_sk);
2537 	msg = buf_msg(buf);
2538 
2539 	/* we lock on new_sk; but lockdep sees the lock on sk */
2540 	lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
2541 
2542 	/*
2543 	 * Reject any stray messages received by new socket
2544 	 * before the socket lock was taken (very, very unlikely)
2545 	 */
2546 	tsk_rej_rx_queue(new_sk);
2547 
2548 	/* Connect new socket to it's peer */
2549 	tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg));
2550 
2551 	tsk_set_importance(new_tsock, msg_importance(msg));
2552 	if (msg_named(msg)) {
2553 		new_tsock->conn_type = msg_nametype(msg);
2554 		new_tsock->conn_instance = msg_nameinst(msg);
2555 	}
2556 
2557 	/*
2558 	 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2559 	 * Respond to 'SYN+' by queuing it on new socket.
2560 	 */
2561 	if (!msg_data_sz(msg)) {
2562 		struct msghdr m = {NULL,};
2563 
2564 		tsk_advance_rx_queue(sk);
2565 		__tipc_sendstream(new_sock, &m, 0);
2566 	} else {
2567 		__skb_dequeue(&sk->sk_receive_queue);
2568 		__skb_queue_head(&new_sk->sk_receive_queue, buf);
2569 		skb_set_owner_r(buf, new_sk);
2570 	}
2571 	release_sock(new_sk);
2572 exit:
2573 	release_sock(sk);
2574 	return res;
2575 }
2576 
2577 /**
2578  * tipc_shutdown - shutdown socket connection
2579  * @sock: socket structure
2580  * @how: direction to close (must be SHUT_RDWR)
2581  *
2582  * Terminates connection (if necessary), then purges socket's receive queue.
2583  *
2584  * Returns 0 on success, errno otherwise
2585  */
2586 static int tipc_shutdown(struct socket *sock, int how)
2587 {
2588 	struct sock *sk = sock->sk;
2589 	int res;
2590 
2591 	if (how != SHUT_RDWR)
2592 		return -EINVAL;
2593 
2594 	lock_sock(sk);
2595 
2596 	trace_tipc_sk_shutdown(sk, NULL, TIPC_DUMP_ALL, " ");
2597 	__tipc_shutdown(sock, TIPC_CONN_SHUTDOWN);
2598 	sk->sk_shutdown = SEND_SHUTDOWN;
2599 
2600 	if (sk->sk_state == TIPC_DISCONNECTING) {
2601 		/* Discard any unreceived messages */
2602 		__skb_queue_purge(&sk->sk_receive_queue);
2603 
2604 		/* Wake up anyone sleeping in poll */
2605 		sk->sk_state_change(sk);
2606 		res = 0;
2607 	} else {
2608 		res = -ENOTCONN;
2609 	}
2610 
2611 	release_sock(sk);
2612 	return res;
2613 }
2614 
2615 static void tipc_sk_check_probing_state(struct sock *sk,
2616 					struct sk_buff_head *list)
2617 {
2618 	struct tipc_sock *tsk = tipc_sk(sk);
2619 	u32 pnode = tsk_peer_node(tsk);
2620 	u32 pport = tsk_peer_port(tsk);
2621 	u32 self = tsk_own_node(tsk);
2622 	u32 oport = tsk->portid;
2623 	struct sk_buff *skb;
2624 
2625 	if (tsk->probe_unacked) {
2626 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2627 		sk->sk_err = ECONNABORTED;
2628 		tipc_node_remove_conn(sock_net(sk), pnode, pport);
2629 		sk->sk_state_change(sk);
2630 		return;
2631 	}
2632 	/* Prepare new probe */
2633 	skb = tipc_msg_create(CONN_MANAGER, CONN_PROBE, INT_H_SIZE, 0,
2634 			      pnode, self, pport, oport, TIPC_OK);
2635 	if (skb)
2636 		__skb_queue_tail(list, skb);
2637 	tsk->probe_unacked = true;
2638 	sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
2639 }
2640 
2641 static void tipc_sk_retry_connect(struct sock *sk, struct sk_buff_head *list)
2642 {
2643 	struct tipc_sock *tsk = tipc_sk(sk);
2644 
2645 	/* Try again later if dest link is congested */
2646 	if (tsk->cong_link_cnt) {
2647 		sk_reset_timer(sk, &sk->sk_timer, msecs_to_jiffies(100));
2648 		return;
2649 	}
2650 	/* Prepare SYN for retransmit */
2651 	tipc_msg_skb_clone(&sk->sk_write_queue, list);
2652 }
2653 
2654 static void tipc_sk_timeout(struct timer_list *t)
2655 {
2656 	struct sock *sk = from_timer(sk, t, sk_timer);
2657 	struct tipc_sock *tsk = tipc_sk(sk);
2658 	u32 pnode = tsk_peer_node(tsk);
2659 	struct sk_buff_head list;
2660 	int rc = 0;
2661 
2662 	skb_queue_head_init(&list);
2663 	bh_lock_sock(sk);
2664 
2665 	/* Try again later if socket is busy */
2666 	if (sock_owned_by_user(sk)) {
2667 		sk_reset_timer(sk, &sk->sk_timer, jiffies + HZ / 20);
2668 		bh_unlock_sock(sk);
2669 		return;
2670 	}
2671 
2672 	if (sk->sk_state == TIPC_ESTABLISHED)
2673 		tipc_sk_check_probing_state(sk, &list);
2674 	else if (sk->sk_state == TIPC_CONNECTING)
2675 		tipc_sk_retry_connect(sk, &list);
2676 
2677 	bh_unlock_sock(sk);
2678 
2679 	if (!skb_queue_empty(&list))
2680 		rc = tipc_node_xmit(sock_net(sk), &list, pnode, tsk->portid);
2681 
2682 	/* SYN messages may cause link congestion */
2683 	if (rc == -ELINKCONG) {
2684 		tipc_dest_push(&tsk->cong_links, pnode, 0);
2685 		tsk->cong_link_cnt = 1;
2686 	}
2687 	sock_put(sk);
2688 }
2689 
2690 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
