xref: /openbmc/linux/net/tipc/socket.c (revision dd5b2498)
1 /*
2  * net/tipc/socket.c: TIPC socket API
3  *
4  * Copyright (c) 2001-2007, 2012-2017, Ericsson AB
5  * Copyright (c) 2004-2008, 2010-2013, Wind River Systems
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions are met:
10  *
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the names of the copyright holders nor the names of its
17  *    contributors may be used to endorse or promote products derived from
18  *    this software without specific prior written permission.
19  *
20  * Alternatively, this software may be distributed under the terms of the
21  * GNU General Public License ("GPL") version 2 as published by the Free
22  * Software Foundation.
23  *
24  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34  * POSSIBILITY OF SUCH DAMAGE.
35  */
36 
37 #include <linux/rhashtable.h>
38 #include <linux/sched/signal.h>
39 
40 #include "core.h"
41 #include "name_table.h"
42 #include "node.h"
43 #include "link.h"
44 #include "name_distr.h"
45 #include "socket.h"
46 #include "bcast.h"
47 #include "netlink.h"
48 #include "group.h"
49 #include "trace.h"
50 
51 #define CONN_TIMEOUT_DEFAULT    8000    /* default connect timeout = 8s */
52 #define CONN_PROBING_INTV	msecs_to_jiffies(3600000)  /* [ms] => 1 h */
53 #define TIPC_FWD_MSG		1
54 #define TIPC_MAX_PORT		0xffffffff
55 #define TIPC_MIN_PORT		1
56 #define TIPC_ACK_RATE		4       /* ACK at 1/4 of of rcv window size */
57 
58 enum {
59 	TIPC_LISTEN = TCP_LISTEN,
60 	TIPC_ESTABLISHED = TCP_ESTABLISHED,
61 	TIPC_OPEN = TCP_CLOSE,
62 	TIPC_DISCONNECTING = TCP_CLOSE_WAIT,
63 	TIPC_CONNECTING = TCP_SYN_SENT,
64 };
65 
66 struct sockaddr_pair {
67 	struct sockaddr_tipc sock;
68 	struct sockaddr_tipc member;
69 };
70 
71 /**
72  * struct tipc_sock - TIPC socket structure
73  * @sk: socket - interacts with 'port' and with user via the socket API
74  * @conn_type: TIPC type used when connection was established
75  * @conn_instance: TIPC instance used when connection was established
76  * @published: non-zero if port has one or more associated names
77  * @max_pkt: maximum packet size "hint" used when building messages sent by port
78  * @portid: unique port identity in TIPC socket hash table
79  * @phdr: preformatted message header used when sending messages
80  * #cong_links: list of congested links
81  * @publications: list of publications for port
82  * @blocking_link: address of the congested link we are currently sleeping on
83  * @pub_count: total # of publications port has made during its lifetime
84  * @conn_timeout: the time we can wait for an unresponded setup request
85  * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue
86  * @cong_link_cnt: number of congested links
87  * @snt_unacked: # messages sent by socket, and not yet acked by peer
88  * @rcv_unacked: # messages read by user, but not yet acked back to peer
89  * @peer: 'connected' peer for dgram/rdm
90  * @node: hash table node
91  * @mc_method: cookie for use between socket and broadcast layer
92  * @rcu: rcu struct for tipc_sock
93  */
94 struct tipc_sock {
95 	struct sock sk;
96 	u32 conn_type;
97 	u32 conn_instance;
98 	int published;
99 	u32 max_pkt;
100 	u32 portid;
101 	struct tipc_msg phdr;
102 	struct list_head cong_links;
103 	struct list_head publications;
104 	u32 pub_count;
105 	atomic_t dupl_rcvcnt;
106 	u16 conn_timeout;
107 	bool probe_unacked;
108 	u16 cong_link_cnt;
109 	u16 snt_unacked;
110 	u16 snd_win;
111 	u16 peer_caps;
112 	u16 rcv_unacked;
113 	u16 rcv_win;
114 	struct sockaddr_tipc peer;
115 	struct rhash_head node;
116 	struct tipc_mc_method mc_method;
117 	struct rcu_head rcu;
118 	struct tipc_group *group;
119 	bool group_is_open;
120 };
121 
122 static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb);
123 static void tipc_data_ready(struct sock *sk);
124 static void tipc_write_space(struct sock *sk);
125 static void tipc_sock_destruct(struct sock *sk);
126 static int tipc_release(struct socket *sock);
127 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags,
128 		       bool kern);
129 static void tipc_sk_timeout(struct timer_list *t);
130 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
131 			   struct tipc_name_seq const *seq);
132 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
133 			    struct tipc_name_seq const *seq);
134 static int tipc_sk_leave(struct tipc_sock *tsk);
135 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid);
136 static int tipc_sk_insert(struct tipc_sock *tsk);
137 static void tipc_sk_remove(struct tipc_sock *tsk);
138 static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz);
139 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dsz);
140 
141 static const struct proto_ops packet_ops;
142 static const struct proto_ops stream_ops;
143 static const struct proto_ops msg_ops;
144 static struct proto tipc_proto;
145 static const struct rhashtable_params tsk_rht_params;
146 
147 static u32 tsk_own_node(struct tipc_sock *tsk)
148 {
149 	return msg_prevnode(&tsk->phdr);
150 }
151 
152 static u32 tsk_peer_node(struct tipc_sock *tsk)
153 {
154 	return msg_destnode(&tsk->phdr);
155 }
156 
157 static u32 tsk_peer_port(struct tipc_sock *tsk)
158 {
159 	return msg_destport(&tsk->phdr);
160 }
161 
162 static  bool tsk_unreliable(struct tipc_sock *tsk)
163 {
164 	return msg_src_droppable(&tsk->phdr) != 0;
165 }
166 
167 static void tsk_set_unreliable(struct tipc_sock *tsk, bool unreliable)
168 {
169 	msg_set_src_droppable(&tsk->phdr, unreliable ? 1 : 0);
170 }
171 
172 static bool tsk_unreturnable(struct tipc_sock *tsk)
173 {
174 	return msg_dest_droppable(&tsk->phdr) != 0;
175 }
176 
177 static void tsk_set_unreturnable(struct tipc_sock *tsk, bool unreturnable)
178 {
179 	msg_set_dest_droppable(&tsk->phdr, unreturnable ? 1 : 0);
180 }
181 
182 static int tsk_importance(struct tipc_sock *tsk)
183 {
184 	return msg_importance(&tsk->phdr);
185 }
186 
187 static int tsk_set_importance(struct tipc_sock *tsk, int imp)
188 {
189 	if (imp > TIPC_CRITICAL_IMPORTANCE)
190 		return -EINVAL;
191 	msg_set_importance(&tsk->phdr, (u32)imp);
192 	return 0;
193 }
194 
195 static struct tipc_sock *tipc_sk(const struct sock *sk)
196 {
197 	return container_of(sk, struct tipc_sock, sk);
198 }
199 
200 static bool tsk_conn_cong(struct tipc_sock *tsk)
201 {
202 	return tsk->snt_unacked > tsk->snd_win;
203 }
204 
205 static u16 tsk_blocks(int len)
206 {
207 	return ((len / FLOWCTL_BLK_SZ) + 1);
208 }
209 
210 /* tsk_blocks(): translate a buffer size in bytes to number of
211  * advertisable blocks, taking into account the ratio truesize(len)/len
212  * We can trust that this ratio is always < 4 for len >= FLOWCTL_BLK_SZ
213  */
214 static u16 tsk_adv_blocks(int len)
215 {
216 	return len / FLOWCTL_BLK_SZ / 4;
217 }
218 
219 /* tsk_inc(): increment counter for sent or received data
220  * - If block based flow control is not supported by peer we
221  *   fall back to message based ditto, incrementing the counter
222  */
223 static u16 tsk_inc(struct tipc_sock *tsk, int msglen)
224 {
225 	if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
226 		return ((msglen / FLOWCTL_BLK_SZ) + 1);
227 	return 1;
228 }
229 
230 /**
231  * tsk_advance_rx_queue - discard first buffer in socket receive queue
232  *
233  * Caller must hold socket lock
234  */
235 static void tsk_advance_rx_queue(struct sock *sk)
236 {
237 	trace_tipc_sk_advance_rx(sk, NULL, TIPC_DUMP_SK_RCVQ, " ");
238 	kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
239 }
240 
241 /* tipc_sk_respond() : send response message back to sender
242  */
243 static void tipc_sk_respond(struct sock *sk, struct sk_buff *skb, int err)
244 {
245 	u32 selector;
246 	u32 dnode;
247 	u32 onode = tipc_own_addr(sock_net(sk));
248 
249 	if (!tipc_msg_reverse(onode, &skb, err))
250 		return;
251 
252 	trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE, "@sk_respond!");
253 	dnode = msg_destnode(buf_msg(skb));
254 	selector = msg_origport(buf_msg(skb));
255 	tipc_node_xmit_skb(sock_net(sk), skb, dnode, selector);
256 }
257 
258 /**
259  * tsk_rej_rx_queue - reject all buffers in socket receive queue
260  *
261  * Caller must hold socket lock
262  */
263 static void tsk_rej_rx_queue(struct sock *sk)
264 {
265 	struct sk_buff *skb;
266 
267 	while ((skb = __skb_dequeue(&sk->sk_receive_queue)))
268 		tipc_sk_respond(sk, skb, TIPC_ERR_NO_PORT);
269 }
270 
271 static bool tipc_sk_connected(struct sock *sk)
272 {
273 	return sk->sk_state == TIPC_ESTABLISHED;
274 }
275 
276 /* tipc_sk_type_connectionless - check if the socket is datagram socket
277  * @sk: socket
278  *
279  * Returns true if connection less, false otherwise
280  */
281 static bool tipc_sk_type_connectionless(struct sock *sk)
282 {
283 	return sk->sk_type == SOCK_RDM || sk->sk_type == SOCK_DGRAM;
284 }
285 
286 /* tsk_peer_msg - verify if message was sent by connected port's peer
287  *
288  * Handles cases where the node's network address has changed from
289  * the default of <0.0.0> to its configured setting.
290  */
291 static bool tsk_peer_msg(struct tipc_sock *tsk, struct tipc_msg *msg)
292 {
293 	struct sock *sk = &tsk->sk;
294 	u32 self = tipc_own_addr(sock_net(sk));
295 	u32 peer_port = tsk_peer_port(tsk);
296 	u32 orig_node, peer_node;
297 
298 	if (unlikely(!tipc_sk_connected(sk)))
299 		return false;
300 
301 	if (unlikely(msg_origport(msg) != peer_port))
302 		return false;
303 
304 	orig_node = msg_orignode(msg);
305 	peer_node = tsk_peer_node(tsk);
306 
307 	if (likely(orig_node == peer_node))
308 		return true;
309 
310 	if (!orig_node && peer_node == self)
311 		return true;
312 
313 	if (!peer_node && orig_node == self)
314 		return true;
315 
316 	return false;
317 }
318 
319 /* tipc_set_sk_state - set the sk_state of the socket
320  * @sk: socket
321  *
322  * Caller must hold socket lock
323  *
324  * Returns 0 on success, errno otherwise
325  */
326 static int tipc_set_sk_state(struct sock *sk, int state)
327 {
328 	int oldsk_state = sk->sk_state;
329 	int res = -EINVAL;
330 
331 	switch (state) {
332 	case TIPC_OPEN:
333 		res = 0;
334 		break;
335 	case TIPC_LISTEN:
336 	case TIPC_CONNECTING:
337 		if (oldsk_state == TIPC_OPEN)
338 			res = 0;
339 		break;
340 	case TIPC_ESTABLISHED:
341 		if (oldsk_state == TIPC_CONNECTING ||
342 		    oldsk_state == TIPC_OPEN)
343 			res = 0;
344 		break;
345 	case TIPC_DISCONNECTING:
346 		if (oldsk_state == TIPC_CONNECTING ||
347 		    oldsk_state == TIPC_ESTABLISHED)
348 			res = 0;
349 		break;
350 	}
351 
352 	if (!res)
353 		sk->sk_state = state;
354 
355 	return res;
356 }
357 
358 static int tipc_sk_sock_err(struct socket *sock, long *timeout)
359 {
360 	struct sock *sk = sock->sk;
361 	int err = sock_error(sk);
362 	int typ = sock->type;
363 
364 	if (err)
365 		return err;
366 	if (typ == SOCK_STREAM || typ == SOCK_SEQPACKET) {
367 		if (sk->sk_state == TIPC_DISCONNECTING)
368 			return -EPIPE;
369 		else if (!tipc_sk_connected(sk))
370 			return -ENOTCONN;
371 	}
372 	if (!*timeout)
373 		return -EAGAIN;
374 	if (signal_pending(current))
375 		return sock_intr_errno(*timeout);
376 
377 	return 0;
378 }
379 
380 #define tipc_wait_for_cond(sock_, timeo_, condition_)			       \
381 ({                                                                             \
382 	DEFINE_WAIT_FUNC(wait_, woken_wake_function);                          \
383 	struct sock *sk_;						       \
384 	int rc_;							       \
385 									       \
386 	while ((rc_ = !(condition_))) {					       \
387 		/* coupled with smp_wmb() in tipc_sk_proto_rcv() */            \
388 		smp_rmb();                                                     \
389 		sk_ = (sock_)->sk;					       \
390 		rc_ = tipc_sk_sock_err((sock_), timeo_);		       \
391 		if (rc_)						       \
392 			break;						       \
393 		add_wait_queue(sk_sleep(sk_), &wait_);                         \
394 		release_sock(sk_);					       \
395 		*(timeo_) = wait_woken(&wait_, TASK_INTERRUPTIBLE, *(timeo_)); \
396 		sched_annotate_sleep();				               \
397 		lock_sock(sk_);						       \
398 		remove_wait_queue(sk_sleep(sk_), &wait_);		       \
399 	}								       \
400 	rc_;								       \
401 })
402 
403 /**
404  * tipc_sk_create - create a TIPC socket
405  * @net: network namespace (must be default network)
406  * @sock: pre-allocated socket structure
407  * @protocol: protocol indicator (must be 0)
408  * @kern: caused by kernel or by userspace?
409  *
410  * This routine creates additional data structures used by the TIPC socket,
411  * initializes them, and links them together.
412  *
413  * Returns 0 on success, errno otherwise
414  */
415 static int tipc_sk_create(struct net *net, struct socket *sock,
416 			  int protocol, int kern)
417 {
418 	const struct proto_ops *ops;
419 	struct sock *sk;
420 	struct tipc_sock *tsk;
421 	struct tipc_msg *msg;
422 
423 	/* Validate arguments */
424 	if (unlikely(protocol != 0))
425 		return -EPROTONOSUPPORT;
426 
427 	switch (sock->type) {
428 	case SOCK_STREAM:
429 		ops = &stream_ops;
430 		break;
431 	case SOCK_SEQPACKET:
432 		ops = &packet_ops;
433 		break;
434 	case SOCK_DGRAM:
435 	case SOCK_RDM:
436 		ops = &msg_ops;
437 		break;
438 	default:
439 		return -EPROTOTYPE;
440 	}
441 
442 	/* Allocate socket's protocol area */
443 	sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto, kern);
444 	if (sk == NULL)
445 		return -ENOMEM;
446 
447 	tsk = tipc_sk(sk);
448 	tsk->max_pkt = MAX_PKT_DEFAULT;
449 	INIT_LIST_HEAD(&tsk->publications);
450 	INIT_LIST_HEAD(&tsk->cong_links);
451 	msg = &tsk->phdr;
452 
453 	/* Finish initializing socket data structures */
454 	sock->ops = ops;
455 	sock_init_data(sock, sk);
456 	tipc_set_sk_state(sk, TIPC_OPEN);
457 	if (tipc_sk_insert(tsk)) {
458 		pr_warn("Socket create failed; port number exhausted\n");
459 		return -EINVAL;
460 	}
461 
462 	/* Ensure tsk is visible before we read own_addr. */
463 	smp_mb();
464 
465 	tipc_msg_init(tipc_own_addr(net), msg, TIPC_LOW_IMPORTANCE,
466 		      TIPC_NAMED_MSG, NAMED_H_SIZE, 0);
467 
468 	msg_set_origport(msg, tsk->portid);
469 	timer_setup(&sk->sk_timer, tipc_sk_timeout, 0);
470 	sk->sk_shutdown = 0;
471 	sk->sk_backlog_rcv = tipc_sk_backlog_rcv;
472 	sk->sk_rcvbuf = sysctl_tipc_rmem[1];
473 	sk->sk_data_ready = tipc_data_ready;
474 	sk->sk_write_space = tipc_write_space;
475 	sk->sk_destruct = tipc_sock_destruct;
476 	tsk->conn_timeout = CONN_TIMEOUT_DEFAULT;
477 	tsk->group_is_open = true;
478 	atomic_set(&tsk->dupl_rcvcnt, 0);
479 
480 	/* Start out with safe limits until we receive an advertised window */
481 	tsk->snd_win = tsk_adv_blocks(RCVBUF_MIN);
482 	tsk->rcv_win = tsk->snd_win;
483 
484 	if (tipc_sk_type_connectionless(sk)) {
485 		tsk_set_unreturnable(tsk, true);
486 		if (sock->type == SOCK_DGRAM)
487 			tsk_set_unreliable(tsk, true);
488 		__skb_queue_head_init(&tsk->mc_method.deferredq);
489 	}
490 
491 	trace_tipc_sk_create(sk, NULL, TIPC_DUMP_NONE, " ");
492 	return 0;
493 }
494 
495 static void tipc_sk_callback(struct rcu_head *head)
496 {
497 	struct tipc_sock *tsk = container_of(head, struct tipc_sock, rcu);
498 
499 	sock_put(&tsk->sk);
500 }
501 
502 /* Caller should hold socket lock for the socket. */
503 static void __tipc_shutdown(struct socket *sock, int error)
504 {
505 	struct sock *sk = sock->sk;
506 	struct tipc_sock *tsk = tipc_sk(sk);
507 	struct net *net = sock_net(sk);
508 	long timeout = CONN_TIMEOUT_DEFAULT;
509 	u32 dnode = tsk_peer_node(tsk);
510 	struct sk_buff *skb;
511 
512 	/* Avoid that hi-prio shutdown msgs bypass msgs in link wakeup queue */
513 	tipc_wait_for_cond(sock, &timeout, (!tsk->cong_link_cnt &&
514 					    !tsk_conn_cong(tsk)));
515 
516 	/* Remove any pending SYN message */
517 	__skb_queue_purge(&sk->sk_write_queue);
518 
519 	/* Reject all unreceived messages, except on an active connection
520 	 * (which disconnects locally & sends a 'FIN+' to peer).
