xref: /openbmc/linux/net/tipc/socket.c (revision b8265621)
1 /*
2  * net/tipc/socket.c: TIPC socket API
3  *
4  * Copyright (c) 2001-2007, 2012-2017, Ericsson AB
5  * Copyright (c) 2004-2008, 2010-2013, Wind River Systems
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions are met:
10  *
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the names of the copyright holders nor the names of its
17  *    contributors may be used to endorse or promote products derived from
18  *    this software without specific prior written permission.
19  *
20  * Alternatively, this software may be distributed under the terms of the
21  * GNU General Public License ("GPL") version 2 as published by the Free
22  * Software Foundation.
23  *
24  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34  * POSSIBILITY OF SUCH DAMAGE.
35  */
36 
37 #include <linux/rhashtable.h>
38 #include <linux/sched/signal.h>
39 
40 #include "core.h"
41 #include "name_table.h"
42 #include "node.h"
43 #include "link.h"
44 #include "name_distr.h"
45 #include "socket.h"
46 #include "bcast.h"
47 #include "netlink.h"
48 #include "group.h"
49 #include "trace.h"
50 
51 #define NAGLE_START_INIT	4
52 #define NAGLE_START_MAX		1024
53 #define CONN_TIMEOUT_DEFAULT    8000    /* default connect timeout = 8s */
54 #define CONN_PROBING_INTV	msecs_to_jiffies(3600000)  /* [ms] => 1 h */
55 #define TIPC_FWD_MSG		1
56 #define TIPC_MAX_PORT		0xffffffff
57 #define TIPC_MIN_PORT		1
58 #define TIPC_ACK_RATE		4       /* ACK at 1/4 of of rcv window size */
59 
60 enum {
61 	TIPC_LISTEN = TCP_LISTEN,
62 	TIPC_ESTABLISHED = TCP_ESTABLISHED,
63 	TIPC_OPEN = TCP_CLOSE,
64 	TIPC_DISCONNECTING = TCP_CLOSE_WAIT,
65 	TIPC_CONNECTING = TCP_SYN_SENT,
66 };
67 
68 struct sockaddr_pair {
69 	struct sockaddr_tipc sock;
70 	struct sockaddr_tipc member;
71 };
72 
73 /**
74  * struct tipc_sock - TIPC socket structure
75  * @sk: socket - interacts with 'port' and with user via the socket API
76  * @conn_type: TIPC type used when connection was established
77  * @conn_instance: TIPC instance used when connection was established
78  * @published: non-zero if port has one or more associated names
79  * @max_pkt: maximum packet size "hint" used when building messages sent by port
80  * @maxnagle: maximum size of msg which can be subject to nagle
81  * @portid: unique port identity in TIPC socket hash table
82  * @phdr: preformatted message header used when sending messages
83  * #cong_links: list of congested links
84  * @publications: list of publications for port
85  * @blocking_link: address of the congested link we are currently sleeping on
86  * @pub_count: total # of publications port has made during its lifetime
87  * @conn_timeout: the time we can wait for an unresponded setup request
88  * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue
89  * @cong_link_cnt: number of congested links
90  * @snt_unacked: # messages sent by socket, and not yet acked by peer
91  * @rcv_unacked: # messages read by user, but not yet acked back to peer
92  * @peer: 'connected' peer for dgram/rdm
93  * @node: hash table node
94  * @mc_method: cookie for use between socket and broadcast layer
95  * @rcu: rcu struct for tipc_sock
96  */
97 struct tipc_sock {
98 	struct sock sk;
99 	u32 conn_type;
100 	u32 conn_instance;
101 	int published;
102 	u32 max_pkt;
103 	u32 maxnagle;
104 	u32 portid;
105 	struct tipc_msg phdr;
106 	struct list_head cong_links;
107 	struct list_head publications;
108 	u32 pub_count;
109 	atomic_t dupl_rcvcnt;
110 	u16 conn_timeout;
111 	bool probe_unacked;
112 	u16 cong_link_cnt;
113 	u16 snt_unacked;
114 	u16 snd_win;
115 	u16 peer_caps;
116 	u16 rcv_unacked;
117 	u16 rcv_win;
118 	struct sockaddr_tipc peer;
119 	struct rhash_head node;
120 	struct tipc_mc_method mc_method;
121 	struct rcu_head rcu;
122 	struct tipc_group *group;
123 	u32 oneway;
124 	u32 nagle_start;
125 	u16 snd_backlog;
126 	u16 msg_acc;
127 	u16 pkt_cnt;
128 	bool expect_ack;
129 	bool nodelay;
130 	bool group_is_open;
131 };
132 
133 static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb);
134 static void tipc_data_ready(struct sock *sk);
135 static void tipc_write_space(struct sock *sk);
136 static void tipc_sock_destruct(struct sock *sk);
137 static int tipc_release(struct socket *sock);
138 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags,
139 		       bool kern);
140 static void tipc_sk_timeout(struct timer_list *t);
141 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
142 			   struct tipc_name_seq const *seq);
143 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
144 			    struct tipc_name_seq const *seq);
145 static int tipc_sk_leave(struct tipc_sock *tsk);
146 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid);
147 static int tipc_sk_insert(struct tipc_sock *tsk);
148 static void tipc_sk_remove(struct tipc_sock *tsk);
149 static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz);
150 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dsz);
151 static void tipc_sk_push_backlog(struct tipc_sock *tsk, bool nagle_ack);
152 
153 static const struct proto_ops packet_ops;
154 static const struct proto_ops stream_ops;
155 static const struct proto_ops msg_ops;
156 static struct proto tipc_proto;
157 static const struct rhashtable_params tsk_rht_params;
158 
159 static u32 tsk_own_node(struct tipc_sock *tsk)
160 {
161 	return msg_prevnode(&tsk->phdr);
162 }
163 
164 static u32 tsk_peer_node(struct tipc_sock *tsk)
165 {
166 	return msg_destnode(&tsk->phdr);
167 }
168 
169 static u32 tsk_peer_port(struct tipc_sock *tsk)
170 {
171 	return msg_destport(&tsk->phdr);
172 }
173 
174 static  bool tsk_unreliable(struct tipc_sock *tsk)
175 {
176 	return msg_src_droppable(&tsk->phdr) != 0;
177 }
178 
179 static void tsk_set_unreliable(struct tipc_sock *tsk, bool unreliable)
180 {
181 	msg_set_src_droppable(&tsk->phdr, unreliable ? 1 : 0);
182 }
183 
184 static bool tsk_unreturnable(struct tipc_sock *tsk)
185 {
186 	return msg_dest_droppable(&tsk->phdr) != 0;
187 }
188 
189 static void tsk_set_unreturnable(struct tipc_sock *tsk, bool unreturnable)
190 {
191 	msg_set_dest_droppable(&tsk->phdr, unreturnable ? 1 : 0);
192 }
193 
194 static int tsk_importance(struct tipc_sock *tsk)
195 {
196 	return msg_importance(&tsk->phdr);
197 }
198 
199 static struct tipc_sock *tipc_sk(const struct sock *sk)
200 {
201 	return container_of(sk, struct tipc_sock, sk);
202 }
203 
204 int tsk_set_importance(struct sock *sk, int imp)
205 {
206 	if (imp > TIPC_CRITICAL_IMPORTANCE)
207 		return -EINVAL;
208 	msg_set_importance(&tipc_sk(sk)->phdr, (u32)imp);
209 	return 0;
210 }
211 
212 static bool tsk_conn_cong(struct tipc_sock *tsk)
213 {
214 	return tsk->snt_unacked > tsk->snd_win;
215 }
216 
217 static u16 tsk_blocks(int len)
218 {
219 	return ((len / FLOWCTL_BLK_SZ) + 1);
220 }
221 
222 /* tsk_blocks(): translate a buffer size in bytes to number of
223  * advertisable blocks, taking into account the ratio truesize(len)/len
224  * We can trust that this ratio is always < 4 for len >= FLOWCTL_BLK_SZ
225  */
226 static u16 tsk_adv_blocks(int len)
227 {
228 	return len / FLOWCTL_BLK_SZ / 4;
229 }
230 
231 /* tsk_inc(): increment counter for sent or received data
232  * - If block based flow control is not supported by peer we
233  *   fall back to message based ditto, incrementing the counter
234  */
235 static u16 tsk_inc(struct tipc_sock *tsk, int msglen)
236 {
237 	if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
238 		return ((msglen / FLOWCTL_BLK_SZ) + 1);
239 	return 1;
240 }
241 
242 /* tsk_set_nagle - enable/disable nagle property by manipulating maxnagle
243  */
244 static void tsk_set_nagle(struct tipc_sock *tsk)
245 {
246 	struct sock *sk = &tsk->sk;
247 
248 	tsk->maxnagle = 0;
249 	if (sk->sk_type != SOCK_STREAM)
250 		return;
251 	if (tsk->nodelay)
252 		return;
253 	if (!(tsk->peer_caps & TIPC_NAGLE))
254 		return;
255 	/* Limit node local buffer size to avoid receive queue overflow */
256 	if (tsk->max_pkt == MAX_MSG_SIZE)
257 		tsk->maxnagle = 1500;
258 	else
259 		tsk->maxnagle = tsk->max_pkt;
260 }
261 
262 /**
263  * tsk_advance_rx_queue - discard first buffer in socket receive queue
264  *
265  * Caller must hold socket lock
266  */
267 static void tsk_advance_rx_queue(struct sock *sk)
268 {
269 	trace_tipc_sk_advance_rx(sk, NULL, TIPC_DUMP_SK_RCVQ, " ");
270 	kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
271 }
272 
273 /* tipc_sk_respond() : send response message back to sender
274  */
275 static void tipc_sk_respond(struct sock *sk, struct sk_buff *skb, int err)
276 {
277 	u32 selector;
278 	u32 dnode;
279 	u32 onode = tipc_own_addr(sock_net(sk));
280 
281 	if (!tipc_msg_reverse(onode, &skb, err))
282 		return;
283 
284 	trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE, "@sk_respond!");
285 	dnode = msg_destnode(buf_msg(skb));
286 	selector = msg_origport(buf_msg(skb));
287 	tipc_node_xmit_skb(sock_net(sk), skb, dnode, selector);
288 }
289 
290 /**
291  * tsk_rej_rx_queue - reject all buffers in socket receive queue
292  *
293  * Caller must hold socket lock
294  */
295 static void tsk_rej_rx_queue(struct sock *sk, int error)
296 {
297 	struct sk_buff *skb;
298 
299 	while ((skb = __skb_dequeue(&sk->sk_receive_queue)))
300 		tipc_sk_respond(sk, skb, error);
301 }
302 
303 static bool tipc_sk_connected(struct sock *sk)
304 {
305 	return sk->sk_state == TIPC_ESTABLISHED;
306 }
307 
308 /* tipc_sk_type_connectionless - check if the socket is datagram socket
309  * @sk: socket
310  *
311  * Returns true if connection less, false otherwise
312  */
313 static bool tipc_sk_type_connectionless(struct sock *sk)
314 {
315 	return sk->sk_type == SOCK_RDM || sk->sk_type == SOCK_DGRAM;
316 }
317 
318 /* tsk_peer_msg - verify if message was sent by connected port's peer
319  *
320  * Handles cases where the node's network address has changed from
321  * the default of <0.0.0> to its configured setting.
322  */
323 static bool tsk_peer_msg(struct tipc_sock *tsk, struct tipc_msg *msg)
324 {
325 	struct sock *sk = &tsk->sk;
326 	u32 self = tipc_own_addr(sock_net(sk));
327 	u32 peer_port = tsk_peer_port(tsk);
328 	u32 orig_node, peer_node;
329 
330 	if (unlikely(!tipc_sk_connected(sk)))
331 		return false;
332 
333 	if (unlikely(msg_origport(msg) != peer_port))
334 		return false;
335 
336 	orig_node = msg_orignode(msg);
337 	peer_node = tsk_peer_node(tsk);
338 
339 	if (likely(orig_node == peer_node))
340 		return true;
341 
342 	if (!orig_node && peer_node == self)
343 		return true;
344 
345 	if (!peer_node && orig_node == self)
346 		return true;
347 
348 	return false;
349 }
350 
351 /* tipc_set_sk_state - set the sk_state of the socket
352  * @sk: socket
353  *
354  * Caller must hold socket lock
355  *
356  * Returns 0 on success, errno otherwise
357  */
358 static int tipc_set_sk_state(struct sock *sk, int state)
359 {
360 	int oldsk_state = sk->sk_state;
361 	int res = -EINVAL;
362 
363 	switch (state) {
364 	case TIPC_OPEN:
365 		res = 0;
366 		break;
367 	case TIPC_LISTEN:
368 	case TIPC_CONNECTING:
369 		if (oldsk_state == TIPC_OPEN)
370 			res = 0;
371 		break;
372 	case TIPC_ESTABLISHED:
373 		if (oldsk_state == TIPC_CONNECTING ||
374 		    oldsk_state == TIPC_OPEN)
375 			res = 0;
376 		break;
377 	case TIPC_DISCONNECTING:
378 		if (oldsk_state == TIPC_CONNECTING ||
379 		    oldsk_state == TIPC_ESTABLISHED)
380 			res = 0;
381 		break;
382 	}
383 
384 	if (!res)
385 		sk->sk_state = state;
386 
387 	return res;
388 }
389 
390 static int tipc_sk_sock_err(struct socket *sock, long *timeout)
391 {
392 	struct sock *sk = sock->sk;
393 	int err = sock_error(sk);
394 	int typ = sock->type;
395 
396 	if (err)
397 		return err;
398 	if (typ == SOCK_STREAM || typ == SOCK_SEQPACKET) {
399 		if (sk->sk_state == TIPC_DISCONNECTING)
400 			return -EPIPE;
401 		else if (!tipc_sk_connected(sk))
402 			return -ENOTCONN;
403 	}
404 	if (!*timeout)
405 		return -EAGAIN;
406 	if (signal_pending(current))
407 		return sock_intr_errno(*timeout);
408 
409 	return 0;
410 }
411 
412 #define tipc_wait_for_cond(sock_, timeo_, condition_)			       \
413 ({                                                                             \
414 	DEFINE_WAIT_FUNC(wait_, woken_wake_function);                          \
415 	struct sock *sk_;						       \
416 	int rc_;							       \
417 									       \
418 	while ((rc_ = !(condition_))) {					       \
419 		/* coupled with smp_wmb() in tipc_sk_proto_rcv() */            \
420 		smp_rmb();                                                     \
421 		sk_ = (sock_)->sk;					       \
422 		rc_ = tipc_sk_sock_err((sock_), timeo_);		       \
423 		if (rc_)						       \
424 			break;						       \
425 		add_wait_queue(sk_sleep(sk_), &wait_);                         \
426 		release_sock(sk_);					       \
427 		*(timeo_) = wait_woken(&wait_, TASK_INTERRUPTIBLE, *(timeo_)); \
428 		sched_annotate_sleep();				               \
429 		lock_sock(sk_);						       \
430 		remove_wait_queue(sk_sleep(sk_), &wait_);		       \
431 	}								       \
432 	rc_;								       \
433 })
434 
435 /**
436  * tipc_sk_create - create a TIPC socket
437  * @net: network namespace (must be default network)
438  * @sock: pre-allocated socket structure
439  * @protocol: protocol indicator (must be 0)
440  * @kern: caused by kernel or by userspace?
441  *
442  * This routine creates additional data structures used by the TIPC socket,
443  * initializes them, and links them together.
444  *
445  * Returns 0 on success, errno otherwise
446  */
447 static int tipc_sk_create(struct net *net, struct socket *sock,
448 			  int protocol, int kern)
449 {
450 	const struct proto_ops *ops;
451 	struct sock *sk;
452 	struct tipc_sock *tsk;
453 	struct tipc_msg *msg;
454 
455 	/* Validate arguments */
456 	if (unlikely(protocol != 0))
457 		return -EPROTONOSUPPORT;
458 
459 	switch (sock->type) {
460 	case SOCK_STREAM:
461 		ops = &stream_ops;
462 		break;
463 	case SOCK_SEQPACKET:
464 		ops = &packet_ops;
465 		break;
466 	case SOCK_DGRAM:
467 	case SOCK_RDM:
468 		ops = &msg_ops;
469 		break;
470 	default:
471 		return -EPROTOTYPE;
472 	}
473 
474 	/* Allocate socket's protocol area */
475 	sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto, kern);
476 	if (sk == NULL)
477 		return -ENOMEM;
478 
479 	tsk = tipc_sk(sk);
480 	tsk->max_pkt = MAX_PKT_DEFAULT;
481 	tsk->maxnagle = 0;
482 	tsk->nagle_start = NAGLE_START_INIT;
483 	INIT_LIST_HEAD(&tsk->publications);
484 	INIT_LIST_HEAD(&tsk->cong_links);
485 	msg = &tsk->phdr;
486 
487 	/* Finish initializing socket data structures */
488 	sock->ops = ops;
489 	sock_init_data(sock, sk);
490 	tipc_set_sk_state(sk, TIPC_OPEN);
491 	if (tipc_sk_insert(tsk)) {
492 		pr_warn("Socket create failed; port number exhausted\n");
493 		return -EINVAL;
494 	}
495 
496 	/* Ensure tsk is visible before we read own_addr. */
497 	smp_mb();
498 
499 	tipc_msg_init(tipc_own_addr(net), msg, TIPC_LOW_IMPORTANCE,
500 		      TIPC_NAMED_MSG, NAMED_H_SIZE, 0);
501 
502 	msg_set_origport(msg, tsk->portid);
503 	timer_setup(&sk->sk_timer, tipc_sk_timeout, 0);
504 	sk->sk_shutdown = 0;
505 	sk->sk_backlog_rcv = tipc_sk_backlog_rcv;
506 	sk->sk_rcvbuf = sysctl_tipc_rmem[1];
507 	sk->sk_data_ready = tipc_data_ready;
508 	sk->sk_write_space = tipc_write_space;
509 	sk->sk_destruct = tipc_sock_destruct;
510 	tsk->conn_timeout = CONN_TIMEOUT_DEFAULT;
511 	tsk->group_is_open = true;
512 	atomic_set(&tsk->dupl_rcvcnt, 0);
513 
514 	/* Start out with safe limits until we receive an advertised window */
515 	tsk->snd_win = tsk_adv_blocks(RCVBUF_MIN);
516 	tsk->rcv_win = tsk->snd_win;
517 
518 	if (tipc_sk_type_connectionless(sk)) {
519 		tsk_set_unreturnable(tsk, true);
520 		if (sock->type == SOCK_DGRAM)
521 			tsk_set_unreliable(tsk, true);
522 	}
523 	__skb_queue_head_init(&tsk->mc_method.deferredq);
524 	trace_tipc_sk_create(sk, NULL, TIPC_DUMP_NONE, " ");
525 	return 0;
526 }
527 
528 static void tipc_sk_callback(struct rcu_head *head)
529 {
530 	struct tipc_sock *tsk = container_of(head, struct tipc_sock, rcu);
531 
532 	sock_put(&tsk->sk);
533 }
534 
535 /* Caller should hold socket lock for the socket. */
536 static void __tipc_shutdown(struct socket *sock, int error)
537 {
538 	struct sock *sk = sock->sk;
539 	struct tipc_sock *tsk = tipc_sk(sk);
540 	struct net *net = sock_net(sk);
541 	long timeout = msecs_to_jiffies(CONN_TIMEOUT_DEFAULT);
542 	u32 dnode = tsk_peer_node(tsk);
543 	struct sk_buff *skb;
544 
545 	/* Avoid that hi-prio shutdown msgs bypass msgs in link wakeup queue */
546 	tipc_wait_for_cond(sock, &timeout, (!tsk->cong_link_cnt &&
547 					    !tsk_conn_cong(tsk)));
548 
549 	/* Push out delayed messages if in Nagle mode */
550 	tipc_sk_push_backlog(tsk, false);
551 	/* Remove pending SYN */
552 	__skb_queue_purge(&sk->sk_write_queue);
553 
554 	/* Remove partially received buffer if any */
555 	skb = skb_peek(&sk->sk_receive_queue);
556 	if (skb && TIPC_SKB_CB(skb)->bytes_read) {
557 		__skb_unlink(skb, &sk->sk_receive_queue);
558 		kfree_skb(skb);
559 	}
560 
561 	/* Reject all unreceived messages if connectionless */
562 	if (tipc_sk_type_connectionless(sk)) {
563 		tsk_rej_rx_queue(sk, error);
564 		return;
565 	}
566 
567 	switch (sk->sk_state) {
568 	case TIPC_CONNECTING:
569 	case TIPC_ESTABLISHED:
570 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
571 		tipc_node_remove_conn(net, dnode, tsk->portid);
572 		/* Send a FIN+/- to its peer */
573 		skb = __skb_dequeue(&sk->sk_receive_queue);
574 		if (skb) {
575 			__skb_queue_purge(&sk->sk_receive_queue);
576 			tipc_sk_respond(sk, skb, error);
577 			break;
578 		}
579 		skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE,
580 				      TIPC_CONN_MSG, SHORT_H_SIZE, 0, dnode,
581 				      tsk_own_node(tsk), tsk_peer_port(tsk),
582 				      tsk->portid, error);
583 		if (skb)
584 			tipc_node_xmit_skb(net, skb, dnode, tsk->portid);
585 		break;
586 	case TIPC_LISTEN:
587 		/* Reject all SYN messages */
588 		tsk_rej_rx_queue(sk, error);
589 		break;
590 	default:
591 		__skb_queue_purge(&sk->sk_receive_queue);
592 		break;
593 	}
594 }
595 
596 /**
597  * tipc_release - destroy a TIPC socket
598  * @sock: socket to destroy
599  *
600  * This routine cleans up any messages that are still queued on the socket.