2691 			   struct tipc_name_seq const *seq)
2692 {
2693 	struct sock *sk = &tsk->sk;
2694 	struct net *net = sock_net(sk);
2695 	struct publication *publ;
2696 	u32 key;
2697 
2698 	if (scope != TIPC_NODE_SCOPE)
2699 		scope = TIPC_CLUSTER_SCOPE;
2700 
2701 	if (tipc_sk_connected(sk))
2702 		return -EINVAL;
2703 	key = tsk->portid + tsk->pub_count + 1;
2704 	if (key == tsk->portid)
2705 		return -EADDRINUSE;
2706 
2707 	publ = tipc_nametbl_publish(net, seq->type, seq->lower, seq->upper,
2708 				    scope, tsk->portid, key);
2709 	if (unlikely(!publ))
2710 		return -EINVAL;
2711 
2712 	list_add(&publ->binding_sock, &tsk->publications);
2713 	tsk->pub_count++;
2714 	tsk->published = 1;
2715 	return 0;
2716 }
2717 
2718 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
2719 			    struct tipc_name_seq const *seq)
2720 {
2721 	struct net *net = sock_net(&tsk->sk);
2722 	struct publication *publ;
2723 	struct publication *safe;
2724 	int rc = -EINVAL;
2725 
2726 	if (scope != TIPC_NODE_SCOPE)
2727 		scope = TIPC_CLUSTER_SCOPE;
2728 
2729 	list_for_each_entry_safe(publ, safe, &tsk->publications, binding_sock) {
2730 		if (seq) {
2731 			if (publ->scope != scope)
2732 				continue;
2733 			if (publ->type != seq->type)
2734 				continue;
2735 			if (publ->lower != seq->lower)
2736 				continue;
2737 			if (publ->upper != seq->upper)
2738 				break;
2739 			tipc_nametbl_withdraw(net, publ->type, publ->lower,
2740 					      publ->upper, publ->key);
2741 			rc = 0;
2742 			break;
2743 		}
2744 		tipc_nametbl_withdraw(net, publ->type, publ->lower,
2745 				      publ->upper, publ->key);
2746 		rc = 0;
2747 	}
2748 	if (list_empty(&tsk->publications))
2749 		tsk->published = 0;
2750 	return rc;
2751 }
2752 
2753 /* tipc_sk_reinit: set non-zero address in all existing sockets
2754  *                 when we go from standalone to network mode.
2755  */
2756 void tipc_sk_reinit(struct net *net)
2757 {
2758 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2759 	struct rhashtable_iter iter;
2760 	struct tipc_sock *tsk;
2761 	struct tipc_msg *msg;
2762 
2763 	rhashtable_walk_enter(&tn->sk_rht, &iter);
2764 
2765 	do {
2766 		rhashtable_walk_start(&iter);
2767 
2768 		while ((tsk = rhashtable_walk_next(&iter)) && !IS_ERR(tsk)) {
2769 			sock_hold(&tsk->sk);
2770 			rhashtable_walk_stop(&iter);
2771 			lock_sock(&tsk->sk);
2772 			msg = &tsk->phdr;
2773 			msg_set_prevnode(msg, tipc_own_addr(net));
2774 			msg_set_orignode(msg, tipc_own_addr(net));
2775 			release_sock(&tsk->sk);
2776 			rhashtable_walk_start(&iter);
2777 			sock_put(&tsk->sk);
2778 		}
2779 
2780 		rhashtable_walk_stop(&iter);
2781 	} while (tsk == ERR_PTR(-EAGAIN));
2782 
2783 	rhashtable_walk_exit(&iter);
2784 }
2785 
2786 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid)
2787 {
2788 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2789 	struct tipc_sock *tsk;
2790 
2791 	rcu_read_lock();
2792 	tsk = rhashtable_lookup_fast(&tn->sk_rht, &portid, tsk_rht_params);
2793 	if (tsk)
2794 		sock_hold(&tsk->sk);
2795 	rcu_read_unlock();
2796 
2797 	return tsk;
2798 }
2799 
2800 static int tipc_sk_insert(struct tipc_sock *tsk)
2801 {
2802 	struct sock *sk = &tsk->sk;
2803 	struct net *net = sock_net(sk);
2804 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2805 	u32 remaining = (TIPC_MAX_PORT - TIPC_MIN_PORT) + 1;
2806 	u32 portid = prandom_u32() % remaining + TIPC_MIN_PORT;
2807 
2808 	while (remaining--) {
2809 		portid++;
2810 		if ((portid < TIPC_MIN_PORT) || (portid > TIPC_MAX_PORT))
2811 			portid = TIPC_MIN_PORT;
2812 		tsk->portid = portid;
2813 		sock_hold(&tsk->sk);
2814 		if (!rhashtable_lookup_insert_fast(&tn->sk_rht, &tsk->node,
2815 						   tsk_rht_params))
2816 			return 0;
2817 		sock_put(&tsk->sk);
2818 	}
2819 
2820 	return -1;
2821 }
2822 
2823 static void tipc_sk_remove(struct tipc_sock *tsk)
2824 {
2825 	struct sock *sk = &tsk->sk;
2826 	struct tipc_net *tn = net_generic(sock_net(sk), tipc_net_id);
2827 
2828 	if (!rhashtable_remove_fast(&tn->sk_rht, &tsk->node, tsk_rht_params)) {
2829 		WARN_ON(refcount_read(&sk->sk_refcnt) == 1);
2830 		__sock_put(sk);
2831 	}
2832 }
2833 
2834 static const struct rhashtable_params tsk_rht_params = {
2835 	.nelem_hint = 192,
2836 	.head_offset = offsetof(struct tipc_sock, node),
2837 	.key_offset = offsetof(struct tipc_sock, portid),
2838 	.key_len = sizeof(u32), /* portid */
2839 	.max_size = 1048576,
2840 	.min_size = 256,
2841 	.automatic_shrinking = true,
2842 };
2843 
2844 int tipc_sk_rht_init(struct net *net)
2845 {
2846 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2847 
2848 	return rhashtable_init(&tn->sk_rht, &tsk_rht_params);