521 	 */
522 	while ((skb = __skb_dequeue(&sk->sk_receive_queue)) != NULL) {
523 		if (TIPC_SKB_CB(skb)->bytes_read) {
524 			kfree_skb(skb);
525 			continue;
526 		}
527 		if (!tipc_sk_type_connectionless(sk) &&
528 		    sk->sk_state != TIPC_DISCONNECTING) {
529 			tipc_set_sk_state(sk, TIPC_DISCONNECTING);
530 			tipc_node_remove_conn(net, dnode, tsk->portid);
531 		}
532 		tipc_sk_respond(sk, skb, error);
533 	}
534 
535 	if (tipc_sk_type_connectionless(sk))
536 		return;
537 
538 	if (sk->sk_state != TIPC_DISCONNECTING) {
539 		skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE,
540 				      TIPC_CONN_MSG, SHORT_H_SIZE, 0, dnode,
541 				      tsk_own_node(tsk), tsk_peer_port(tsk),
542 				      tsk->portid, error);
543 		if (skb)
544 			tipc_node_xmit_skb(net, skb, dnode, tsk->portid);
545 		tipc_node_remove_conn(net, dnode, tsk->portid);
546 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
547 	}
548 }
549 
550 /**
551  * tipc_release - destroy a TIPC socket
552  * @sock: socket to destroy
553  *
554  * This routine cleans up any messages that are still queued on the socket.
555  * For DGRAM and RDM socket types, all queued messages are rejected.
556  * For SEQPACKET and STREAM socket types, the first message is rejected
557  * and any others are discarded.  (If the first message on a STREAM socket
558  * is partially-read, it is discarded and the next one is rejected instead.)
559  *
560  * NOTE: Rejected messages are not necessarily returned to the sender!  They
561  * are returned or discarded according to the "destination droppable" setting
562  * specified for the message by the sender.
563  *
564  * Returns 0 on success, errno otherwise
565  */
566 static int tipc_release(struct socket *sock)
567 {
568 	struct sock *sk = sock->sk;
569 	struct tipc_sock *tsk;
570 
571 	/*
572 	 * Exit if socket isn't fully initialized (occurs when a failed accept()
573 	 * releases a pre-allocated child socket that was never used)
574 	 */
575 	if (sk == NULL)
576 		return 0;
577 
578 	tsk = tipc_sk(sk);
579 	lock_sock(sk);
580 
581 	trace_tipc_sk_release(sk, NULL, TIPC_DUMP_ALL, " ");
582 	__tipc_shutdown(sock, TIPC_ERR_NO_PORT);
583 	sk->sk_shutdown = SHUTDOWN_MASK;
584 	tipc_sk_leave(tsk);
585 	tipc_sk_withdraw(tsk, 0, NULL);
586 	__skb_queue_purge(&tsk->mc_method.deferredq);
587 	sk_stop_timer(sk, &sk->sk_timer);
588 	tipc_sk_remove(tsk);
589 
590 	sock_orphan(sk);
591 	/* Reject any messages that accumulated in backlog queue */
592 	release_sock(sk);
593 	tipc_dest_list_purge(&tsk->cong_links);
594 	tsk->cong_link_cnt = 0;
595 	call_rcu(&tsk->rcu, tipc_sk_callback);
596 	sock->sk = NULL;
597 
598 	return 0;
599 }
600 
601 /**
602  * tipc_bind - associate or disassocate TIPC name(s) with a socket
603  * @sock: socket structure
604  * @uaddr: socket address describing name(s) and desired operation
605  * @uaddr_len: size of socket address data structure
606  *
607  * Name and name sequence binding is indicated using a positive scope value;
608  * a negative scope value unbinds the specified name.  Specifying no name
609  * (i.e. a socket address length of 0) unbinds all names from the socket.
610  *
611  * Returns 0 on success, errno otherwise
612  *
613  * NOTE: This routine doesn't need to take the socket lock since it doesn't
614  *       access any non-constant socket information.
615  */
616 static int tipc_bind(struct socket *sock, struct sockaddr *uaddr,
617 		     int uaddr_len)
618 {
619 	struct sock *sk = sock->sk;
620 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
621 	struct tipc_sock *tsk = tipc_sk(sk);
622 	int res = -EINVAL;
623 
624 	lock_sock(sk);
625 	if (unlikely(!uaddr_len)) {
626 		res = tipc_sk_withdraw(tsk, 0, NULL);
627 		goto exit;
628 	}
629 	if (tsk->group) {
630 		res = -EACCES;
631 		goto exit;
632 	}
633 	if (uaddr_len < sizeof(struct sockaddr_tipc)) {
634 		res = -EINVAL;
635 		goto exit;
636 	}
637 	if (addr->family != AF_TIPC) {
638 		res = -EAFNOSUPPORT;
639 		goto exit;
640 	}
641 
642 	if (addr->addrtype == TIPC_ADDR_NAME)
643 		addr->addr.nameseq.upper = addr->addr.nameseq.lower;
644 	else if (addr->addrtype != TIPC_ADDR_NAMESEQ) {
645 		res = -EAFNOSUPPORT;
646 		goto exit;
647 	}
648 
649 	if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) &&
650 	    (addr->addr.nameseq.type != TIPC_TOP_SRV) &&
651 	    (addr->addr.nameseq.type != TIPC_CFG_SRV)) {
652 		res = -EACCES;
653 		goto exit;
654 	}
655 
656 	res = (addr->scope >= 0) ?
657 		tipc_sk_publish(tsk, addr->scope, &addr->addr.nameseq) :
658 		tipc_sk_withdraw(tsk, -addr->scope, &addr->addr.nameseq);
659 exit:
660 	release_sock(sk);
661 	return res;
662 }
663 
664 /**
665  * tipc_getname - get port ID of socket or peer socket
666  * @sock: socket structure
667  * @uaddr: area for returned socket address
668  * @uaddr_len: area for returned length of socket address
669  * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
670  *
671  * Returns 0 on success, errno otherwise
672  *
673  * NOTE: This routine doesn't need to take the socket lock since it only
674  *       accesses socket information that is unchanging (or which changes in
675  *       a completely predictable manner).
676  */
677 static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
678 			int peer)
679 {
680 	struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
681 	struct sock *sk = sock->sk;
682 	struct tipc_sock *tsk = tipc_sk(sk);
683 
684 	memset(addr, 0, sizeof(*addr));
685 	if (peer) {
686 		if ((!tipc_sk_connected(sk)) &&
687 		    ((peer != 2) || (sk->sk_state != TIPC_DISCONNECTING)))
688 			return -ENOTCONN;
689 		addr->addr.id.ref = tsk_peer_port(tsk);
690 		addr->addr.id.node = tsk_peer_node(tsk);
691 	} else {
692 		addr->addr.id.ref = tsk->portid;
693 		addr->addr.id.node = tipc_own_addr(sock_net(sk));
694 	}
695 
696 	addr->addrtype = TIPC_ADDR_ID;
697 	addr->family = AF_TIPC;
698 	addr->scope = 0;
699 	addr->addr.name.domain = 0;
700 
701 	return sizeof(*addr);
702 }
703 
704 /**
705  * tipc_poll - read and possibly block on pollmask
706  * @file: file structure associated with the socket
707  * @sock: socket for which to calculate the poll bits
708  * @wait: ???
709  *
710  * Returns pollmask value
711  *
712  * COMMENTARY:
713  * It appears that the usual socket locking mechanisms are not useful here
714  * since the pollmask info is potentially out-of-date the moment this routine
715  * exits.  TCP and other protocols seem to rely on higher level poll routines
716  * to handle any preventable race conditions, so TIPC will do the same ...
717  *
718  * IMPORTANT: The fact that a read or write operation is indicated does NOT
719  * imply that the operation will succeed, merely that it should be performed
720  * and will not block.
721  */
722 static __poll_t tipc_poll(struct file *file, struct socket *sock,
723 			      poll_table *wait)
724 {
725 	struct sock *sk = sock->sk;
726 	struct tipc_sock *tsk = tipc_sk(sk);
727 	__poll_t revents = 0;
728 
729 	sock_poll_wait(file, sock, wait);
730 	trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " ");
731 
732 	if (sk->sk_shutdown & RCV_SHUTDOWN)
733 		revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
734 	if (sk->sk_shutdown == SHUTDOWN_MASK)
735 		revents |= EPOLLHUP;
736 
737 	switch (sk->sk_state) {
738 	case TIPC_ESTABLISHED:
739 	case TIPC_CONNECTING:
740 		if (!tsk->cong_link_cnt && !tsk_conn_cong(tsk))
741 			revents |= EPOLLOUT;
742 		/* fall through */
743 	case TIPC_LISTEN:
744 		if (!skb_queue_empty(&sk->sk_receive_queue))
745 			revents |= EPOLLIN | EPOLLRDNORM;
746 		break;
747 	case TIPC_OPEN:
748 		if (tsk->group_is_open && !tsk->cong_link_cnt)
749 			revents |= EPOLLOUT;
750 		if (!tipc_sk_type_connectionless(sk))
751 			break;
752 		if (skb_queue_empty(&sk->sk_receive_queue))
753 			break;
754 		revents |= EPOLLIN | EPOLLRDNORM;
755 		break;
756 	case TIPC_DISCONNECTING:
757 		revents = EPOLLIN | EPOLLRDNORM | EPOLLHUP;
758 		break;
759 	}
760 	return revents;
761 }
762 
763 /**
764  * tipc_sendmcast - send multicast message
765  * @sock: socket structure
766  * @seq: destination address
767  * @msg: message to send
768  * @dlen: length of data to send
769  * @timeout: timeout to wait for wakeup
770  *
771  * Called from function tipc_sendmsg(), which has done all sanity checks
772  * Returns the number of bytes sent on success, or errno
773  */
774 static int tipc_sendmcast(struct  socket *sock, struct tipc_name_seq *seq,
775 			  struct msghdr *msg, size_t dlen, long timeout)
776 {
777 	struct sock *sk = sock->sk;
778 	struct tipc_sock *tsk = tipc_sk(sk);
779 	struct tipc_msg *hdr = &tsk->phdr;
780 	struct net *net = sock_net(sk);
781 	int mtu = tipc_bcast_get_mtu(net);
782 	struct tipc_mc_method *method = &tsk->mc_method;
783 	struct sk_buff_head pkts;
784 	struct tipc_nlist dsts;
785 	int rc;
786 
787 	if (tsk->group)
788 		return -EACCES;
789 
790 	/* Block or return if any destination link is congested */
791 	rc = tipc_wait_for_cond(sock, &timeout, !tsk->cong_link_cnt);
792 	if (unlikely(rc))
793 		return rc;
794 
795 	/* Lookup destination nodes */
796 	tipc_nlist_init(&dsts, tipc_own_addr(net));
797 	tipc_nametbl_lookup_dst_nodes(net, seq->type, seq->lower,
798 				      seq->upper, &dsts);
799 	if (!dsts.local && !dsts.remote)
800 		return -EHOSTUNREACH;
801 
802 	/* Build message header */
803 	msg_set_type(hdr, TIPC_MCAST_MSG);
804 	msg_set_hdr_sz(hdr, MCAST_H_SIZE);
805 	msg_set_lookup_scope(hdr, TIPC_CLUSTER_SCOPE);
806 	msg_set_destport(hdr, 0);
807 	msg_set_destnode(hdr, 0);
808 	msg_set_nametype(hdr, seq->type);
809 	msg_set_namelower(hdr, seq->lower);
810 	msg_set_nameupper(hdr, seq->upper);
811 
812 	/* Build message as chain of buffers */
813 	skb_queue_head_init(&pkts);
814 	rc = tipc_msg_build(hdr, msg, 0, dlen, mtu, &pkts);
815 
816 	/* Send message if build was successful */
817 	if (unlikely(rc == dlen)) {
818 		trace_tipc_sk_sendmcast(sk, skb_peek(&pkts),
819 					TIPC_DUMP_SK_SNDQ, " ");
820 		rc = tipc_mcast_xmit(net, &pkts, method, &dsts,
821 				     &tsk->cong_link_cnt);
822 	}
823 
824 	tipc_nlist_purge(&dsts);
825 
826 	return rc ? rc : dlen;
827 }
828 
829 /**
830  * tipc_send_group_msg - send a message to a member in the group
831  * @net: network namespace
832  * @m: message to send
833  * @mb: group member
834  * @dnode: destination node
835  * @dport: destination port
836  * @dlen: total length of message data
837  */
838 static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
839 			       struct msghdr *m, struct tipc_member *mb,
840 			       u32 dnode, u32 dport, int dlen)
841 {
842 	u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group);
843 	struct tipc_mc_method *method = &tsk->mc_method;
844 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
845 	struct tipc_msg *hdr = &tsk->phdr;
846 	struct sk_buff_head pkts;
847 	int mtu, rc;
848 
849 	/* Complete message header */
850 	msg_set_type(hdr, TIPC_GRP_UCAST_MSG);
851 	msg_set_hdr_sz(hdr, GROUP_H_SIZE);
852 	msg_set_destport(hdr, dport);
853 	msg_set_destnode(hdr, dnode);
854 	msg_set_grp_bc_seqno(hdr, bc_snd_nxt);
855 
856 	/* Build message as chain of buffers */
857 	skb_queue_head_init(&pkts);
858 	mtu = tipc_node_get_mtu(net, dnode, tsk->portid);
859 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
860 	if (unlikely(rc != dlen))
861 		return rc;
862 
863 	/* Send message */
864 	rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
865 	if (unlikely(rc == -ELINKCONG)) {
866 		tipc_dest_push(&tsk->cong_links, dnode, 0);
867 		tsk->cong_link_cnt++;
868 	}
869 
870 	/* Update send window */
871 	tipc_group_update_member(mb, blks);
872 
873 	/* A broadcast sent within next EXPIRE period must follow same path */
874 	method->rcast = true;
875 	method->mandatory = true;
876 	return dlen;
877 }
878 
879 /**
880  * tipc_send_group_unicast - send message to a member in the group
881  * @sock: socket structure
882  * @m: message to send
883  * @dlen: total length of message data
884  * @timeout: timeout to wait for wakeup
885  *
886  * Called from function tipc_sendmsg(), which has done all sanity checks
887  * Returns the number of bytes sent on success, or errno
888  */
889 static int tipc_send_group_unicast(struct socket *sock, struct msghdr *m,
890 				   int dlen, long timeout)
891 {
892 	struct sock *sk = sock->sk;
893 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
894 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
895 	struct tipc_sock *tsk = tipc_sk(sk);
896 	struct net *net = sock_net(sk);
897 	struct tipc_member *mb = NULL;
898 	u32 node, port;
899 	int rc;
900 
901 	node = dest->addr.id.node;
902 	port = dest->addr.id.ref;
903 	if (!port && !node)
904 		return -EHOSTUNREACH;
905 
906 	/* Block or return if destination link or member is congested */
907 	rc = tipc_wait_for_cond(sock, &timeout,
908 				!tipc_dest_find(&tsk->cong_links, node, 0) &&
909 				tsk->group &&
910 				!tipc_group_cong(tsk->group, node, port, blks,
911 						 &mb));
912 	if (unlikely(rc))
913 		return rc;
914 
915 	if (unlikely(!mb))
916 		return -EHOSTUNREACH;
917 
918 	rc = tipc_send_group_msg(net, tsk, m, mb, node, port, dlen);
919 
920 	return rc ? rc : dlen;
921 }
922 
923 /**
924  * tipc_send_group_anycast - send message to any member with given identity
925  * @sock: socket structure
926  * @m: message to send
927  * @dlen: total length of message data
928  * @timeout: timeout to wait for wakeup
929  *
930  * Called from function tipc_sendmsg(), which has done all sanity checks
931  * Returns the number of bytes sent on success, or errno
932  */
933 static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m,
934 				   int dlen, long timeout)
935 {
936 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
937 	struct sock *sk = sock->sk;
938 	struct tipc_sock *tsk = tipc_sk(sk);
939 	struct list_head *cong_links = &tsk->cong_links;
940 	int blks = tsk_blocks(GROUP_H_SIZE + dlen);
941 	struct tipc_msg *hdr = &tsk->phdr;
942 	struct tipc_member *first = NULL;
943 	struct tipc_member *mbr = NULL;
944 	struct net *net = sock_net(sk);
945 	u32 node, port, exclude;
946 	struct list_head dsts;
947 	u32 type, inst, scope;
948 	int lookups = 0;
949 	int dstcnt, rc;
950 	bool cong;
951 
952 	INIT_LIST_HEAD(&dsts);
953 
954 	type = msg_nametype(hdr);
955 	inst = dest->addr.name.name.instance;
956 	scope = msg_lookup_scope(hdr);
957 
958 	while (++lookups < 4) {
959 		exclude = tipc_group_exclude(tsk->group);
960 
961 		first = NULL;
962 
963 		/* Look for a non-congested destination member, if any */
964 		while (1) {
965 			if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
966 						 &dstcnt, exclude, false))
967 				return -EHOSTUNREACH;
968 			tipc_dest_pop(&dsts, &node, &port);