601  * For DGRAM and RDM socket types, all queued messages are rejected.
602  * For SEQPACKET and STREAM socket types, the first message is rejected
603  * and any others are discarded.  (If the first message on a STREAM socket
604  * is partially-read, it is discarded and the next one is rejected instead.)
605  *
606  * NOTE: Rejected messages are not necessarily returned to the sender!  They
607  * are returned or discarded according to the "destination droppable" setting
608  * specified for the message by the sender.
609  *
610  * Returns 0 on success, errno otherwise
611  */
612 static int tipc_release(struct socket *sock)
613 {
614 	struct sock *sk = sock->sk;
615 	struct tipc_sock *tsk;
616 
617 	/*
618 	 * Exit if socket isn't fully initialized (occurs when a failed accept()
619 	 * releases a pre-allocated child socket that was never used)
620 	 */
621 	if (sk == NULL)
622 		return 0;
623 
624 	tsk = tipc_sk(sk);
625 	lock_sock(sk);
626 
627 	trace_tipc_sk_release(sk, NULL, TIPC_DUMP_ALL, " ");
628 	__tipc_shutdown(sock, TIPC_ERR_NO_PORT);
629 	sk->sk_shutdown = SHUTDOWN_MASK;
630 	tipc_sk_leave(tsk);
631 	tipc_sk_withdraw(tsk, 0, NULL);
632 	__skb_queue_purge(&tsk->mc_method.deferredq);
633 	sk_stop_timer(sk, &sk->sk_timer);
634 	tipc_sk_remove(tsk);
635 
636 	sock_orphan(sk);
637 	/* Reject any messages that accumulated in backlog queue */
638 	release_sock(sk);
639 	tipc_dest_list_purge(&tsk->cong_links);
640 	tsk->cong_link_cnt = 0;
641 	call_rcu(&tsk->rcu, tipc_sk_callback);
642 	sock->sk = NULL;
643 
644 	return 0;
645 }
646 
647 /**
648  * tipc_bind - associate or disassocate TIPC name(s) with a socket
649  * @sock: socket structure
650  * @uaddr: socket address describing name(s) and desired operation
651  * @uaddr_len: size of socket address data structure
652  *
653  * Name and name sequence binding is indicated using a positive scope value;
654  * a negative scope value unbinds the specified name.  Specifying no name
655  * (i.e. a socket address length of 0) unbinds all names from the socket.
656  *
657  * Returns 0 on success, errno otherwise
658  *
659  * NOTE: This routine doesn't need to take the socket lock since it doesn't
660  *       access any non-constant socket information.
661  */
662 static int tipc_bind(struct socket *sock, struct sockaddr *uaddr,
663 		     int uaddr_len)
664 {
665 	struct sock *sk = sock->sk;
666 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
667 	struct tipc_sock *tsk = tipc_sk(sk);
668 	int res = -EINVAL;
669 
670 	lock_sock(sk);
671 	if (unlikely(!uaddr_len)) {
672 		res = tipc_sk_withdraw(tsk, 0, NULL);
673 		goto exit;
674 	}
675 	if (tsk->group) {
676 		res = -EACCES;
677 		goto exit;
678 	}
679 	if (uaddr_len < sizeof(struct sockaddr_tipc)) {
680 		res = -EINVAL;
681 		goto exit;
682 	}
683 	if (addr->family != AF_TIPC) {
684 		res = -EAFNOSUPPORT;
685 		goto exit;
686 	}
687 
688 	if (addr->addrtype == TIPC_ADDR_NAME)
689 		addr->addr.nameseq.upper = addr->addr.nameseq.lower;
690 	else if (addr->addrtype != TIPC_ADDR_NAMESEQ) {
691 		res = -EAFNOSUPPORT;
692 		goto exit;
693 	}
694 
695 	if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) &&
696 	    (addr->addr.nameseq.type != TIPC_TOP_SRV) &&
697 	    (addr->addr.nameseq.type != TIPC_CFG_SRV)) {
698 		res = -EACCES;
699 		goto exit;
700 	}
701 
702 	res = (addr->scope >= 0) ?
703 		tipc_sk_publish(tsk, addr->scope, &addr->addr.nameseq) :
704 		tipc_sk_withdraw(tsk, -addr->scope, &addr->addr.nameseq);
705 exit:
706 	release_sock(sk);
707 	return res;
708 }
709 
710 /**
711  * tipc_getname - get port ID of socket or peer socket
712  * @sock: socket structure
713  * @uaddr: area for returned socket address
714  * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
715  *
716  * Returns 0 on success, errno otherwise
717  *
718  * NOTE: This routine doesn't need to take the socket lock since it only
719  *       accesses socket information that is unchanging (or which changes in
720  *       a completely predictable manner).
721  */
722 static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
723 			int peer)
724 {
725 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
726 	struct sock *sk = sock->sk;
727 	struct tipc_sock *tsk = tipc_sk(sk);
728 
729 	memset(addr, 0, sizeof(*addr));
730 	if (peer) {
731 		if ((!tipc_sk_connected(sk)) &&
732 		    ((peer != 2) || (sk->sk_state != TIPC_DISCONNECTING)))
733 			return -ENOTCONN;
734 		addr->addr.id.ref = tsk_peer_port(tsk);
735 		addr->addr.id.node = tsk_peer_node(tsk);
736 	} else {
737 		addr->addr.id.ref = tsk->portid;
738 		addr->addr.id.node = tipc_own_addr(sock_net(sk));
739 	}
740 
741 	addr->addrtype = TIPC_ADDR_ID;
742 	addr->family = AF_TIPC;
743 	addr->scope = 0;
744 	addr->addr.name.domain = 0;
745 
746 	return sizeof(*addr);
747 }
748 
749 /**
750  * tipc_poll - read and possibly block on pollmask
751  * @file: file structure associated with the socket
752  * @sock: socket for which to calculate the poll bits
753  * @wait: ???
754  *
755  * Returns pollmask value
756  *
757  * COMMENTARY:
758  * It appears that the usual socket locking mechanisms are not useful here
759  * since the pollmask info is potentially out-of-date the moment this routine
760  * exits.  TCP and other protocols seem to rely on higher level poll routines
761  * to handle any preventable race conditions, so TIPC will do the same ...
762  *
763  * IMPORTANT: The fact that a read or write operation is indicated does NOT
764  * imply that the operation will succeed, merely that it should be performed
765  * and will not block.
766  */
767 static __poll_t tipc_poll(struct file *file, struct socket *sock,
768 			      poll_table *wait)
769 {
770 	struct sock *sk = sock->sk;
771 	struct tipc_sock *tsk = tipc_sk(sk);
772 	__poll_t revents = 0;
773 
774 	sock_poll_wait(file, sock, wait);
775 	trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " ");
776 
777 	if (sk->sk_shutdown & RCV_SHUTDOWN)
778 		revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
779 	if (sk->sk_shutdown == SHUTDOWN_MASK)
780 		revents |= EPOLLHUP;
781 
782 	switch (sk->sk_state) {
783 	case TIPC_ESTABLISHED:
784 		if (!tsk->cong_link_cnt && !tsk_conn_cong(tsk))
785 			revents |= EPOLLOUT;
786 		/* fall through */
787 	case TIPC_LISTEN:
788 	case TIPC_CONNECTING:
789 		if (!skb_queue_empty_lockless(&sk->sk_receive_queue))
790 			revents |= EPOLLIN | EPOLLRDNORM;
791 		break;
792 	case TIPC_OPEN:
793 		if (tsk->group_is_open && !tsk->cong_link_cnt)
794 			revents |= EPOLLOUT;
795 		if (!tipc_sk_type_connectionless(sk))
796 			break;
797 		if (skb_queue_empty_lockless(&sk->sk_receive_queue))
798 			break;
799 		revents |= EPOLLIN | EPOLLRDNORM;
800 		break;
801 	case TIPC_DISCONNECTING:
802 		revents = EPOLLIN | EPOLLRDNORM | EPOLLHUP;
803 		break;
804 	}
805 	return revents;
806 }
807 
808 /**
809  * tipc_sendmcast - send multicast message
810  * @sock: socket structure
811  * @seq: destination address
812  * @msg: message to send
813  * @dlen: length of data to send
814  * @timeout: timeout to wait for wakeup
815  *
816  * Called from function tipc_sendmsg(), which has done all sanity checks
817  * Returns the number of bytes sent on success, or errno
818  */
819 static int tipc_sendmcast(struct  socket *sock, struct tipc_name_seq *seq,
820 			  struct msghdr *msg, size_t dlen, long timeout)
821 {
822 	struct sock *sk = sock->sk;
823 	struct tipc_sock *tsk = tipc_sk(sk);
824 	struct tipc_msg *hdr = &tsk->phdr;
825 	struct net *net = sock_net(sk);
826 	int mtu = tipc_bcast_get_mtu(net);
827 	struct tipc_mc_method *method = &tsk->mc_method;
828 	struct sk_buff_head pkts;
829 	struct tipc_nlist dsts;
830 	int rc;
831 
832 	if (tsk->group)
833 		return -EACCES;
834 
835 	/* Block or return if any destination link is congested */
836 	rc = tipc_wait_for_cond(sock, &timeout, !tsk->cong_link_cnt);
837 	if (unlikely(rc))
838 		return rc;
839 
840 	/* Lookup destination nodes */
841 	tipc_nlist_init(&dsts, tipc_own_addr(net));
842 	tipc_nametbl_lookup_dst_nodes(net, seq->type, seq->lower,
843 				      seq->upper, &dsts);
844 	if (!dsts.local && !dsts.remote)
845 		return -EHOSTUNREACH;
846 
847 	/* Build message header */
848 	msg_set_type(hdr, TIPC_MCAST_MSG);
849 	msg_set_hdr_sz(hdr, MCAST_H_SIZE);
850 	msg_set_lookup_scope(hdr, TIPC_CLUSTER_SCOPE);
851 	msg_set_destport(hdr, 0);
852 	msg_set_destnode(hdr, 0);
853 	msg_set_nametype(hdr, seq->type);
854 	msg_set_namelower(hdr, seq->lower);
855 	msg_set_nameupper(hdr, seq->upper);
856 
857 	/* Build message as chain of buffers */
858 	__skb_queue_head_init(&pkts);
859 	rc = tipc_msg_build(hdr, msg, 0, dlen, mtu, &pkts);
860 
861 	/* Send message if build was successful */
862 	if (unlikely(rc == dlen)) {
863 		trace_tipc_sk_sendmcast(sk, skb_peek(&pkts),
864 					TIPC_DUMP_SK_SNDQ, " ");
865 		rc = tipc_mcast_xmit(net, &pkts, method, &dsts,
866 				     &tsk->cong_link_cnt);
867 	}
868 
869 	tipc_nlist_purge(&dsts);
870 
871 	return rc ? rc : dlen;
872 }
873 
874 /**
875  * tipc_send_group_msg - send a message to a member in the group
876  * @net: network namespace
877  * @m: message to send
878  * @mb: group member
879  * @dnode: destination node
880  * @dport: destination port
881  * @dlen: total length of message data
882  */
883 static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
884 			       struct msghdr *m, struct tipc_member *mb,
885 			       u32 dnode, u32 dport, int dlen)
886 {
887 	u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group);
888 	struct tipc_mc_method *method = &tsk->mc_method;
889 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
890 	struct tipc_msg *hdr = &tsk->phdr;
891 	struct sk_buff_head pkts;
892 	int mtu, rc;
893 
894 	/* Complete message header */
895 	msg_set_type(hdr, TIPC_GRP_UCAST_MSG);
896 	msg_set_hdr_sz(hdr, GROUP_H_SIZE);
897 	msg_set_destport(hdr, dport);
898 	msg_set_destnode(hdr, dnode);
899 	msg_set_grp_bc_seqno(hdr, bc_snd_nxt);
900 
901 	/* Build message as chain of buffers */
902 	__skb_queue_head_init(&pkts);
903 	mtu = tipc_node_get_mtu(net, dnode, tsk->portid, false);
904 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
905 	if (unlikely(rc != dlen))
906 		return rc;
907 
908 	/* Send message */
909 	rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
910 	if (unlikely(rc == -ELINKCONG)) {
911 		tipc_dest_push(&tsk->cong_links, dnode, 0);
912 		tsk->cong_link_cnt++;
913 	}
914 
915 	/* Update send window */
916 	tipc_group_update_member(mb, blks);
917 
918 	/* A broadcast sent within next EXPIRE period must follow same path */
919 	method->rcast = true;
920 	method->mandatory = true;
921 	return dlen;
922 }
923 
924 /**
925  * tipc_send_group_unicast - send message to a member in the group
926  * @sock: socket structure
927  * @m: message to send
928  * @dlen: total length of message data
929  * @timeout: timeout to wait for wakeup
930  *
931  * Called from function tipc_sendmsg(), which has done all sanity checks
932  * Returns the number of bytes sent on success, or errno
933  */
934 static int tipc_send_group_unicast(struct socket *sock, struct msghdr *m,
935 				   int dlen, long timeout)
936 {
937 	struct sock *sk = sock->sk;
938 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
939 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
940 	struct tipc_sock *tsk = tipc_sk(sk);
941 	struct net *net = sock_net(sk);
942 	struct tipc_member *mb = NULL;
943 	u32 node, port;
944 	int rc;
945 
946 	node = dest->addr.id.node;
947 	port = dest->addr.id.ref;
948 	if (!port && !node)
949 		return -EHOSTUNREACH;
950 
951 	/* Block or return if destination link or member is congested */
952 	rc = tipc_wait_for_cond(sock, &timeout,
953 				!tipc_dest_find(&tsk->cong_links, node, 0) &&
954 				tsk->group &&
955 				!tipc_group_cong(tsk->group, node, port, blks,
956 						 &mb));
957 	if (unlikely(rc))
958 		return rc;
959 
960 	if (unlikely(!mb))
961 		return -EHOSTUNREACH;
962 
963 	rc = tipc_send_group_msg(net, tsk, m, mb, node, port, dlen);
964 
965 	return rc ? rc : dlen;
966 }
967 
968 /**
969  * tipc_send_group_anycast - send message to any member with given identity
970  * @sock: socket structure
971  * @m: message to send
972  * @dlen: total length of message data
973  * @timeout: timeout to wait for wakeup
974  *
975  * Called from function tipc_sendmsg(), which has done all sanity checks
976  * Returns the number of bytes sent on success, or errno
977  */
978 static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m,
979 				   int dlen, long timeout)
980 {
981 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
982 	struct sock *sk = sock->sk;
983 	struct tipc_sock *tsk = tipc_sk(sk);
984 	struct list_head *cong_links = &tsk->cong_links;
985 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
986 	struct tipc_msg *hdr = &tsk->phdr;
987 	struct tipc_member *first = NULL;
988 	struct tipc_member *mbr = NULL;
989 	struct net *net = sock_net(sk);
990 	u32 node, port, exclude;
991 	struct list_head dsts;
992 	u32 type, inst, scope;
993 	int lookups = 0;
994 	int dstcnt, rc;
995 	bool cong;
996 
997 	INIT_LIST_HEAD(&dsts);
998 
999 	type = msg_nametype(hdr);
1000 	inst = dest->addr.name.name.instance;
1001 	scope = msg_lookup_scope(hdr);
1002 
1003 	while (++lookups < 4) {
1004 		exclude = tipc_group_exclude(tsk->group);
1005 
1006 		first = NULL;
1007 
1008 		/* Look for a non-congested destination member, if any */
1009 		while (1) {
1010 			if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
1011 						 &dstcnt, exclude, false))
1012 				return -EHOSTUNREACH;
1013 			tipc_dest_pop(&dsts, &node, &port);
1014 			cong = tipc_group_cong(tsk->group, node, port, blks,
1015 					       &mbr);
1016 			if (!cong)
1017 				break;
1018 			if (mbr == first)
1019 				break;
1020 			if (!first)
1021 				first = mbr;
1022 		}
1023 
1024 		/* Start over if destination was not in member list */
1025 		if (unlikely(!mbr))
1026 			continue;
1027 
1028 		if (likely(!cong && !tipc_dest_find(cong_links, node, 0)))
1029 			break;
1030 
1031 		/* Block or return if destination link or member is congested */
1032 		rc = tipc_wait_for_cond(sock, &timeout,
1033 					!tipc_dest_find(cong_links, node, 0) &&
1034 					tsk->group &&
1035 					!tipc_group_cong(tsk->group, node, port,
1036 							 blks, &mbr));
1037 		if (unlikely(rc))
1038 			return rc;
1039 
1040 		/* Send, unless destination disappeared while waiting */
1041 		if (likely(mbr))
1042 			break;
1043 	}
1044 
1045 	if (unlikely(lookups >= 4))
1046 		return -EHOSTUNREACH;
1047 
1048 	rc = tipc_send_group_msg(net, tsk, m, mbr, node, port, dlen);
1049 
1050 	return rc ? rc : dlen;
1051 }
1052 
1053 /**
1054  * tipc_send_group_bcast - send message to all members in communication group
1055  * @sock: socket structure
1056  * @m: message to send
1057  * @dlen: total length of message data
1058  * @timeout: timeout to wait for wakeup
1059  *
1060  * Called from function tipc_sendmsg(), which has done all sanity checks
1061  * Returns the number of bytes sent on success, or errno
1062  */
1063 static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