2849 }
2850 
2851 void tipc_sk_rht_destroy(struct net *net)
2852 {
2853 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2854 
2855 	/* Wait for socket readers to complete */
2856 	synchronize_net();
2857 
2858 	rhashtable_destroy(&tn->sk_rht);
2859 }
2860 
2861 static int tipc_sk_join(struct tipc_sock *tsk, struct tipc_group_req *mreq)
2862 {
2863 	struct net *net = sock_net(&tsk->sk);
2864 	struct tipc_group *grp = tsk->group;
2865 	struct tipc_msg *hdr = &tsk->phdr;
2866 	struct tipc_name_seq seq;
2867 	int rc;
2868 
2869 	if (mreq->type < TIPC_RESERVED_TYPES)
2870 		return -EACCES;
2871 	if (mreq->scope > TIPC_NODE_SCOPE)
2872 		return -EINVAL;
2873 	if (grp)
2874 		return -EACCES;
2875 	grp = tipc_group_create(net, tsk->portid, mreq, &tsk->group_is_open);
2876 	if (!grp)
2877 		return -ENOMEM;
2878 	tsk->group = grp;
2879 	msg_set_lookup_scope(hdr, mreq->scope);
2880 	msg_set_nametype(hdr, mreq->type);
2881 	msg_set_dest_droppable(hdr, true);
2882 	seq.type = mreq->type;
2883 	seq.lower = mreq->instance;
2884 	seq.upper = seq.lower;
2885 	tipc_nametbl_build_group(net, grp, mreq->type, mreq->scope);
2886 	rc = tipc_sk_publish(tsk, mreq->scope, &seq);
2887 	if (rc) {
2888 		tipc_group_delete(net, grp);
2889 		tsk->group = NULL;
2890 		return rc;
2891 	}
2892 	/* Eliminate any risk that a broadcast overtakes sent JOINs */
2893 	tsk->mc_method.rcast = true;
2894 	tsk->mc_method.mandatory = true;
2895 	tipc_group_join(net, grp, &tsk->sk.sk_rcvbuf);
2896 	return rc;
2897 }
2898 
2899 static int tipc_sk_leave(struct tipc_sock *tsk)
2900 {
2901 	struct net *net = sock_net(&tsk->sk);
2902 	struct tipc_group *grp = tsk->group;
2903 	struct tipc_name_seq seq;
2904 	int scope;
2905 
2906 	if (!grp)
2907 		return -EINVAL;
2908 	tipc_group_self(grp, &seq, &scope);
2909 	tipc_group_delete(net, grp);
2910 	tsk->group = NULL;
2911 	tipc_sk_withdraw(tsk, scope, &seq);
2912 	return 0;
2913 }
2914 
2915 /**
2916  * tipc_setsockopt - set socket option
2917  * @sock: socket structure
2918  * @lvl: option level
2919  * @opt: option identifier
2920  * @ov: pointer to new option value
2921  * @ol: length of option value
2922  *
2923  * For stream sockets only, accepts and ignores all IPPROTO_TCP options
2924  * (to ease compatibility).
2925  *
2926  * Returns 0 on success, errno otherwise
2927  */
2928 static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
2929 			   char __user *ov, unsigned int ol)
2930 {
2931 	struct sock *sk = sock->sk;
2932 	struct tipc_sock *tsk = tipc_sk(sk);
2933 	struct tipc_group_req mreq;
2934 	u32 value = 0;
2935 	int res = 0;
2936 
2937 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
2938 		return 0;
2939 	if (lvl != SOL_TIPC)
2940 		return -ENOPROTOOPT;
2941 
2942 	switch (opt) {
2943 	case TIPC_IMPORTANCE:
2944 	case TIPC_SRC_DROPPABLE:
2945 	case TIPC_DEST_DROPPABLE:
2946 	case TIPC_CONN_TIMEOUT:
2947 		if (ol < sizeof(value))
2948 			return -EINVAL;
2949 		if (get_user(value, (u32 __user *)ov))
2950 			return -EFAULT;
2951 		break;
2952 	case TIPC_GROUP_JOIN:
2953 		if (ol < sizeof(mreq))
2954 			return -EINVAL;
2955 		if (copy_from_user(&mreq, ov, sizeof(mreq)))
2956 			return -EFAULT;
2957 		break;
2958 	default:
2959 		if (ov || ol)
2960 			return -EINVAL;
2961 	}
2962 
2963 	lock_sock(sk);
2964 
2965 	switch (opt) {
2966 	case TIPC_IMPORTANCE:
2967 		res = tsk_set_importance(tsk, value);
2968 		break;
2969 	case TIPC_SRC_DROPPABLE:
2970 		if (sock->type != SOCK_STREAM)
2971 			tsk_set_unreliable(tsk, value);
2972 		else
2973 			res = -ENOPROTOOPT;
2974 		break;
2975 	case TIPC_DEST_DROPPABLE:
2976 		tsk_set_unreturnable(tsk, value);
2977 		break;
2978 	case TIPC_CONN_TIMEOUT:
2979 		tipc_sk(sk)->conn_timeout = value;
2980 		break;
2981 	case TIPC_MCAST_BROADCAST:
2982 		tsk->mc_method.rcast = false;
2983 		tsk->mc_method.mandatory = true;
2984 		break;
2985 	case TIPC_MCAST_REPLICAST:
2986 		tsk->mc_method.rcast = true;
2987 		tsk->mc_method.mandatory = true;
2988 		break;
2989 	case TIPC_GROUP_JOIN:
2990 		res = tipc_sk_join(tsk, &mreq);
2991 		break;
2992 	case TIPC_GROUP_LEAVE:
2993 		res = tipc_sk_leave(tsk);
2994 		break;
2995 	default:
2996 		res = -EINVAL;
2997 	}
2998 
2999 	release_sock(sk);
3000 
3001 	return res;
3002 }
3003 
3004 /**
3005  * tipc_getsockopt - get socket option
3006  * @sock: socket structure
3007  * @lvl: option level
3008  * @opt: option identifier
3009  * @ov: receptacle for option value
3010  * @ol: receptacle for length of option value
3011  *
3012  * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
3013  * (to ease compatibility).