969 			cong = tipc_group_cong(tsk->group, node, port, blks,
970 					       &mbr);
971 			if (!cong)
972 				break;
973 			if (mbr == first)
974 				break;
975 			if (!first)
976 				first = mbr;
977 		}
978 
979 		/* Start over if destination was not in member list */
980 		if (unlikely(!mbr))
981 			continue;
982 
983 		if (likely(!cong && !tipc_dest_find(cong_links, node, 0)))
984 			break;
985 
986 		/* Block or return if destination link or member is congested */
987 		rc = tipc_wait_for_cond(sock, &timeout,
988 					!tipc_dest_find(cong_links, node, 0) &&
989 					tsk->group &&
990 					!tipc_group_cong(tsk->group, node, port,
991 							 blks, &mbr));
992 		if (unlikely(rc))
993 			return rc;
994 
995 		/* Send, unless destination disappeared while waiting */
996 		if (likely(mbr))
997 			break;
998 	}
999 
1000 	if (unlikely(lookups >= 4))
1001 		return -EHOSTUNREACH;
1002 
1003 	rc = tipc_send_group_msg(net, tsk, m, mbr, node, port, dlen);
1004 
1005 	return rc ? rc : dlen;
1006 }
1007 
1008 /**
1009  * tipc_send_group_bcast - send message to all members in communication group
1010  * @sk: socket structure
1011  * @m: message to send
1012  * @dlen: total length of message data
1013  * @timeout: timeout to wait for wakeup
1014  *
1015  * Called from function tipc_sendmsg(), which has done all sanity checks
1016  * Returns the number of bytes sent on success, or errno
1017  */
1018 static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
1019 				 int dlen, long timeout)
1020 {
1021 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1022 	struct sock *sk = sock->sk;
1023 	struct net *net = sock_net(sk);
1024 	struct tipc_sock *tsk = tipc_sk(sk);
1025 	struct tipc_nlist *dsts;
1026 	struct tipc_mc_method *method = &tsk->mc_method;
1027 	bool ack = method->mandatory && method->rcast;
1028 	int blks = tsk_blocks(MCAST_H_SIZE + dlen);
1029 	struct tipc_msg *hdr = &tsk->phdr;
1030 	int mtu = tipc_bcast_get_mtu(net);
1031 	struct sk_buff_head pkts;
1032 	int rc = -EHOSTUNREACH;
1033 
1034 	/* Block or return if any destination link or member is congested */
1035 	rc = tipc_wait_for_cond(sock, &timeout,
1036 				!tsk->cong_link_cnt && tsk->group &&
1037 				!tipc_group_bc_cong(tsk->group, blks));
1038 	if (unlikely(rc))
1039 		return rc;
1040 
1041 	dsts = tipc_group_dests(tsk->group);
1042 	if (!dsts->local && !dsts->remote)
1043 		return -EHOSTUNREACH;
1044 
1045 	/* Complete message header */
1046 	if (dest) {
1047 		msg_set_type(hdr, TIPC_GRP_MCAST_MSG);
1048 		msg_set_nameinst(hdr, dest->addr.name.name.instance);
1049 	} else {
1050 		msg_set_type(hdr, TIPC_GRP_BCAST_MSG);
1051 		msg_set_nameinst(hdr, 0);
1052 	}
1053 	msg_set_hdr_sz(hdr, GROUP_H_SIZE);
1054 	msg_set_destport(hdr, 0);
1055 	msg_set_destnode(hdr, 0);
1056 	msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(tsk->group));
1057 
1058 	/* Avoid getting stuck with repeated forced replicasts */
1059 	msg_set_grp_bc_ack_req(hdr, ack);
1060 
1061 	/* Build message as chain of buffers */
1062 	skb_queue_head_init(&pkts);
1063 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1064 	if (unlikely(rc != dlen))
1065 		return rc;
1066 
1067 	/* Send message */
1068 	rc = tipc_mcast_xmit(net, &pkts, method, dsts, &tsk->cong_link_cnt);
1069 	if (unlikely(rc))
1070 		return rc;
1071 
1072 	/* Update broadcast sequence number and send windows */
1073 	tipc_group_update_bc_members(tsk->group, blks, ack);
1074 
1075 	/* Broadcast link is now free to choose method for next broadcast */
1076 	method->mandatory = false;
1077 	method->expires = jiffies;
1078 
1079 	return dlen;
1080 }
1081 
1082 /**
1083  * tipc_send_group_mcast - send message to all members with given identity
1084  * @sock: socket structure
1085  * @m: message to send
1086  * @dlen: total length of message data
1087  * @timeout: timeout to wait for wakeup
1088  *
1089  * Called from function tipc_sendmsg(), which has done all sanity checks
1090  * Returns the number of bytes sent on success, or errno
1091  */
1092 static int tipc_send_group_mcast(struct socket *sock, struct msghdr *m,
1093 				 int dlen, long timeout)
1094 {
1095 	struct sock *sk = sock->sk;
1096 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1097 	struct tipc_sock *tsk = tipc_sk(sk);
1098 	struct tipc_group *grp = tsk->group;
1099 	struct tipc_msg *hdr = &tsk->phdr;
1100 	struct net *net = sock_net(sk);
1101 	u32 type, inst, scope, exclude;
1102 	struct list_head dsts;
1103 	u32 dstcnt;
1104 
1105 	INIT_LIST_HEAD(&dsts);
1106 
1107 	type = msg_nametype(hdr);
1108 	inst = dest->addr.name.name.instance;
1109 	scope = msg_lookup_scope(hdr);
1110 	exclude = tipc_group_exclude(grp);
1111 
1112 	if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
1113 				 &dstcnt, exclude, true))
1114 		return -EHOSTUNREACH;
1115 
1116 	if (dstcnt == 1) {
1117 		tipc_dest_pop(&dsts, &dest->addr.id.node, &dest->addr.id.ref);
1118 		return tipc_send_group_unicast(sock, m, dlen, timeout);
1119 	}
1120 
1121 	tipc_dest_list_purge(&dsts);
1122 	return tipc_send_group_bcast(sock, m, dlen, timeout);
1123 }
1124 
1125 /**
1126  * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
1127  * @arrvq: queue with arriving messages, to be cloned after destination lookup
1128  * @inputq: queue with cloned messages, delivered to socket after dest lookup
1129  *
1130  * Multi-threaded: parallel calls with reference to same queues may occur
1131  */
1132 void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
1133 		       struct sk_buff_head *inputq)
1134 {
1135 	u32 self = tipc_own_addr(net);
1136 	u32 type, lower, upper, scope;
1137 	struct sk_buff *skb, *_skb;
1138 	u32 portid, onode;
1139 	struct sk_buff_head tmpq;
1140 	struct list_head dports;
1141 	struct tipc_msg *hdr;
1142 	int user, mtyp, hlen;
1143 	bool exact;
1144 
1145 	__skb_queue_head_init(&tmpq);
1146 	INIT_LIST_HEAD(&dports);
1147 
1148 	skb = tipc_skb_peek(arrvq, &inputq->lock);
1149 	for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) {
1150 		hdr = buf_msg(skb);
1151 		user = msg_user(hdr);
1152 		mtyp = msg_type(hdr);
1153 		hlen = skb_headroom(skb) + msg_hdr_sz(hdr);
1154 		onode = msg_orignode(hdr);
1155 		type = msg_nametype(hdr);
1156 
1157 		if (mtyp == TIPC_GRP_UCAST_MSG || user == GROUP_PROTOCOL) {
1158 			spin_lock_bh(&inputq->lock);
1159 			if (skb_peek(arrvq) == skb) {
1160 				__skb_dequeue(arrvq);
1161 				__skb_queue_tail(inputq, skb);
1162 			}
1163 			kfree_skb(skb);
1164 			spin_unlock_bh(&inputq->lock);
1165 			continue;
1166 		}
1167 
1168 		/* Group messages require exact scope match */
1169 		if (msg_in_group(hdr)) {
1170 			lower = 0;
1171 			upper = ~0;
1172 			scope = msg_lookup_scope(hdr);
1173 			exact = true;
1174 		} else {
1175 			/* TIPC_NODE_SCOPE means "any scope" in this context */
1176 			if (onode == self)
1177 				scope = TIPC_NODE_SCOPE;
1178 			else
1179 				scope = TIPC_CLUSTER_SCOPE;
1180 			exact = false;
1181 			lower = msg_namelower(hdr);
1182 			upper = msg_nameupper(hdr);
1183 		}
1184 
1185 		/* Create destination port list: */
1186 		tipc_nametbl_mc_lookup(net, type, lower, upper,
1187 				       scope, exact, &dports);
1188 
1189 		/* Clone message per destination */
1190 		while (tipc_dest_pop(&dports, NULL, &portid)) {
1191 			_skb = __pskb_copy(skb, hlen, GFP_ATOMIC);
1192 			if (_skb) {
1193 				msg_set_destport(buf_msg(_skb), portid);
1194 				__skb_queue_tail(&tmpq, _skb);
1195 				continue;
1196 			}
1197 			pr_warn("Failed to clone mcast rcv buffer\n");
1198 		}
1199 		/* Append to inputq if not already done by other thread */
1200 		spin_lock_bh(&inputq->lock);
1201 		if (skb_peek(arrvq) == skb) {
1202 			skb_queue_splice_tail_init(&tmpq, inputq);
1203 			kfree_skb(__skb_dequeue(arrvq));
1204 		}
1205 		spin_unlock_bh(&inputq->lock);
1206 		__skb_queue_purge(&tmpq);
1207 		kfree_skb(skb);
1208 	}
1209 	tipc_sk_rcv(net, inputq);
1210 }
1211 
1212 /**
1213  * tipc_sk_conn_proto_rcv - receive a connection mng protocol message
1214  * @tsk: receiving socket
1215  * @skb: pointer to message buffer.
1216  */
1217 static void tipc_sk_conn_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb,
1218 				   struct sk_buff_head *inputq,
1219 				   struct sk_buff_head *xmitq)
1220 {
1221 	struct tipc_msg *hdr = buf_msg(skb);
1222 	u32 onode = tsk_own_node(tsk);
1223 	struct sock *sk = &tsk->sk;
1224 	int mtyp = msg_type(hdr);
1225 	bool conn_cong;
1226 
1227 	/* Ignore if connection cannot be validated: */
1228 	if (!tsk_peer_msg(tsk, hdr)) {
1229 		trace_tipc_sk_drop_msg(sk, skb, TIPC_DUMP_NONE, "@proto_rcv!");
1230 		goto exit;
1231 	}
1232 
1233 	if (unlikely(msg_errcode(hdr))) {
1234 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
1235 		tipc_node_remove_conn(sock_net(sk), tsk_peer_node(tsk),
1236 				      tsk_peer_port(tsk));
1237 		sk->sk_state_change(sk);
1238 
1239 		/* State change is ignored if socket already awake,
1240 		 * - convert msg to abort msg and add to inqueue
1241 		 */
1242 		msg_set_user(hdr, TIPC_CRITICAL_IMPORTANCE);
1243 		msg_set_type(hdr, TIPC_CONN_MSG);
1244 		msg_set_size(hdr, BASIC_H_SIZE);
1245 		msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1246 		__skb_queue_tail(inputq, skb);
1247 		return;
1248 	}
1249 
1250 	tsk->probe_unacked = false;
1251 
1252 	if (mtyp == CONN_PROBE) {
1253 		msg_set_type(hdr, CONN_PROBE_REPLY);
1254 		if (tipc_msg_reverse(onode, &skb, TIPC_OK))
1255 			__skb_queue_tail(xmitq, skb);
1256 		return;
1257 	} else if (mtyp == CONN_ACK) {
1258 		conn_cong = tsk_conn_cong(tsk);
1259 		tsk->snt_unacked -= msg_conn_ack(hdr);
1260 		if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1261 			tsk->snd_win = msg_adv_win(hdr);
1262 		if (conn_cong)
1263 			sk->sk_write_space(sk);
1264 	} else if (mtyp != CONN_PROBE_REPLY) {
1265 		pr_warn("Received unknown CONN_PROTO msg\n");
1266 	}
1267 exit:
1268 	kfree_skb(skb);
1269 }
1270 
1271 /**
1272  * tipc_sendmsg - send message in connectionless manner
1273  * @sock: socket structure
1274  * @m: message to send
1275  * @dsz: amount of user data to be sent
1276  *
1277  * Message must have an destination specified explicitly.
1278  * Used for SOCK_RDM and SOCK_DGRAM messages,
1279  * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
1280  * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
1281  *
1282  * Returns the number of bytes sent on success, or errno otherwise
1283  */
1284 static int tipc_sendmsg(struct socket *sock,
1285 			struct msghdr *m, size_t dsz)
1286 {
1287 	struct sock *sk = sock->sk;
1288 	int ret;
1289 
1290 	lock_sock(sk);
1291 	ret = __tipc_sendmsg(sock, m, dsz);
1292 	release_sock(sk);
1293 
1294 	return ret;
1295 }
1296 
1297 static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)
1298 {
1299 	struct sock *sk = sock->sk;
1300 	struct net *net = sock_net(sk);
1301 	struct tipc_sock *tsk = tipc_sk(sk);
1302 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1303 	long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1304 	struct list_head *clinks = &tsk->cong_links;
1305 	bool syn = !tipc_sk_type_connectionless(sk);
1306 	struct tipc_group *grp = tsk->group;
1307 	struct tipc_msg *hdr = &tsk->phdr;
1308 	struct tipc_name_seq *seq;
1309 	struct sk_buff_head pkts;
1310 	u32 dport, dnode = 0;
1311 	u32 type, inst;
1312 	int mtu, rc;
1313 
1314 	if (unlikely(dlen > TIPC_MAX_USER_MSG_SIZE))
1315 		return -EMSGSIZE;
1316 
1317 	if (likely(dest)) {
1318 		if (unlikely(m->msg_namelen < sizeof(*dest)))
1319 			return -EINVAL;
1320 		if (unlikely(dest->family != AF_TIPC))
1321 			return -EINVAL;
1322 	}
1323 
1324 	if (grp) {
1325 		if (!dest)
1326 			return tipc_send_group_bcast(sock, m, dlen, timeout);
1327 		if (dest->addrtype == TIPC_ADDR_NAME)
1328 			return tipc_send_group_anycast(sock, m, dlen, timeout);
1329 		if (dest->addrtype == TIPC_ADDR_ID)
1330 			return tipc_send_group_unicast(sock, m, dlen, timeout);
1331 		if (dest->addrtype == TIPC_ADDR_MCAST)
1332 			return tipc_send_group_mcast(sock, m, dlen, timeout);
1333 		return -EINVAL;
1334 	}
1335 
1336 	if (unlikely(!dest)) {
1337 		dest = &tsk->peer;
1338 		if (!syn && dest->family != AF_TIPC)
1339 			return -EDESTADDRREQ;
1340 	}
1341 
1342 	if (unlikely(syn)) {
1343 		if (sk->sk_state == TIPC_LISTEN)
1344 			return -EPIPE;
1345 		if (sk->sk_state != TIPC_OPEN)
1346 			return -EISCONN;
1347 		if (tsk->published)
1348 			return -EOPNOTSUPP;
1349 		if (dest->addrtype == TIPC_ADDR_NAME) {
1350 			tsk->conn_type = dest->addr.name.name.type;
1351 			tsk->conn_instance = dest->addr.name.name.instance;
1352 		}
1353 		msg_set_syn(hdr, 1);
1354 	}
1355 
1356 	seq = &dest->addr.nameseq;
1357 	if (dest->addrtype == TIPC_ADDR_MCAST)
1358 		return tipc_sendmcast(sock, seq, m, dlen, timeout);
1359 
1360 	if (dest->addrtype == TIPC_ADDR_NAME) {
1361 		type = dest->addr.name.name.type;
1362 		inst = dest->addr.name.name.instance;
1363 		dnode = dest->addr.name.domain;
1364 		msg_set_type(hdr, TIPC_NAMED_MSG);
1365 		msg_set_hdr_sz(hdr, NAMED_H_SIZE);
1366 		msg_set_nametype(hdr, type);
1367 		msg_set_nameinst(hdr, inst);
1368 		msg_set_lookup_scope(hdr, tipc_node2scope(dnode));
1369 		dport = tipc_nametbl_translate(net, type, inst, &dnode);
1370 		msg_set_destnode(hdr, dnode);
1371 		msg_set_destport(hdr, dport);
1372 		if (unlikely(!dport && !dnode))
1373 			return -EHOSTUNREACH;
1374 	} else if (dest->addrtype == TIPC_ADDR_ID) {
1375 		dnode = dest->addr.id.node;
1376 		msg_set_type(hdr, TIPC_DIRECT_MSG);
1377 		msg_set_lookup_scope(hdr, 0);
1378 		msg_set_destnode(hdr, dnode);
1379 		msg_set_destport(hdr, dest->addr.id.ref);
1380 		msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1381 	} else {
1382 		return -EINVAL;
1383 	}
1384 
1385 	/* Block or return if destination link is congested */
1386 	rc = tipc_wait_for_cond(sock, &timeout,
1387 				!tipc_dest_find(clinks, dnode, 0));
1388 	if (unlikely(rc))
1389 		return rc;
1390 
1391 	skb_queue_head_init(&pkts);
1392 	mtu = tipc_node_get_mtu(net, dnode, tsk->portid);
1393 	rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1394 	if (unlikely(rc != dlen))
1395 		return rc;
1396 	if (unlikely(syn && !tipc_msg_skb_clone(&pkts, &sk->sk_write_queue)))
1397 		return -ENOMEM;
1398 
1399 	trace_tipc_sk_sendmsg(sk, skb_peek(&pkts), TIPC_DUMP_SK_SNDQ, " ");
1400 	rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
1401 	if (unlikely(rc == -ELINKCONG)) {
1402 		tipc_dest_push(clinks, dnode, 0);
1403 		tsk->cong_link_cnt++;
1404 		rc = 0;
1405 	}
1406 
1407 	if (unlikely(syn && !rc))
1408 		tipc_set_sk_state(sk, TIPC_CONNECTING);
1409 
1410 	return rc ? rc : dlen;
1411 }
1412 
1413 /**
1414  * tipc_sendstream - send stream-oriented data
1415  * @sock: socket structure
1416  * @m: data to send
1417  * @dsz: total length of data to be transmitted
1418  *
1419  * Used for SOCK_STREAM data.