1064 				 int dlen, long timeout)
1065 {
1066 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1067 	struct sock *sk = sock->sk;
1068 	struct net *net = sock_net(sk);
1069 	struct tipc_sock *tsk = tipc_sk(sk);
1070 	struct tipc_nlist *dsts;
1071 	struct tipc_mc_method *method = &tsk->mc_method;
1072 	bool ack = method->mandatory && method->rcast;
1073 	int blks = tsk_blocks(MCAST_H_SIZE + dlen);
1074 	struct tipc_msg *hdr = &tsk->phdr;
1075 	int mtu = tipc_bcast_get_mtu(net);
1076 	struct sk_buff_head pkts;
1077 	int rc = -EHOSTUNREACH;
1078 
1079 	/* Block or return if any destination link or member is congested */
1080 	rc = tipc_wait_for_cond(sock, &timeout,
1081 				!tsk->cong_link_cnt && tsk->group &&
1082 				!tipc_group_bc_cong(tsk->group, blks));
1083 	if (unlikely(rc))
1084 		return rc;
1085 
1086 	dsts = tipc_group_dests(tsk->group);
1087 	if (!dsts->local && !dsts->remote)
1088 		return -EHOSTUNREACH;
1089 
1090 	/* Complete message header */
1091 	if (dest) {
1092 		msg_set_type(hdr, TIPC_GRP_MCAST_MSG);
1093 		msg_set_nameinst(hdr, dest->addr.name.name.instance);
1094 	} else {
1095 		msg_set_type(hdr, TIPC_GRP_BCAST_MSG);
1096 		msg_set_nameinst(hdr, 0);
1097 	}
1098 	msg_set_hdr_sz(hdr, GROUP_H_SIZE);
1099 	msg_set_destport(hdr, 0);
1100 	msg_set_destnode(hdr, 0);
1101 	msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(tsk->group));
1102 
1103 	/* Avoid getting stuck with repeated forced replicasts */
1104 	msg_set_grp_bc_ack_req(hdr, ack);
1105 
1106 	/* Build message as chain of buffers */
1107 	__skb_queue_head_init(&pkts);
1108 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1109 	if (unlikely(rc != dlen))
1110 		return rc;
1111 
1112 	/* Send message */
1113 	rc = tipc_mcast_xmit(net, &pkts, method, dsts, &tsk->cong_link_cnt);
1114 	if (unlikely(rc))
1115 		return rc;
1116 
1117 	/* Update broadcast sequence number and send windows */
1118 	tipc_group_update_bc_members(tsk->group, blks, ack);
1119 
1120 	/* Broadcast link is now free to choose method for next broadcast */
1121 	method->mandatory = false;
1122 	method->expires = jiffies;
1123 
1124 	return dlen;
1125 }
1126 
1127 /**
1128  * tipc_send_group_mcast - send message to all members with given identity
1129  * @sock: socket structure
1130  * @m: message to send
1131  * @dlen: total length of message data
1132  * @timeout: timeout to wait for wakeup
1133  *
1134  * Called from function tipc_sendmsg(), which has done all sanity checks
1135  * Returns the number of bytes sent on success, or errno
1136  */
1137 static int tipc_send_group_mcast(struct socket *sock, struct msghdr *m,
1138 				 int dlen, long timeout)
1139 {
1140 	struct sock *sk = sock->sk;
1141 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1142 	struct tipc_sock *tsk = tipc_sk(sk);
1143 	struct tipc_group *grp = tsk->group;
1144 	struct tipc_msg *hdr = &tsk->phdr;
1145 	struct net *net = sock_net(sk);
1146 	u32 type, inst, scope, exclude;
1147 	struct list_head dsts;
1148 	u32 dstcnt;
1149 
1150 	INIT_LIST_HEAD(&dsts);
1151 
1152 	type = msg_nametype(hdr);
1153 	inst = dest->addr.name.name.instance;
1154 	scope = msg_lookup_scope(hdr);
1155 	exclude = tipc_group_exclude(grp);
1156 
1157 	if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
1158 				 &dstcnt, exclude, true))
1159 		return -EHOSTUNREACH;
1160 
1161 	if (dstcnt == 1) {
1162 		tipc_dest_pop(&dsts, &dest->addr.id.node, &dest->addr.id.ref);
1163 		return tipc_send_group_unicast(sock, m, dlen, timeout);
1164 	}
1165 
1166 	tipc_dest_list_purge(&dsts);
1167 	return tipc_send_group_bcast(sock, m, dlen, timeout);
1168 }
1169 
1170 /**
1171  * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
1172  * @arrvq: queue with arriving messages, to be cloned after destination lookup
1173  * @inputq: queue with cloned messages, delivered to socket after dest lookup
1174  *
1175  * Multi-threaded: parallel calls with reference to same queues may occur
1176  */
1177 void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
1178 		       struct sk_buff_head *inputq)
1179 {
1180 	u32 self = tipc_own_addr(net);
1181 	u32 type, lower, upper, scope;
1182 	struct sk_buff *skb, *_skb;
1183 	u32 portid, onode;
1184 	struct sk_buff_head tmpq;
1185 	struct list_head dports;
1186 	struct tipc_msg *hdr;
1187 	int user, mtyp, hlen;
1188 	bool exact;
1189 
1190 	__skb_queue_head_init(&tmpq);
1191 	INIT_LIST_HEAD(&dports);
1192 
1193 	skb = tipc_skb_peek(arrvq, &inputq->lock);
1194 	for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) {
1195 		hdr = buf_msg(skb);
1196 		user = msg_user(hdr);
1197 		mtyp = msg_type(hdr);
1198 		hlen = skb_headroom(skb) + msg_hdr_sz(hdr);
1199 		onode = msg_orignode(hdr);
1200 		type = msg_nametype(hdr);
1201 
1202 		if (mtyp == TIPC_GRP_UCAST_MSG || user == GROUP_PROTOCOL) {
1203 			spin_lock_bh(&inputq->lock);
1204 			if (skb_peek(arrvq) == skb) {
1205 				__skb_dequeue(arrvq);
1206 				__skb_queue_tail(inputq, skb);
1207 			}
1208 			kfree_skb(skb);
1209 			spin_unlock_bh(&inputq->lock);
1210 			continue;
1211 		}
1212 
1213 		/* Group messages require exact scope match */
1214 		if (msg_in_group(hdr)) {
1215 			lower = 0;
1216 			upper = ~0;
1217 			scope = msg_lookup_scope(hdr);
1218 			exact = true;
1219 		} else {
1220 			/* TIPC_NODE_SCOPE means "any scope" in this context */
1221 			if (onode == self)
1222 				scope = TIPC_NODE_SCOPE;
1223 			else
1224 				scope = TIPC_CLUSTER_SCOPE;
1225 			exact = false;
1226 			lower = msg_namelower(hdr);
1227 			upper = msg_nameupper(hdr);
1228 		}
1229 
1230 		/* Create destination port list: */
1231 		tipc_nametbl_mc_lookup(net, type, lower, upper,
1232 				       scope, exact, &dports);
1233 
1234 		/* Clone message per destination */
1235 		while (tipc_dest_pop(&dports, NULL, &portid)) {
1236 			_skb = __pskb_copy(skb, hlen, GFP_ATOMIC);
1237 			if (_skb) {
1238 				msg_set_destport(buf_msg(_skb), portid);
1239 				__skb_queue_tail(&tmpq, _skb);
1240 				continue;
1241 			}
1242 			pr_warn("Failed to clone mcast rcv buffer\n");
1243 		}
1244 		/* Append to inputq if not already done by other thread */
1245 		spin_lock_bh(&inputq->lock);
1246 		if (skb_peek(arrvq) == skb) {
1247 			skb_queue_splice_tail_init(&tmpq, inputq);
1248 			kfree_skb(__skb_dequeue(arrvq));
1249 		}
1250 		spin_unlock_bh(&inputq->lock);
1251 		__skb_queue_purge(&tmpq);
1252 		kfree_skb(skb);
1253 	}
1254 	tipc_sk_rcv(net, inputq);
1255 }
1256 
1257 /* tipc_sk_push_backlog(): send accumulated buffers in socket write queue
1258  *                         when socket is in Nagle mode
1259  */
1260 static void tipc_sk_push_backlog(struct tipc_sock *tsk, bool nagle_ack)
1261 {
1262 	struct sk_buff_head *txq = &tsk->sk.sk_write_queue;
1263 	struct sk_buff *skb = skb_peek_tail(txq);
1264 	struct net *net = sock_net(&tsk->sk);
1265 	u32 dnode = tsk_peer_node(tsk);
1266 	int rc;
1267 
1268 	if (nagle_ack) {
1269 		tsk->pkt_cnt += skb_queue_len(txq);
1270 		if (!tsk->pkt_cnt || tsk->msg_acc / tsk->pkt_cnt < 2) {
1271 			tsk->oneway = 0;
1272 			if (tsk->nagle_start < NAGLE_START_MAX)
1273 				tsk->nagle_start *= 2;
1274 			tsk->expect_ack = false;
1275 			pr_debug("tsk %10u: bad nagle %u -> %u, next start %u!\n",
1276 				 tsk->portid, tsk->msg_acc, tsk->pkt_cnt,
1277 				 tsk->nagle_start);
1278 		} else {
1279 			tsk->nagle_start = NAGLE_START_INIT;
1280 			if (skb) {
1281 				msg_set_ack_required(buf_msg(skb));
1282 				tsk->expect_ack = true;
1283 			} else {
1284 				tsk->expect_ack = false;
1285 			}
1286 		}
1287 		tsk->msg_acc = 0;
1288 		tsk->pkt_cnt = 0;
1289 	}
1290 
1291 	if (!skb || tsk->cong_link_cnt)
1292 		return;
1293 
1294 	/* Do not send SYN again after congestion */
1295 	if (msg_is_syn(buf_msg(skb)))
1296 		return;
1297 
1298 	if (tsk->msg_acc)
1299 		tsk->pkt_cnt += skb_queue_len(txq);
1300 	tsk->snt_unacked += tsk->snd_backlog;
1301 	tsk->snd_backlog = 0;
1302 	rc = tipc_node_xmit(net, txq, dnode, tsk->portid);
1303 	if (rc == -ELINKCONG)
1304 		tsk->cong_link_cnt = 1;
1305 }
1306 
1307 /**
1308  * tipc_sk_conn_proto_rcv - receive a connection mng protocol message
1309  * @tsk: receiving socket
1310  * @skb: pointer to message buffer.
1311  */
1312 static void tipc_sk_conn_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb,
1313 				   struct sk_buff_head *inputq,
1314 				   struct sk_buff_head *xmitq)
1315 {
1316 	struct tipc_msg *hdr = buf_msg(skb);
1317 	u32 onode = tsk_own_node(tsk);
1318 	struct sock *sk = &tsk->sk;
1319 	int mtyp = msg_type(hdr);
1320 	bool was_cong;
1321 
1322 	/* Ignore if connection cannot be validated: */
1323 	if (!tsk_peer_msg(tsk, hdr)) {
1324 		trace_tipc_sk_drop_msg(sk, skb, TIPC_DUMP_NONE, "@proto_rcv!");
1325 		goto exit;
1326 	}
1327 
1328 	if (unlikely(msg_errcode(hdr))) {
1329 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
1330 		tipc_node_remove_conn(sock_net(sk), tsk_peer_node(tsk),
1331 				      tsk_peer_port(tsk));
1332 		sk->sk_state_change(sk);
1333 
1334 		/* State change is ignored if socket already awake,
1335 		 * - convert msg to abort msg and add to inqueue
1336 		 */
1337 		msg_set_user(hdr, TIPC_CRITICAL_IMPORTANCE);
1338 		msg_set_type(hdr, TIPC_CONN_MSG);
1339 		msg_set_size(hdr, BASIC_H_SIZE);
1340 		msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1341 		__skb_queue_tail(inputq, skb);
1342 		return;
1343 	}
1344 
1345 	tsk->probe_unacked = false;
1346 
1347 	if (mtyp == CONN_PROBE) {
1348 		msg_set_type(hdr, CONN_PROBE_REPLY);
1349 		if (tipc_msg_reverse(onode, &skb, TIPC_OK))
1350 			__skb_queue_tail(xmitq, skb);
1351 		return;
1352 	} else if (mtyp == CONN_ACK) {
1353 		was_cong = tsk_conn_cong(tsk);
1354 		tipc_sk_push_backlog(tsk, msg_nagle_ack(hdr));
1355 		tsk->snt_unacked -= msg_conn_ack(hdr);
1356 		if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1357 			tsk->snd_win = msg_adv_win(hdr);
1358 		if (was_cong && !tsk_conn_cong(tsk))
1359 			sk->sk_write_space(sk);
1360 	} else if (mtyp != CONN_PROBE_REPLY) {
1361 		pr_warn("Received unknown CONN_PROTO msg\n");
1362 	}
1363 exit:
1364 	kfree_skb(skb);
1365 }
1366 
1367 /**
1368  * tipc_sendmsg - send message in connectionless manner
1369  * @sock: socket structure
1370  * @m: message to send
1371  * @dsz: amount of user data to be sent
1372  *
1373  * Message must have an destination specified explicitly.
1374  * Used for SOCK_RDM and SOCK_DGRAM messages,
1375  * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
1376  * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
1377  *
1378  * Returns the number of bytes sent on success, or errno otherwise
1379  */
1380 static int tipc_sendmsg(struct socket *sock,
1381 			struct msghdr *m, size_t dsz)
1382 {
1383 	struct sock *sk = sock->sk;
1384 	int ret;
1385 
1386 	lock_sock(sk);
1387 	ret = __tipc_sendmsg(sock, m, dsz);
1388 	release_sock(sk);
1389 
1390 	return ret;
1391 }
1392 
1393 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)
1394 {
1395 	struct sock *sk = sock->sk;
1396 	struct net *net = sock_net(sk);
1397 	struct tipc_sock *tsk = tipc_sk(sk);
1398 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1399 	long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1400 	struct list_head *clinks = &tsk->cong_links;
1401 	bool syn = !tipc_sk_type_connectionless(sk);
1402 	struct tipc_group *grp = tsk->group;
1403 	struct tipc_msg *hdr = &tsk->phdr;
1404 	struct tipc_name_seq *seq;
1405 	struct sk_buff_head pkts;
1406 	u32 dport = 0, dnode = 0;
1407 	u32 type = 0, inst = 0;
1408 	int mtu, rc;
1409 
1410 	if (unlikely(dlen > TIPC_MAX_USER_MSG_SIZE))
1411 		return -EMSGSIZE;
1412 
1413 	if (likely(dest)) {
1414 		if (unlikely(m->msg_namelen < sizeof(*dest)))
1415 			return -EINVAL;
1416 		if (unlikely(dest->family != AF_TIPC))
1417 			return -EINVAL;
1418 	}
1419 
1420 	if (grp) {
1421 		if (!dest)
1422 			return tipc_send_group_bcast(sock, m, dlen, timeout);
1423 		if (dest->addrtype == TIPC_ADDR_NAME)
1424 			return tipc_send_group_anycast(sock, m, dlen, timeout);
1425 		if (dest->addrtype == TIPC_ADDR_ID)
1426 			return tipc_send_group_unicast(sock, m, dlen, timeout);
1427 		if (dest->addrtype == TIPC_ADDR_MCAST)
1428 			return tipc_send_group_mcast(sock, m, dlen, timeout);
1429 		return -EINVAL;
1430 	}
1431 
1432 	if (unlikely(!dest)) {
1433 		dest = &tsk->peer;
1434 		if (!syn && dest->family != AF_TIPC)
1435 			return -EDESTADDRREQ;
1436 	}
1437 
1438 	if (unlikely(syn)) {
1439 		if (sk->sk_state == TIPC_LISTEN)
1440 			return -EPIPE;
1441 		if (sk->sk_state != TIPC_OPEN)
1442 			return -EISCONN;
1443 		if (tsk->published)
1444 			return -EOPNOTSUPP;
1445 		if (dest->addrtype == TIPC_ADDR_NAME) {
1446 			tsk->conn_type = dest->addr.name.name.type;
1447 			tsk->conn_instance = dest->addr.name.name.instance;
1448 		}
1449 		msg_set_syn(hdr, 1);
1450 	}
1451 
1452 	seq = &dest->addr.nameseq;
1453 	if (dest->addrtype == TIPC_ADDR_MCAST)
1454 		return tipc_sendmcast(sock, seq, m, dlen, timeout);
1455 
1456 	if (dest->addrtype == TIPC_ADDR_NAME) {
1457 		type = dest->addr.name.name.type;
1458 		inst = dest->addr.name.name.instance;
1459 		dnode = dest->addr.name.domain;
1460 		dport = tipc_nametbl_translate(net, type, inst, &dnode);
1461 		if (unlikely(!dport && !dnode))
1462 			return -EHOSTUNREACH;
1463 	} else if (dest->addrtype == TIPC_ADDR_ID) {
1464 		dnode = dest->addr.id.node;
1465 	} else {
1466 		return -EINVAL;
1467 	}
1468 
1469 	/* Block or return if destination link is congested */
1470 	rc = tipc_wait_for_cond(sock, &timeout,
1471 				!tipc_dest_find(clinks, dnode, 0));
1472 	if (unlikely(rc))
1473 		return rc;
1474 
1475 	if (dest->addrtype == TIPC_ADDR_NAME) {
1476 		msg_set_type(hdr, TIPC_NAMED_MSG);
1477 		msg_set_hdr_sz(hdr, NAMED_H_SIZE);
1478 		msg_set_nametype(hdr, type);
1479 		msg_set_nameinst(hdr, inst);
1480 		msg_set_lookup_scope(hdr, tipc_node2scope(dnode));
1481 		msg_set_destnode(hdr, dnode);
1482 		msg_set_destport(hdr, dport);
1483 	} else { /* TIPC_ADDR_ID */
1484 		msg_set_type(hdr, TIPC_DIRECT_MSG);
1485 		msg_set_lookup_scope(hdr, 0);
1486 		msg_set_destnode(hdr, dnode);
1487 		msg_set_destport(hdr, dest->addr.id.ref);
1488 		msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1489 	}
1490 
1491 	__skb_queue_head_init(&pkts);
1492 	mtu = tipc_node_get_mtu(net, dnode, tsk->portid, true);
1493 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1494 	if (unlikely(rc != dlen))
1495 		return rc;
1496 	if (unlikely(syn && !tipc_msg_skb_clone(&pkts, &sk->sk_write_queue))) {
1497 		__skb_queue_purge(&pkts);
1498 		return -ENOMEM;
1499 	}
1500 
1501 	trace_tipc_sk_sendmsg(sk, skb_peek(&pkts), TIPC_DUMP_SK_SNDQ, " ");
1502 	rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
1503 	if (unlikely(rc == -ELINKCONG)) {
1504 		tipc_dest_push(clinks, dnode, 0);
1505 		tsk->cong_link_cnt++;
1506 		rc = 0;
1507 	}
1508 
1509 	if (unlikely(syn && !rc))
1510 		tipc_set_sk_state(sk, TIPC_CONNECTING);
1511 
1512 	return rc ? rc : dlen;
1513 }
1514 
1515 /**
1516  * tipc_sendstream - send stream-oriented data
1517  * @sock: socket structure
1518  * @m: data to send
1519  * @dsz: total length of data to be transmitted
1520  *
1521  * Used for SOCK_STREAM data.