3014  *
3015  * Returns 0 on success, errno otherwise
3016  */
3017 static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
3018 			   char __user *ov, int __user *ol)
3019 {
3020 	struct sock *sk = sock->sk;
3021 	struct tipc_sock *tsk = tipc_sk(sk);
3022 	struct tipc_name_seq seq;
3023 	int len, scope;
3024 	u32 value;
3025 	int res;
3026 
3027 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3028 		return put_user(0, ol);
3029 	if (lvl != SOL_TIPC)
3030 		return -ENOPROTOOPT;
3031 	res = get_user(len, ol);
3032 	if (res)
3033 		return res;
3034 
3035 	lock_sock(sk);
3036 
3037 	switch (opt) {
3038 	case TIPC_IMPORTANCE:
3039 		value = tsk_importance(tsk);
3040 		break;
3041 	case TIPC_SRC_DROPPABLE:
3042 		value = tsk_unreliable(tsk);
3043 		break;
3044 	case TIPC_DEST_DROPPABLE:
3045 		value = tsk_unreturnable(tsk);
3046 		break;
3047 	case TIPC_CONN_TIMEOUT:
3048 		value = tsk->conn_timeout;
3049 		/* no need to set "res", since already 0 at this point */
3050 		break;
3051 	case TIPC_NODE_RECVQ_DEPTH:
3052 		value = 0; /* was tipc_queue_size, now obsolete */
3053 		break;
3054 	case TIPC_SOCK_RECVQ_DEPTH:
3055 		value = skb_queue_len(&sk->sk_receive_queue);
3056 		break;
3057 	case TIPC_GROUP_JOIN:
3058 		seq.type = 0;
3059 		if (tsk->group)
3060 			tipc_group_self(tsk->group, &seq, &scope);
3061 		value = seq.type;
3062 		break;
3063 	default:
3064 		res = -EINVAL;
3065 	}
3066 
3067 	release_sock(sk);
3068 
3069 	if (res)
3070 		return res;	/* "get" failed */
3071 
3072 	if (len < sizeof(value))
3073 		return -EINVAL;
3074 
3075 	if (copy_to_user(ov, &value, sizeof(value)))
3076 		return -EFAULT;
3077 
3078 	return put_user(sizeof(value), ol);
3079 }
3080 
3081 static int tipc_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
3082 {
3083 	struct net *net = sock_net(sock->sk);
3084 	struct tipc_sioc_nodeid_req nr = {0};
3085 	struct tipc_sioc_ln_req lnr;
3086 	void __user *argp = (void __user *)arg;
3087 
3088 	switch (cmd) {
3089 	case SIOCGETLINKNAME:
3090 		if (copy_from_user(&lnr, argp, sizeof(lnr)))
3091 			return -EFAULT;
3092 		if (!tipc_node_get_linkname(net,
3093 					    lnr.bearer_id & 0xffff, lnr.peer,
3094 					    lnr.linkname, TIPC_MAX_LINK_NAME)) {
3095 			if (copy_to_user(argp, &lnr, sizeof(lnr)))
3096 				return -EFAULT;
3097 			return 0;
3098 		}
3099 		return -EADDRNOTAVAIL;
3100 	case SIOCGETNODEID:
3101 		if (copy_from_user(&nr, argp, sizeof(nr)))
3102 			return -EFAULT;
3103 		if (!tipc_node_get_id(net, nr.peer, nr.node_id))
3104 			return -EADDRNOTAVAIL;
3105 		if (copy_to_user(argp, &nr, sizeof(nr)))
3106 			return -EFAULT;
3107 		return 0;
3108 	default:
3109 		return -ENOIOCTLCMD;
3110 	}
3111 }
3112 
3113 static int tipc_socketpair(struct socket *sock1, struct socket *sock2)
3114 {
3115 	struct tipc_sock *tsk2 = tipc_sk(sock2->sk);
3116 	struct tipc_sock *tsk1 = tipc_sk(sock1->sk);
3117 	u32 onode = tipc_own_addr(sock_net(sock1->sk));
3118 
3119 	tsk1->peer.family = AF_TIPC;
3120 	tsk1->peer.addrtype = TIPC_ADDR_ID;
3121 	tsk1->peer.scope = TIPC_NODE_SCOPE;
3122 	tsk1->peer.addr.id.ref = tsk2->portid;
3123 	tsk1->peer.addr.id.node = onode;
3124 	tsk2->peer.family = AF_TIPC;
3125 	tsk2->peer.addrtype = TIPC_ADDR_ID;
3126 	tsk2->peer.scope = TIPC_NODE_SCOPE;
3127 	tsk2->peer.addr.id.ref = tsk1->portid;
3128 	tsk2->peer.addr.id.node = onode;
3129 
3130 	tipc_sk_finish_conn(tsk1, tsk2->portid, onode);
3131 	tipc_sk_finish_conn(tsk2, tsk1->portid, onode);
3132 	return 0;
3133 }
3134 
3135 /* Protocol switches for the various types of TIPC sockets */
3136 
3137 static const struct proto_ops msg_ops = {
3138 	.owner		= THIS_MODULE,
3139 	.family		= AF_TIPC,
3140 	.release	= tipc_release,
3141 	.bind		= tipc_bind,
3142 	.connect	= tipc_connect,
3143 	.socketpair	= tipc_socketpair,
3144 	.accept		= sock_no_accept,
3145 	.getname	= tipc_getname,
3146 	.poll		= tipc_poll,
3147 	.ioctl		= tipc_ioctl,
3148 	.listen		= sock_no_listen,
3149 	.shutdown	= tipc_shutdown,
3150 	.setsockopt	= tipc_setsockopt,
3151 	.getsockopt	= tipc_getsockopt,
3152 	.sendmsg	= tipc_sendmsg,
3153 	.recvmsg	= tipc_recvmsg,
3154 	.mmap		= sock_no_mmap,
3155 	.sendpage	= sock_no_sendpage
3156 };
3157 
3158 static const struct proto_ops packet_ops = {
3159 	.owner		= THIS_MODULE,
3160 	.family		= AF_TIPC,
3161 	.release	= tipc_release,
3162 	.bind		= tipc_bind,
3163 	.connect	= tipc_connect,
3164 	.socketpair	= tipc_socketpair,
3165 	.accept		= tipc_accept,
3166 	.getname	= tipc_getname,
3167 	.poll		= tipc_poll,
3168 	.ioctl		= tipc_ioctl,
3169 	.listen		= tipc_listen,
3170 	.shutdown	= tipc_shutdown,
3171 	.setsockopt	= tipc_setsockopt,
3172 	.getsockopt	= tipc_getsockopt,
3173 	.sendmsg	= tipc_send_packet,
3174 	.recvmsg	= tipc_recvmsg,
3175 	.mmap		= sock_no_mmap,
3176 	.sendpage	= sock_no_sendpage
3177 };
3178 
3179 static const struct proto_ops stream_ops = {
3180 	.owner		= THIS_MODULE,
3181 	.family		= AF_TIPC,
3182 	.release	= tipc_release,
3183 	.bind		= tipc_bind,
3184 	.connect	= tipc_connect,
3185 	.socketpair	= tipc_socketpair,
3186 	.accept		= tipc_accept,
3187 	.getname	= tipc_getname,