1420  *
1421  * Returns the number of bytes sent on success (or partial success),
1422  * or errno if no data sent
1423  */
1424 static int tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz)
1425 {
1426 	struct sock *sk = sock->sk;
1427 	int ret;
1428 
1429 	lock_sock(sk);
1430 	ret = __tipc_sendstream(sock, m, dsz);
1431 	release_sock(sk);
1432 
1433 	return ret;
1434 }
1435 
1436 static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dlen)
1437 {
1438 	struct sock *sk = sock->sk;
1439 	DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1440 	long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1441 	struct tipc_sock *tsk = tipc_sk(sk);
1442 	struct tipc_msg *hdr = &tsk->phdr;
1443 	struct net *net = sock_net(sk);
1444 	struct sk_buff_head pkts;
1445 	u32 dnode = tsk_peer_node(tsk);
1446 	int send, sent = 0;
1447 	int rc = 0;
1448 
1449 	skb_queue_head_init(&pkts);
1450 
1451 	if (unlikely(dlen > INT_MAX))
1452 		return -EMSGSIZE;
1453 
1454 	/* Handle implicit connection setup */
1455 	if (unlikely(dest)) {
1456 		rc = __tipc_sendmsg(sock, m, dlen);
1457 		if (dlen && dlen == rc) {
1458 			tsk->peer_caps = tipc_node_get_capabilities(net, dnode);
1459 			tsk->snt_unacked = tsk_inc(tsk, dlen + msg_hdr_sz(hdr));
1460 		}
1461 		return rc;
1462 	}
1463 
1464 	do {
1465 		rc = tipc_wait_for_cond(sock, &timeout,
1466 					(!tsk->cong_link_cnt &&
1467 					 !tsk_conn_cong(tsk) &&
1468 					 tipc_sk_connected(sk)));
1469 		if (unlikely(rc))
1470 			break;
1471 
1472 		send = min_t(size_t, dlen - sent, TIPC_MAX_USER_MSG_SIZE);
1473 		rc = tipc_msg_build(hdr, m, sent, send, tsk->max_pkt, &pkts);
1474 		if (unlikely(rc != send))
1475 			break;
1476 
1477 		trace_tipc_sk_sendstream(sk, skb_peek(&pkts),
1478 					 TIPC_DUMP_SK_SNDQ, " ");
1479 		rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
1480 		if (unlikely(rc == -ELINKCONG)) {
1481 			tsk->cong_link_cnt = 1;
1482 			rc = 0;
1483 		}
1484 		if (likely(!rc)) {
1485 			tsk->snt_unacked += tsk_inc(tsk, send + MIN_H_SIZE);
1486 			sent += send;
1487 		}
1488 	} while (sent < dlen && !rc);
1489 
1490 	return sent ? sent : rc;
1491 }
1492 
1493 /**
1494  * tipc_send_packet - send a connection-oriented message
1495  * @sock: socket structure
1496  * @m: message to send
1497  * @dsz: length of data to be transmitted
1498  *
1499  * Used for SOCK_SEQPACKET messages.
1500  *
1501  * Returns the number of bytes sent on success, or errno otherwise
1502  */
1503 static int tipc_send_packet(struct socket *sock, struct msghdr *m, size_t dsz)
1504 {
1505 	if (dsz > TIPC_MAX_USER_MSG_SIZE)
1506 		return -EMSGSIZE;
1507 
1508 	return tipc_sendstream(sock, m, dsz);
1509 }
1510 
1511 /* tipc_sk_finish_conn - complete the setup of a connection
1512  */
1513 static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port,
1514 				u32 peer_node)
1515 {
1516 	struct sock *sk = &tsk->sk;
1517 	struct net *net = sock_net(sk);
1518 	struct tipc_msg *msg = &tsk->phdr;
1519 
1520 	msg_set_syn(msg, 0);
1521 	msg_set_destnode(msg, peer_node);
1522 	msg_set_destport(msg, peer_port);
1523 	msg_set_type(msg, TIPC_CONN_MSG);
1524 	msg_set_lookup_scope(msg, 0);
1525 	msg_set_hdr_sz(msg, SHORT_H_SIZE);
1526 
1527 	sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
1528 	tipc_set_sk_state(sk, TIPC_ESTABLISHED);
1529 	tipc_node_add_conn(net, peer_node, tsk->portid, peer_port);
1530 	tsk->max_pkt = tipc_node_get_mtu(net, peer_node, tsk->portid);
1531 	tsk->peer_caps = tipc_node_get_capabilities(net, peer_node);
1532 	__skb_queue_purge(&sk->sk_write_queue);
1533 	if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1534 		return;
1535 
1536 	/* Fall back to message based flow control */
1537 	tsk->rcv_win = FLOWCTL_MSG_WIN;
1538 	tsk->snd_win = FLOWCTL_MSG_WIN;
1539 }
1540 
1541 /**
1542  * tipc_sk_set_orig_addr - capture sender's address for received message
1543  * @m: descriptor for message info
1544  * @hdr: received message header
1545  *
1546  * Note: Address is not captured if not requested by receiver.
1547  */
1548 static void tipc_sk_set_orig_addr(struct msghdr *m, struct sk_buff *skb)
1549 {
1550 	DECLARE_SOCKADDR(struct sockaddr_pair *, srcaddr, m->msg_name);
1551 	struct tipc_msg *hdr = buf_msg(skb);
1552 
1553 	if (!srcaddr)
1554 		return;
1555 
1556 	srcaddr->sock.family = AF_TIPC;
1557 	srcaddr->sock.addrtype = TIPC_ADDR_ID;
1558 	srcaddr->sock.scope = 0;
1559 	srcaddr->sock.addr.id.ref = msg_origport(hdr);
1560 	srcaddr->sock.addr.id.node = msg_orignode(hdr);
1561 	srcaddr->sock.addr.name.domain = 0;
1562 	m->msg_namelen = sizeof(struct sockaddr_tipc);
1563 
1564 	if (!msg_in_group(hdr))
1565 		return;
1566 
1567 	/* Group message users may also want to know sending member's id */
1568 	srcaddr->member.family = AF_TIPC;
1569 	srcaddr->member.addrtype = TIPC_ADDR_NAME;
1570 	srcaddr->member.scope = 0;
1571 	srcaddr->member.addr.name.name.type = msg_nametype(hdr);
1572 	srcaddr->member.addr.name.name.instance = TIPC_SKB_CB(skb)->orig_member;
1573 	srcaddr->member.addr.name.domain = 0;
1574 	m->msg_namelen = sizeof(*srcaddr);
1575 }
1576 
1577 /**
1578  * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
1579  * @m: descriptor for message info
1580  * @skb: received message buffer
1581  * @tsk: TIPC port associated with message
1582  *
1583  * Note: Ancillary data is not captured if not requested by receiver.
1584  *
1585  * Returns 0 if successful, otherwise errno
1586  */
1587 static int tipc_sk_anc_data_recv(struct msghdr *m, struct sk_buff *skb,
1588 				 struct tipc_sock *tsk)
1589 {
1590 	struct tipc_msg *msg;
1591 	u32 anc_data[3];
1592 	u32 err;
1593 	u32 dest_type;
1594 	int has_name;
1595 	int res;
1596 
1597 	if (likely(m->msg_controllen == 0))
1598 		return 0;
1599 	msg = buf_msg(skb);
1600 
1601 	/* Optionally capture errored message object(s) */
1602 	err = msg ? msg_errcode(msg) : 0;
1603 	if (unlikely(err)) {
1604 		anc_data[0] = err;
1605 		anc_data[1] = msg_data_sz(msg);
1606 		res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
1607 		if (res)
1608 			return res;
1609 		if (anc_data[1]) {
1610 			if (skb_linearize(skb))
1611 				return -ENOMEM;
1612 			msg = buf_msg(skb);
1613 			res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
1614 				       msg_data(msg));
1615 			if (res)
1616 				return res;
1617 		}
1618 	}
1619 
1620 	/* Optionally capture message destination object */
1621 	dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
1622 	switch (dest_type) {
1623 	case TIPC_NAMED_MSG:
1624 		has_name = 1;
1625 		anc_data[0] = msg_nametype(msg);
1626 		anc_data[1] = msg_namelower(msg);
1627 		anc_data[2] = msg_namelower(msg);
1628 		break;
1629 	case TIPC_MCAST_MSG:
1630 		has_name = 1;
1631 		anc_data[0] = msg_nametype(msg);
1632 		anc_data[1] = msg_namelower(msg);
1633 		anc_data[2] = msg_nameupper(msg);
1634 		break;
1635 	case TIPC_CONN_MSG:
1636 		has_name = (tsk->conn_type != 0);
1637 		anc_data[0] = tsk->conn_type;
1638 		anc_data[1] = tsk->conn_instance;
1639 		anc_data[2] = tsk->conn_instance;
1640 		break;
1641 	default:
1642 		has_name = 0;
1643 	}
1644 	if (has_name) {
1645 		res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
1646 		if (res)
1647 			return res;
1648 	}
1649 
1650 	return 0;
1651 }
1652 
1653 static void tipc_sk_send_ack(struct tipc_sock *tsk)
1654 {
1655 	struct sock *sk = &tsk->sk;
1656 	struct net *net = sock_net(sk);
1657 	struct sk_buff *skb = NULL;
1658 	struct tipc_msg *msg;
1659 	u32 peer_port = tsk_peer_port(tsk);
1660 	u32 dnode = tsk_peer_node(tsk);
1661 
1662 	if (!tipc_sk_connected(sk))
1663 		return;
1664 	skb = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0,
1665 			      dnode, tsk_own_node(tsk), peer_port,
1666 			      tsk->portid, TIPC_OK);
1667 	if (!skb)
1668 		return;
1669 	msg = buf_msg(skb);
1670 	msg_set_conn_ack(msg, tsk->rcv_unacked);
1671 	tsk->rcv_unacked = 0;
1672 
1673 	/* Adjust to and advertize the correct window limit */
1674 	if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) {
1675 		tsk->rcv_win = tsk_adv_blocks(tsk->sk.sk_rcvbuf);
1676 		msg_set_adv_win(msg, tsk->rcv_win);
1677 	}
1678 	tipc_node_xmit_skb(net, skb, dnode, msg_link_selector(msg));
1679 }
1680 
1681 static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop)
1682 {
1683 	struct sock *sk = sock->sk;
1684 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
1685 	long timeo = *timeop;
1686 	int err = sock_error(sk);
1687 
1688 	if (err)
1689 		return err;
1690 
1691 	for (;;) {
1692 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1693 			if (sk->sk_shutdown & RCV_SHUTDOWN) {
1694 				err = -ENOTCONN;
1695 				break;
1696 			}
1697 			add_wait_queue(sk_sleep(sk), &wait);
1698 			release_sock(sk);
1699 			timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
1700 			sched_annotate_sleep();
1701 			lock_sock(sk);
1702 			remove_wait_queue(sk_sleep(sk), &wait);
1703 		}
1704 		err = 0;
1705 		if (!skb_queue_empty(&sk->sk_receive_queue))
1706 			break;
1707 		err = -EAGAIN;
1708 		if (!timeo)
1709 			break;
1710 		err = sock_intr_errno(timeo);
1711 		if (signal_pending(current))
1712 			break;
1713 
1714 		err = sock_error(sk);
1715 		if (err)
1716 			break;
1717 	}
1718 	*timeop = timeo;
1719 	return err;
1720 }
1721 
1722 /**
1723  * tipc_recvmsg - receive packet-oriented message
1724  * @m: descriptor for message info
1725  * @buflen: length of user buffer area
1726  * @flags: receive flags
1727  *
1728  * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1729  * If the complete message doesn't fit in user area, truncate it.
1730  *
1731  * Returns size of returned message data, errno otherwise
1732  */
1733 static int tipc_recvmsg(struct socket *sock, struct msghdr *m,
1734 			size_t buflen,	int flags)
1735 {
1736 	struct sock *sk = sock->sk;
1737 	bool connected = !tipc_sk_type_connectionless(sk);
1738 	struct tipc_sock *tsk = tipc_sk(sk);
1739 	int rc, err, hlen, dlen, copy;
1740 	struct sk_buff_head xmitq;
1741 	struct tipc_msg *hdr;
1742 	struct sk_buff *skb;
1743 	bool grp_evt;
1744 	long timeout;
1745 
1746 	/* Catch invalid receive requests */
1747 	if (unlikely(!buflen))
1748 		return -EINVAL;
1749 
1750 	lock_sock(sk);
1751 	if (unlikely(connected && sk->sk_state == TIPC_OPEN)) {
1752 		rc = -ENOTCONN;
1753 		goto exit;
1754 	}
1755 	timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1756 
1757 	/* Step rcv queue to first msg with data or error; wait if necessary */
1758 	do {
1759 		rc = tipc_wait_for_rcvmsg(sock, &timeout);
1760 		if (unlikely(rc))
1761 			goto exit;
1762 		skb = skb_peek(&sk->sk_receive_queue);
1763 		hdr = buf_msg(skb);
1764 		dlen = msg_data_sz(hdr);
1765 		hlen = msg_hdr_sz(hdr);
1766 		err = msg_errcode(hdr);
1767 		grp_evt = msg_is_grp_evt(hdr);
1768 		if (likely(dlen || err))
1769 			break;
1770 		tsk_advance_rx_queue(sk);
1771 	} while (1);
1772 
1773 	/* Collect msg meta data, including error code and rejected data */
1774 	tipc_sk_set_orig_addr(m, skb);
1775 	rc = tipc_sk_anc_data_recv(m, skb, tsk);
1776 	if (unlikely(rc))
1777 		goto exit;
1778 	hdr = buf_msg(skb);
1779 
1780 	/* Capture data if non-error msg, otherwise just set return value */
1781 	if (likely(!err)) {
1782 		copy = min_t(int, dlen, buflen);
1783 		if (unlikely(copy != dlen))
1784 			m->msg_flags |= MSG_TRUNC;
1785 		rc = skb_copy_datagram_msg(skb, hlen, m, copy);
1786 	} else {
1787 		copy = 0;
1788 		rc = 0;
1789 		if (err != TIPC_CONN_SHUTDOWN && connected && !m->msg_control)
1790 			rc = -ECONNRESET;
1791 	}
1792 	if (unlikely(rc))
1793 		goto exit;
1794 
1795 	/* Mark message as group event if applicable */
1796 	if (unlikely(grp_evt)) {
1797 		if (msg_grp_evt(hdr) == TIPC_WITHDRAWN)
1798 			m->msg_flags |= MSG_EOR;
1799 		m->msg_flags |= MSG_OOB;
1800 		copy = 0;
1801 	}
1802 
1803 	/* Caption of data or error code/rejected data was successful */
1804 	if (unlikely(flags & MSG_PEEK))
1805 		goto exit;
1806 
1807 	/* Send group flow control advertisement when applicable */
1808 	if (tsk->group && msg_in_group(hdr) && !grp_evt) {
1809 		skb_queue_head_init(&xmitq);
1810 		tipc_group_update_rcv_win(tsk->group, tsk_blocks(hlen + dlen),
1811 					  msg_orignode(hdr), msg_origport(hdr),
1812 					  &xmitq);
1813 		tipc_node_distr_xmit(sock_net(sk), &xmitq);
1814 	}
1815 
1816 	tsk_advance_rx_queue(sk);
1817 
1818 	if (likely(!connected))
1819 		goto exit;
1820 
1821 	/* Send connection flow control advertisement when applicable */
1822 	tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
1823 	if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
1824 		tipc_sk_send_ack(tsk);
1825 exit:
1826 	release_sock(sk);
1827 	return rc ? rc : copy;
1828 }
1829 
1830 /**
1831  * tipc_recvstream - receive stream-oriented data
1832  * @m: descriptor for message info
1833  * @buflen: total size of user buffer area
1834  * @flags: receive flags
1835  *
1836  * Used for SOCK_STREAM messages only.  If not enough data is available
1837  * will optionally wait for more; never truncates data.
1838  *
1839  * Returns size of returned message data, errno otherwise
1840  */
1841 static int tipc_recvstream(struct socket *sock, struct msghdr *m,
1842 			   size_t buflen, int flags)
1843 {
1844 	struct sock *sk = sock->sk;
1845 	struct tipc_sock *tsk = tipc_sk(sk);
1846 	struct sk_buff *skb;
1847 	struct tipc_msg *hdr;
1848 	struct tipc_skb_cb *skb_cb;
1849 	bool peek = flags & MSG_PEEK;
1850 	int offset, required, copy, copied = 0;
1851 	int hlen, dlen, err, rc;
1852 	long timeout;
1853 
1854 	/* Catch invalid receive attempts */
1855 	if (unlikely(!buflen))
1856 		return -EINVAL;
1857 
1858 	lock_sock(sk);
1859 
1860 	if (unlikely(sk->sk_state == TIPC_OPEN)) {
1861 		rc = -ENOTCONN;
1862 		goto exit;
1863 	}
1864 	required = sock_rcvlowat(sk, flags & MSG_WAITALL, buflen);
1865 	timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1866 
1867 	do {
1868 		/* Look at first msg in receive queue; wait if necessary */
1869 		rc = tipc_wait_for_rcvmsg(sock, &timeout);
1870 		if (unlikely(rc))
1871 			break;
1872 		skb = skb_peek(&sk->sk_receive_queue);
1873 		skb_cb = TIPC_SKB_CB(skb);
1874 		hdr = buf_msg(skb);
1875 		dlen = msg_data_sz(hdr);
1876 		hlen = msg_hdr_sz(hdr);
1877 		err = msg_errcode(hdr);
1878 
1879 		/* Discard any empty non-errored (SYN-) message */
1880 		if (unlikely(!dlen && !err)) {
1881 			tsk_advance_rx_queue(sk);
1882 			continue;
1883 		}
1884 
1885 		/* Collect msg meta data, incl. error code and rejected data */
1886 		if (!copied) {
1887 			tipc_sk_set_orig_addr(m, skb);
1888 			rc = tipc_sk_anc_data_recv(m, skb, tsk);
1889 			if (rc)
1890 				break;
1891 			hdr = buf_msg(skb);
1892 		}
1893 
1894 		/* Copy data if msg ok, otherwise return error/partial data */
1895 		if (likely(!err)) {
1896 			offset = skb_cb->bytes_read;
1897 			copy = min_t(int, dlen - offset, buflen - copied);
1898 			rc = skb_copy_datagram_msg(skb, hlen + offset, m, copy);
1899 			if (unlikely(rc))
1900 				break;
1901 			copied += copy;
1902 			offset += copy;
1903 			if (unlikely(offset < dlen)) {
1904 				if (!peek)
1905 					skb_cb->bytes_read = offset;
1906 				break;
1907 			}
1908 		} else {
1909 			rc = 0;
1910 			if ((err != TIPC_CONN_SHUTDOWN) && !m->msg_control)
1911 				rc = -ECONNRESET;
1912 			if (copied || rc)
1913 				break;
1914 		}
1915 
1916 		if (unlikely(peek))
1917 			break;
1918 
1919 		tsk_advance_rx_queue(sk);
1920 
1921 		/* Send connection flow control advertisement when applicable */
1922 		tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
1923 		if (unlikely(tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE))
1924 			tipc_sk_send_ack(tsk);
1925 
1926 		/* Exit if all requested data or FIN/error received */
1927 		if (copied == buflen || err)
1928 			break;
1929 
1930 	} while (!skb_queue_empty(&sk->sk_receive_queue) || copied < required);
1931 exit:
1932 	release_sock(sk);
1933 	return copied ? copied : rc;
1934 }
1935 
1936 /**
1937  * tipc_write_space - wake up thread if port congestion is released
1938  * @sk: socket
1939  */
1940 static void tipc_write_space(struct sock *sk)
1941 {
1942 	struct socket_wq *wq;
1943 
1944 	rcu_read_lock();
1945 	wq = rcu_dereference(sk->sk_wq);
1946 	if (skwq_has_sleeper(wq))
1947 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLOUT |
1948 						EPOLLWRNORM | EPOLLWRBAND);
1949 	rcu_read_unlock();
1950 }
1951 
1952 /**
1953  * tipc_data_ready - wake up threads to indicate messages have been received
1954  * @sk: socket
1955  * @len: the length of messages
1956  */
1957 static void tipc_data_ready(struct sock *sk)
1958 {
1959 	struct socket_wq *wq;
1960 
1961 	rcu_read_lock();
1962 	wq = rcu_dereference(sk->sk_wq);
1963 	if (skwq_has_sleeper(wq))
1964 		wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN |
1965 						EPOLLRDNORM | EPOLLRDBAND);
1966 	rcu_read_unlock();
1967 }
1968 
1969 static void tipc_sock_destruct(struct sock *sk)
1970 {
1971 	__skb_queue_purge(&sk->sk_receive_queue);
1972 }
1973 
1974 static void tipc_sk_proto_rcv(struct sock *sk,
1975 			      struct sk_buff_head *inputq,
1976 			      struct sk_buff_head *xmitq)
1977 {
1978 	struct sk_buff *skb = __skb_dequeue(inputq);
1979 	struct tipc_sock *tsk = tipc_sk(sk);
1980 	struct tipc_msg *hdr = buf_msg(skb);
1981 	struct tipc_group *grp = tsk->group;
1982 	bool wakeup = false;
1983 
1984 	switch (msg_user(hdr)) {
1985 	case CONN_MANAGER:
1986 		tipc_sk_conn_proto_rcv(tsk, skb, inputq, xmitq);
1987 		return;
1988 	case SOCK_WAKEUP:
1989 		tipc_dest_del(&tsk->cong_links, msg_orignode(hdr), 0);
1990 		/* coupled with smp_rmb() in tipc_wait_for_cond() */
1991 		smp_wmb();
1992 		tsk->cong_link_cnt--;
1993 		wakeup = true;
1994 		break;
1995 	case GROUP_PROTOCOL:
1996 		tipc_group_proto_rcv(grp, &wakeup, hdr, inputq, xmitq);
1997 		break;
1998 	case TOP_SRV:
1999 		tipc_group_member_evt(tsk->group, &wakeup, &sk->sk_rcvbuf,
2000 				      hdr, inputq, xmitq);
2001 		break;
2002 	default:
2003 		break;
2004 	}
2005 
2006 	if (wakeup)
2007 		sk->sk_write_space(sk);
2008 
2009 	kfree_skb(skb);
2010 }
2011 
2012 /**
2013  * tipc_sk_filter_connect - check incoming message for a connection-based socket
2014  * @tsk: TIPC socket
2015  * @skb: pointer to message buffer.