1522  *
1523  * Returns the number of bytes sent on success (or partial success),
1524  * or errno if no data sent
1525  */
1526 static int tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz)
1527 {
1528 	struct sock *sk = sock->sk;
1529 	int ret;
1530 
1531 	lock_sock(sk);
1532 	ret = __tipc_sendstream(sock, m, dsz);
1533 	release_sock(sk);
1534 
1535 	return ret;
1536 }
1537 
1538 static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dlen)
1539 {
1540 	struct sock *sk = sock->sk;
1541 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1542 	long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1543 	struct sk_buff_head *txq = &sk->sk_write_queue;
1544 	struct tipc_sock *tsk = tipc_sk(sk);
1545 	struct tipc_msg *hdr = &tsk->phdr;
1546 	struct net *net = sock_net(sk);
1547 	struct sk_buff *skb;
1548 	u32 dnode = tsk_peer_node(tsk);
1549 	int maxnagle = tsk->maxnagle;
1550 	int maxpkt = tsk->max_pkt;
1551 	int send, sent = 0;
1552 	int blocks, rc = 0;
1553 
1554 	if (unlikely(dlen > INT_MAX))
1555 		return -EMSGSIZE;
1556 
1557 	/* Handle implicit connection setup */
1558 	if (unlikely(dest)) {
1559 		rc = __tipc_sendmsg(sock, m, dlen);
1560 		if (dlen && dlen == rc) {
1561 			tsk->peer_caps = tipc_node_get_capabilities(net, dnode);
1562 			tsk->snt_unacked = tsk_inc(tsk, dlen + msg_hdr_sz(hdr));
1563 		}
1564 		return rc;
1565 	}
1566 
1567 	do {
1568 		rc = tipc_wait_for_cond(sock, &timeout,
1569 					(!tsk->cong_link_cnt &&
1570 					 !tsk_conn_cong(tsk) &&
1571 					 tipc_sk_connected(sk)));
1572 		if (unlikely(rc))
1573 			break;
1574 		send = min_t(size_t, dlen - sent, TIPC_MAX_USER_MSG_SIZE);
1575 		blocks = tsk->snd_backlog;
1576 		if (tsk->oneway++ >= tsk->nagle_start && maxnagle &&
1577 		    send <= maxnagle) {
1578 			rc = tipc_msg_append(hdr, m, send, maxnagle, txq);
1579 			if (unlikely(rc < 0))
1580 				break;
1581 			blocks += rc;
1582 			tsk->msg_acc++;
1583 			if (blocks <= 64 && tsk->expect_ack) {
1584 				tsk->snd_backlog = blocks;
1585 				sent += send;
1586 				break;
1587 			} else if (blocks > 64) {
1588 				tsk->pkt_cnt += skb_queue_len(txq);
1589 			} else {
1590 				skb = skb_peek_tail(txq);
1591 				if (skb) {
1592 					msg_set_ack_required(buf_msg(skb));
1593 					tsk->expect_ack = true;
1594 				} else {
1595 					tsk->expect_ack = false;
1596 				}
1597 				tsk->msg_acc = 0;
1598 				tsk->pkt_cnt = 0;
1599 			}
1600 		} else {
1601 			rc = tipc_msg_build(hdr, m, sent, send, maxpkt, txq);
1602 			if (unlikely(rc != send))
1603 				break;
1604 			blocks += tsk_inc(tsk, send + MIN_H_SIZE);
1605 		}
1606 		trace_tipc_sk_sendstream(sk, skb_peek(txq),
1607 					 TIPC_DUMP_SK_SNDQ, " ");
1608 		rc = tipc_node_xmit(net, txq, dnode, tsk->portid);
1609 		if (unlikely(rc == -ELINKCONG)) {
1610 			tsk->cong_link_cnt = 1;
1611 			rc = 0;
1612 		}
1613 		if (likely(!rc)) {
1614 			tsk->snt_unacked += blocks;
1615 			tsk->snd_backlog = 0;
1616 			sent += send;
1617 		}
1618 	} while (sent < dlen && !rc);
1619 
1620 	return sent ? sent : rc;
1621 }
1622 
1623 /**
1624  * tipc_send_packet - send a connection-oriented message
1625  * @sock: socket structure
1626  * @m: message to send
1627  * @dsz: length of data to be transmitted
1628  *
1629  * Used for SOCK_SEQPACKET messages.
1630  *
1631  * Returns the number of bytes sent on success, or errno otherwise
1632  */
1633 static int tipc_send_packet(struct socket *sock, struct msghdr *m, size_t dsz)
1634 {
1635 	if (dsz > TIPC_MAX_USER_MSG_SIZE)
1636 		return -EMSGSIZE;
1637 
1638 	return tipc_sendstream(sock, m, dsz);
1639 }
1640 
1641 /* tipc_sk_finish_conn - complete the setup of a connection
1642  */
1643 static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port,
1644 				u32 peer_node)
1645 {
1646 	struct sock *sk = &tsk->sk;
1647 	struct net *net = sock_net(sk);
1648 	struct tipc_msg *msg = &tsk->phdr;
1649 
1650 	msg_set_syn(msg, 0);
1651 	msg_set_destnode(msg, peer_node);
1652 	msg_set_destport(msg, peer_port);
1653 	msg_set_type(msg, TIPC_CONN_MSG);
1654 	msg_set_lookup_scope(msg, 0);
1655 	msg_set_hdr_sz(msg, SHORT_H_SIZE);
1656 
1657 	sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
1658 	tipc_set_sk_state(sk, TIPC_ESTABLISHED);
1659 	tipc_node_add_conn(net, peer_node, tsk->portid, peer_port);
1660 	tsk->max_pkt = tipc_node_get_mtu(net, peer_node, tsk->portid, true);
1661 	tsk->peer_caps = tipc_node_get_capabilities(net, peer_node);
1662 	tsk_set_nagle(tsk);
1663 	__skb_queue_purge(&sk->sk_write_queue);
1664 	if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1665 		return;
1666 
1667 	/* Fall back to message based flow control */
1668 	tsk->rcv_win = FLOWCTL_MSG_WIN;
1669 	tsk->snd_win = FLOWCTL_MSG_WIN;
1670 }
1671 
1672 /**
1673  * tipc_sk_set_orig_addr - capture sender's address for received message
1674  * @m: descriptor for message info
1675  * @skb: received message
1676  *
1677  * Note: Address is not captured if not requested by receiver.
1678  */
1679 static void tipc_sk_set_orig_addr(struct msghdr *m, struct sk_buff *skb)
1680 {
1681 	DECLARE_SOCKADDR(struct sockaddr_pair *, srcaddr, m->msg_name);
1682 	struct tipc_msg *hdr = buf_msg(skb);
1683 
1684 	if (!srcaddr)
1685 		return;
1686 
1687 	srcaddr->sock.family = AF_TIPC;
1688 	srcaddr->sock.addrtype = TIPC_ADDR_ID;
1689 	srcaddr->sock.scope = 0;
1690 	srcaddr->sock.addr.id.ref = msg_origport(hdr);
1691 	srcaddr->sock.addr.id.node = msg_orignode(hdr);
1692 	srcaddr->sock.addr.name.domain = 0;
1693 	m->msg_namelen = sizeof(struct sockaddr_tipc);
1694 
1695 	if (!msg_in_group(hdr))
1696 		return;
1697 
1698 	/* Group message users may also want to know sending member's id */
1699 	srcaddr->member.family = AF_TIPC;
1700 	srcaddr->member.addrtype = TIPC_ADDR_NAME;
1701 	srcaddr->member.scope = 0;
1702 	srcaddr->member.addr.name.name.type = msg_nametype(hdr);
1703 	srcaddr->member.addr.name.name.instance = TIPC_SKB_CB(skb)->orig_member;
1704 	srcaddr->member.addr.name.domain = 0;
1705 	m->msg_namelen = sizeof(*srcaddr);
1706 }
1707 
1708 /**
1709  * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
1710  * @m: descriptor for message info
1711  * @skb: received message buffer
1712  * @tsk: TIPC port associated with message
1713  *
1714  * Note: Ancillary data is not captured if not requested by receiver.
1715  *
1716  * Returns 0 if successful, otherwise errno
1717  */
1718 static int tipc_sk_anc_data_recv(struct msghdr *m, struct sk_buff *skb,
1719 				 struct tipc_sock *tsk)
1720 {
1721 	struct tipc_msg *msg;
1722 	u32 anc_data[3];
1723 	u32 err;
1724 	u32 dest_type;
1725 	int has_name;
1726 	int res;
1727 
1728 	if (likely(m->msg_controllen == 0))
1729 		return 0;
1730 	msg = buf_msg(skb);
1731 
1732 	/* Optionally capture errored message object(s) */
1733 	err = msg ? msg_errcode(msg) : 0;
1734 	if (unlikely(err)) {
1735 		anc_data[0] = err;
1736 		anc_data[1] = msg_data_sz(msg);
1737 		res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
1738 		if (res)
1739 			return res;
1740 		if (anc_data[1]) {
1741 			if (skb_linearize(skb))
1742 				return -ENOMEM;
1743 			msg = buf_msg(skb);
1744 			res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
1745 				       msg_data(msg));
1746 			if (res)
1747 				return res;
1748 		}
1749 	}
1750 
1751 	/* Optionally capture message destination object */
1752 	dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
1753 	switch (dest_type) {
1754 	case TIPC_NAMED_MSG:
1755 		has_name = 1;
1756 		anc_data[0] = msg_nametype(msg);
1757 		anc_data[1] = msg_namelower(msg);
1758 		anc_data[2] = msg_namelower(msg);
1759 		break;
1760 	case TIPC_MCAST_MSG:
1761 		has_name = 1;
1762 		anc_data[0] = msg_nametype(msg);
1763 		anc_data[1] = msg_namelower(msg);
1764 		anc_data[2] = msg_nameupper(msg);
1765 		break;
1766 	case TIPC_CONN_MSG:
1767 		has_name = (tsk->conn_type != 0);
1768 		anc_data[0] = tsk->conn_type;
1769 		anc_data[1] = tsk->conn_instance;
1770 		anc_data[2] = tsk->conn_instance;
1771 		break;
1772 	default:
1773 		has_name = 0;
1774 	}
1775 	if (has_name) {
1776 		res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
1777 		if (res)
1778 			return res;
1779 	}
1780 
1781 	return 0;
1782 }
1783 
1784 static struct sk_buff *tipc_sk_build_ack(struct tipc_sock *tsk)
1785 {
1786 	struct sock *sk = &tsk->sk;
1787 	struct sk_buff *skb = NULL;
1788 	struct tipc_msg *msg;
1789 	u32 peer_port = tsk_peer_port(tsk);
1790 	u32 dnode = tsk_peer_node(tsk);
1791 
1792 	if (!tipc_sk_connected(sk))
1793 		return NULL;
1794 	skb = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0,
1795 			      dnode, tsk_own_node(tsk), peer_port,
1796 			      tsk->portid, TIPC_OK);
1797 	if (!skb)
1798 		return NULL;
1799 	msg = buf_msg(skb);
1800 	msg_set_conn_ack(msg, tsk->rcv_unacked);
1801 	tsk->rcv_unacked = 0;
1802 
1803 	/* Adjust to and advertize the correct window limit */
1804 	if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) {
1805 		tsk->rcv_win = tsk_adv_blocks(tsk->sk.sk_rcvbuf);
1806 		msg_set_adv_win(msg, tsk->rcv_win);
1807 	}
1808 	return skb;
1809 }
1810 
1811 static void tipc_sk_send_ack(struct tipc_sock *tsk)
1812 {
1813 	struct sk_buff *skb;
1814 
1815 	skb = tipc_sk_build_ack(tsk);
1816 	if (!skb)
1817 		return;
1818 
1819 	tipc_node_xmit_skb(sock_net(&tsk->sk), skb, tsk_peer_node(tsk),
1820 			   msg_link_selector(buf_msg(skb)));
1821 }
1822 
1823 static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop)
1824 {
1825 	struct sock *sk = sock->sk;
1826 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
1827 	long timeo = *timeop;
1828 	int err = sock_error(sk);
1829 
1830 	if (err)
1831 		return err;
1832 
1833 	for (;;) {
1834 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1835 			if (sk->sk_shutdown & RCV_SHUTDOWN) {
1836 				err = -ENOTCONN;
1837 				break;
1838 			}
1839 			add_wait_queue(sk_sleep(sk), &wait);
1840 			release_sock(sk);
1841 			timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
1842 			sched_annotate_sleep();
1843 			lock_sock(sk);
1844 			remove_wait_queue(sk_sleep(sk), &wait);
1845 		}
1846 		err = 0;
1847 		if (!skb_queue_empty(&sk->sk_receive_queue))
1848 			break;
1849 		err = -EAGAIN;
1850 		if (!timeo)
1851 			break;
1852 		err = sock_intr_errno(timeo);
1853 		if (signal_pending(current))
1854 			break;
1855 
1856 		err = sock_error(sk);
1857 		if (err)
1858 			break;
1859 	}
1860 	*timeop = timeo;
1861 	return err;
1862 }
1863 
1864 /**
1865  * tipc_recvmsg - receive packet-oriented message
1866  * @m: descriptor for message info
1867  * @buflen: length of user buffer area
1868  * @flags: receive flags
1869  *
1870  * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1871  * If the complete message doesn't fit in user area, truncate it.
1872  *
1873  * Returns size of returned message data, errno otherwise
1874  */
1875 static int tipc_recvmsg(struct socket *sock, struct msghdr *m,
1876 			size_t buflen,	int flags)
1877 {
1878 	struct sock *sk = sock->sk;
1879 	bool connected = !tipc_sk_type_connectionless(sk);
1880 	struct tipc_sock *tsk = tipc_sk(sk);
1881 	int rc, err, hlen, dlen, copy;
1882 	struct sk_buff_head xmitq;
1883 	struct tipc_msg *hdr;
1884 	struct sk_buff *skb;
1885 	bool grp_evt;
1886 	long timeout;
1887 
1888 	/* Catch invalid receive requests */
1889 	if (unlikely(!buflen))
1890 		return -EINVAL;
1891 
1892 	lock_sock(sk);
1893 	if (unlikely(connected && sk->sk_state == TIPC_OPEN)) {
1894 		rc = -ENOTCONN;
1895 		goto exit;
1896 	}
1897 	timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1898 
1899 	/* Step rcv queue to first msg with data or error; wait if necessary */
1900 	do {
1901 		rc = tipc_wait_for_rcvmsg(sock, &timeout);
1902 		if (unlikely(rc))
1903 			goto exit;
1904 		skb = skb_peek(&sk->sk_receive_queue);
1905 		hdr = buf_msg(skb);
1906 		dlen = msg_data_sz(hdr);
1907 		hlen = msg_hdr_sz(hdr);
1908 		err = msg_errcode(hdr);
1909 		grp_evt = msg_is_grp_evt(hdr);
1910 		if (likely(dlen || err))
1911 			break;
1912 		tsk_advance_rx_queue(sk);
1913 	} while (1);
1914 
1915 	/* Collect msg meta data, including error code and rejected data */
1916 	tipc_sk_set_orig_addr(m, skb);
1917 	rc = tipc_sk_anc_data_recv(m, skb, tsk);
1918 	if (unlikely(rc))
1919 		goto exit;
1920 	hdr = buf_msg(skb);
1921 
1922 	/* Capture data if non-error msg, otherwise just set return value */
1923 	if (likely(!err)) {
1924 		copy = min_t(int, dlen, buflen);
1925 		if (unlikely(copy != dlen))
1926 			m->msg_flags |= MSG_TRUNC;
1927 		rc = skb_copy_datagram_msg(skb, hlen, m, copy);
1928 	} else {
1929 		copy = 0;
1930 		rc = 0;
1931 		if (err != TIPC_CONN_SHUTDOWN && connected && !m->msg_control)
1932 			rc = -ECONNRESET;
1933 	}
1934 	if (unlikely(rc))
1935 		goto exit;
1936 
1937 	/* Mark message as group event if applicable */
1938 	if (unlikely(grp_evt)) {
1939 		if (msg_grp_evt(hdr) == TIPC_WITHDRAWN)
1940 			m->msg_flags |= MSG_EOR;
1941 		m->msg_flags |= MSG_OOB;
1942 		copy = 0;
1943 	}
1944 
1945 	/* Caption of data or error code/rejected data was successful */
1946 	if (unlikely(flags & MSG_PEEK))
1947 		goto exit;
1948 
1949 	/* Send group flow control advertisement when applicable */
1950 	if (tsk->group && msg_in_group(hdr) && !grp_evt) {
1951 		__skb_queue_head_init(&xmitq);
1952 		tipc_group_update_rcv_win(tsk->group, tsk_blocks(hlen + dlen),
1953 					  msg_orignode(hdr), msg_origport(hdr),
1954 					  &xmitq);
1955 		tipc_node_distr_xmit(sock_net(sk), &xmitq);
1956 	}
1957 
1958 	tsk_advance_rx_queue(sk);
1959 
1960 	if (likely(!connected))
1961 		goto exit;
1962 
1963 	/* Send connection flow control advertisement when applicable */
1964 	tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
1965 	if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
1966 		tipc_sk_send_ack(tsk);
1967 exit:
1968 	release_sock(sk);
1969 	return rc ? rc : copy;
1970 }
1971 
1972 /**
1973  * tipc_recvstream - receive stream-oriented data
1974  * @m: descriptor for message info
1975  * @buflen: total size of user buffer area
1976  * @flags: receive flags
1977  *
1978  * Used for SOCK_STREAM messages only.  If not enough data is available
1979  * will optionally wait for more; never truncates data.