3188 	.poll		= tipc_poll,
3189 	.ioctl		= tipc_ioctl,
3190 	.listen		= tipc_listen,
3191 	.shutdown	= tipc_shutdown,
3192 	.setsockopt	= tipc_setsockopt,
3193 	.getsockopt	= tipc_getsockopt,
3194 	.sendmsg	= tipc_sendstream,
3195 	.recvmsg	= tipc_recvstream,
3196 	.mmap		= sock_no_mmap,
3197 	.sendpage	= sock_no_sendpage
3198 };
3199 
3200 static const struct net_proto_family tipc_family_ops = {
3201 	.owner		= THIS_MODULE,
3202 	.family		= AF_TIPC,
3203 	.create		= tipc_sk_create
3204 };
3205 
3206 static struct proto tipc_proto = {
3207 	.name		= "TIPC",
3208 	.owner		= THIS_MODULE,
3209 	.obj_size	= sizeof(struct tipc_sock),
3210 	.sysctl_rmem	= sysctl_tipc_rmem
3211 };
3212 
3213 /**
3214  * tipc_socket_init - initialize TIPC socket interface
3215  *
3216  * Returns 0 on success, errno otherwise
3217  */
3218 int tipc_socket_init(void)
3219 {
3220 	int res;
3221 
3222 	res = proto_register(&tipc_proto, 1);
3223 	if (res) {
3224 		pr_err("Failed to register TIPC protocol type\n");
3225 		goto out;
3226 	}
3227 
3228 	res = sock_register(&tipc_family_ops);
3229 	if (res) {
3230 		pr_err("Failed to register TIPC socket type\n");
3231 		proto_unregister(&tipc_proto);
3232 		goto out;
3233 	}
3234  out:
3235 	return res;
3236 }
3237 
3238 /**
3239  * tipc_socket_stop - stop TIPC socket interface
3240  */
3241 void tipc_socket_stop(void)
3242 {
3243 	sock_unregister(tipc_family_ops.family);
3244 	proto_unregister(&tipc_proto);
3245 }
3246 
3247 /* Caller should hold socket lock for the passed tipc socket. */
3248 static int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk)
3249 {
3250 	u32 peer_node;
3251 	u32 peer_port;
3252 	struct nlattr *nest;
3253 
3254 	peer_node = tsk_peer_node(tsk);
3255 	peer_port = tsk_peer_port(tsk);
3256 
3257 	nest = nla_nest_start(skb, TIPC_NLA_SOCK_CON);
3258 
3259 	if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node))
3260 		goto msg_full;
3261 	if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port))
3262 		goto msg_full;
3263 
3264 	if (tsk->conn_type != 0) {
3265 		if (nla_put_flag(skb, TIPC_NLA_CON_FLAG))
3266 			goto msg_full;
3267 		if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, tsk->conn_type))
3268 			goto msg_full;
3269 		if (nla_put_u32(skb, TIPC_NLA_CON_INST, tsk->conn_instance))
3270 			goto msg_full;
3271 	}
3272 	nla_nest_end(skb, nest);
3273 
3274 	return 0;
3275 
3276 msg_full:
3277 	nla_nest_cancel(skb, nest);
3278 
3279 	return -EMSGSIZE;
3280 }
3281 
3282 static int __tipc_nl_add_sk_info(struct sk_buff *skb, struct tipc_sock
3283 			  *tsk)
3284 {
3285 	struct net *net = sock_net(skb->sk);
3286 	struct sock *sk = &tsk->sk;
3287 
3288 	if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->portid) ||
3289 	    nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tipc_own_addr(net)))
3290 		return -EMSGSIZE;
3291 
3292 	if (tipc_sk_connected(sk)) {
3293 		if (__tipc_nl_add_sk_con(skb, tsk))
3294 			return -EMSGSIZE;
3295 	} else if (!list_empty(&tsk->publications)) {
3296 		if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL))
3297 			return -EMSGSIZE;
3298 	}
3299 	return 0;
3300 }
3301 
3302 /* Caller should hold socket lock for the passed tipc socket. */
3303 static int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb,
3304 			    struct tipc_sock *tsk)
3305 {
3306 	struct nlattr *attrs;
3307 	void *hdr;
3308 
3309 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
3310 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_SOCK_GET);
3311 	if (!hdr)
3312 		goto msg_cancel;
3313 
3314 	attrs = nla_nest_start(skb, TIPC_NLA_SOCK);
3315 	if (!attrs)
3316 		goto genlmsg_cancel;
3317 
3318 	if (__tipc_nl_add_sk_info(skb, tsk))
3319 		goto attr_msg_cancel;
3320 
3321 	nla_nest_end(skb, attrs);
3322 	genlmsg_end(skb, hdr);
3323 
3324 	return 0;
3325 
3326 attr_msg_cancel:
3327 	nla_nest_cancel(skb, attrs);
3328 genlmsg_cancel:
3329 	genlmsg_cancel(skb, hdr);
3330 msg_cancel:
3331 	return -EMSGSIZE;
3332 }
3333 
3334 int tipc_nl_sk_walk(struct sk_buff *skb, struct netlink_callback *cb,
3335 		    int (*skb_handler)(struct sk_buff *skb,
3336 				       struct netlink_callback *cb,
3337 				       struct tipc_sock *tsk))
3338 {
3339 	struct rhashtable_iter *iter = (void *)cb->args[4];
3340 	struct tipc_sock *tsk;
3341 	int err;
3342 
3343 	rhashtable_walk_start(iter);
3344 	while ((tsk = rhashtable_walk_next(iter)) != NULL) {
3345 		if (IS_ERR(tsk)) {
3346 			err = PTR_ERR(tsk);
3347 			if (err == -EAGAIN) {
3348 				err = 0;
3349 				continue;
3350 			}
3351 			break;
3352 		}
3353 
3354 		sock_hold(&tsk->sk);
3355 		rhashtable_walk_stop(iter);
3356 		lock_sock(&tsk->sk);
3357 		err = skb_handler(skb, cb, tsk);
3358 		if (err) {
3359 			release_sock(&tsk->sk);
3360 			sock_put(&tsk->sk);
3361 			goto out;
3362 		}
3363 		release_sock(&tsk->sk);
3364 		rhashtable_walk_start(iter);
3365 		sock_put(&tsk->sk);
3366 	}
3367 	rhashtable_walk_stop(iter);
3368 out:
3369 	return skb->len;
3370 }
3371 EXPORT_SYMBOL(tipc_nl_sk_walk);
3372 
3373 int tipc_dump_start(struct netlink_callback *cb)
3374 {
3375 	return __tipc_dump_start(cb, sock_net(cb->skb->sk));