2016  * Returns true if message should be added to receive queue, false otherwise
2017  */
2018 static bool tipc_sk_filter_connect(struct tipc_sock *tsk, struct sk_buff *skb)
2019 {
2020 	struct sock *sk = &tsk->sk;
2021 	struct net *net = sock_net(sk);
2022 	struct tipc_msg *hdr = buf_msg(skb);
2023 	bool con_msg = msg_connected(hdr);
2024 	u32 pport = tsk_peer_port(tsk);
2025 	u32 pnode = tsk_peer_node(tsk);
2026 	u32 oport = msg_origport(hdr);
2027 	u32 onode = msg_orignode(hdr);
2028 	int err = msg_errcode(hdr);
2029 	unsigned long delay;
2030 
2031 	if (unlikely(msg_mcast(hdr)))
2032 		return false;
2033 
2034 	switch (sk->sk_state) {
2035 	case TIPC_CONNECTING:
2036 		/* Setup ACK */
2037 		if (likely(con_msg)) {
2038 			if (err)
2039 				break;
2040 			tipc_sk_finish_conn(tsk, oport, onode);
2041 			msg_set_importance(&tsk->phdr, msg_importance(hdr));
2042 			/* ACK+ message with data is added to receive queue */
2043 			if (msg_data_sz(hdr))
2044 				return true;
2045 			/* Empty ACK-, - wake up sleeping connect() and drop */
2046 			sk->sk_data_ready(sk);
2047 			msg_set_dest_droppable(hdr, 1);
2048 			return false;
2049 		}
2050 		/* Ignore connectionless message if not from listening socket */
2051 		if (oport != pport || onode != pnode)
2052 			return false;
2053 
2054 		/* Rejected SYN */
2055 		if (err != TIPC_ERR_OVERLOAD)
2056 			break;
2057 
2058 		/* Prepare for new setup attempt if we have a SYN clone */
2059 		if (skb_queue_empty(&sk->sk_write_queue))
2060 			break;
2061 		get_random_bytes(&delay, 2);
2062 		delay %= (tsk->conn_timeout / 4);
2063 		delay = msecs_to_jiffies(delay + 100);
2064 		sk_reset_timer(sk, &sk->sk_timer, jiffies + delay);
2065 		return false;
2066 	case TIPC_OPEN:
2067 	case TIPC_DISCONNECTING:
2068 		return false;
2069 	case TIPC_LISTEN:
2070 		/* Accept only SYN message */
2071 		if (!msg_is_syn(hdr) &&
2072 		    tipc_node_get_capabilities(net, onode) & TIPC_SYN_BIT)
2073 			return false;
2074 		if (!con_msg && !err)
2075 			return true;
2076 		return false;
2077 	case TIPC_ESTABLISHED:
2078 		/* Accept only connection-based messages sent by peer */
2079 		if (likely(con_msg && !err && pport == oport && pnode == onode))
2080 			return true;
2081 		if (!tsk_peer_msg(tsk, hdr))
2082 			return false;
2083 		if (!err)
2084 			return true;
2085 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2086 		tipc_node_remove_conn(net, pnode, tsk->portid);
2087 		sk->sk_state_change(sk);
2088 		return true;
2089 	default:
2090 		pr_err("Unknown sk_state %u\n", sk->sk_state);
2091 	}
2092 	/* Abort connection setup attempt */
2093 	tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2094 	sk->sk_err = ECONNREFUSED;
2095 	sk->sk_state_change(sk);
2096 	return true;
2097 }
2098 
2099 /**
2100  * rcvbuf_limit - get proper overload limit of socket receive queue
2101  * @sk: socket
2102  * @skb: message
2103  *
2104  * For connection oriented messages, irrespective of importance,
2105  * default queue limit is 2 MB.
2106  *
2107  * For connectionless messages, queue limits are based on message
2108  * importance as follows:
2109  *
2110  * TIPC_LOW_IMPORTANCE       (2 MB)
2111  * TIPC_MEDIUM_IMPORTANCE    (4 MB)
2112  * TIPC_HIGH_IMPORTANCE      (8 MB)
2113  * TIPC_CRITICAL_IMPORTANCE  (16 MB)
2114  *
2115  * Returns overload limit according to corresponding message importance
2116  */
2117 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *skb)
2118 {
2119 	struct tipc_sock *tsk = tipc_sk(sk);
2120 	struct tipc_msg *hdr = buf_msg(skb);
2121 
2122 	if (unlikely(msg_in_group(hdr)))
2123 		return sk->sk_rcvbuf;
2124 
2125 	if (unlikely(!msg_connected(hdr)))
2126 		return sk->sk_rcvbuf << msg_importance(hdr);
2127 
2128 	if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
2129 		return sk->sk_rcvbuf;
2130 
2131 	return FLOWCTL_MSG_LIM;
2132 }
2133 
2134 /**
2135  * tipc_sk_filter_rcv - validate incoming message
2136  * @sk: socket
2137  * @skb: pointer to message.
2138  *
2139  * Enqueues message on receive queue if acceptable; optionally handles
2140  * disconnect indication for a connected socket.
2141  *
2142  * Called with socket lock already taken
2143  *
2144  */
2145 static void tipc_sk_filter_rcv(struct sock *sk, struct sk_buff *skb,
2146 			       struct sk_buff_head *xmitq)
2147 {
2148 	bool sk_conn = !tipc_sk_type_connectionless(sk);
2149 	struct tipc_sock *tsk = tipc_sk(sk);
2150 	struct tipc_group *grp = tsk->group;
2151 	struct tipc_msg *hdr = buf_msg(skb);
2152 	struct net *net = sock_net(sk);
2153 	struct sk_buff_head inputq;
2154 	int mtyp = msg_type(hdr);
2155 	int limit, err = TIPC_OK;
2156 
2157 	trace_tipc_sk_filter_rcv(sk, skb, TIPC_DUMP_ALL, " ");
2158 	TIPC_SKB_CB(skb)->bytes_read = 0;
2159 	__skb_queue_head_init(&inputq);
2160 	__skb_queue_tail(&inputq, skb);
2161 
2162 	if (unlikely(!msg_isdata(hdr)))
2163 		tipc_sk_proto_rcv(sk, &inputq, xmitq);
2164 
2165 	if (unlikely(grp))
2166 		tipc_group_filter_msg(grp, &inputq, xmitq);
2167 
2168 	if (unlikely(!grp) && mtyp == TIPC_MCAST_MSG)
2169 		tipc_mcast_filter_msg(net, &tsk->mc_method.deferredq, &inputq);
2170 
2171 	/* Validate and add to receive buffer if there is space */
2172 	while ((skb = __skb_dequeue(&inputq))) {
2173 		hdr = buf_msg(skb);
2174 		limit = rcvbuf_limit(sk, skb);
2175 		if ((sk_conn && !tipc_sk_filter_connect(tsk, skb)) ||
2176 		    (!sk_conn && msg_connected(hdr)) ||
2177 		    (!grp && msg_in_group(hdr)))
2178 			err = TIPC_ERR_NO_PORT;
2179 		else if (sk_rmem_alloc_get(sk) + skb->truesize >= limit) {
2180 			trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL,
2181 					   "err_overload2!");
2182 			atomic_inc(&sk->sk_drops);
2183 			err = TIPC_ERR_OVERLOAD;
2184 		}
2185 
2186 		if (unlikely(err)) {
2187 			if (tipc_msg_reverse(tipc_own_addr(net), &skb, err)) {
2188 				trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE,
2189 						      "@filter_rcv!");
2190 				__skb_queue_tail(xmitq, skb);
2191 			}
2192 			err = TIPC_OK;
2193 			continue;
2194 		}
2195 		__skb_queue_tail(&sk->sk_receive_queue, skb);
2196 		skb_set_owner_r(skb, sk);
2197 		trace_tipc_sk_overlimit2(sk, skb, TIPC_DUMP_ALL,
2198 					 "rcvq >90% allocated!");
2199 		sk->sk_data_ready(sk);
2200 	}
2201 }
2202 
2203 /**
2204  * tipc_sk_backlog_rcv - handle incoming message from backlog queue
2205  * @sk: socket
2206  * @skb: message
2207  *
2208  * Caller must hold socket lock
2209  */
2210 static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb)
2211 {
2212 	unsigned int before = sk_rmem_alloc_get(sk);
2213 	struct sk_buff_head xmitq;
2214 	unsigned int added;
2215 
2216 	__skb_queue_head_init(&xmitq);
2217 
2218 	tipc_sk_filter_rcv(sk, skb, &xmitq);
2219 	added = sk_rmem_alloc_get(sk) - before;
2220 	atomic_add(added, &tipc_sk(sk)->dupl_rcvcnt);
2221 
2222 	/* Send pending response/rejected messages, if any */
2223 	tipc_node_distr_xmit(sock_net(sk), &xmitq);
2224 	return 0;
2225 }
2226 
2227 /**
2228  * tipc_sk_enqueue - extract all buffers with destination 'dport' from
2229  *                   inputq and try adding them to socket or backlog queue
2230  * @inputq: list of incoming buffers with potentially different destinations
2231  * @sk: socket where the buffers should be enqueued
2232  * @dport: port number for the socket
2233  *
2234  * Caller must hold socket lock
2235  */
2236 static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk,
2237 			    u32 dport, struct sk_buff_head *xmitq)
2238 {
2239 	unsigned long time_limit = jiffies + 2;
2240 	struct sk_buff *skb;
2241 	unsigned int lim;
2242 	atomic_t *dcnt;
2243 	u32 onode;
2244 
2245 	while (skb_queue_len(inputq)) {
2246 		if (unlikely(time_after_eq(jiffies, time_limit)))
2247 			return;
2248 
2249 		skb = tipc_skb_dequeue(inputq, dport);
2250 		if (unlikely(!skb))
2251 			return;
2252 
2253 		/* Add message directly to receive queue if possible */
2254 		if (!sock_owned_by_user(sk)) {
2255 			tipc_sk_filter_rcv(sk, skb, xmitq);
2256 			continue;
2257 		}
2258 
2259 		/* Try backlog, compensating for double-counted bytes */
2260 		dcnt = &tipc_sk(sk)->dupl_rcvcnt;
2261 		if (!sk->sk_backlog.len)
2262 			atomic_set(dcnt, 0);
2263 		lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
2264 		if (likely(!sk_add_backlog(sk, skb, lim))) {
2265 			trace_tipc_sk_overlimit1(sk, skb, TIPC_DUMP_ALL,
2266 						 "bklg & rcvq >90% allocated!");
2267 			continue;
2268 		}
2269 
2270 		trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL, "err_overload!");
2271 		/* Overload => reject message back to sender */
2272 		onode = tipc_own_addr(sock_net(sk));
2273 		atomic_inc(&sk->sk_drops);
2274 		if (tipc_msg_reverse(onode, &skb, TIPC_ERR_OVERLOAD)) {
2275 			trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_ALL,
2276 					      "@sk_enqueue!");
2277 			__skb_queue_tail(xmitq, skb);
2278 		}
2279 		break;
2280 	}
2281 }
2282 
2283 /**
2284  * tipc_sk_rcv - handle a chain of incoming buffers
2285  * @inputq: buffer list containing the buffers
2286  * Consumes all buffers in list until inputq is empty
2287  * Note: may be called in multiple threads referring to the same queue
2288  */
2289 void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq)
2290 {
2291 	struct sk_buff_head xmitq;
2292 	u32 dnode, dport = 0;
2293 	int err;
2294 	struct tipc_sock *tsk;
2295 	struct sock *sk;
2296 	struct sk_buff *skb;
2297 
2298 	__skb_queue_head_init(&xmitq);
2299 	while (skb_queue_len(inputq)) {
2300 		dport = tipc_skb_peek_port(inputq, dport);
2301 		tsk = tipc_sk_lookup(net, dport);
2302 
2303 		if (likely(tsk)) {
2304 			sk = &tsk->sk;
2305 			if (likely(spin_trylock_bh(&sk->sk_lock.slock))) {
2306 				tipc_sk_enqueue(inputq, sk, dport, &xmitq);
2307 				spin_unlock_bh(&sk->sk_lock.slock);
2308 			}
2309 			/* Send pending response/rejected messages, if any */
2310 			tipc_node_distr_xmit(sock_net(sk), &xmitq);
2311 			sock_put(sk);
2312 			continue;
2313 		}
2314 		/* No destination socket => dequeue skb if still there */
2315 		skb = tipc_skb_dequeue(inputq, dport);
2316 		if (!skb)
2317 			return;
2318 
2319 		/* Try secondary lookup if unresolved named message */
2320 		err = TIPC_ERR_NO_PORT;
2321 		if (tipc_msg_lookup_dest(net, skb, &err))
2322 			goto xmit;
2323 
2324 		/* Prepare for message rejection */
2325 		if (!tipc_msg_reverse(tipc_own_addr(net), &skb, err))
2326 			continue;
2327 
2328 		trace_tipc_sk_rej_msg(NULL, skb, TIPC_DUMP_NONE, "@sk_rcv!");
2329 xmit:
2330 		dnode = msg_destnode(buf_msg(skb));
2331 		tipc_node_xmit_skb(net, skb, dnode, dport);
2332 	}
2333 }
2334 
2335 static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
2336 {
2337 	DEFINE_WAIT_FUNC(wait, woken_wake_function);
2338 	struct sock *sk = sock->sk;
2339 	int done;
2340 
2341 	do {
2342 		int err = sock_error(sk);
2343 		if (err)
2344 			return err;
2345 		if (!*timeo_p)
2346 			return -ETIMEDOUT;
2347 		if (signal_pending(current))
2348 			return sock_intr_errno(*timeo_p);
2349 
2350 		add_wait_queue(sk_sleep(sk), &wait);
2351 		done = sk_wait_event(sk, timeo_p,
2352 				     sk->sk_state != TIPC_CONNECTING, &wait);
2353 		remove_wait_queue(sk_sleep(sk), &wait);
2354 	} while (!done);
2355 	return 0;
2356 }
2357 
2358 static bool tipc_sockaddr_is_sane(struct sockaddr_tipc *addr)
2359 {
2360 	if (addr->family != AF_TIPC)
2361 		return false;
2362 	if (addr->addrtype == TIPC_SERVICE_RANGE)
2363 		return (addr->addr.nameseq.lower <= addr->addr.nameseq.upper);
2364 	return (addr->addrtype == TIPC_SERVICE_ADDR ||
2365 		addr->addrtype == TIPC_SOCKET_ADDR);
2366 }
2367 
2368 /**
2369  * tipc_connect - establish a connection to another TIPC port
2370  * @sock: socket structure
2371  * @dest: socket address for destination port
2372  * @destlen: size of socket address data structure
2373  * @flags: file-related flags associated with socket
2374  *
2375  * Returns 0 on success, errno otherwise
2376  */
2377 static int tipc_connect(struct socket *sock, struct sockaddr *dest,
2378 			int destlen, int flags)
2379 {
2380 	struct sock *sk = sock->sk;
2381 	struct tipc_sock *tsk = tipc_sk(sk);
2382 	struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
2383 	struct msghdr m = {NULL,};
2384 	long timeout = (flags & O_NONBLOCK) ? 0 : tsk->conn_timeout;
2385 	int previous;
2386 	int res = 0;
2387 
2388 	if (destlen != sizeof(struct sockaddr_tipc))
2389 		return -EINVAL;
2390 
2391 	lock_sock(sk);
2392 
2393 	if (tsk->group) {
2394 		res = -EINVAL;
2395 		goto exit;
2396 	}
2397 
2398 	if (dst->family == AF_UNSPEC) {
2399 		memset(&tsk->peer, 0, sizeof(struct sockaddr_tipc));
2400 		if (!tipc_sk_type_connectionless(sk))
2401 			res = -EINVAL;
2402 		goto exit;
2403 	}
2404 	if (!tipc_sockaddr_is_sane(dst)) {
2405 		res = -EINVAL;
2406 		goto exit;
2407 	}
2408 	/* DGRAM/RDM connect(), just save the destaddr */
2409 	if (tipc_sk_type_connectionless(sk)) {
2410 		memcpy(&tsk->peer, dest, destlen);
2411 		goto exit;
2412 	} else if (dst->addrtype == TIPC_SERVICE_RANGE) {
2413 		res = -EINVAL;
2414 		goto exit;
2415 	}
2416 
2417 	previous = sk->sk_state;
2418 
2419 	switch (sk->sk_state) {
2420 	case TIPC_OPEN:
2421 		/* Send a 'SYN-' to destination */
2422 		m.msg_name = dest;
2423 		m.msg_namelen = destlen;
2424 
2425 		/* If connect is in non-blocking case, set MSG_DONTWAIT to
2426 		 * indicate send_msg() is never blocked.