1980  *
1981  * Returns size of returned message data, errno otherwise
1982  */
1983 static int tipc_recvstream(struct socket *sock, struct msghdr *m,
1984 			   size_t buflen, int flags)
1985 {
1986 	struct sock *sk = sock->sk;
1987 	struct tipc_sock *tsk = tipc_sk(sk);
1988 	struct sk_buff *skb;
1989 	struct tipc_msg *hdr;
1990 	struct tipc_skb_cb *skb_cb;
1991 	bool peek = flags & MSG_PEEK;
1992 	int offset, required, copy, copied = 0;
1993 	int hlen, dlen, err, rc;
1994 	long timeout;
1995 
1996 	/* Catch invalid receive attempts */
1997 	if (unlikely(!buflen))
1998 		return -EINVAL;
1999 
2000 	lock_sock(sk);
2001 
2002 	if (unlikely(sk->sk_state == TIPC_OPEN)) {
2003 		rc = -ENOTCONN;
2004 		goto exit;
2005 	}
2006 	required = sock_rcvlowat(sk, flags & MSG_WAITALL, buflen);
2007 	timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
2008 
2009 	do {
2010 		/* Look at first msg in receive queue; wait if necessary */
2011 		rc = tipc_wait_for_rcvmsg(sock, &timeout);
2012 		if (unlikely(rc))
2013 			break;
2014 		skb = skb_peek(&sk->sk_receive_queue);
2015 		skb_cb = TIPC_SKB_CB(skb);
2016 		hdr = buf_msg(skb);
2017 		dlen = msg_data_sz(hdr);
2018 		hlen = msg_hdr_sz(hdr);
2019 		err = msg_errcode(hdr);
2020 
2021 		/* Discard any empty non-errored (SYN-) message */
2022 		if (unlikely(!dlen && !err)) {
2023 			tsk_advance_rx_queue(sk);
2024 			continue;
2025 		}
2026 
2027 		/* Collect msg meta data, incl. error code and rejected data */
2028 		if (!copied) {
2029 			tipc_sk_set_orig_addr(m, skb);
2030 			rc = tipc_sk_anc_data_recv(m, skb, tsk);
2031 			if (rc)
2032 				break;
2033 			hdr = buf_msg(skb);
2034 		}
2035 
2036 		/* Copy data if msg ok, otherwise return error/partial data */
2037 		if (likely(!err)) {
2038 			offset = skb_cb->bytes_read;
2039 			copy = min_t(int, dlen - offset, buflen - copied);
2040 			rc = skb_copy_datagram_msg(skb, hlen + offset, m, copy);
2041 			if (unlikely(rc))
2042 				break;
2043 			copied += copy;
2044 			offset += copy;
2045 			if (unlikely(offset < dlen)) {
2046 				if (!peek)
2047 					skb_cb->bytes_read = offset;
2048 				break;
2049 			}
2050 		} else {
2051 			rc = 0;
2052 			if ((err != TIPC_CONN_SHUTDOWN) && !m->msg_control)
2053 				rc = -ECONNRESET;
2054 			if (copied || rc)
2055 				break;
2056 		}
2057 
2058 		if (unlikely(peek))
2059 			break;
2060 
2061 		tsk_advance_rx_queue(sk);
2062 
2063 		/* Send connection flow control advertisement when applicable */
2064 		tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
2065 		if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
2066 			tipc_sk_send_ack(tsk);
2067 
2068 		/* Exit if all requested data or FIN/error received */
2069 		if (copied == buflen || err)
2070 			break;
2071 
2072 	} while (!skb_queue_empty(&sk->sk_receive_queue) || copied < required);
2073 exit:
2074 	release_sock(sk);
2075 	return copied ? copied : rc;
2076 }
2077 
2078 /**
2079  * tipc_write_space - wake up thread if port congestion is released
2080  * @sk: socket
2081  */
2082 static void tipc_write_space(struct sock *sk)
2083 {
2084 	struct socket_wq *wq;
2085 
2086 	rcu_read_lock();
2087 	wq = rcu_dereference(sk->sk_wq);
2088 	if (skwq_has_sleeper(wq))
2089 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLOUT |
2090 						EPOLLWRNORM | EPOLLWRBAND);
2091 	rcu_read_unlock();
2092 }
2093 
2094 /**
2095  * tipc_data_ready - wake up threads to indicate messages have been received
2096  * @sk: socket
2097  */
2098 static void tipc_data_ready(struct sock *sk)
2099 {
2100 	struct socket_wq *wq;
2101 
2102 	rcu_read_lock();
2103 	wq = rcu_dereference(sk->sk_wq);
2104 	if (skwq_has_sleeper(wq))
2105 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN |
2106 						EPOLLRDNORM | EPOLLRDBAND);
2107 	rcu_read_unlock();
2108 }
2109 
2110 static void tipc_sock_destruct(struct sock *sk)
2111 {
2112 	__skb_queue_purge(&sk->sk_receive_queue);
2113 }
2114 
2115 static void tipc_sk_proto_rcv(struct sock *sk,
2116 			      struct sk_buff_head *inputq,
2117 			      struct sk_buff_head *xmitq)
2118 {
2119 	struct sk_buff *skb = __skb_dequeue(inputq);
2120 	struct tipc_sock *tsk = tipc_sk(sk);
2121 	struct tipc_msg *hdr = buf_msg(skb);
2122 	struct tipc_group *grp = tsk->group;
2123 	bool wakeup = false;
2124 
2125 	switch (msg_user(hdr)) {
2126 	case CONN_MANAGER:
2127 		tipc_sk_conn_proto_rcv(tsk, skb, inputq, xmitq);
2128 		return;
2129 	case SOCK_WAKEUP:
2130 		tipc_dest_del(&tsk->cong_links, msg_orignode(hdr), 0);
2131 		/* coupled with smp_rmb() in tipc_wait_for_cond() */
2132 		smp_wmb();
2133 		tsk->cong_link_cnt--;
2134 		wakeup = true;
2135 		tipc_sk_push_backlog(tsk, false);
2136 		break;
2137 	case GROUP_PROTOCOL:
2138 		tipc_group_proto_rcv(grp, &wakeup, hdr, inputq, xmitq);
2139 		break;
2140 	case TOP_SRV:
2141 		tipc_group_member_evt(tsk->group, &wakeup, &sk->sk_rcvbuf,
2142 				      hdr, inputq, xmitq);
2143 		break;
2144 	default:
2145 		break;
2146 	}
2147 
2148 	if (wakeup)
2149 		sk->sk_write_space(sk);
2150 
2151 	kfree_skb(skb);
2152 }
2153 
2154 /**
2155  * tipc_sk_filter_connect - check incoming message for a connection-based socket
2156  * @tsk: TIPC socket
2157  * @skb: pointer to message buffer.
2158  * @xmitq: for Nagle ACK if any
2159  * Returns true if message should be added to receive queue, false otherwise
2160  */
2161 static bool tipc_sk_filter_connect(struct tipc_sock *tsk, struct sk_buff *skb,
2162 				   struct sk_buff_head *xmitq)
2163 {
2164 	struct sock *sk = &tsk->sk;
2165 	struct net *net = sock_net(sk);
2166 	struct tipc_msg *hdr = buf_msg(skb);
2167 	bool con_msg = msg_connected(hdr);
2168 	u32 pport = tsk_peer_port(tsk);
2169 	u32 pnode = tsk_peer_node(tsk);
2170 	u32 oport = msg_origport(hdr);
2171 	u32 onode = msg_orignode(hdr);
2172 	int err = msg_errcode(hdr);
2173 	unsigned long delay;
2174 
2175 	if (unlikely(msg_mcast(hdr)))
2176 		return false;
2177 	tsk->oneway = 0;
2178 
2179 	switch (sk->sk_state) {
2180 	case TIPC_CONNECTING:
2181 		/* Setup ACK */
2182 		if (likely(con_msg)) {
2183 			if (err)
2184 				break;
2185 			tipc_sk_finish_conn(tsk, oport, onode);
2186 			msg_set_importance(&tsk->phdr, msg_importance(hdr));
2187 			/* ACK+ message with data is added to receive queue */
2188 			if (msg_data_sz(hdr))
2189 				return true;
2190 			/* Empty ACK-, - wake up sleeping connect() and drop */
2191 			sk->sk_state_change(sk);
2192 			msg_set_dest_droppable(hdr, 1);
2193 			return false;
2194 		}
2195 		/* Ignore connectionless message if not from listening socket */
2196 		if (oport != pport || onode != pnode)
2197 			return false;
2198 
2199 		/* Rejected SYN */
2200 		if (err != TIPC_ERR_OVERLOAD)
2201 			break;
2202 
2203 		/* Prepare for new setup attempt if we have a SYN clone */
2204 		if (skb_queue_empty(&sk->sk_write_queue))
2205 			break;
2206 		get_random_bytes(&delay, 2);
2207 		delay %= (tsk->conn_timeout / 4);
2208 		delay = msecs_to_jiffies(delay + 100);
2209 		sk_reset_timer(sk, &sk->sk_timer, jiffies + delay);
2210 		return false;
2211 	case TIPC_OPEN:
2212 	case TIPC_DISCONNECTING:
2213 		return false;
2214 	case TIPC_LISTEN:
2215 		/* Accept only SYN message */
2216 		if (!msg_is_syn(hdr) &&
2217 		    tipc_node_get_capabilities(net, onode) & TIPC_SYN_BIT)
2218 			return false;
2219 		if (!con_msg && !err)
2220 			return true;
2221 		return false;
2222 	case TIPC_ESTABLISHED:
2223 		if (!skb_queue_empty(&sk->sk_write_queue))
2224 			tipc_sk_push_backlog(tsk, false);
2225 		/* Accept only connection-based messages sent by peer */
2226 		if (likely(con_msg && !err && pport == oport &&
2227 			   pnode == onode)) {
2228 			if (msg_ack_required(hdr)) {
2229 				struct sk_buff *skb;
2230 
2231 				skb = tipc_sk_build_ack(tsk);
2232 				if (skb) {
2233 					msg_set_nagle_ack(buf_msg(skb));
2234 					__skb_queue_tail(xmitq, skb);
2235 				}
2236 			}
2237 			return true;
2238 		}
2239 		if (!tsk_peer_msg(tsk, hdr))
2240 			return false;
2241 		if (!err)
2242 			return true;
2243 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2244 		tipc_node_remove_conn(net, pnode, tsk->portid);
2245 		sk->sk_state_change(sk);
2246 		return true;
2247 	default:
2248 		pr_err("Unknown sk_state %u\n", sk->sk_state);
2249 	}
2250 	/* Abort connection setup attempt */
2251 	tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2252 	sk->sk_err = ECONNREFUSED;
2253 	sk->sk_state_change(sk);
2254 	return true;
2255 }
2256 
2257 /**
2258  * rcvbuf_limit - get proper overload limit of socket receive queue
2259  * @sk: socket
2260  * @skb: message
2261  *
2262  * For connection oriented messages, irrespective of importance,
2263  * default queue limit is 2 MB.
2264  *
2265  * For connectionless messages, queue limits are based on message
2266  * importance as follows:
2267  *
2268  * TIPC_LOW_IMPORTANCE       (2 MB)
2269  * TIPC_MEDIUM_IMPORTANCE    (4 MB)
2270  * TIPC_HIGH_IMPORTANCE      (8 MB)
2271  * TIPC_CRITICAL_IMPORTANCE  (16 MB)
2272  *
2273  * Returns overload limit according to corresponding message importance
2274  */
2275 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *skb)
2276 {
2277 	struct tipc_sock *tsk = tipc_sk(sk);
2278 	struct tipc_msg *hdr = buf_msg(skb);
2279 
2280 	if (unlikely(msg_in_group(hdr)))
2281 		return READ_ONCE(sk->sk_rcvbuf);
2282 
2283 	if (unlikely(!msg_connected(hdr)))
2284 		return READ_ONCE(sk->sk_rcvbuf) << msg_importance(hdr);
2285 
2286 	if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
2287 		return READ_ONCE(sk->sk_rcvbuf);
2288 
2289 	return FLOWCTL_MSG_LIM;
2290 }
2291 
2292 /**
2293  * tipc_sk_filter_rcv - validate incoming message
2294  * @sk: socket
2295  * @skb: pointer to message.
2296  *
2297  * Enqueues message on receive queue if acceptable; optionally handles
2298  * disconnect indication for a connected socket.
2299  *
2300  * Called with socket lock already taken
2301  *
2302  */
2303 static void tipc_sk_filter_rcv(struct sock *sk, struct sk_buff *skb,
2304 			       struct sk_buff_head *xmitq)
2305 {
2306 	bool sk_conn = !tipc_sk_type_connectionless(sk);
2307 	struct tipc_sock *tsk = tipc_sk(sk);
2308 	struct tipc_group *grp = tsk->group;
2309 	struct tipc_msg *hdr = buf_msg(skb);
2310 	struct net *net = sock_net(sk);
2311 	struct sk_buff_head inputq;
2312 	int mtyp = msg_type(hdr);
2313 	int limit, err = TIPC_OK;
2314 
2315 	trace_tipc_sk_filter_rcv(sk, skb, TIPC_DUMP_ALL, " ");
2316 	TIPC_SKB_CB(skb)->bytes_read = 0;
2317 	__skb_queue_head_init(&inputq);
2318 	__skb_queue_tail(&inputq, skb);
2319 
2320 	if (unlikely(!msg_isdata(hdr)))
2321 		tipc_sk_proto_rcv(sk, &inputq, xmitq);
2322 
2323 	if (unlikely(grp))
2324 		tipc_group_filter_msg(grp, &inputq, xmitq);
2325 
2326 	if (unlikely(!grp) && mtyp == TIPC_MCAST_MSG)
2327 		tipc_mcast_filter_msg(net, &tsk->mc_method.deferredq, &inputq);
2328 
2329 	/* Validate and add to receive buffer if there is space */
2330 	while ((skb = __skb_dequeue(&inputq))) {
2331 		hdr = buf_msg(skb);
2332 		limit = rcvbuf_limit(sk, skb);
2333 		if ((sk_conn && !tipc_sk_filter_connect(tsk, skb, xmitq)) ||
2334 		    (!sk_conn && msg_connected(hdr)) ||
2335 		    (!grp && msg_in_group(hdr)))
2336 			err = TIPC_ERR_NO_PORT;
2337 		else if (sk_rmem_alloc_get(sk) + skb->truesize >= limit) {
2338 			trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL,
2339 					   "err_overload2!");
2340 			atomic_inc(&sk->sk_drops);
2341 			err = TIPC_ERR_OVERLOAD;
2342 		}
2343 
2344 		if (unlikely(err)) {
2345 			if (tipc_msg_reverse(tipc_own_addr(net), &skb, err)) {
2346 				trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE,
2347 						      "@filter_rcv!");
2348 				__skb_queue_tail(xmitq, skb);
2349 			}
2350 			err = TIPC_OK;
2351 			continue;
2352 		}
2353 		__skb_queue_tail(&sk->sk_receive_queue, skb);
2354 		skb_set_owner_r(skb, sk);
2355 		trace_tipc_sk_overlimit2(sk, skb, TIPC_DUMP_ALL,
2356 					 "rcvq >90% allocated!");
2357 		sk->sk_data_ready(sk);
2358 	}
2359 }
2360 
2361 /**
2362  * tipc_sk_backlog_rcv - handle incoming message from backlog queue
2363  * @sk: socket
2364  * @skb: message
2365  *
2366  * Caller must hold socket lock
2367  */
2368 static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb)
2369 {
2370 	unsigned int before = sk_rmem_alloc_get(sk);
2371 	struct sk_buff_head xmitq;
2372 	unsigned int added;
2373 
2374 	__skb_queue_head_init(&xmitq);
2375 
2376 	tipc_sk_filter_rcv(sk, skb, &xmitq);
2377 	added = sk_rmem_alloc_get(sk) - before;
2378 	atomic_add(added, &tipc_sk(sk)->dupl_rcvcnt);
2379 
2380 	/* Send pending response/rejected messages, if any */
2381 	tipc_node_distr_xmit(sock_net(sk), &xmitq);
2382 	return 0;
2383 }
2384 
2385 /**
2386  * tipc_sk_enqueue - extract all buffers with destination 'dport' from
2387  *                   inputq and try adding them to socket or backlog queue
2388  * @inputq: list of incoming buffers with potentially different destinations
2389  * @sk: socket where the buffers should be enqueued
2390  * @dport: port number for the socket
2391  *
2392  * Caller must hold socket lock
2393  */
2394 static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk,
2395 			    u32 dport, struct sk_buff_head *xmitq)
2396 {
2397 	unsigned long time_limit = jiffies + 2;
2398 	struct sk_buff *skb;
2399 	unsigned int lim;
2400 	atomic_t *dcnt;
2401 	u32 onode;
2402 
2403 	while (skb_queue_len(inputq)) {
2404 		if (unlikely(time_after_eq(jiffies, time_limit)))
2405 			return;
2406 
2407 		skb = tipc_skb_dequeue(inputq, dport);
2408 		if (unlikely(!skb))
2409 			return;
2410 
2411 		/* Add message directly to receive queue if possible */
2412 		if (!sock_owned_by_user(sk)) {
2413 			tipc_sk_filter_rcv(sk, skb, xmitq);
2414 			continue;
2415 		}
2416 
2417 		/* Try backlog, compensating for double-counted bytes */
2418 		dcnt = &tipc_sk(sk)->dupl_rcvcnt;
2419 		if (!sk->sk_backlog.len)
2420 			atomic_set(dcnt, 0);
2421 		lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
2422 		if (likely(!sk_add_backlog(sk, skb, lim))) {
2423 			trace_tipc_sk_overlimit1(sk, skb, TIPC_DUMP_ALL,
2424 						 "bklg & rcvq >90% allocated!");
2425 			continue;
2426 		}
2427 
2428 		trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL, "err_overload!");
2429 		/* Overload => reject message back to sender */
2430 		onode = tipc_own_addr(sock_net(sk));
2431 		atomic_inc(&sk->sk_drops);
2432 		if (tipc_msg_reverse(onode, &skb, TIPC_ERR_OVERLOAD)) {
2433 			trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_ALL,
2434 					      "@sk_enqueue!");
2435 			__skb_queue_tail(xmitq, skb);
2436 		}
2437 		break;
2438 	}
2439 }
2440 
2441 /**
2442  * tipc_sk_rcv - handle a chain of incoming buffers
2443  * @inputq: buffer list containing the buffers
2444  * Consumes all buffers in list until inputq is empty
2445  * Note: may be called in multiple threads referring to the same queue
2446  */
2447 void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq)
2448 {
2449 	struct sk_buff_head xmitq;
2450 	u32 dnode, dport = 0;
2451 	int err;
2452 	struct tipc_sock *tsk;
2453 	struct sock *sk;
2454 	struct sk_buff *skb;
2455 
2456 	__skb_queue_head_init(&xmitq);
2457 	while (skb_queue_len(inputq)) {
2458 		dport = tipc_skb_peek_port(inputq, dport);
2459 		tsk = tipc_sk_lookup(net, dport);
2460 
2461 		if (likely(tsk)) {
2462 			sk = &tsk->sk;
2463 			if (likely(spin_trylock_bh(&sk->sk_lock.slock))) {
2464 				tipc_sk_enqueue(inputq, sk, dport, &xmitq);
2465 				spin_unlock_bh(&sk->sk_lock.slock);
2466 			}
2467 			/* Send pending response/rejected messages, if any */
2468 			tipc_node_distr_xmit(sock_net(sk), &xmitq);
2469 			sock_put(sk);
2470 			continue;
2471 		}
2472 		/* No destination socket => dequeue skb if still there */
2473 		skb = tipc_skb_dequeue(inputq, dport);
2474 		if (!skb)
2475 			return;
2476 
2477 		/* Try secondary lookup if unresolved named message */
2478 		err = TIPC_ERR_NO_PORT;
2479 		if (tipc_msg_lookup_dest(net, skb, &err))
2480 			goto xmit;
2481 
2482 		/* Prepare for message rejection */
2483 		if (!tipc_msg_reverse(tipc_own_addr(net), &skb, err))
2484 			continue;
2485 
2486 		trace_tipc_sk_rej_msg(NULL, skb, TIPC_DUMP_NONE, "@sk_rcv!");
2487 xmit:
2488 		dnode = msg_destnode(buf_msg(skb));
2489 		tipc_node_xmit_skb(net, skb, dnode, dport);
2490 	}
2491 }
2492 
2493 static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
2494 {
2495 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
2496 	struct sock *sk = sock->sk;
2497 	int done;
2498 
2499 	do {
2500 		int err = sock_error(sk);
2501 		if (err)
2502 			return err;
2503 		if (!*timeo_p)
2504 			return -ETIMEDOUT;
2505 		if (signal_pending(current))
2506 			return sock_intr_errno(*timeo_p);
2507 		if (sk->sk_state == TIPC_DISCONNECTING)
2508 			break;
2509 
2510 		add_wait_queue(sk_sleep(sk), &wait);
2511 		done = sk_wait_event(sk, timeo_p, tipc_sk_connected(sk),
2512 				     &wait);
2513 		remove_wait_queue(sk_sleep(sk), &wait);
2514 	} while (!done);
2515 	return 0;
2516 }
2517 
2518 static bool tipc_sockaddr_is_sane(struct sockaddr_tipc *addr)
2519 {
2520 	if (addr->family != AF_TIPC)
2521 		return false;
2522 	if (addr->addrtype == TIPC_SERVICE_RANGE)
2523 		return (addr->addr.nameseq.lower <= addr->addr.nameseq.upper);
2524 	return (addr->addrtype == TIPC_SERVICE_ADDR ||
2525 		addr->addrtype == TIPC_SOCKET_ADDR);
2526 }
2527 
2528 /**
2529  * tipc_connect - establish a connection to another TIPC port
2530  * @sock: socket structure
2531  * @dest: socket address for destination port
2532  * @destlen: size of socket address data structure
2533  * @flags: file-related flags associated with socket
2534  *
2535  * Returns 0 on success, errno otherwise
2536  */
2537 static int tipc_connect(struct socket *sock, struct sockaddr *dest,
2538 			int destlen, int flags)
2539 {
2540 	struct sock *sk = sock->sk;
2541 	struct tipc_sock *tsk = tipc_sk(sk);
2542 	struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
2543 	struct msghdr m = {NULL,};
2544 	long timeout = (flags & O_NONBLOCK) ? 0 : tsk->conn_timeout;
2545 	int previous;
2546 	int res = 0;
2547 
2548 	if (destlen != sizeof(struct sockaddr_tipc))
2549 		return -EINVAL;
2550 
2551 	lock_sock(sk);
2552 
2553 	if (tsk->group) {
2554 		res = -EINVAL;
2555 		goto exit;
2556 	}
2557 
2558 	if (dst->family == AF_UNSPEC) {
2559 		memset(&tsk->peer, 0, sizeof(struct sockaddr_tipc));
2560 		if (!tipc_sk_type_connectionless(sk))
2561 			res = -EINVAL;
2562 		goto exit;
2563 	}
2564 	if (!tipc_sockaddr_is_sane(dst)) {
2565 		res = -EINVAL;
2566 		goto exit;
2567 	}
2568 	/* DGRAM/RDM connect(), just save the destaddr */
2569 	if (tipc_sk_type_connectionless(sk)) {
2570 		memcpy(&tsk->peer, dest, destlen);
2571 		goto exit;
2572 	} else if (dst->addrtype == TIPC_SERVICE_RANGE) {
2573 		res = -EINVAL;
2574 		goto exit;
2575 	}
2576 
2577 	previous = sk->sk_state;
2578 
2579 	switch (sk->sk_state) {
2580 	case TIPC_OPEN:
2581 		/* Send a 'SYN-' to destination */
2582 		m.msg_name = dest;
2583 		m.msg_namelen = destlen;
2584 
2585 		/* If connect is in non-blocking case, set MSG_DONTWAIT to
2586 		 * indicate send_msg() is never blocked.