3376 }
3377 EXPORT_SYMBOL(tipc_dump_start);
3378 
3379 int __tipc_dump_start(struct netlink_callback *cb, struct net *net)
3380 {
3381 	/* tipc_nl_name_table_dump() uses cb->args[0...3]. */
3382 	struct rhashtable_iter *iter = (void *)cb->args[4];
3383 	struct tipc_net *tn = tipc_net(net);
3384 
3385 	if (!iter) {
3386 		iter = kmalloc(sizeof(*iter), GFP_KERNEL);
3387 		if (!iter)
3388 			return -ENOMEM;
3389 
3390 		cb->args[4] = (long)iter;
3391 	}
3392 
3393 	rhashtable_walk_enter(&tn->sk_rht, iter);
3394 	return 0;
3395 }
3396 
3397 int tipc_dump_done(struct netlink_callback *cb)
3398 {
3399 	struct rhashtable_iter *hti = (void *)cb->args[4];
3400 
3401 	rhashtable_walk_exit(hti);
3402 	kfree(hti);
3403 	return 0;
3404 }
3405 EXPORT_SYMBOL(tipc_dump_done);
3406 
3407 int tipc_sk_fill_sock_diag(struct sk_buff *skb, struct netlink_callback *cb,
3408 			   struct tipc_sock *tsk, u32 sk_filter_state,
3409 			   u64 (*tipc_diag_gen_cookie)(struct sock *sk))
3410 {
3411 	struct sock *sk = &tsk->sk;
3412 	struct nlattr *attrs;
3413 	struct nlattr *stat;
3414 
3415 	/*filter response w.r.t sk_state*/
3416 	if (!(sk_filter_state & (1 << sk->sk_state)))
3417 		return 0;
3418 
3419 	attrs = nla_nest_start(skb, TIPC_NLA_SOCK);
3420 	if (!attrs)
3421 		goto msg_cancel;
3422 
3423 	if (__tipc_nl_add_sk_info(skb, tsk))
3424 		goto attr_msg_cancel;
3425 
3426 	if (nla_put_u32(skb, TIPC_NLA_SOCK_TYPE, (u32)sk->sk_type) ||
3427 	    nla_put_u32(skb, TIPC_NLA_SOCK_TIPC_STATE, (u32)sk->sk_state) ||
3428 	    nla_put_u32(skb, TIPC_NLA_SOCK_INO, sock_i_ino(sk)) ||
3429 	    nla_put_u32(skb, TIPC_NLA_SOCK_UID,
3430 			from_kuid_munged(sk_user_ns(NETLINK_CB(cb->skb).sk),
3431 					 sock_i_uid(sk))) ||
3432 	    nla_put_u64_64bit(skb, TIPC_NLA_SOCK_COOKIE,
3433 			      tipc_diag_gen_cookie(sk),
3434 			      TIPC_NLA_SOCK_PAD))
3435 		goto attr_msg_cancel;
3436 
3437 	stat = nla_nest_start(skb, TIPC_NLA_SOCK_STAT);
3438 	if (!stat)
3439 		goto attr_msg_cancel;
3440 
3441 	if (nla_put_u32(skb, TIPC_NLA_SOCK_STAT_RCVQ,
3442 			skb_queue_len(&sk->sk_receive_queue)) ||
3443 	    nla_put_u32(skb, TIPC_NLA_SOCK_STAT_SENDQ,
3444 			skb_queue_len(&sk->sk_write_queue)) ||
3445 	    nla_put_u32(skb, TIPC_NLA_SOCK_STAT_DROP,
3446 			atomic_read(&sk->sk_drops)))
3447 		goto stat_msg_cancel;
3448 
3449 	if (tsk->cong_link_cnt &&
3450 	    nla_put_flag(skb, TIPC_NLA_SOCK_STAT_LINK_CONG))
3451 		goto stat_msg_cancel;
3452 
3453 	if (tsk_conn_cong(tsk) &&
3454 	    nla_put_flag(skb, TIPC_NLA_SOCK_STAT_CONN_CONG))
3455 		goto stat_msg_cancel;
3456 
3457 	nla_nest_end(skb, stat);
3458 
3459 	if (tsk->group)
3460 		if (tipc_group_fill_sock_diag(tsk->group, skb))
3461 			goto stat_msg_cancel;
3462 
3463 	nla_nest_end(skb, attrs);
3464 
3465 	return 0;
3466 
3467 stat_msg_cancel:
3468 	nla_nest_cancel(skb, stat);
3469 attr_msg_cancel:
3470 	nla_nest_cancel(skb, attrs);
3471 msg_cancel:
3472 	return -EMSGSIZE;
3473 }
3474 EXPORT_SYMBOL(tipc_sk_fill_sock_diag);
3475 
3476 int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb)
3477 {
3478 	return tipc_nl_sk_walk(skb, cb, __tipc_nl_add_sk);
3479 }
3480 
3481 /* Caller should hold socket lock for the passed tipc socket. */
3482 static int __tipc_nl_add_sk_publ(struct sk_buff *skb,
3483 				 struct netlink_callback *cb,
3484 				 struct publication *publ)
3485 {
3486 	void *hdr;
3487 	struct nlattr *attrs;
3488 
3489 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
3490 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_PUBL_GET);
3491 	if (!hdr)
3492 		goto msg_cancel;
3493 
3494 	attrs = nla_nest_start(skb, TIPC_NLA_PUBL);
3495 	if (!attrs)
3496 		goto genlmsg_cancel;
3497 
3498 	if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key))
3499 		goto attr_msg_cancel;
3500 	if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->type))
3501 		goto attr_msg_cancel;
3502 	if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->lower))
3503 		goto attr_msg_cancel;
3504 	if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->upper))
3505 		goto attr_msg_cancel;
3506 
3507 	nla_nest_end(skb, attrs);
3508 	genlmsg_end(skb, hdr);
3509 
3510 	return 0;
3511 
3512 attr_msg_cancel:
3513 	nla_nest_cancel(skb, attrs);
3514 genlmsg_cancel:
3515 	genlmsg_cancel(skb, hdr);
3516 msg_cancel:
3517 	return -EMSGSIZE;
3518 }
3519 
3520 /* Caller should hold socket lock for the passed tipc socket. */
3521 static int __tipc_nl_list_sk_publ(struct sk_buff *skb,
3522 				  struct netlink_callback *cb,
3523 				  struct tipc_sock *tsk, u32 *last_publ)
3524 {
3525 	int err;
3526 	struct publication *p;
3527 
3528 	if (*last_publ) {
3529 		list_for_each_entry(p, &tsk->publications, binding_sock) {
3530 			if (p->key == *last_publ)
3531 				break;
3532 		}
3533 		if (p->key != *last_publ) {
3534 			/* We never set seq or call nl_dump_check_consistent()
3535 			 * this means that setting prev_seq here will cause the
3536 			 * consistence check to fail in the netlink callback
3537 			 * handler. Resulting in the last NLMSG_DONE message
3538 			 * having the NLM_F_DUMP_INTR flag set.