2427 		 */
2428 		if (!timeout)
2429 			m.msg_flags = MSG_DONTWAIT;
2430 
2431 		res = __tipc_sendmsg(sock, &m, 0);
2432 		if ((res < 0) && (res != -EWOULDBLOCK))
2433 			goto exit;
2434 
2435 		/* Just entered TIPC_CONNECTING state; the only
2436 		 * difference is that return value in non-blocking
2437 		 * case is EINPROGRESS, rather than EALREADY.
2438 		 */
2439 		res = -EINPROGRESS;
2440 		/* fall through */
2441 	case TIPC_CONNECTING:
2442 		if (!timeout) {
2443 			if (previous == TIPC_CONNECTING)
2444 				res = -EALREADY;
2445 			goto exit;
2446 		}
2447 		timeout = msecs_to_jiffies(timeout);
2448 		/* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
2449 		res = tipc_wait_for_connect(sock, &timeout);
2450 		break;
2451 	case TIPC_ESTABLISHED:
2452 		res = -EISCONN;
2453 		break;
2454 	default:
2455 		res = -EINVAL;
2456 	}
2457 
2458 exit:
2459 	release_sock(sk);
2460 	return res;
2461 }
2462 
2463 /**
2464  * tipc_listen - allow socket to listen for incoming connections
2465  * @sock: socket structure
2466  * @len: (unused)
2467  *
2468  * Returns 0 on success, errno otherwise
2469  */
2470 static int tipc_listen(struct socket *sock, int len)
2471 {
2472 	struct sock *sk = sock->sk;
2473 	int res;
2474 
2475 	lock_sock(sk);
2476 	res = tipc_set_sk_state(sk, TIPC_LISTEN);
2477 	release_sock(sk);
2478 
2479 	return res;
2480 }
2481 
2482 static int tipc_wait_for_accept(struct socket *sock, long timeo)
2483 {
2484 	struct sock *sk = sock->sk;
2485 	DEFINE_WAIT(wait);
2486 	int err;
2487 
2488 	/* True wake-one mechanism for incoming connections: only
2489 	 * one process gets woken up, not the 'whole herd'.
2490 	 * Since we do not 'race & poll' for established sockets
2491 	 * anymore, the common case will execute the loop only once.
2492 	*/
2493 	for (;;) {
2494 		prepare_to_wait_exclusive(sk_sleep(sk), &wait,
2495 					  TASK_INTERRUPTIBLE);
2496 		if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
2497 			release_sock(sk);
2498 			timeo = schedule_timeout(timeo);
2499 			lock_sock(sk);
2500 		}
2501 		err = 0;
2502 		if (!skb_queue_empty(&sk->sk_receive_queue))
2503 			break;
2504 		err = -EAGAIN;
2505 		if (!timeo)
2506 			break;
2507 		err = sock_intr_errno(timeo);
2508 		if (signal_pending(current))
2509 			break;
2510 	}
2511 	finish_wait(sk_sleep(sk), &wait);
2512 	return err;
2513 }
2514 
2515 /**
2516  * tipc_accept - wait for connection request
2517  * @sock: listening socket
2518  * @newsock: new socket that is to be connected
2519  * @flags: file-related flags associated with socket
2520  *
2521  * Returns 0 on success, errno otherwise
2522  */
2523 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags,
2524 		       bool kern)
2525 {
2526 	struct sock *new_sk, *sk = sock->sk;
2527 	struct sk_buff *buf;
2528 	struct tipc_sock *new_tsock;
2529 	struct tipc_msg *msg;
2530 	long timeo;
2531 	int res;
2532 
2533 	lock_sock(sk);
2534 
2535 	if (sk->sk_state != TIPC_LISTEN) {
2536 		res = -EINVAL;
2537 		goto exit;
2538 	}
2539 	timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
2540 	res = tipc_wait_for_accept(sock, timeo);
2541 	if (res)
2542 		goto exit;
2543 
2544 	buf = skb_peek(&sk->sk_receive_queue);
2545 
2546 	res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, kern);
2547 	if (res)
2548 		goto exit;
2549 	security_sk_clone(sock->sk, new_sock->sk);
2550 
2551 	new_sk = new_sock->sk;
2552 	new_tsock = tipc_sk(new_sk);
2553 	msg = buf_msg(buf);
2554 
2555 	/* we lock on new_sk; but lockdep sees the lock on sk */
2556 	lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
2557 
2558 	/*
2559 	 * Reject any stray messages received by new socket
2560 	 * before the socket lock was taken (very, very unlikely)
2561 	 */
2562 	tsk_rej_rx_queue(new_sk);
2563 
2564 	/* Connect new socket to it's peer */
2565 	tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg));
2566 
2567 	tsk_set_importance(new_tsock, msg_importance(msg));
2568 	if (msg_named(msg)) {
2569 		new_tsock->conn_type = msg_nametype(msg);
2570 		new_tsock->conn_instance = msg_nameinst(msg);
2571 	}
2572 
2573 	/*
2574 	 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2575 	 * Respond to 'SYN+' by queuing it on new socket.
2576 	 */
2577 	if (!msg_data_sz(msg)) {
2578 		struct msghdr m = {NULL,};
2579 
2580 		tsk_advance_rx_queue(sk);
2581 		__tipc_sendstream(new_sock, &m, 0);
2582 	} else {
2583 		__skb_dequeue(&sk->sk_receive_queue);
2584 		__skb_queue_head(&new_sk->sk_receive_queue, buf);
2585 		skb_set_owner_r(buf, new_sk);
2586 	}
2587 	release_sock(new_sk);
2588 exit:
2589 	release_sock(sk);
2590 	return res;
2591 }
2592 
2593 /**
2594  * tipc_shutdown - shutdown socket connection
2595  * @sock: socket structure
2596  * @how: direction to close (must be SHUT_RDWR)
2597  *
2598  * Terminates connection (if necessary), then purges socket's receive queue.
2599  *
2600  * Returns 0 on success, errno otherwise
2601  */
2602 static int tipc_shutdown(struct socket *sock, int how)
2603 {
2604 	struct sock *sk = sock->sk;
2605 	int res;
2606 
2607 	if (how != SHUT_RDWR)
2608 		return -EINVAL;
2609 
2610 	lock_sock(sk);
2611 
2612 	trace_tipc_sk_shutdown(sk, NULL, TIPC_DUMP_ALL, " ");
2613 	__tipc_shutdown(sock, TIPC_CONN_SHUTDOWN);
2614 	sk->sk_shutdown = SEND_SHUTDOWN;
2615 
2616 	if (sk->sk_state == TIPC_DISCONNECTING) {
2617 		/* Discard any unreceived messages */
2618 		__skb_queue_purge(&sk->sk_receive_queue);
2619 
2620 		/* Wake up anyone sleeping in poll */
2621 		sk->sk_state_change(sk);
2622 		res = 0;
2623 	} else {
2624 		res = -ENOTCONN;
2625 	}
2626 
2627 	release_sock(sk);
2628 	return res;
2629 }
2630 
2631 static void tipc_sk_check_probing_state(struct sock *sk,
2632 					struct sk_buff_head *list)
2633 {
2634 	struct tipc_sock *tsk = tipc_sk(sk);
2635 	u32 pnode = tsk_peer_node(tsk);
2636 	u32 pport = tsk_peer_port(tsk);
2637 	u32 self = tsk_own_node(tsk);
2638 	u32 oport = tsk->portid;
2639 	struct sk_buff *skb;
2640 
2641 	if (tsk->probe_unacked) {
2642 		tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2643 		sk->sk_err = ECONNABORTED;
2644 		tipc_node_remove_conn(sock_net(sk), pnode, pport);
2645 		sk->sk_state_change(sk);
2646 		return;
2647 	}
2648 	/* Prepare new probe */
2649 	skb = tipc_msg_create(CONN_MANAGER, CONN_PROBE, INT_H_SIZE, 0,
2650 			      pnode, self, pport, oport, TIPC_OK);
2651 	if (skb)
2652 		__skb_queue_tail(list, skb);
2653 	tsk->probe_unacked = true;
2654 	sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
2655 }
2656 
2657 static void tipc_sk_retry_connect(struct sock *sk, struct sk_buff_head *list)
2658 {
2659 	struct tipc_sock *tsk = tipc_sk(sk);
2660 
2661 	/* Try again later if dest link is congested */
2662 	if (tsk->cong_link_cnt) {
2663 		sk_reset_timer(sk, &sk->sk_timer, msecs_to_jiffies(100));
2664 		return;
2665 	}
2666 	/* Prepare SYN for retransmit */
2667 	tipc_msg_skb_clone(&sk->sk_write_queue, list);
2668 }
2669 
2670 static void tipc_sk_timeout(struct timer_list *t)
2671 {
2672 	struct sock *sk = from_timer(sk, t, sk_timer);
2673 	struct tipc_sock *tsk = tipc_sk(sk);
2674 	u32 pnode = tsk_peer_node(tsk);
2675 	struct sk_buff_head list;
2676 	int rc = 0;
2677 
2678 	skb_queue_head_init(&list);
2679 	bh_lock_sock(sk);
2680 
2681 	/* Try again later if socket is busy */
2682 	if (sock_owned_by_user(sk)) {
2683 		sk_reset_timer(sk, &sk->sk_timer, jiffies + HZ / 20);
2684 		bh_unlock_sock(sk);
2685 		return;
2686 	}
2687 
2688 	if (sk->sk_state == TIPC_ESTABLISHED)
2689 		tipc_sk_check_probing_state(sk, &list);
2690 	else if (sk->sk_state == TIPC_CONNECTING)
2691 		tipc_sk_retry_connect(sk, &list);
2692 
2693 	bh_unlock_sock(sk);
2694 
2695 	if (!skb_queue_empty(&list))
2696 		rc = tipc_node_xmit(sock_net(sk), &list, pnode, tsk->portid);
2697 
2698 	/* SYN messages may cause link congestion */
2699 	if (rc == -ELINKCONG) {
2700 		tipc_dest_push(&tsk->cong_links, pnode, 0);
2701 		tsk->cong_link_cnt = 1;
2702 	}
2703 	sock_put(sk);
2704 }
2705 
2706 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
2707 			   struct tipc_name_seq const *seq)
2708 {
2709 	struct sock *sk = &tsk->sk;
2710 	struct net *net = sock_net(sk);
2711 	struct publication *publ;
2712 	u32 key;
2713 
2714 	if (scope != TIPC_NODE_SCOPE)
2715 		scope = TIPC_CLUSTER_SCOPE;
2716 
2717 	if (tipc_sk_connected(sk))
2718 		return -EINVAL;
2719 	key = tsk->portid + tsk->pub_count + 1;
2720 	if (key == tsk->portid)
2721 		return -EADDRINUSE;
2722 
2723 	publ = tipc_nametbl_publish(net, seq->type, seq->lower, seq->upper,
2724 				    scope, tsk->portid, key);
2725 	if (unlikely(!publ))
2726 		return -EINVAL;
2727 
2728 	list_add(&publ->binding_sock, &tsk->publications);
2729 	tsk->pub_count++;
2730 	tsk->published = 1;
2731 	return 0;
2732 }
2733 
2734 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
2735 			    struct tipc_name_seq const *seq)
2736 {
2737 	struct net *net = sock_net(&tsk->sk);
2738 	struct publication *publ;
2739 	struct publication *safe;
2740 	int rc = -EINVAL;
2741 
2742 	if (scope != TIPC_NODE_SCOPE)
2743 		scope = TIPC_CLUSTER_SCOPE;
2744 
2745 	list_for_each_entry_safe(publ, safe, &tsk->publications, binding_sock) {
2746 		if (seq) {
2747 			if (publ->scope != scope)
2748 				continue;
2749 			if (publ->type != seq->type)
2750 				continue;
2751 			if (publ->lower != seq->lower)
2752 				continue;
2753 			if (publ->upper != seq->upper)
2754 				break;
2755 			tipc_nametbl_withdraw(net, publ->type, publ->lower,
2756 					      publ->upper, publ->key);
2757 			rc = 0;
2758 			break;
2759 		}
2760 		tipc_nametbl_withdraw(net, publ->type, publ->lower,
2761 				      publ->upper, publ->key);
2762 		rc = 0;
2763 	}
2764 	if (list_empty(&tsk->publications))
2765 		tsk->published = 0;
2766 	return rc;
2767 }
2768 
2769 /* tipc_sk_reinit: set non-zero address in all existing sockets
2770  *                 when we go from standalone to network mode.
2771  */
2772 void tipc_sk_reinit(struct net *net)
2773 {
2774 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2775 	struct rhashtable_iter iter;
2776 	struct tipc_sock *tsk;
2777 	struct tipc_msg *msg;
2778 
2779 	rhashtable_walk_enter(&tn->sk_rht, &iter);
2780 
2781 	do {
2782 		rhashtable_walk_start(&iter);
2783 
2784 		while ((tsk = rhashtable_walk_next(&iter)) && !IS_ERR(tsk)) {
2785 			sock_hold(&tsk->sk);
2786 			rhashtable_walk_stop(&iter);
2787 			lock_sock(&tsk->sk);
2788 			msg = &tsk->phdr;
2789 			msg_set_prevnode(msg, tipc_own_addr(net));
2790 			msg_set_orignode(msg, tipc_own_addr(net));
2791 			release_sock(&tsk->sk);
2792 			rhashtable_walk_start(&iter);
2793 			sock_put(&tsk->sk);
2794 		}
2795 
2796 		rhashtable_walk_stop(&iter);
2797 	} while (tsk == ERR_PTR(-EAGAIN));
2798 
2799 	rhashtable_walk_exit(&iter);
2800 }
2801 
2802 static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid)
2803 {
2804 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2805 	struct tipc_sock *tsk;
2806 
2807 	rcu_read_lock();
2808 	tsk = rhashtable_lookup_fast(&tn->sk_rht, &portid, tsk_rht_params);
2809 	if (tsk)
2810 		sock_hold(&tsk->sk);
2811 	rcu_read_unlock();
2812 
2813 	return tsk;
2814 }
2815 
2816 static int tipc_sk_insert(struct tipc_sock *tsk)
2817 {
2818 	struct sock *sk = &tsk->sk;
2819 	struct net *net = sock_net(sk);
2820 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2821 	u32 remaining = (TIPC_MAX_PORT - TIPC_MIN_PORT) + 1;
2822 	u32 portid = prandom_u32() % remaining + TIPC_MIN_PORT;
2823 
2824 	while (remaining--) {
2825 		portid++;
2826 		if ((portid < TIPC_MIN_PORT) || (portid > TIPC_MAX_PORT))
2827 			portid = TIPC_MIN_PORT;
2828 		tsk->portid = portid;
2829 		sock_hold(&tsk->sk);
2830 		if (!rhashtable_lookup_insert_fast(&tn->sk_rht, &tsk->node,
2831 						   tsk_rht_params))
2832 			return 0;
2833 		sock_put(&tsk->sk);
2834 	}
2835 
2836 	return -1;
2837 }
2838 
2839 static void tipc_sk_remove(struct tipc_sock *tsk)
2840 {
2841 	struct sock *sk = &tsk->sk;
2842 	struct tipc_net *tn = net_generic(sock_net(sk), tipc_net_id);
2843 
2844 	if (!rhashtable_remove_fast(&tn->sk_rht, &tsk->node, tsk_rht_params)) {
2845 		WARN_ON(refcount_read(&sk->sk_refcnt) == 1);
2846 		__sock_put(sk);
2847 	}
2848 }
2849 
2850 static const struct rhashtable_params tsk_rht_params = {
2851 	.nelem_hint = 192,
2852 	.head_offset = offsetof(struct tipc_sock, node),
2853 	.key_offset = offsetof(struct tipc_sock, portid),
2854 	.key_len = sizeof(u32), /* portid */
2855 	.max_size = 1048576,
2856 	.min_size = 256,
2857 	.automatic_shrinking = true,
2858 };
2859 
2860 int tipc_sk_rht_init(struct net *net)
2861 {
2862 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2863 
2864 	return rhashtable_init(&tn->sk_rht, &tsk_rht_params);
2865 }
2866 
2867 void tipc_sk_rht_destroy(struct net *net)
2868 {
2869 	struct tipc_net *tn = net_generic(net, tipc_net_id);
2870 
2871 	/* Wait for socket readers to complete */
2872 	synchronize_net();
2873 
2874 	rhashtable_destroy(&tn->sk_rht);
2875 }
2876 
2877 static int tipc_sk_join(struct tipc_sock *tsk, struct tipc_group_req *mreq)
2878 {
2879 	struct net *net = sock_net(&tsk->sk);
2880 	struct tipc_group *grp = tsk->group;
2881 	struct tipc_msg *hdr = &tsk->phdr;
2882 	struct tipc_name_seq seq;
2883 	int rc;
2884 
2885 	if (mreq->type < TIPC_RESERVED_TYPES)
2886 		return -EACCES;
2887 	if (mreq->scope > TIPC_NODE_SCOPE)
2888 		return -EINVAL;
2889 	if (grp)
2890 		return -EACCES;
2891 	grp = tipc_group_create(net, tsk->portid, mreq, &tsk->group_is_open);
2892 	if (!grp)
2893 		return -ENOMEM;
2894 	tsk->group = grp;
2895 	msg_set_lookup_scope(hdr, mreq->scope);
2896 	msg_set_nametype(hdr, mreq->type);
2897 	msg_set_dest_droppable(hdr, true);
2898 	seq.type = mreq->type;
2899 	seq.lower = mreq->instance;
2900 	seq.upper = seq.lower;
2901 	tipc_nametbl_build_group(net, grp, mreq->type, mreq->scope);
2902 	rc = tipc_sk_publish(tsk, mreq->scope, &seq);
2903 	if (rc) {
2904 		tipc_group_delete(net, grp);
2905 		tsk->group = NULL;
2906 		return rc;
2907 	}
2908 	/* Eliminate any risk that a broadcast overtakes sent JOINs */
2909 	tsk->mc_method.rcast = true;
2910 	tsk->mc_method.mandatory = true;
2911 	tipc_group_join(net, grp, &tsk->sk.sk_rcvbuf);
2912 	return rc;
2913 }
2914 
2915 static int tipc_sk_leave(struct tipc_sock *tsk)
2916 {
2917 	struct net *net = sock_net(&tsk->sk);
2918 	struct tipc_group *grp = tsk->group;
2919 	struct tipc_name_seq seq;
2920 	int scope;
2921 
2922 	if (!grp)
2923 		return -EINVAL;
2924 	tipc_group_self(grp, &seq, &scope);
2925 	tipc_group_delete(net, grp);
2926 	tsk->group = NULL;
2927 	tipc_sk_withdraw(tsk, scope, &seq);
2928 	return 0;
2929 }
2930 
2931 /**
2932  * tipc_setsockopt - set socket option
2933  * @sock: socket structure
2934  * @lvl: option level
2935  * @opt: option identifier
2936  * @ov: pointer to new option value
2937  * @ol: length of option value
2938  *
2939  * For stream sockets only, accepts and ignores all IPPROTO_TCP options
2940  * (to ease compatibility).