2587 		 */
2588 		if (!timeout)
2589 			m.msg_flags = MSG_DONTWAIT;
2590 
2591 		res = __tipc_sendmsg(sock, &m, 0);
2592 		if ((res < 0) && (res != -EWOULDBLOCK))
2593 			goto exit;
2594 
2595 		/* Just entered TIPC_CONNECTING state; the only
2596 		 * difference is that return value in non-blocking
2597 		 * case is EINPROGRESS, rather than EALREADY.
2598 		 */
2599 		res = -EINPROGRESS;
2600 		/* fall through */
2601 	case TIPC_CONNECTING:
2602 		if (!timeout) {
2603 			if (previous == TIPC_CONNECTING)
2604 				res = -EALREADY;
2605 			goto exit;
2606 		}
2607 		timeout = msecs_to_jiffies(timeout);
2608 		/* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
2609 		res = tipc_wait_for_connect(sock, &timeout);
2610 		break;
2611 	case TIPC_ESTABLISHED:
2612 		res = -EISCONN;
2613 		break;
2614 	default:
2615 		res = -EINVAL;
2616 	}
2617 
2618 exit:
2619 	release_sock(sk);
2620 	return res;
2621 }
2622 
2623 /**
2624  * tipc_listen - allow socket to listen for incoming connections
2625  * @sock: socket structure
2626  * @len: (unused)
2627  *
2628  * Returns 0 on success, errno otherwise
2629  */
2630 static int tipc_listen(struct socket *sock, int len)
2631 {
2632 	struct sock *sk = sock->sk;
2633 	int res;
2634 
2635 	lock_sock(sk);
2636 	res = tipc_set_sk_state(sk, TIPC_LISTEN);
2637 	release_sock(sk);
2638 
2639 	return res;
2640 }
2641 
2642 static int tipc_wait_for_accept(struct socket *sock, long timeo)
2643 {
2644 	struct sock *sk = sock->sk;
2645 	DEFINE_WAIT(wait);
2646 	int err;
2647 
2648 	/* True wake-one mechanism for incoming connections: only
2649 	 * one process gets woken up, not the 'whole herd'.
2650 	 * Since we do not 'race & poll' for established sockets
2651 	 * anymore, the common case will execute the loop only once.
2652 	*/
2653 	for (;;) {
2654 		prepare_to_wait_exclusive(sk_sleep(sk), &wait,
2655 					  TASK_INTERRUPTIBLE);
2656 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
2657 			release_sock(sk);
2658 			timeo = schedule_timeout(timeo);
2659 			lock_sock(sk);
2660 		}
2661 		err = 0;
2662 		if (!skb_queue_empty(&sk->sk_receive_queue))
2663 			break;
2664 		err = -EAGAIN;
2665 		if (!timeo)
2666 			break;
2667 		err = sock_intr_errno(timeo);
2668 		if (signal_pending(current))
2669 			break;
2670 	}
2671 	finish_wait(sk_sleep(sk), &wait);
2672 	return err;
2673 }
2674 
2675 /**
2676  * tipc_accept - wait for connection request
2677  * @sock: listening socket
2678  * @new_sock: new socket that is to be connected
2679  * @flags: file-related flags associated with socket
2680  *
2681  * Returns 0 on success, errno otherwise
2682  */
2683 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags,
2684 		       bool kern)
2685 {
2686 	struct sock *new_sk, *sk = sock->sk;
2687 	struct sk_buff *buf;
2688 	struct tipc_sock *new_tsock;
2689 	struct tipc_msg *msg;
2690 	long timeo;
2691 	int res;
2692 
2693 	lock_sock(sk);
2694 
2695 	if (sk->sk_state != TIPC_LISTEN) {
2696 		res = -EINVAL;
2697 		goto exit;
2698 	}
2699 	timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
2700 	res = tipc_wait_for_accept(sock, timeo);
2701 	if (res)
2702 		goto exit;
2703 
2704 	buf = skb_peek(&sk->sk_receive_queue);
2705 
2706 	res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, kern);
2707 	if (res)
2708 		goto exit;
2709 	security_sk_clone(sock->sk, new_sock->sk);
2710 
2711 	new_sk = new_sock->sk;
2712 	new_tsock = tipc_sk(new_sk);
2713 	msg = buf_msg(buf);
2714 
2715 	/* we lock on new_sk; but lockdep sees the lock on sk */
2716 	lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
2717 
2718 	/*
2719 	 * Reject any stray messages received by new socket
2720 	 * before the socket lock was taken (very, very unlikely)
2721 	 */
2722 	tsk_rej_rx_queue(new_sk, TIPC_ERR_NO_PORT);
2723 
2724 	/* Connect new socket to it's peer */
2725 	tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg));
2726 
2727 	tsk_set_importance(new_sk, msg_importance(msg));
2728 	if (msg_named(msg)) {
2729 		new_tsock->conn_type = msg_nametype(msg);
2730 		new_tsock->conn_instance = msg_nameinst(msg);
2731 	}
2732 
2733 	/*
2734 	 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2735 	 * Respond to 'SYN+' by queuing it on new socket.
2736 	 */
2737 	if (!msg_data_sz(msg)) {
2738 		struct msghdr m = {NULL,};
2739 
2740 		tsk_advance_rx_queue(sk);
2741 		__tipc_sendstream(new_sock, &m, 0);
2742 	} else {
2743 		__skb_dequeue(&sk->sk_receive_queue);
2744 		__skb_queue_head(&new_sk->sk_receive_queue, buf);
2745 		skb_set_owner_r(buf, new_sk);
2746 	}
2747 	release_sock(new_sk);
2748 exit:
2749 	release_sock(sk);
2750 	return res;
2751 }
2752 
2753 /**
2754  * tipc_shutdown - shutdown socket connection
2755  * @sock: socket structure
2756  * @how: direction to close (must be SHUT_RDWR)
2757  *
2758  * Terminates connection (if necessary), then purges socket's receive queue.
2759  *
2760  * Returns 0 on success, errno otherwise
2761  */
2762 static int tipc_shutdown(struct socket *sock, int how)
2763 {
2764 	struct sock *sk = sock->sk;
2765 	int res;
2766 
2767 	if (how != SHUT_RDWR)
2768 		return -EINVAL;
2769 
2770 	lock_sock(sk);
2771 
2772 	trace_tipc_sk_shutdown(sk, NULL, TIPC_DUMP_ALL, " ");
2773 	__tipc_shutdown(sock, TIPC_CONN_SHUTDOWN);
2774 	sk->sk_shutdown = SEND_SHUTDOWN;
2775 
2776 	if (sk->sk_state == TIPC_DISCONNECTING) {
2777 		/* Discard any unreceived messages */
2778 		__skb_queue_purge(&sk->sk_receive_queue);
2779 
2780 		/* Wake up anyone sleeping in poll */
2781 		sk->sk_state_change(sk);
2782 		res = 0;
2783 	} else {
2784 		res = -ENOTCONN;
2785 	}
2786 
2787 	release_sock(sk);
2788 	return res;
2789 }
2790 
2791 static void tipc_sk_check_probing_state(struct sock *sk,
2792 					struct sk_buff_head *list)
2793 {
2794 	struct tipc_sock *tsk = tipc_sk(sk);
2795 	u32 pnode = tsk_peer_node(tsk);
2796 	u32 pport = tsk_peer_port(tsk);
2797 	u32 self = tsk_own_node(tsk);
2798 	u32 oport = tsk->portid;
2799 	struct sk_buff *skb;
2800 
2801 	if (tsk->probe_unacked) {
2802 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2803 		sk->sk_err = ECONNABORTED;
2804 		tipc_node_remove_conn(sock_net(sk), pnode, pport);
2805 		sk->sk_state_change(sk);
2806 		return;
2807 	}
2808 	/* Prepare new probe */
2809 	skb = tipc_msg_create(CONN_MANAGER, CONN_PROBE, INT_H_SIZE, 0,
2810 			      pnode, self, pport, oport, TIPC_OK);
2811 	if (skb)
2812 		__skb_queue_tail(list, skb);
2813 	tsk->probe_unacked = true;
2814 	sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
2815 }
2816 
2817 static void tipc_sk_retry_connect(struct sock *sk, struct sk_buff_head *list)
2818 {
2819 	struct tipc_sock *tsk = tipc_sk(sk);
2820 
2821 	/* Try again later if dest link is congested */
2822 	if (tsk->cong_link_cnt) {
2823 		sk_reset_timer(sk, &sk->sk_timer, msecs_to_jiffies(100));
2824 		return;
2825 	}
2826 	/* Prepare SYN for retransmit */
2827 	tipc_msg_skb_clone(&sk->sk_write_queue, list);
2828 }
2829 
2830 static void tipc_sk_timeout(struct timer_list *t)
2831 {
2832 	struct sock *sk = from_timer(sk, t, sk_timer);
2833 	struct tipc_sock *tsk = tipc_sk(sk);
2834 	u32 pnode = tsk_peer_node(tsk);
2835 	struct sk_buff_head list;
2836 	int rc = 0;
2837 
2838 	__skb_queue_head_init(&list);
2839 	bh_lock_sock(sk);
2840 
2841 	/* Try again later if socket is busy */
2842 	if (sock_owned_by_user(sk)) {
2843 		sk_reset_timer(sk, &sk->sk_timer, jiffies + HZ / 20);
2844 		bh_unlock_sock(sk);
2845 		sock_put(sk);
2846 		return;
2847 	}
2848 
2849 	if (sk->sk_state == TIPC_ESTABLISHED)
2850 		tipc_sk_check_probing_state(sk, &list);
2851 	else if (sk->sk_state == TIPC_CONNECTING)
2852 		tipc_sk_retry_connect(sk, &list);
2853 
2854 	bh_unlock_sock(sk);
2855 
2856 	if (!skb_queue_empty(&list))
2857 		rc = tipc_node_xmit(sock_net(sk), &list, pnode, tsk->portid);
2858 
2859 	/* SYN messages may cause link congestion */
2860 	if (rc == -ELINKCONG) {
2861 		tipc_dest_push(&tsk->cong_links, pnode, 0);
2862 		tsk->cong_link_cnt = 1;
2863 	}
2864 	sock_put(sk);
2865 }
2866 
2867 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
2868 			   struct tipc_name_seq const *seq)
2869 {
2870 	struct sock *sk = &tsk->sk;
2871 	struct net *net = sock_net(sk);
2872 	struct publication *publ;
2873 	u32 key;
2874 
2875 	if (scope != TIPC_NODE_SCOPE)
2876 		scope = TIPC_CLUSTER_SCOPE;
2877 
2878 	if (tipc_sk_connected(sk))
2879 		return -EINVAL;
2880 	key = tsk->portid + tsk->pub_count + 1;
2881 	if (key == tsk->portid)
2882 		return -EADDRINUSE;
2883 
2884 	publ = tipc_nametbl_publish(net, seq->type, seq->lower, seq->upper,
2885 				    scope, tsk->portid, key);
2886 	if (unlikely(!publ))
2887 		return -EINVAL;
2888 
2889 	list_add(&publ->binding_sock, &tsk->publications);
2890 	tsk->pub_count++;
2891 	tsk->published = 1;
2892 	return 0;
2893 }
2894 
2895 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
2896 			    struct tipc_name_seq const *seq)
2897 {
2898 	struct net *net = sock_net(&tsk->sk);
2899 	struct publication *publ;
2900 	struct publication *safe;
2901 	int rc = -EINVAL;
2902 
2903 	if (scope != TIPC_NODE_SCOPE)
2904 		scope = TIPC_CLUSTER_SCOPE;
2905 
2906 	list_for_each_entry_safe(publ, safe, &tsk->publications, binding_sock) {
2907 		if (seq) {
2908 			if (publ->scope != scope)
2909 				continue;
2910 			if (publ->type != seq->type)
2911 				continue;
2912 			if (publ->lower != seq->lower)
2913 				continue;
2914 			if (publ->upper != seq->upper)
2915 				break;
2916 			tipc_nametbl_withdraw(net, publ->type, publ->lower,
2917 					      publ->upper, publ->key);
2918 			rc = 0;
2919 			break;
2920 		}
2921 		tipc_nametbl_withdraw(net, publ->type, publ->lower,
2922 				      publ->upper, publ->key);
2923 		rc = 0;
2924 	}
2925 	if (list_empty(&tsk->publications))
2926 		tsk->published = 0;
2927 	return rc;
2928 }
2929 
2930 /* tipc_sk_reinit: set non-zero address in all existing sockets
2931  *                 when we go from standalone to network mode.