3539 			 */
3540 			cb->prev_seq = 1;
3541 			*last_publ = 0;
3542 			return -EPIPE;
3543 		}
3544 	} else {
3545 		p = list_first_entry(&tsk->publications, struct publication,
3546 				     binding_sock);
3547 	}
3548 
3549 	list_for_each_entry_from(p, &tsk->publications, binding_sock) {
3550 		err = __tipc_nl_add_sk_publ(skb, cb, p);
3551 		if (err) {
3552 			*last_publ = p->key;
3553 			return err;
3554 		}
3555 	}
3556 	*last_publ = 0;
3557 
3558 	return 0;
3559 }
3560 
3561 int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)
3562 {
3563 	int err;
3564 	u32 tsk_portid = cb->args[0];
3565 	u32 last_publ = cb->args[1];
3566 	u32 done = cb->args[2];
3567 	struct net *net = sock_net(skb->sk);
3568 	struct tipc_sock *tsk;
3569 
3570 	if (!tsk_portid) {
3571 		struct nlattr **attrs;
3572 		struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1];
3573 
3574 		err = tipc_nlmsg_parse(cb->nlh, &attrs);
3575 		if (err)
3576 			return err;
3577 
3578 		if (!attrs[TIPC_NLA_SOCK])
3579 			return -EINVAL;
3580 
3581 		err = nla_parse_nested(sock, TIPC_NLA_SOCK_MAX,
3582 				       attrs[TIPC_NLA_SOCK],
3583 				       tipc_nl_sock_policy, NULL);
3584 		if (err)
3585 			return err;
3586 
3587 		if (!sock[TIPC_NLA_SOCK_REF])
3588 			return -EINVAL;
3589 
3590 		tsk_portid = nla_get_u32(sock[TIPC_NLA_SOCK_REF]);
3591 	}
3592 
3593 	if (done)
3594 		return 0;
3595 
3596 	tsk = tipc_sk_lookup(net, tsk_portid);
3597 	if (!tsk)
3598 		return -EINVAL;
3599 
3600 	lock_sock(&tsk->sk);
3601 	err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ);
3602 	if (!err)
3603 		done = 1;
3604 	release_sock(&tsk->sk);
3605 	sock_put(&tsk->sk);
3606 
3607 	cb->args[0] = tsk_portid;
3608 	cb->args[1] = last_publ;
3609 	cb->args[2] = done;
3610 
3611 	return skb->len;
3612 }
3613 
3614 /**
3615  * tipc_sk_filtering - check if a socket should be traced
3616  * @sk: the socket to be examined
3617  * @sysctl_tipc_sk_filter[]: the socket tuple for filtering,
3618  *  (portid, sock type, name type, name lower, name upper)
3619  *
3620  * Returns true if the socket meets the socket tuple data
3621  * (value 0 = 'any') or when there is no tuple set (all = 0),
3622  * otherwise false
3623  */
3624 bool tipc_sk_filtering(struct sock *sk)
3625 {
3626 	struct tipc_sock *tsk;
3627 	struct publication *p;
3628 	u32 _port, _sktype, _type, _lower, _upper;
3629 	u32 type = 0, lower = 0, upper = 0;
3630 
3631 	if (!sk)
3632 		return true;
3633 
3634 	tsk = tipc_sk(sk);
3635 
3636 	_port = sysctl_tipc_sk_filter[0];
3637 	_sktype = sysctl_tipc_sk_filter[1];
3638 	_type = sysctl_tipc_sk_filter[2];
3639 	_lower = sysctl_tipc_sk_filter[3];
3640 	_upper = sysctl_tipc_sk_filter[4];
3641 
3642 	if (!_port && !_sktype && !_type && !_lower && !_upper)
3643 		return true;
3644 
3645 	if (_port)
3646 		return (_port == tsk->portid);
3647 
3648 	if (_sktype && _sktype != sk->sk_type)
3649 		return false;
3650 
3651 	if (tsk->published) {
3652 		p = list_first_entry_or_null(&tsk->publications,
3653 					     struct publication, binding_sock);
3654 		if (p) {
3655 			type = p->type;
3656 			lower = p->lower;
3657 			upper = p->upper;
3658 		}
3659 	}
3660 
3661 	if (!tipc_sk_type_connectionless(sk)) {
3662 		type = tsk->conn_type;
3663 		lower = tsk->conn_instance;
3664 		upper = tsk->conn_instance;
3665 	}
3666 
3667 	if ((_type && _type != type) || (_lower && _lower != lower) ||
3668 	    (_upper && _upper != upper))
3669 		return false;
3670 
3671 	return true;
3672 }
3673 
3674 u32 tipc_sock_get_portid(struct sock *sk)
3675 {
3676 	return (sk) ? (tipc_sk(sk))->portid : 0;
3677 }
3678 
3679 /**
3680  * tipc_sk_overlimit1 - check if socket rx queue is about to be overloaded,
3681  *			both the rcv and backlog queues are considered
3682  * @sk: tipc sk to be checked
3683  * @skb: tipc msg to be checked
3684  *
3685  * Returns true if the socket rx queue allocation is > 90%, otherwise false
3686  */
3687 
3688 bool tipc_sk_overlimit1(struct sock *sk, struct sk_buff *skb)
3689 {
3690 	atomic_t *dcnt = &tipc_sk(sk)->dupl_rcvcnt;
3691 	unsigned int lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
3692 	unsigned int qsize = sk->sk_backlog.len + sk_rmem_alloc_get(sk);
3693 
3694 	return (qsize > lim * 90 / 100);
3695 }
3696 
3697 /**
3698  * tipc_sk_overlimit2 - check if socket rx queue is about to be overloaded,
3699  *			only the rcv queue is considered
3700  * @sk: tipc sk to be checked
3701  * @skb: tipc msg to be checked
3702  *
3703  * Returns true if the socket rx queue allocation is > 90%, otherwise false
3704  */
3705 
3706 bool tipc_sk_overlimit2(struct sock *sk, struct sk_buff *skb)
3707 {
3708 	unsigned int lim = rcvbuf_limit(sk, skb);
3709 	unsigned int qsize = sk_rmem_alloc_get(sk);
3710 
3711 	return (qsize > lim * 90 / 100);
3712 }
3713 
3714 /**
3715  * tipc_sk_dump - dump TIPC socket
3716  * @sk: tipc sk to be dumped
3717  * @dqueues: bitmask to decide if any socket queue to be dumped?