2941  *
2942  * Returns 0 on success, errno otherwise
2943  */
2944 static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
2945 			   char __user *ov, unsigned int ol)
2946 {
2947 	struct sock *sk = sock->sk;
2948 	struct tipc_sock *tsk = tipc_sk(sk);
2949 	struct tipc_group_req mreq;
2950 	u32 value = 0;
2951 	int res = 0;
2952 
2953 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
2954 		return 0;
2955 	if (lvl != SOL_TIPC)
2956 		return -ENOPROTOOPT;
2957 
2958 	switch (opt) {
2959 	case TIPC_IMPORTANCE:
2960 	case TIPC_SRC_DROPPABLE:
2961 	case TIPC_DEST_DROPPABLE:
2962 	case TIPC_CONN_TIMEOUT:
2963 		if (ol < sizeof(value))
2964 			return -EINVAL;
2965 		if (get_user(value, (u32 __user *)ov))
2966 			return -EFAULT;
2967 		break;
2968 	case TIPC_GROUP_JOIN:
2969 		if (ol < sizeof(mreq))
2970 			return -EINVAL;
2971 		if (copy_from_user(&mreq, ov, sizeof(mreq)))
2972 			return -EFAULT;
2973 		break;
2974 	default:
2975 		if (ov || ol)
2976 			return -EINVAL;
2977 	}
2978 
2979 	lock_sock(sk);
2980 
2981 	switch (opt) {
2982 	case TIPC_IMPORTANCE:
2983 		res = tsk_set_importance(tsk, value);
2984 		break;
2985 	case TIPC_SRC_DROPPABLE:
2986 		if (sock->type != SOCK_STREAM)
2987 			tsk_set_unreliable(tsk, value);
2988 		else
2989 			res = -ENOPROTOOPT;
2990 		break;
2991 	case TIPC_DEST_DROPPABLE:
2992 		tsk_set_unreturnable(tsk, value);
2993 		break;
2994 	case TIPC_CONN_TIMEOUT:
2995 		tipc_sk(sk)->conn_timeout = value;
2996 		break;
2997 	case TIPC_MCAST_BROADCAST:
2998 		tsk->mc_method.rcast = false;
2999 		tsk->mc_method.mandatory = true;
3000 		break;
3001 	case TIPC_MCAST_REPLICAST:
3002 		tsk->mc_method.rcast = true;
3003 		tsk->mc_method.mandatory = true;
3004 		break;
3005 	case TIPC_GROUP_JOIN:
3006 		res = tipc_sk_join(tsk, &mreq);
3007 		break;
3008 	case TIPC_GROUP_LEAVE:
3009 		res = tipc_sk_leave(tsk);
3010 		break;
3011 	default:
3012 		res = -EINVAL;
3013 	}
3014 
3015 	release_sock(sk);
3016 
3017 	return res;
3018 }
3019 
3020 /**
3021  * tipc_getsockopt - get socket option
3022  * @sock: socket structure
3023  * @lvl: option level
3024  * @opt: option identifier
3025  * @ov: receptacle for option value
3026  * @ol: receptacle for length of option value
3027  *
3028  * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
3029  * (to ease compatibility).
3030  *
3031  * Returns 0 on success, errno otherwise
3032  */
3033 static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
3034 			   char __user *ov, int __user *ol)
3035 {
3036 	struct sock *sk = sock->sk;
3037 	struct tipc_sock *tsk = tipc_sk(sk);
3038 	struct tipc_name_seq seq;
3039 	int len, scope;
3040 	u32 value;
3041 	int res;
3042 
3043 	if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3044 		return put_user(0, ol);
3045 	if (lvl != SOL_TIPC)
3046 		return -ENOPROTOOPT;
3047 	res = get_user(len, ol);
3048 	if (res)
3049 		return res;
3050 
3051 	lock_sock(sk);
3052 
3053 	switch (opt) {
3054 	case TIPC_IMPORTANCE:
3055 		value = tsk_importance(tsk);
3056 		break;
3057 	case TIPC_SRC_DROPPABLE:
3058 		value = tsk_unreliable(tsk);
3059 		break;
3060 	case TIPC_DEST_DROPPABLE:
3061 		value = tsk_unreturnable(tsk);
3062 		break;
3063 	case TIPC_CONN_TIMEOUT:
3064 		value = tsk->conn_timeout;
3065 		/* no need to set "res", since already 0 at this point */
3066 		break;
3067 	case TIPC_NODE_RECVQ_DEPTH:
3068 		value = 0; /* was tipc_queue_size, now obsolete */
3069 		break;
3070 	case TIPC_SOCK_RECVQ_DEPTH:
3071 		value = skb_queue_len(&sk->sk_receive_queue);
3072 		break;
3073 	case TIPC_GROUP_JOIN:
3074 		seq.type = 0;
3075 		if (tsk->group)
3076 			tipc_group_self(tsk->group, &seq, &scope);
3077 		value = seq.type;
3078 		break;
3079 	default:
3080 		res = -EINVAL;
3081 	}
3082 
3083 	release_sock(sk);
3084 
3085 	if (res)
3086 		return res;	/* "get" failed */
3087 
3088 	if (len < sizeof(value))
3089 		return -EINVAL;
3090 
3091 	if (copy_to_user(ov, &value, sizeof(value)))
3092 		return -EFAULT;
3093 
3094 	return put_user(sizeof(value), ol);
3095 }
3096 
3097 static int tipc_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
3098 {
3099 	struct net *net = sock_net(sock->sk);
3100 	struct tipc_sioc_nodeid_req nr = {0};
3101 	struct tipc_sioc_ln_req lnr;
3102 	void __user *argp = (void __user *)arg;
3103 
3104 	switch (cmd) {
3105 	case SIOCGETLINKNAME:
3106 		if (copy_from_user(&lnr, argp, sizeof(lnr)))
3107 			return -EFAULT;
3108 		if (!tipc_node_get_linkname(net,
3109 					    lnr.bearer_id & 0xffff, lnr.peer,
3110 					    lnr.linkname, TIPC_MAX_LINK_NAME)) {
3111 			if (copy_to_user(argp, &lnr, sizeof(lnr)))
3112 				return -EFAULT;
3113 			return 0;
3114 		}
3115 		return -EADDRNOTAVAIL;
3116 	case SIOCGETNODEID:
3117 		if (copy_from_user(&nr, argp, sizeof(nr)))
3118 			return -EFAULT;
3119 		if (!tipc_node_get_id(net, nr.peer, nr.node_id))
3120 			return -EADDRNOTAVAIL;
3121 		if (copy_to_user(argp, &nr, sizeof(nr)))
3122 			return -EFAULT;
3123 		return 0;
3124 	default:
3125 		return -ENOIOCTLCMD;
3126 	}
3127 }
3128 
3129 static int tipc_socketpair(struct socket *sock1, struct socket *sock2)
3130 {
3131 	struct tipc_sock *tsk2 = tipc_sk(sock2->sk);
3132 	struct tipc_sock *tsk1 = tipc_sk(sock1->sk);
3133 	u32 onode = tipc_own_addr(sock_net(sock1->sk));
3134 
3135 	tsk1->peer.family = AF_TIPC;
3136 	tsk1->peer.addrtype = TIPC_ADDR_ID;
3137 	tsk1->peer.scope = TIPC_NODE_SCOPE;
3138 	tsk1->peer.addr.id.ref = tsk2->portid;
3139 	tsk1->peer.addr.id.node = onode;
3140 	tsk2->peer.family = AF_TIPC;
3141 	tsk2->peer.addrtype = TIPC_ADDR_ID;
3142 	tsk2->peer.scope = TIPC_NODE_SCOPE;
3143 	tsk2->peer.addr.id.ref = tsk1->portid;
3144 	tsk2->peer.addr.id.node = onode;
3145 
3146 	tipc_sk_finish_conn(tsk1, tsk2->portid, onode);
3147 	tipc_sk_finish_conn(tsk2, tsk1->portid, onode);
3148 	return 0;
3149 }
3150 
3151 /* Protocol switches for the various types of TIPC sockets */
3152 
3153 static const struct proto_ops msg_ops = {
3154 	.owner		= THIS_MODULE,
3155 	.family		= AF_TIPC,
3156 	.release	= tipc_release,
3157 	.bind		= tipc_bind,
3158 	.connect	= tipc_connect,
3159 	.socketpair	= tipc_socketpair,
3160 	.accept		= sock_no_accept,
3161 	.getname	= tipc_getname,
3162 	.poll		= tipc_poll,
3163 	.ioctl		= tipc_ioctl,
3164 	.listen		= sock_no_listen,
3165 	.shutdown	= tipc_shutdown,
3166 	.setsockopt	= tipc_setsockopt,
3167 	.getsockopt	= tipc_getsockopt,
3168 	.sendmsg	= tipc_sendmsg,
3169 	.recvmsg	= tipc_recvmsg,
3170 	.mmap		= sock_no_mmap,
3171 	.sendpage	= sock_no_sendpage
3172 };
3173 
3174 static const struct proto_ops packet_ops = {
3175 	.owner		= THIS_MODULE,
3176 	.family		= AF_TIPC,
3177 	.release	= tipc_release,
3178 	.bind		= tipc_bind,
3179 	.connect	= tipc_connect,
3180 	.socketpair	= tipc_socketpair,
3181 	.accept		= tipc_accept,
3182 	.getname	= tipc_getname,
3183 	.poll		= tipc_poll,
3184 	.ioctl		= tipc_ioctl,
3185 	.listen		= tipc_listen,
3186 	.shutdown	= tipc_shutdown,
3187 	.setsockopt	= tipc_setsockopt,
3188 	.getsockopt	= tipc_getsockopt,
3189 	.sendmsg	= tipc_send_packet,
3190 	.recvmsg	= tipc_recvmsg,
3191 	.mmap		= sock_no_mmap,
3192 	.sendpage	= sock_no_sendpage
3193 };
3194 
3195 static const struct proto_ops stream_ops = {
3196 	.owner		= THIS_MODULE,
3197 	.family		= AF_TIPC,
3198 	.release	= tipc_release,
3199 	.bind		= tipc_bind,
3200 	.connect	= tipc_connect,
3201 	.socketpair	= tipc_socketpair,
3202 	.accept		= tipc_accept,
3203 	.getname	= tipc_getname,
3204 	.poll		= tipc_poll,
3205 	.ioctl		= tipc_ioctl,
3206 	.listen		= tipc_listen,
3207 	.shutdown	= tipc_shutdown,
3208 	.setsockopt	= tipc_setsockopt,
3209 	.getsockopt	= tipc_getsockopt,
3210 	.sendmsg	= tipc_sendstream,
3211 	.recvmsg	= tipc_recvstream,
3212 	.mmap		= sock_no_mmap,
3213 	.sendpage	= sock_no_sendpage
3214 };
3215 
3216 static const struct net_proto_family tipc_family_ops = {
3217 	.owner		= THIS_MODULE,
3218 	.family		= AF_TIPC,
3219 	.create		= tipc_sk_create
3220 };
3221 
3222 static struct proto tipc_proto = {
3223 	.name		= "TIPC",
3224 	.owner		= THIS_MODULE,
3225 	.obj_size	= sizeof(struct tipc_sock),
3226 	.sysctl_rmem	= sysctl_tipc_rmem
3227 };
3228 
3229 /**
3230  * tipc_socket_init - initialize TIPC socket interface
3231  *
3232  * Returns 0 on success, errno otherwise
3233  */
3234 int tipc_socket_init(void)
3235 {
3236 	int res;
3237 
3238 	res = proto_register(&tipc_proto, 1);
3239 	if (res) {
3240 		pr_err("Failed to register TIPC protocol type\n");
3241 		goto out;
3242 	}
3243 
3244 	res = sock_register(&tipc_family_ops);
3245 	if (res) {
3246 		pr_err("Failed to register TIPC socket type\n");
3247 		proto_unregister(&tipc_proto);
3248 		goto out;
3249 	}
3250  out:
3251 	return res;
3252 }
3253 
3254 /**
3255  * tipc_socket_stop - stop TIPC socket interface
3256  */
3257 void tipc_socket_stop(void)
3258 {
3259 	sock_unregister(tipc_family_ops.family);
3260 	proto_unregister(&tipc_proto);
3261 }
3262 
3263 /* Caller should hold socket lock for the passed tipc socket. */
3264 static int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk)
3265 {
3266 	u32 peer_node;
3267 	u32 peer_port;
3268 	struct nlattr *nest;
3269 
3270 	peer_node = tsk_peer_node(tsk);
3271 	peer_port = tsk_peer_port(tsk);
3272 
3273 	nest = nla_nest_start(skb, TIPC_NLA_SOCK_CON);
3274 	if (!nest)
3275 		return -EMSGSIZE;
3276 
3277 	if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node))
3278 		goto msg_full;
3279 	if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port))
3280 		goto msg_full;
3281 
3282 	if (tsk->conn_type != 0) {
3283 		if (nla_put_flag(skb, TIPC_NLA_CON_FLAG))
3284 			goto msg_full;
3285 		if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, tsk->conn_type))
3286 			goto msg_full;
3287 		if (nla_put_u32(skb, TIPC_NLA_CON_INST, tsk->conn_instance))
3288 			goto msg_full;
3289 	}
3290 	nla_nest_end(skb, nest);
3291 
3292 	return 0;
3293 
3294 msg_full:
3295 	nla_nest_cancel(skb, nest);
3296 
3297 	return -EMSGSIZE;
3298 }
3299 
3300 static int __tipc_nl_add_sk_info(struct sk_buff *skb, struct tipc_sock
3301 			  *tsk)
3302 {
3303 	struct net *net = sock_net(skb->sk);
3304 	struct sock *sk = &tsk->sk;
3305 
3306 	if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->portid) ||
3307 	    nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tipc_own_addr(net)))
3308 		return -EMSGSIZE;
3309 
3310 	if (tipc_sk_connected(sk)) {
3311 		if (__tipc_nl_add_sk_con(skb, tsk))
3312 			return -EMSGSIZE;
3313 	} else if (!list_empty(&tsk->publications)) {
3314 		if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL))
3315 			return -EMSGSIZE;
3316 	}
3317 	return 0;
3318 }
3319 
3320 /* Caller should hold socket lock for the passed tipc socket. */
3321 static int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb,
3322 			    struct tipc_sock *tsk)
3323 {
3324 	struct nlattr *attrs;
3325 	void *hdr;
3326 
3327 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
3328 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_SOCK_GET);
3329 	if (!hdr)
3330 		goto msg_cancel;
3331 
3332 	attrs = nla_nest_start(skb, TIPC_NLA_SOCK);
3333 	if (!attrs)
3334 		goto genlmsg_cancel;
3335 
3336 	if (__tipc_nl_add_sk_info(skb, tsk))
3337 		goto attr_msg_cancel;
3338 
3339 	nla_nest_end(skb, attrs);
3340 	genlmsg_end(skb, hdr);
3341 
3342 	return 0;
3343 
3344 attr_msg_cancel:
3345 	nla_nest_cancel(skb, attrs);
3346 genlmsg_cancel:
3347 	genlmsg_cancel(skb, hdr);
3348 msg_cancel:
3349 	return -EMSGSIZE;
3350 }
3351 
3352 int tipc_nl_sk_walk(struct sk_buff *skb, struct netlink_callback *cb,
3353 		    int (*skb_handler)(struct sk_buff *skb,
3354 				       struct netlink_callback *cb,
3355 				       struct tipc_sock *tsk))
3356 {
3357 	struct rhashtable_iter *iter = (void *)cb->args[4];
3358 	struct tipc_sock *tsk;
3359 	int err;
3360 
3361 	rhashtable_walk_start(iter);
3362 	while ((tsk = rhashtable_walk_next(iter)) != NULL) {
3363 		if (IS_ERR(tsk)) {
3364 			err = PTR_ERR(tsk);
3365 			if (err == -EAGAIN) {
3366 				err = 0;
3367 				continue;
3368 			}
3369 			break;
3370 		}
3371 
3372 		sock_hold(&tsk->sk);
3373 		rhashtable_walk_stop(iter);
3374 		lock_sock(&tsk->sk);
3375 		err = skb_handler(skb, cb, tsk);
3376 		if (err) {
3377 			release_sock(&tsk->sk);
3378 			sock_put(&tsk->sk);
3379 			goto out;
3380 		}
3381 		release_sock(&tsk->sk);
3382 		rhashtable_walk_start(iter);
3383 		sock_put(&tsk->sk);
3384 	}
3385 	rhashtable_walk_stop(iter);
3386 out:
3387 	return skb->len;
3388 }
3389 EXPORT_SYMBOL(tipc_nl_sk_walk);
3390 
3391 int tipc_dump_start(struct netlink_callback *cb)
3392 {
3393 	return __tipc_dump_start(cb, sock_net(cb->skb->sk));
3394 }
3395 EXPORT_SYMBOL(tipc_dump_start);
3396 
3397 int __tipc_dump_start(struct netlink_callback *cb, struct net *net)
3398 {
3399 	/* tipc_nl_name_table_dump() uses cb->args[0...3]. */
3400 	struct rhashtable_iter *iter = (void *)cb->args[4];
3401 	struct tipc_net *tn = tipc_net(net);
3402 
3403 	if (!iter) {
3404 		iter = kmalloc(sizeof(*iter), GFP_KERNEL);
3405 		if (!iter)
3406 			return -ENOMEM;
3407 
3408 		cb->args[4] = (long)iter;
3409 	}
3410 