2932  */
2933 void tipc_sk_reinit(struct net *net)
2934 {
2935 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2936 	struct rhashtable_iter iter;
2937 	struct tipc_sock *tsk;
2938 	struct tipc_msg *msg;
2939 
2940 	rhashtable_walk_enter(&tn->sk_rht, &iter);
2941 
2942 	do {
2943 		rhashtable_walk_start(&iter);
2944 
2945 		while ((tsk = rhashtable_walk_next(&iter)) && !IS_ERR(tsk)) {
2946 			sock_hold(&tsk->sk);
2947 			rhashtable_walk_stop(&iter);
2948 			lock_sock(&tsk->sk);
2949 			msg = &tsk->phdr;
2950 			msg_set_prevnode(msg, tipc_own_addr(net));
2951 			msg_set_orignode(msg, tipc_own_addr(net));
2952 			release_sock(&tsk->sk);
2953 			rhashtable_walk_start(&iter);
2954 			sock_put(&tsk->sk);
2955 		}
2956 
2957 		rhashtable_walk_stop(&iter);
2958 	} while (tsk == ERR_PTR(-EAGAIN));
2959 
2960 	rhashtable_walk_exit(&iter);
2961 }
2962 
2963 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid)
2964 {
2965 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2966 	struct tipc_sock *tsk;
2967 
2968 	rcu_read_lock();
2969 	tsk = rhashtable_lookup(&tn->sk_rht, &portid, tsk_rht_params);
2970 	if (tsk)
2971 		sock_hold(&tsk->sk);
2972 	rcu_read_unlock();
2973 
2974 	return tsk;
2975 }
2976 
2977 static int tipc_sk_insert(struct tipc_sock *tsk)
2978 {
2979 	struct sock *sk = &tsk->sk;
2980 	struct net *net = sock_net(sk);
2981 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2982 	u32 remaining = (TIPC_MAX_PORT - TIPC_MIN_PORT) + 1;
2983 	u32 portid = prandom_u32() % remaining + TIPC_MIN_PORT;
2984 
2985 	while (remaining--) {
2986 		portid++;
2987 		if ((portid < TIPC_MIN_PORT) || (portid > TIPC_MAX_PORT))
2988 			portid = TIPC_MIN_PORT;
2989 		tsk->portid = portid;
2990 		sock_hold(&tsk->sk);
2991 		if (!rhashtable_lookup_insert_fast(&tn->sk_rht, &tsk->node,
2992 						   tsk_rht_params))
2993 			return 0;
2994 		sock_put(&tsk->sk);
2995 	}
2996 
2997 	return -1;
2998 }
2999 
3000 static void tipc_sk_remove(struct tipc_sock *tsk)
3001 {
3002 	struct sock *sk = &tsk->sk;
3003 	struct tipc_net *tn = net_generic(sock_net(sk), tipc_net_id);
3004 
3005 	if (!rhashtable_remove_fast(&tn->sk_rht, &tsk->node, tsk_rht_params)) {
3006 		WARN_ON(refcount_read(&sk->sk_refcnt) == 1);
3007 		__sock_put(sk);
3008 	}
3009 }
3010 
3011 static const struct rhashtable_params tsk_rht_params = {
3012 	.nelem_hint = 192,
3013 	.head_offset = offsetof(struct tipc_sock, node),
3014 	.key_offset = offsetof(struct tipc_sock, portid),
3015 	.key_len = sizeof(u32), /* portid */
3016 	.max_size = 1048576,
3017 	.min_size = 256,
3018 	.automatic_shrinking = true,
3019 };
3020 
3021 int tipc_sk_rht_init(struct net *net)
3022 {
3023 	struct tipc_net *tn = net_generic(net, tipc_net_id);
3024 
3025 	return rhashtable_init(&tn->sk_rht, &tsk_rht_params);
3026 }
3027 
3028 void tipc_sk_rht_destroy(struct net *net)
3029 {
3030 	struct tipc_net *tn = net_generic(net, tipc_net_id);
3031 
3032 	/* Wait for socket readers to complete */
3033 	synchronize_net();
3034 
3035 	rhashtable_destroy(&tn->sk_rht);
3036 }
3037 
3038 static int tipc_sk_join(struct tipc_sock *tsk, struct tipc_group_req *mreq)
3039 {
3040 	struct net *net = sock_net(&tsk->sk);
3041 	struct tipc_group *grp = tsk->group;
3042 	struct tipc_msg *hdr = &tsk->phdr;
3043 	struct tipc_name_seq seq;
3044 	int rc;
3045 
3046 	if (mreq->type < TIPC_RESERVED_TYPES)
3047 		return -EACCES;
3048 	if (mreq->scope > TIPC_NODE_SCOPE)
3049 		return -EINVAL;
3050 	if (grp)
3051 		return -EACCES;
3052 	grp = tipc_group_create(net, tsk->portid, mreq, &tsk->group_is_open);
3053 	if (!grp)
3054 		return -ENOMEM;
3055 	tsk->group = grp;
3056 	msg_set_lookup_scope(hdr, mreq->scope);
3057 	msg_set_nametype(hdr, mreq->type);
3058 	msg_set_dest_droppable(hdr, true);
3059 	seq.type = mreq->type;
3060 	seq.lower = mreq->instance;
3061 	seq.upper = seq.lower;
3062 	tipc_nametbl_build_group(net, grp, mreq->type, mreq->scope);
3063 	rc = tipc_sk_publish(tsk, mreq->scope, &seq);
3064 	if (rc) {
3065 		tipc_group_delete(net, grp);
3066 		tsk->group = NULL;
3067 		return rc;
3068 	}
3069 	/* Eliminate any risk that a broadcast overtakes sent JOINs */
3070 	tsk->mc_method.rcast = true;
3071 	tsk->mc_method.mandatory = true;
3072 	tipc_group_join(net, grp, &tsk->sk.sk_rcvbuf);
3073 	return rc;
3074 }
3075 
3076 static int tipc_sk_leave(struct tipc_sock *tsk)
3077 {
3078 	struct net *net = sock_net(&tsk->sk);
3079 	struct tipc_group *grp = tsk->group;
3080 	struct tipc_name_seq seq;
3081 	int scope;
3082 
3083 	if (!grp)
3084 		return -EINVAL;
3085 	tipc_group_self(grp, &seq, &scope);
3086 	tipc_group_delete(net, grp);
3087 	tsk->group = NULL;
3088 	tipc_sk_withdraw(tsk, scope, &seq);
3089 	return 0;
3090 }
3091 
3092 /**
3093  * tipc_setsockopt - set socket option
3094  * @sock: socket structure
3095  * @lvl: option level
3096  * @opt: option identifier
3097  * @ov: pointer to new option value
3098  * @ol: length of option value
3099  *
3100  * For stream sockets only, accepts and ignores all IPPROTO_TCP options
3101  * (to ease compatibility).
3102  *
3103  * Returns 0 on success, errno otherwise
3104  */
3105 static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
3106 			   sockptr_t ov, unsigned int ol)
3107 {
3108 	struct sock *sk = sock->sk;
3109 	struct tipc_sock *tsk = tipc_sk(sk);
3110 	struct tipc_group_req mreq;
3111 	u32 value = 0;
3112 	int res = 0;
3113 
3114 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3115 		return 0;
3116 	if (lvl != SOL_TIPC)
3117 		return -ENOPROTOOPT;
3118 
3119 	switch (opt) {
3120 	case TIPC_IMPORTANCE:
3121 	case TIPC_SRC_DROPPABLE:
3122 	case TIPC_DEST_DROPPABLE:
3123 	case TIPC_CONN_TIMEOUT:
3124 	case TIPC_NODELAY:
3125 		if (ol < sizeof(value))
3126 			return -EINVAL;
3127 		if (copy_from_sockptr(&value, ov, sizeof(u32)))
3128 			return -EFAULT;
3129 		break;
3130 	case TIPC_GROUP_JOIN:
3131 		if (ol < sizeof(mreq))
3132 			return -EINVAL;
3133 		if (copy_from_sockptr(&mreq, ov, sizeof(mreq)))
3134 			return -EFAULT;
3135 		break;
3136 	default:
3137 		if (!sockptr_is_null(ov) || ol)
3138 			return -EINVAL;
3139 	}
3140 
3141 	lock_sock(sk);
3142 
3143 	switch (opt) {
3144 	case TIPC_IMPORTANCE:
3145 		res = tsk_set_importance(sk, value);
3146 		break;
3147 	case TIPC_SRC_DROPPABLE:
3148 		if (sock->type != SOCK_STREAM)
3149 			tsk_set_unreliable(tsk, value);
3150 		else
3151 			res = -ENOPROTOOPT;
3152 		break;
3153 	case TIPC_DEST_DROPPABLE:
3154 		tsk_set_unreturnable(tsk, value);
3155 		break;
3156 	case TIPC_CONN_TIMEOUT:
3157 		tipc_sk(sk)->conn_timeout = value;
3158 		break;
3159 	case TIPC_MCAST_BROADCAST:
3160 		tsk->mc_method.rcast = false;
3161 		tsk->mc_method.mandatory = true;
3162 		break;
3163 	case TIPC_MCAST_REPLICAST:
3164 		tsk->mc_method.rcast = true;
3165 		tsk->mc_method.mandatory = true;
3166 		break;
3167 	case TIPC_GROUP_JOIN:
3168 		res = tipc_sk_join(tsk, &mreq);
3169 		break;
3170 	case TIPC_GROUP_LEAVE:
3171 		res = tipc_sk_leave(tsk);
3172 		break;
3173 	case TIPC_NODELAY:
3174 		tsk->nodelay = !!value;
3175 		tsk_set_nagle(tsk);
3176 		break;
3177 	default:
3178 		res = -EINVAL;
3179 	}
3180 
3181 	release_sock(sk);
3182 
3183 	return res;
3184 }
3185 
3186 /**
3187  * tipc_getsockopt - get socket option
3188  * @sock: socket structure
3189  * @lvl: option level
3190  * @opt: option identifier
3191  * @ov: receptacle for option value
3192  * @ol: receptacle for length of option value
3193  *
3194  * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
3195  * (to ease compatibility).
3196  *
3197  * Returns 0 on success, errno otherwise
3198  */
3199 static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
3200 			   char __user *ov, int __user *ol)
3201 {
3202 	struct sock *sk = sock->sk;
3203 	struct tipc_sock *tsk = tipc_sk(sk);
3204 	struct tipc_name_seq seq;
3205 	int len, scope;
3206 	u32 value;
3207 	int res;
3208 
3209 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3210 		return put_user(0, ol);
3211 	if (lvl != SOL_TIPC)
3212 		return -ENOPROTOOPT;
3213 	res = get_user(len, ol);
3214 	if (res)
3215 		return res;
3216 
3217 	lock_sock(sk);
3218 
3219 	switch (opt) {
3220 	case TIPC_IMPORTANCE:
3221 		value = tsk_importance(tsk);
3222 		break;
3223 	case TIPC_SRC_DROPPABLE:
3224 		value = tsk_unreliable(tsk);
3225 		break;
3226 	case TIPC_DEST_DROPPABLE:
3227 		value = tsk_unreturnable(tsk);
3228 		break;
3229 	case TIPC_CONN_TIMEOUT:
3230 		value = tsk->conn_timeout;
3231 		/* no need to set "res", since already 0 at this point */
3232 		break;
3233 	case TIPC_NODE_RECVQ_DEPTH:
3234 		value = 0; /* was tipc_queue_size, now obsolete */
3235 		break;
3236 	case TIPC_SOCK_RECVQ_DEPTH:
3237 		value = skb_queue_len(&sk->sk_receive_queue);
3238 		break;
3239 	case TIPC_SOCK_RECVQ_USED:
3240 		value = sk_rmem_alloc_get(sk);
3241 		break;
3242 	case TIPC_GROUP_JOIN:
3243 		seq.type = 0;
3244 		if (tsk->group)
3245 			tipc_group_self(tsk->group, &seq, &scope);
3246 		value = seq.type;
3247 		break;
3248 	default:
3249 		res = -EINVAL;
3250 	}
3251 
3252 	release_sock(sk);
3253 
3254 	if (res)
3255 		return res;	/* "get" failed */
3256 
3257 	if (len < sizeof(value))
3258 		return -EINVAL;
3259 
3260 	if (copy_to_user(ov, &value, sizeof(value)))
3261 		return -EFAULT;
3262 
3263 	return put_user(sizeof(value), ol);
3264 }
3265 
3266 static int tipc_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
3267 {
3268 	struct net *net = sock_net(sock->sk);
3269 	struct tipc_sioc_nodeid_req nr = {0};
3270 	struct tipc_sioc_ln_req lnr;
3271 	void __user *argp = (void __user *)arg;
3272 
3273 	switch (cmd) {
3274 	case SIOCGETLINKNAME:
3275 		if (copy_from_user(&lnr, argp, sizeof(lnr)))
3276 			return -EFAULT;
3277 		if (!tipc_node_get_linkname(net,
3278 					    lnr.bearer_id & 0xffff, lnr.peer,
3279 					    lnr.linkname, TIPC_MAX_LINK_NAME)) {
3280 			if (copy_to_user(argp, &lnr, sizeof(lnr)))
3281 				return -EFAULT;
3282 			return 0;
3283 		}
3284 		return -EADDRNOTAVAIL;
3285 	case SIOCGETNODEID:
3286 		if (copy_from_user(&nr, argp, sizeof(nr)))
3287 			return -EFAULT;
3288 		if (!tipc_node_get_id(net, nr.peer, nr.node_id))
3289 			return -EADDRNOTAVAIL;
3290 		if (copy_to_user(argp, &nr, sizeof(nr)))
3291 			return -EFAULT;
3292 		return 0;
3293 	default:
3294 		return -ENOIOCTLCMD;
3295 	}
3296 }
3297 
3298 static int tipc_socketpair(struct socket *sock1, struct socket *sock2)
3299 {
3300 	struct tipc_sock *tsk2 = tipc_sk(sock2->sk);
3301 	struct tipc_sock *tsk1 = tipc_sk(sock1->sk);
3302 	u32 onode = tipc_own_addr(sock_net(sock1->sk));
3303 
3304 	tsk1->peer.family = AF_TIPC;
3305 	tsk1->peer.addrtype = TIPC_ADDR_ID;
3306 	tsk1->peer.scope = TIPC_NODE_SCOPE;
3307 	tsk1->peer.addr.id.ref = tsk2->portid;
3308 	tsk1->peer.addr.id.node = onode;
3309 	tsk2->peer.family = AF_TIPC;
3310 	tsk2->peer.addrtype = TIPC_ADDR_ID;
3311 	tsk2->peer.scope = TIPC_NODE_SCOPE;
3312 	tsk2->peer.addr.id.ref = tsk1->portid;
3313 	tsk2->peer.addr.id.node = onode;
3314 
3315 	tipc_sk_finish_conn(tsk1, tsk2->portid, onode);
3316 	tipc_sk_finish_conn(tsk2, tsk1->portid, onode);
3317 	return 0;
3318 }
3319 
3320 /* Protocol switches for the various types of TIPC sockets */
3321 
3322 static const struct proto_ops msg_ops = {
3323 	.owner		= THIS_MODULE,
3324 	.family		= AF_TIPC,
3325 	.release	= tipc_release,
3326 	.bind		= tipc_bind,
3327 	.connect	= tipc_connect,
3328 	.socketpair	= tipc_socketpair,
3329 	.accept		= sock_no_accept,
3330 	.getname	= tipc_getname,
3331 	.poll		= tipc_poll,
3332 	.ioctl		= tipc_ioctl,
3333 	.listen		= sock_no_listen,
3334 	.shutdown	= tipc_shutdown,
3335 	.setsockopt	= tipc_setsockopt,
3336 	.getsockopt	= tipc_getsockopt,
3337 	.sendmsg	= tipc_sendmsg,
3338 	.recvmsg	= tipc_recvmsg,
3339 	.mmap		= sock_no_mmap,
3340 	.sendpage	= sock_no_sendpage
3341 };
3342 
3343 static const struct proto_ops packet_ops = {
3344 	.owner		= THIS_MODULE,
3345 	.family		= AF_TIPC,
3346 	.release	= tipc_release,
3347 	.bind		= tipc_bind,
3348 	.connect	= tipc_connect,
3349 	.socketpair	= tipc_socketpair,
3350 	.accept		= tipc_accept,
3351 	.getname	= tipc_getname,
3352 	.poll		= tipc_poll,
3353 	.ioctl		= tipc_ioctl,
3354 	.listen		= tipc_listen,
3355 	.shutdown	= tipc_shutdown,
3356 	.setsockopt	= tipc_setsockopt,
3357 	.getsockopt	= tipc_getsockopt,
3358 	.sendmsg	= tipc_send_packet,
3359 	.recvmsg	= tipc_recvmsg,
3360 	.mmap		= sock_no_mmap,
3361 	.sendpage	= sock_no_sendpage
3362 };
3363 
3364 static const struct proto_ops stream_ops = {
3365 	.owner		= THIS_MODULE,
3366 	.family		= AF_TIPC,
3367 	.release	= tipc_release,
3368 	.bind		= tipc_bind,
3369 	.connect	= tipc_connect,
3370 	.socketpair	= tipc_socketpair,
3371 	.accept		= tipc_accept,
3372 	.getname	= tipc_getname,
3373 	.poll		= tipc_poll,
3374 	.ioctl		= tipc_ioctl,
3375 	.listen		= tipc_listen,
3376 	.shutdown	= tipc_shutdown,
3377 	.setsockopt	= tipc_setsockopt,
3378 	.getsockopt	= tipc_getsockopt,
3379 	.sendmsg	= tipc_sendstream,
3380 	.recvmsg	= tipc_recvstream,
3381 	.mmap		= sock_no_mmap,
3382 	.sendpage	= sock_no_sendpage
3383 };
3384 
3385 static const struct net_proto_family tipc_family_ops = {
3386 	.owner		= THIS_MODULE,
3387 	.family		= AF_TIPC,
3388 	.create		= tipc_sk_create
3389 };
3390 
3391 static struct proto tipc_proto = {
3392 	.name		= "TIPC",
3393 	.owner		= THIS_MODULE,
3394 	.obj_size	= sizeof(struct tipc_sock),
3395 	.sysctl_rmem	= sysctl_tipc_rmem
3396 };
3397 
3398 /**
3399  * tipc_socket_init - initialize TIPC socket interface
3400  *
3401  * Returns 0 on success, errno otherwise
3402  */
3403 int tipc_socket_init(void)
3404 {
3405 	int res;
3406 
3407 	res = proto_register(&tipc_proto, 1);
3408 	if (res) {
3409 		pr_err("Failed to register TIPC protocol type\n");
3410 		goto out;
3411 	}
3412 
3413 	res = sock_register(&tipc_family_ops);
3414 	if (res) {
3415 		pr_err("Failed to register TIPC socket type\n");
3416 		proto_unregister(&tipc_proto);
3417 		goto out;
3418 	}
3419  out:
3420 	return res;
3421 }
3422 
3423 /**
3424  * tipc_socket_stop - stop TIPC socket interface
3425  */
3426 void tipc_socket_stop(void)
3427 {
3428 	sock_unregister(tipc_family_ops.family);
3429 	proto_unregister(&tipc_proto);
3430 }
3431 
3432 /* Caller should hold socket lock for the passed tipc socket. */
3433 static int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk)
3434 {
3435 	u32 peer_node;
3436 	u32 peer_port;
3437 	struct nlattr *nest;
3438 
3439 	peer_node = tsk_peer_node(tsk);
3440 	peer_port = tsk_peer_port(tsk);
3441 
3442 	nest = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_CON);
3443 	if (!nest)
3444 		return -EMSGSIZE;
3445 
3446 	if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node))
3447 		goto msg_full;
3448 	if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port))
3449 		goto msg_full;
3450 
3451 	if (tsk->conn_type != 0) {
3452 		if (nla_put_flag(skb, TIPC_NLA_CON_FLAG))
3453 			goto msg_full;
3454 		if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, tsk->conn_type))
3455 			goto msg_full;
3456 		if (nla_put_u32(skb, TIPC_NLA_CON_INST, tsk->conn_instance))
3457 			goto msg_full;
3458 	}
3459 	nla_nest_end(skb, nest);
3460 
3461 	return 0;
3462 
3463 msg_full:
3464 	nla_nest_cancel(skb, nest);
3465 
3466 	return -EMSGSIZE;
3467 }
3468 
3469 static int __tipc_nl_add_sk_info(struct sk_buff *skb, struct tipc_sock
3470 			  *tsk)
3471 {
3472 	struct net *net = sock_net(skb->sk);
3473 	struct sock *sk = &tsk->sk;
3474 
3475 	if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->portid) ||
3476 	    nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tipc_own_addr(net)))
3477 		return -EMSGSIZE;
3478 
3479 	if (tipc_sk_connected(sk)) {
3480 		if (__tipc_nl_add_sk_con(skb, tsk))
3481 			return -EMSGSIZE;