3718  *           - TIPC_DUMP_NONE: don't dump socket queues
3719  *           - TIPC_DUMP_SK_SNDQ: dump socket send queue
3720  *           - TIPC_DUMP_SK_RCVQ: dump socket rcv queue
3721  *           - TIPC_DUMP_SK_BKLGQ: dump socket backlog queue
3722  *           - TIPC_DUMP_ALL: dump all the socket queues above
3723  * @buf: returned buffer of dump data in format
3724  */
3725 int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)
3726 {
3727 	int i = 0;
3728 	size_t sz = (dqueues) ? SK_LMAX : SK_LMIN;
3729 	struct tipc_sock *tsk;
3730 	struct publication *p;
3731 	bool tsk_connected;
3732 
3733 	if (!sk) {
3734 		i += scnprintf(buf, sz, "sk data: (null)\n");
3735 		return i;
3736 	}
3737 
3738 	tsk = tipc_sk(sk);
3739 	tsk_connected = !tipc_sk_type_connectionless(sk);
3740 
3741 	i += scnprintf(buf, sz, "sk data: %u", sk->sk_type);
3742 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_state);
3743 	i += scnprintf(buf + i, sz - i, " %x", tsk_own_node(tsk));
3744 	i += scnprintf(buf + i, sz - i, " %u", tsk->portid);
3745 	i += scnprintf(buf + i, sz - i, " | %u", tsk_connected);
3746 	if (tsk_connected) {
3747 		i += scnprintf(buf + i, sz - i, " %x", tsk_peer_node(tsk));
3748 		i += scnprintf(buf + i, sz - i, " %u", tsk_peer_port(tsk));
3749 		i += scnprintf(buf + i, sz - i, " %u", tsk->conn_type);
3750 		i += scnprintf(buf + i, sz - i, " %u", tsk->conn_instance);
3751 	}
3752 	i += scnprintf(buf + i, sz - i, " | %u", tsk->published);
3753 	if (tsk->published) {
3754 		p = list_first_entry_or_null(&tsk->publications,
3755 					     struct publication, binding_sock);
3756 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->type : 0);
3757 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->lower : 0);
3758 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->upper : 0);
3759 	}
3760 	i += scnprintf(buf + i, sz - i, " | %u", tsk->snd_win);
3761 	i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_win);
3762 	i += scnprintf(buf + i, sz - i, " %u", tsk->max_pkt);
3763 	i += scnprintf(buf + i, sz - i, " %x", tsk->peer_caps);
3764 	i += scnprintf(buf + i, sz - i, " %u", tsk->cong_link_cnt);
3765 	i += scnprintf(buf + i, sz - i, " %u", tsk->snt_unacked);
3766 	i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_unacked);
3767 	i += scnprintf(buf + i, sz - i, " %u", atomic_read(&tsk->dupl_rcvcnt));
3768 	i += scnprintf(buf + i, sz - i, " %u", sk->sk_shutdown);
3769 	i += scnprintf(buf + i, sz - i, " | %d", sk_wmem_alloc_get(sk));
3770 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_sndbuf);
3771 	i += scnprintf(buf + i, sz - i, " | %d", sk_rmem_alloc_get(sk));
3772 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_rcvbuf);
3773 	i += scnprintf(buf + i, sz - i, " | %d\n", sk->sk_backlog.len);
3774 
3775 	if (dqueues & TIPC_DUMP_SK_SNDQ) {
3776 		i += scnprintf(buf + i, sz - i, "sk_write_queue: ");
3777 		i += tipc_list_dump(&sk->sk_write_queue, false, buf + i);
3778 	}
3779 
3780 	if (dqueues & TIPC_DUMP_SK_RCVQ) {
3781 		i += scnprintf(buf + i, sz - i, "sk_receive_queue: ");
3782 		i += tipc_list_dump(&sk->sk_receive_queue, false, buf + i);
3783 	}
3784 
3785 	if (dqueues & TIPC_DUMP_SK_BKLGQ) {
3786 		i += scnprintf(buf + i, sz - i, "sk_backlog:\n  head ");
3787 		i += tipc_skb_dump(sk->sk_backlog.head, false, buf + i);
3788 		if (sk->sk_backlog.tail != sk->sk_backlog.head) {
3789 			i += scnprintf(buf + i, sz - i, "  tail ");
3790 			i += tipc_skb_dump(sk->sk_backlog.tail, false,
3791 					   buf + i);
3792 		}
3793 	}
3794 
3795 	return i;
3796 }
3797