3411 	rhashtable_walk_enter(&tn->sk_rht, iter);
3412 	return 0;
3413 }
3414 
3415 int tipc_dump_done(struct netlink_callback *cb)
3416 {
3417 	struct rhashtable_iter *hti = (void *)cb->args[4];
3418 
3419 	rhashtable_walk_exit(hti);
3420 	kfree(hti);
3421 	return 0;
3422 }
3423 EXPORT_SYMBOL(tipc_dump_done);
3424 
3425 int tipc_sk_fill_sock_diag(struct sk_buff *skb, struct netlink_callback *cb,
3426 			   struct tipc_sock *tsk, u32 sk_filter_state,
3427 			   u64 (*tipc_diag_gen_cookie)(struct sock *sk))
3428 {
3429 	struct sock *sk = &tsk->sk;
3430 	struct nlattr *attrs;
3431 	struct nlattr *stat;
3432 
3433 	/*filter response w.r.t sk_state*/
3434 	if (!(sk_filter_state & (1 << sk->sk_state)))
3435 		return 0;
3436 
3437 	attrs = nla_nest_start(skb, TIPC_NLA_SOCK);
3438 	if (!attrs)
3439 		goto msg_cancel;
3440 
3441 	if (__tipc_nl_add_sk_info(skb, tsk))
3442 		goto attr_msg_cancel;
3443 
3444 	if (nla_put_u32(skb, TIPC_NLA_SOCK_TYPE, (u32)sk->sk_type) ||
3445 	    nla_put_u32(skb, TIPC_NLA_SOCK_TIPC_STATE, (u32)sk->sk_state) ||
3446 	    nla_put_u32(skb, TIPC_NLA_SOCK_INO, sock_i_ino(sk)) ||
3447 	    nla_put_u32(skb, TIPC_NLA_SOCK_UID,
3448 			from_kuid_munged(sk_user_ns(NETLINK_CB(cb->skb).sk),
3449 					 sock_i_uid(sk))) ||
3450 	    nla_put_u64_64bit(skb, TIPC_NLA_SOCK_COOKIE,
3451 			      tipc_diag_gen_cookie(sk),
3452 			      TIPC_NLA_SOCK_PAD))
3453 		goto attr_msg_cancel;
3454 
3455 	stat = nla_nest_start(skb, TIPC_NLA_SOCK_STAT);
3456 	if (!stat)
3457 		goto attr_msg_cancel;
3458 
3459 	if (nla_put_u32(skb, TIPC_NLA_SOCK_STAT_RCVQ,
3460 			skb_queue_len(&sk->sk_receive_queue)) ||
3461 	    nla_put_u32(skb, TIPC_NLA_SOCK_STAT_SENDQ,
3462 			skb_queue_len(&sk->sk_write_queue)) ||
3463 	    nla_put_u32(skb, TIPC_NLA_SOCK_STAT_DROP,
3464 			atomic_read(&sk->sk_drops)))
3465 		goto stat_msg_cancel;
3466 
3467 	if (tsk->cong_link_cnt &&
3468 	    nla_put_flag(skb, TIPC_NLA_SOCK_STAT_LINK_CONG))
3469 		goto stat_msg_cancel;
3470 
3471 	if (tsk_conn_cong(tsk) &&
3472 	    nla_put_flag(skb, TIPC_NLA_SOCK_STAT_CONN_CONG))
3473 		goto stat_msg_cancel;
3474 
3475 	nla_nest_end(skb, stat);
3476 
3477 	if (tsk->group)
3478 		if (tipc_group_fill_sock_diag(tsk->group, skb))
3479 			goto stat_msg_cancel;
3480 
3481 	nla_nest_end(skb, attrs);
3482 
3483 	return 0;
3484 
3485 stat_msg_cancel:
3486 	nla_nest_cancel(skb, stat);
3487 attr_msg_cancel:
3488 	nla_nest_cancel(skb, attrs);
3489 msg_cancel:
3490 	return -EMSGSIZE;
3491 }
3492 EXPORT_SYMBOL(tipc_sk_fill_sock_diag);
3493 
3494 int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb)
3495 {
3496 	return tipc_nl_sk_walk(skb, cb, __tipc_nl_add_sk);
3497 }
3498 
3499 /* Caller should hold socket lock for the passed tipc socket. */
3500 static int __tipc_nl_add_sk_publ(struct sk_buff *skb,
3501 				 struct netlink_callback *cb,
3502 				 struct publication *publ)
3503 {
3504 	void *hdr;
3505 	struct nlattr *attrs;
3506 
3507 	hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
3508 			  &tipc_genl_family, NLM_F_MULTI, TIPC_NL_PUBL_GET);
3509 	if (!hdr)
3510 		goto msg_cancel;
3511 
3512 	attrs = nla_nest_start(skb, TIPC_NLA_PUBL);
3513 	if (!attrs)
3514 		goto genlmsg_cancel;
3515 
3516 	if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key))
3517 		goto attr_msg_cancel;
3518 	if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->type))
3519 		goto attr_msg_cancel;
3520 	if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->lower))
3521 		goto attr_msg_cancel;
3522 	if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->upper))
3523 		goto attr_msg_cancel;
3524 
3525 	nla_nest_end(skb, attrs);
3526 	genlmsg_end(skb, hdr);
3527 
3528 	return 0;
3529 
3530 attr_msg_cancel:
3531 	nla_nest_cancel(skb, attrs);
3532 genlmsg_cancel:
3533 	genlmsg_cancel(skb, hdr);
3534 msg_cancel:
3535 	return -EMSGSIZE;
3536 }
3537 
3538 /* Caller should hold socket lock for the passed tipc socket. */
3539 static int __tipc_nl_list_sk_publ(struct sk_buff *skb,
3540 				  struct netlink_callback *cb,
3541 				  struct tipc_sock *tsk, u32 *last_publ)
3542 {
3543 	int err;
3544 	struct publication *p;
3545 
3546 	if (*last_publ) {
3547 		list_for_each_entry(p, &tsk->publications, binding_sock) {
3548 			if (p->key == *last_publ)
3549 				break;
3550 		}
3551 		if (p->key != *last_publ) {
3552 			/* We never set seq or call nl_dump_check_consistent()
3553 			 * this means that setting prev_seq here will cause the
3554 			 * consistence check to fail in the netlink callback
3555 			 * handler. Resulting in the last NLMSG_DONE message
3556 			 * having the NLM_F_DUMP_INTR flag set.
3557 			 */
3558 			cb->prev_seq = 1;
3559 			*last_publ = 0;
3560 			return -EPIPE;
3561 		}
3562 	} else {
3563 		p = list_first_entry(&tsk->publications, struct publication,
3564 				     binding_sock);
3565 	}
3566 
3567 	list_for_each_entry_from(p, &tsk->publications, binding_sock) {
3568 		err = __tipc_nl_add_sk_publ(skb, cb, p);
3569 		if (err) {
3570 			*last_publ = p->key;
3571 			return err;
3572 		}
3573 	}
3574 	*last_publ = 0;
3575 
3576 	return 0;
3577 }
3578 
3579 int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)
3580 {
3581 	int err;
3582 	u32 tsk_portid = cb->args[0];
3583 	u32 last_publ = cb->args[1];
3584 	u32 done = cb->args[2];
3585 	struct net *net = sock_net(skb->sk);
3586 	struct tipc_sock *tsk;
3587 
3588 	if (!tsk_portid) {
3589 		struct nlattr **attrs;
3590 		struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1];
3591 
3592 		err = tipc_nlmsg_parse(cb->nlh, &attrs);
3593 		if (err)
3594 			return err;
3595 
3596 		if (!attrs[TIPC_NLA_SOCK])
3597 			return -EINVAL;
3598 
3599 		err = nla_parse_nested(sock, TIPC_NLA_SOCK_MAX,
3600 				       attrs[TIPC_NLA_SOCK],
3601 				       tipc_nl_sock_policy, NULL);
3602 		if (err)
3603 			return err;
3604 
3605 		if (!sock[TIPC_NLA_SOCK_REF])
3606 			return -EINVAL;
3607 
3608 		tsk_portid = nla_get_u32(sock[TIPC_NLA_SOCK_REF]);
3609 	}
3610 
3611 	if (done)
3612 		return 0;
3613 
3614 	tsk = tipc_sk_lookup(net, tsk_portid);
3615 	if (!tsk)
3616 		return -EINVAL;
3617 
3618 	lock_sock(&tsk->sk);
3619 	err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ);
3620 	if (!err)
3621 		done = 1;
3622 	release_sock(&tsk->sk);
3623 	sock_put(&tsk->sk);
3624 
3625 	cb->args[0] = tsk_portid;
3626 	cb->args[1] = last_publ;
3627 	cb->args[2] = done;
3628 
3629 	return skb->len;
3630 }
3631 
3632 /**
3633  * tipc_sk_filtering - check if a socket should be traced
3634  * @sk: the socket to be examined
3635  * @sysctl_tipc_sk_filter[]: the socket tuple for filtering,
3636  *  (portid, sock type, name type, name lower, name upper)
3637  *
3638  * Returns true if the socket meets the socket tuple data
3639  * (value 0 = 'any') or when there is no tuple set (all = 0),
3640  * otherwise false
3641  */
3642 bool tipc_sk_filtering(struct sock *sk)
3643 {
3644 	struct tipc_sock *tsk;
3645 	struct publication *p;
3646 	u32 _port, _sktype, _type, _lower, _upper;
3647 	u32 type = 0, lower = 0, upper = 0;
3648 
3649 	if (!sk)
3650 		return true;
3651 
3652 	tsk = tipc_sk(sk);
3653 
3654 	_port = sysctl_tipc_sk_filter[0];
3655 	_sktype = sysctl_tipc_sk_filter[1];
3656 	_type = sysctl_tipc_sk_filter[2];
3657 	_lower = sysctl_tipc_sk_filter[3];
3658 	_upper = sysctl_tipc_sk_filter[4];
3659 
3660 	if (!_port && !_sktype && !_type && !_lower && !_upper)
3661 		return true;
3662 
3663 	if (_port)
3664 		return (_port == tsk->portid);
3665 
3666 	if (_sktype && _sktype != sk->sk_type)
3667 		return false;
3668 
3669 	if (tsk->published) {
3670 		p = list_first_entry_or_null(&tsk->publications,
3671 					     struct publication, binding_sock);
3672 		if (p) {
3673 			type = p->type;
3674 			lower = p->lower;
3675 			upper = p->upper;
3676 		}
3677 	}
3678 
3679 	if (!tipc_sk_type_connectionless(sk)) {
3680 		type = tsk->conn_type;
3681 		lower = tsk->conn_instance;
3682 		upper = tsk->conn_instance;
3683 	}
3684 
3685 	if ((_type && _type != type) || (_lower && _lower != lower) ||
3686 	    (_upper && _upper != upper))
3687 		return false;
3688 
3689 	return true;
3690 }
3691 
3692 u32 tipc_sock_get_portid(struct sock *sk)
3693 {
3694 	return (sk) ? (tipc_sk(sk))->portid : 0;
3695 }
3696 
3697 /**
3698  * tipc_sk_overlimit1 - check if socket rx queue is about to be overloaded,
3699  *			both the rcv and backlog queues are considered
3700  * @sk: tipc sk to be checked
3701  * @skb: tipc msg to be checked
3702  *
3703  * Returns true if the socket rx queue allocation is > 90%, otherwise false
3704  */
3705 
3706 bool tipc_sk_overlimit1(struct sock *sk, struct sk_buff *skb)
3707 {
3708 	atomic_t *dcnt = &tipc_sk(sk)->dupl_rcvcnt;
3709 	unsigned int lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
3710 	unsigned int qsize = sk->sk_backlog.len + sk_rmem_alloc_get(sk);
3711 
3712 	return (qsize > lim * 90 / 100);
3713 }
3714 
3715 /**
3716  * tipc_sk_overlimit2 - check if socket rx queue is about to be overloaded,
3717  *			only the rcv queue is considered
3718  * @sk: tipc sk to be checked
3719  * @skb: tipc msg to be checked
3720  *
3721  * Returns true if the socket rx queue allocation is > 90%, otherwise false
3722  */
3723 
3724 bool tipc_sk_overlimit2(struct sock *sk, struct sk_buff *skb)
3725 {
3726 	unsigned int lim = rcvbuf_limit(sk, skb);
3727 	unsigned int qsize = sk_rmem_alloc_get(sk);
3728 
3729 	return (qsize > lim * 90 / 100);
3730 }
3731 
3732 /**
3733  * tipc_sk_dump - dump TIPC socket
3734  * @sk: tipc sk to be dumped
3735  * @dqueues: bitmask to decide if any socket queue to be dumped?
3736  *           - TIPC_DUMP_NONE: don't dump socket queues
3737  *           - TIPC_DUMP_SK_SNDQ: dump socket send queue
3738  *           - TIPC_DUMP_SK_RCVQ: dump socket rcv queue
3739  *           - TIPC_DUMP_SK_BKLGQ: dump socket backlog queue
3740  *           - TIPC_DUMP_ALL: dump all the socket queues above
3741  * @buf: returned buffer of dump data in format
3742  */
3743 int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)
3744 {
3745 	int i = 0;
3746 	size_t sz = (dqueues) ? SK_LMAX : SK_LMIN;
3747 	struct tipc_sock *tsk;
3748 	struct publication *p;
3749 	bool tsk_connected;
3750 
3751 	if (!sk) {
3752 		i += scnprintf(buf, sz, "sk data: (null)\n");
3753 		return i;
3754 	}
3755 
3756 	tsk = tipc_sk(sk);
3757 	tsk_connected = !tipc_sk_type_connectionless(sk);
3758 
3759 	i += scnprintf(buf, sz, "sk data: %u", sk->sk_type);
3760 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_state);
3761 	i += scnprintf(buf + i, sz - i, " %x", tsk_own_node(tsk));
3762 	i += scnprintf(buf + i, sz - i, " %u", tsk->portid);
3763 	i += scnprintf(buf + i, sz - i, " | %u", tsk_connected);
3764 	if (tsk_connected) {
3765 		i += scnprintf(buf + i, sz - i, " %x", tsk_peer_node(tsk));
3766 		i += scnprintf(buf + i, sz - i, " %u", tsk_peer_port(tsk));
3767 		i += scnprintf(buf + i, sz - i, " %u", tsk->conn_type);
3768 		i += scnprintf(buf + i, sz - i, " %u", tsk->conn_instance);
3769 	}
3770 	i += scnprintf(buf + i, sz - i, " | %u", tsk->published);
3771 	if (tsk->published) {
3772 		p = list_first_entry_or_null(&tsk->publications,
3773 					     struct publication, binding_sock);
3774 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->type : 0);
3775 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->lower : 0);
3776 		i += scnprintf(buf + i, sz - i, " %u", (p) ? p->upper : 0);
3777 	}
3778 	i += scnprintf(buf + i, sz - i, " | %u", tsk->snd_win);
3779 	i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_win);
3780 	i += scnprintf(buf + i, sz - i, " %u", tsk->max_pkt);
3781 	i += scnprintf(buf + i, sz - i, " %x", tsk->peer_caps);
3782 	i += scnprintf(buf + i, sz - i, " %u", tsk->cong_link_cnt);
3783 	i += scnprintf(buf + i, sz - i, " %u", tsk->snt_unacked);
3784 	i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_unacked);
3785 	i += scnprintf(buf + i, sz - i, " %u", atomic_read(&tsk->dupl_rcvcnt));
3786 	i += scnprintf(buf + i, sz - i, " %u", sk->sk_shutdown);
3787 	i += scnprintf(buf + i, sz - i, " | %d", sk_wmem_alloc_get(sk));
3788 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_sndbuf);
3789 	i += scnprintf(buf + i, sz - i, " | %d", sk_rmem_alloc_get(sk));
3790 	i += scnprintf(buf + i, sz - i, " %d", sk->sk_rcvbuf);
3791 	i += scnprintf(buf + i, sz - i, " | %d\n", sk->sk_backlog.len);
3792 
3793 	if (dqueues & TIPC_DUMP_SK_SNDQ) {
3794 		i += scnprintf(buf + i, sz - i, "sk_write_queue: ");
3795 		i += tipc_list_dump(&sk->sk_write_queue, false, buf + i);
3796 	}
3797 
3798 	if (dqueues & TIPC_DUMP_SK_RCVQ) {
3799 		i += scnprintf(buf + i, sz - i, "sk_receive_queue: ");
3800 		i += tipc_list_dump(&sk->sk_receive_queue, false, buf + i);
3801 	}
3802 
3803 	if (dqueues & TIPC_DUMP_SK_BKLGQ) {
3804 		i += scnprintf(buf + i, sz - i, "sk_backlog:\n  head ");
3805 		i += tipc_skb_dump(sk->sk_backlog.head, false, buf + i);
3806 		if (sk->sk_backlog.tail != sk->sk_backlog.head) {
3807 			i += scnprintf(buf + i, sz - i, "  tail ");
3808 			i += tipc_skb_dump(sk->sk_backlog.tail, false,
3809 					   buf + i);
3810 		}
3811 	}
3812 
3813 	return i;
3814 }
3815