3482 	} else if (!list_empty(&tsk->publications)) {
3483 		if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL))
3484 			return -EMSGSIZE;
3485 	}
3486 	return 0;
3487 }
3488 
3489 /* Caller should hold socket lock for the passed tipc socket. */
3490 static int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb,
3491 			    struct tipc_sock *tsk)
3492 {
3493 	struct nlattr *attrs;
3494 	void *hdr;
3495 
3496 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
3497 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_SOCK_GET);
3498 	if (!hdr)
3499 		goto msg_cancel;
3500 
3501 	attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK);
3502 	if (!attrs)
3503 		goto genlmsg_cancel;
3504 
3505 	if (__tipc_nl_add_sk_info(skb, tsk))
3506 		goto attr_msg_cancel;
3507 
3508 	nla_nest_end(skb, attrs);
3509 	genlmsg_end(skb, hdr);
3510 
3511 	return 0;
3512 
3513 attr_msg_cancel:
3514 	nla_nest_cancel(skb, attrs);
3515 genlmsg_cancel:
3516 	genlmsg_cancel(skb, hdr);
3517 msg_cancel:
3518 	return -EMSGSIZE;
3519 }
3520 
3521 int tipc_nl_sk_walk(struct sk_buff *skb, struct netlink_callback *cb,
3522 		    int (*skb_handler)(struct sk_buff *skb,
3523 				       struct netlink_callback *cb,
3524 				       struct tipc_sock *tsk))
3525 {
3526 	struct rhashtable_iter *iter = (void *)cb->args[4];
3527 	struct tipc_sock *tsk;
3528 	int err;
3529 
3530 	rhashtable_walk_start(iter);
3531 	while ((tsk = rhashtable_walk_next(iter)) != NULL) {
3532 		if (IS_ERR(tsk)) {
3533 			err = PTR_ERR(tsk);
3534 			if (err == -EAGAIN) {
3535 				err = 0;
3536 				continue;
3537 			}
3538 			break;
3539 		}
3540 
3541 		sock_hold(&tsk->sk);
3542 		rhashtable_walk_stop(iter);
3543 		lock_sock(&tsk->sk);
3544 		err = skb_handler(skb, cb, tsk);
3545 		if (err) {
3546 			release_sock(&tsk->sk);
3547 			sock_put(&tsk->sk);
3548 			goto out;
3549 		}
3550 		release_sock(&tsk->sk);
3551 		rhashtable_walk_start(iter);
3552 		sock_put(&tsk->sk);
3553 	}
3554 	rhashtable_walk_stop(iter);
3555 out:
3556 	return skb->len;
3557 }
3558 EXPORT_SYMBOL(tipc_nl_sk_walk);
3559 
3560 int tipc_dump_start(struct netlink_callback *cb)
3561 {
3562 	return __tipc_dump_start(cb, sock_net(cb->skb->sk));
3563 }
3564 EXPORT_SYMBOL(tipc_dump_start);
3565 
3566 int __tipc_dump_start(struct netlink_callback *cb, struct net *net)
3567 {
3568 	/* tipc_nl_name_table_dump() uses cb->args[0...3]. */
3569 	struct rhashtable_iter *iter = (void *)cb->args[4];
3570 	struct tipc_net *tn = tipc_net(net);
3571 
3572 	if (!iter) {
3573 		iter = kmalloc(sizeof(*iter), GFP_KERNEL);
3574 		if (!iter)
3575 			return -ENOMEM;
3576 
3577 		cb->args[4] = (long)iter;
3578 	}
3579 
3580 	rhashtable_walk_enter(&tn->sk_rht, iter);
3581 	return 0;
3582 }
3583 
3584 int tipc_dump_done(struct netlink_callback *cb)
3585 {
3586 	struct rhashtable_iter *hti = (void *)cb->args[4];
3587 
3588 	rhashtable_walk_exit(hti);
3589 	kfree(hti);
3590 	return 0;
3591 }
3592 EXPORT_SYMBOL(tipc_dump_done);
3593 
3594 int tipc_sk_fill_sock_diag(struct sk_buff *skb, struct netlink_callback *cb,
3595 			   struct tipc_sock *tsk, u32 sk_filter_state,
3596 			   u64 (*tipc_diag_gen_cookie)(struct sock *sk))
3597 {
3598 	struct sock *sk = &tsk->sk;
3599 	struct nlattr *attrs;
3600 	struct nlattr *stat;
3601 
3602 	/*filter response w.r.t sk_state*/
3603 	if (!(sk_filter_state & (1 << sk->sk_state)))
3604 		return 0;
3605 
3606 	attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK);
3607 	if (!attrs)
3608 		goto msg_cancel;
3609 
3610 	if (__tipc_nl_add_sk_info(skb, tsk))
3611 		goto attr_msg_cancel;
3612 
3613 	if (nla_put_u32(skb, TIPC_NLA_SOCK_TYPE, (u32)sk->sk_type) ||
3614 	    nla_put_u32(skb, TIPC_NLA_SOCK_TIPC_STATE, (u32)sk->sk_state) ||
3615 	    nla_put_u32(skb, TIPC_NLA_SOCK_INO, sock_i_ino(sk)) ||
3616 	    nla_put_u32(skb, TIPC_NLA_SOCK_UID,
3617 			from_kuid_munged(sk_user_ns(NETLINK_CB(cb->skb).sk),
3618 					 sock_i_uid(sk))) ||
3619 	    nla_put_u64_64bit(skb, TIPC_NLA_SOCK_COOKIE,
3620 			      tipc_diag_gen_cookie(sk),
3621 			      TIPC_NLA_SOCK_PAD))
3622 		goto attr_msg_cancel;
3623 
3624 	stat = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_STAT);
3625 	if (!stat)
3626 		goto attr_msg_cancel;
3627 
3628 	if (nla_put_u32(skb, TIPC_NLA_SOCK_STAT_RCVQ,
3629 			skb_queue_len(&sk->sk_receive_queue)) ||
3630 	    nla_put_u32(skb, TIPC_NLA_SOCK_STAT_SENDQ,
3631 			skb_queue_len(&sk->sk_write_queue)) ||
3632 	    nla_put_u32(skb, TIPC_NLA_SOCK_STAT_DROP,
3633 			atomic_read(&sk->sk_drops)))
3634 		goto stat_msg_cancel;
3635 
3636 	if (tsk->cong_link_cnt &&
3637 	    nla_put_flag(skb, TIPC_NLA_SOCK_STAT_LINK_CONG))
3638 		goto stat_msg_cancel;
3639 
3640 	if (tsk_conn_cong(tsk) &&
3641 	    nla_put_flag(skb, TIPC_NLA_SOCK_STAT_CONN_CONG))
3642 		goto stat_msg_cancel;
3643 
3644 	nla_nest_end(skb, stat);
3645 
3646 	if (tsk->group)
3647 		if (tipc_group_fill_sock_diag(tsk->group, skb))
3648 			goto stat_msg_cancel;
3649 
3650 	nla_nest_end(skb, attrs);
3651 
3652 	return 0;
3653 
3654 stat_msg_cancel:
3655 	nla_nest_cancel(skb, stat);
3656 attr_msg_cancel:
3657 	nla_nest_cancel(skb, attrs);
3658 msg_cancel:
3659 	return -EMSGSIZE;
3660 }
3661 EXPORT_SYMBOL(tipc_sk_fill_sock_diag);
3662 
3663 int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb)
3664 {
3665 	return tipc_nl_sk_walk(skb, cb, __tipc_nl_add_sk);
3666 }
3667 
3668 /* Caller should hold socket lock for the passed tipc socket. */
3669 static int __tipc_nl_add_sk_publ(struct sk_buff *skb,
3670 				 struct netlink_callback *cb,
3671 				 struct publication *publ)
3672 {
3673 	void *hdr;
3674 	struct nlattr *attrs;
3675 
3676 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
3677 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_PUBL_GET);
3678 	if (!hdr)
3679 		goto msg_cancel;
3680 
3681 	attrs = nla_nest_start_noflag(skb, TIPC_NLA_PUBL);
3682 	if (!attrs)
3683 		goto genlmsg_cancel;
3684 
3685 	if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key))
3686 		goto attr_msg_cancel;
3687 	if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->type))
3688 		goto attr_msg_cancel;
3689 	if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->lower))
3690 		goto attr_msg_cancel;
3691 	if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->upper))
3692 		goto attr_msg_cancel;
3693 
3694 	nla_nest_end(skb, attrs);
3695 	genlmsg_end(skb, hdr);
3696 
3697 	return 0;
3698 
3699 attr_msg_cancel:
3700 	nla_nest_cancel(skb, attrs);
3701 genlmsg_cancel:
3702 	genlmsg_cancel(skb, hdr);
3703 msg_cancel:
3704 	return -EMSGSIZE;
3705 }
3706 
3707 /* Caller should hold socket lock for the passed tipc socket. */
3708 static int __tipc_nl_list_sk_publ(struct sk_buff *skb,
3709 				  struct netlink_callback *cb,
3710 				  struct tipc_sock *tsk, u32 *last_publ)
3711 {
3712 	int err;
3713 	struct publication *p;
3714 
3715 	if (*last_publ) {
3716 		list_for_each_entry(p, &tsk->publications, binding_sock) {
3717 			if (p->key == *last_publ)
3718 				break;
3719 		}
3720 		if (p->key != *last_publ) {
3721 			/* We never set seq or call nl_dump_check_consistent()
3722 			 * this means that setting prev_seq here will cause the
3723 			 * consistence check to fail in the netlink callback
3724 			 * handler. Resulting in the last NLMSG_DONE message
3725 			 * having the NLM_F_DUMP_INTR flag set.
3726 			 */
3727 			cb->prev_seq = 1;
3728 			*last_publ = 0;
3729 			return -EPIPE;
3730 		}
3731 	} else {
3732 		p = list_first_entry(&tsk->publications, struct publication,
3733 				     binding_sock);
3734 	}
3735 
3736 	list_for_each_entry_from(p, &tsk->publications, binding_sock) {
3737 		err = __tipc_nl_add_sk_publ(skb, cb, p);
3738 		if (err) {
3739 			*last_publ = p->key;
3740 			return err;
3741 		}
3742 	}
3743 	*last_publ = 0;
3744 
3745 	return 0;
3746 }
3747 
3748 int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)
3749 {
3750 	int err;
3751 	u32 tsk_portid = cb->args[0];
3752 	u32 last_publ = cb->args[1];
3753 	u32 done = cb->args[2];
3754 	struct net *net = sock_net(skb->sk);
3755 	struct tipc_sock *tsk;
3756 
3757 	if (!tsk_portid) {
3758 		struct nlattr **attrs = genl_dumpit_info(cb)->attrs;
3759 		struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1];
3760 
3761 		if (!attrs[TIPC_NLA_SOCK])
3762 			return -EINVAL;
3763 
3764 		err = nla_parse_nested_deprecated(sock, TIPC_NLA_SOCK_MAX,
3765 						  attrs[TIPC_NLA_SOCK],
3766 						  tipc_nl_sock_policy, NULL);
3767 		if (err)
3768 			return err;
3769 
3770 		if (!sock[TIPC_NLA_SOCK_REF])
3771 			return -EINVAL;
3772 
3773 		tsk_portid = nla_get_u32(sock[TIPC_NLA_SOCK_REF]);
3774 	}
3775 
3776 	if (done)
3777 		return 0;
3778 
3779 	tsk = tipc_sk_lookup(net, tsk_portid);
3780 	if (!tsk)
3781 		return -EINVAL;
3782 
3783 	lock_sock(&tsk->sk);
3784 	err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ);
3785 	if (!err)
3786 		done = 1;
3787 	release_sock(&tsk->sk);
3788 	sock_put(&tsk->sk);
3789 
3790 	cb->args[0] = tsk_portid;
3791 	cb->args[1] = last_publ;
3792 	cb->args[2] = done;
3793 
3794 	return skb->len;
3795 }
3796 
3797 /**
3798  * tipc_sk_filtering - check if a socket should be traced
3799  * @sk: the socket to be examined
3800  * @sysctl_tipc_sk_filter[]: the socket tuple for filtering,
3801  *  (portid, sock type, name type, name lower, name upper)
3802  *
3803  * Returns true if the socket meets the socket tuple data
3804  * (value 0 = 'any') or when there is no tuple set (all = 0),
3805  * otherwise false
3806  */
3807 bool tipc_sk_filtering(struct sock *sk)
3808 {
3809 	struct tipc_sock *tsk;
3810 	struct publication *p;
3811 	u32 _port, _sktype, _type, _lower, _upper;
3812 	u32 type = 0, lower = 0, upper = 0;
3813 
3814 	if (!sk)
3815 		return true;
3816 
3817 	tsk = tipc_sk(sk);
3818 
3819 	_port = sysctl_tipc_sk_filter[0];
3820 	_sktype = sysctl_tipc_sk_filter[1];
3821 	_type = sysctl_tipc_sk_filter[2];
3822 	_lower = sysctl_tipc_sk_filter[3];
3823 	_upper = sysctl_tipc_sk_filter[4];
3824 
3825 	if (!_port && !_sktype && !_type && !_lower && !_upper)
3826 		return true;
3827 
3828 	if (_port)
3829 		return (_port == tsk->portid);
3830 
3831 	if (_sktype && _sktype != sk->sk_type)
3832 		return false;
3833 
3834 	if (tsk->published) {
3835 		p = list_first_entry_or_null(&tsk->publications,
3836 					     struct publication, binding_sock);
3837 		if (p) {
3838 			type = p->type;
3839 			lower = p->lower;
3840 			upper = p->upper;
3841 		}
3842 	}
3843 
3844 	if (!tipc_sk_type_connectionless(sk)) {
3845 		type = tsk->conn_type;
3846 		lower = tsk->conn_instance;
3847 		upper = tsk->conn_instance;
3848 	}
3849 
3850 	if ((_type && _type != type) || (_lower && _lower != lower) ||
3851 	    (_upper && _upper != upper))
3852 		return false;
3853 
3854 	return true;
3855 }
3856 
3857 u32 tipc_sock_get_portid(struct sock *sk)
3858 {
3859 	return (sk) ? (tipc_sk(sk))->portid : 0;
3860 }
3861 
3862 /**
3863  * tipc_sk_overlimit1 - check if socket rx queue is about to be overloaded,
3864  *			both the rcv and backlog queues are considered
3865  * @sk: tipc sk to be checked
3866  * @skb: tipc msg to be checked
3867  *
3868  * Returns true if the socket rx queue allocation is > 90%, otherwise false
3869  */
3870 
3871 bool tipc_sk_overlimit1(struct sock *sk, struct sk_buff *skb)
3872 {
3873 	atomic_t *dcnt = &tipc_sk(sk)->dupl_rcvcnt;
3874 	unsigned int lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
3875 	unsigned int qsize = sk->sk_backlog.len + sk_rmem_alloc_get(sk);
3876 
3877 	return (qsize > lim * 90 / 100);
3878 }
3879 
3880 /**
3881  * tipc_sk_overlimit2 - check if socket rx queue is about to be overloaded,
3882  *			only the rcv queue is considered
3883  * @sk: tipc sk to be checked
3884  * @skb: tipc msg to be checked
3885  *
3886  * Returns true if the socket rx queue allocation is > 90%, otherwise false
3887  */
3888 
3889 bool tipc_sk_overlimit2(struct sock *sk, struct sk_buff *skb)
3890 {
3891 	unsigned int lim = rcvbuf_limit(sk, skb);
3892 	unsigned int qsize = sk_rmem_alloc_get(sk);
3893 
3894 	return (qsize > lim * 90 / 100);
3895 }
3896 
3897 /**
3898  * tipc_sk_dump - dump TIPC socket
3899  * @sk: tipc sk to be dumped
3900  * @dqueues: bitmask to decide if any socket queue to be dumped?
3901  *           - TIPC_DUMP_NONE: don't dump socket queues
3902  *           - TIPC_DUMP_SK_SNDQ: dump socket send queue
3903  *           - TIPC_DUMP_SK_RCVQ: dump socket rcv queue
3904  *           - TIPC_DUMP_SK_BKLGQ: dump socket backlog queue
3905  *           - TIPC_DUMP_ALL: dump all the socket queues above
3906  * @buf: returned buffer of dump data in format
3907  */
3908 int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)
3909 {
3910 	int i = 0;
3911 	size_t sz = (dqueues) ? SK_LMAX : SK_LMIN;
3912 	struct tipc_sock *tsk;
3913 	struct publication *p;
3914 	bool tsk_connected;
3915 
3916 	if (!sk) {
3917 		i += scnprintf(buf, sz, "sk data: (null)\n");
3918 		return i;
3919 	}
3920 
3921 	tsk = tipc_sk(sk);
3922 	tsk_connected = !tipc_sk_type_connectionless(sk);
3923 
3924 	i += scnprintf(buf, sz, "sk data: %u", sk->sk_type);
3925 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_state);
3926 	i += scnprintf(buf + i, sz - i, " %x", tsk_own_node(tsk));
3927 	i += scnprintf(buf + i, sz - i, " %u", tsk->portid);
3928 	i += scnprintf(buf + i, sz - i, " | %u", tsk_connected);
3929 	if (tsk_connected) {
3930 		i += scnprintf(buf + i, sz - i, " %x", tsk_peer_node(tsk));
3931 		i += scnprintf(buf + i, sz - i, " %u", tsk_peer_port(tsk));
3932 		i += scnprintf(buf + i, sz - i, " %u", tsk->conn_type);
3933 		i += scnprintf(buf + i, sz - i, " %u", tsk->conn_instance);
3934 	}
3935 	i += scnprintf(buf + i, sz - i, " | %u", tsk->published);
3936 	if (tsk->published) {
3937 		p = list_first_entry_or_null(&tsk->publications,
3938 					     struct publication, binding_sock);
3939 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->type : 0);
3940 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->lower : 0);
3941 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->upper : 0);
3942 	}
3943 	i += scnprintf(buf + i, sz - i, " | %u", tsk->snd_win);
3944 	i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_win);
3945 	i += scnprintf(buf + i, sz - i, " %u", tsk->max_pkt);
3946 	i += scnprintf(buf + i, sz - i, " %x", tsk->peer_caps);
3947 	i += scnprintf(buf + i, sz - i, " %u", tsk->cong_link_cnt);
3948 	i += scnprintf(buf + i, sz - i, " %u", tsk->snt_unacked);
3949 	i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_unacked);
3950 	i += scnprintf(buf + i, sz - i, " %u", atomic_read(&tsk->dupl_rcvcnt));
3951 	i += scnprintf(buf + i, sz - i, " %u", sk->sk_shutdown);
3952 	i += scnprintf(buf + i, sz - i, " | %d", sk_wmem_alloc_get(sk));
3953 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_sndbuf);
3954 	i += scnprintf(buf + i, sz - i, " | %d", sk_rmem_alloc_get(sk));
3955 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_rcvbuf);
3956 	i += scnprintf(buf + i, sz - i, " | %d\n", READ_ONCE(sk->sk_backlog.len));
3957 
3958 	if (dqueues & TIPC_DUMP_SK_SNDQ) {
3959 		i += scnprintf(buf + i, sz - i, "sk_write_queue: ");
3960 		i += tipc_list_dump(&sk->sk_write_queue, false, buf + i);
3961 	}
3962 
3963 	if (dqueues & TIPC_DUMP_SK_RCVQ) {
3964 		i += scnprintf(buf + i, sz - i, "sk_receive_queue: ");
3965 		i += tipc_list_dump(&sk->sk_receive_queue, false, buf + i);
3966 	}
3967 
3968 	if (dqueues & TIPC_DUMP_SK_BKLGQ) {
3969 		i += scnprintf(buf + i, sz - i, "sk_backlog:\n  head ");
3970 		i += tipc_skb_dump(sk->sk_backlog.head, false, buf + i);
3971 		if (sk->sk_backlog.tail != sk->sk_backlog.head) {
3972 			i += scnprintf(buf + i, sz - i, "  tail ");
3973 			i += tipc_skb_dump(sk->sk_backlog.tail, false,
3974 					   buf + i);
3975 		}
3976 	}
3977 
3978 	return i;
3979 }
3980