xref: /openbmc/linux/net/tipc/name_table.c (revision b868a02e)
1 /*
2  * net/tipc/name_table.c: TIPC name table code
3  *
4  * Copyright (c) 2000-2006, 2014-2018, Ericsson AB
5  * Copyright (c) 2004-2008, 2010-2014, Wind River Systems
6  * Copyright (c) 2020-2021, Red Hat Inc
7  * All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions are met:
11  *
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. Neither the names of the copyright holders nor the names of its
18  *    contributors may be used to endorse or promote products derived from
19  *    this software without specific prior written permission.
20  *
21  * Alternatively, this software may be distributed under the terms of the
22  * GNU General Public License ("GPL") version 2 as published by the Free
23  * Software Foundation.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
26  * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
29  * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
35  * POSSIBILITY OF SUCH DAMAGE.
36  */
37 
38 #include <net/sock.h>
39 #include <linux/list_sort.h>
40 #include <linux/rbtree_augmented.h>
41 #include "core.h"
42 #include "netlink.h"
43 #include "name_table.h"
44 #include "name_distr.h"
45 #include "subscr.h"
46 #include "bcast.h"
47 #include "addr.h"
48 #include "node.h"
49 #include "group.h"
50 
51 /**
52  * struct service_range - container for all bindings of a service range
53  * @lower: service range lower bound
54  * @upper: service range upper bound
55  * @tree_node: member of service range RB tree
56  * @max: largest 'upper' in this node subtree
57  * @local_publ: list of identical publications made from this node
58  *   Used by closest_first lookup and multicast lookup algorithm
59  * @all_publ: all publications identical to this one, whatever node and scope
60  *   Used by round-robin lookup algorithm
61  */
62 struct service_range {
63 	u32 lower;
64 	u32 upper;
65 	struct rb_node tree_node;
66 	u32 max;
67 	struct list_head local_publ;
68 	struct list_head all_publ;
69 };
70 
71 /**
72  * struct tipc_service - container for all published instances of a service type
73  * @type: 32 bit 'type' value for service
74  * @publ_cnt: increasing counter for publications in this service
75  * @ranges: rb tree containing all service ranges for this service
76  * @service_list: links to adjacent name ranges in hash chain
77  * @subscriptions: list of subscriptions for this service type
78  * @lock: spinlock controlling access to pertaining service ranges/publications
79  * @rcu: RCU callback head used for deferred freeing
80  */
81 struct tipc_service {
82 	u32 type;
83 	u32 publ_cnt;
84 	struct rb_root ranges;
85 	struct hlist_node service_list;
86 	struct list_head subscriptions;
87 	spinlock_t lock; /* Covers service range list */
88 	struct rcu_head rcu;
89 };
90 
91 #define service_range_upper(sr) ((sr)->upper)
92 RB_DECLARE_CALLBACKS_MAX(static, sr_callbacks,
93 			 struct service_range, tree_node, u32, max,
94 			 service_range_upper)
95 
96 #define service_range_entry(rbtree_node)				\
97 	(container_of(rbtree_node, struct service_range, tree_node))
98 
99 #define service_range_overlap(sr, start, end)				\
100 	((sr)->lower <= (end) && (sr)->upper >= (start))
101 
102 /**
103  * service_range_foreach_match - iterate over tipc service rbtree for each
104  *                               range match
105  * @sr: the service range pointer as a loop cursor
106  * @sc: the pointer to tipc service which holds the service range rbtree
107  * @start: beginning of the search range (end >= start) for matching
108  * @end: end of the search range (end >= start) for matching
109  */
110 #define service_range_foreach_match(sr, sc, start, end)			\
111 	for (sr = service_range_match_first((sc)->ranges.rb_node,	\
112 					    start,			\
113 					    end);			\
114 	     sr;							\
115 	     sr = service_range_match_next(&(sr)->tree_node,		\
116 					   start,			\
117 					   end))
118 
119 /**
120  * service_range_match_first - find first service range matching a range
121  * @n: the root node of service range rbtree for searching
122  * @start: beginning of the search range (end >= start) for matching
123  * @end: end of the search range (end >= start) for matching
124  *
125  * Return: the leftmost service range node in the rbtree that overlaps the
126  * specific range if any. Otherwise, returns NULL.
127  */
128 static struct service_range *service_range_match_first(struct rb_node *n,
129 						       u32 start, u32 end)
130 {
131 	struct service_range *sr;
132 	struct rb_node *l, *r;
133 
134 	/* Non overlaps in tree at all? */
135 	if (!n || service_range_entry(n)->max < start)
136 		return NULL;
137 
138 	while (n) {
139 		l = n->rb_left;
140 		if (l && service_range_entry(l)->max >= start) {
141 			/* A leftmost overlap range node must be one in the left
142 			 * subtree. If not, it has lower > end, then nodes on
143 			 * the right side cannot satisfy the condition either.
144 			 */
145 			n = l;
146 			continue;
147 		}
148 
149 		/* No one in the left subtree can match, return if this node is
150 		 * an overlap i.e. leftmost.
151 		 */
152 		sr = service_range_entry(n);
153 		if (service_range_overlap(sr, start, end))
154 			return sr;
155 
156 		/* Ok, try to lookup on the right side */
157 		r = n->rb_right;
158 		if (sr->lower <= end &&
159 		    r && service_range_entry(r)->max >= start) {
160 			n = r;
161 			continue;
162 		}
163 		break;
164 	}
165 
166 	return NULL;
167 }
168 
169 /**
170  * service_range_match_next - find next service range matching a range
171  * @n: a node in service range rbtree from which the searching starts
172  * @start: beginning of the search range (end >= start) for matching
173  * @end: end of the search range (end >= start) for matching
174  *
175  * Return: the next service range node to the given node in the rbtree that
176  * overlaps the specific range if any. Otherwise, returns NULL.
177  */
178 static struct service_range *service_range_match_next(struct rb_node *n,
179 						      u32 start, u32 end)
180 {
181 	struct service_range *sr;
182 	struct rb_node *p, *r;
183 
184 	while (n) {
185 		r = n->rb_right;
186 		if (r && service_range_entry(r)->max >= start)
187 			/* A next overlap range node must be one in the right
188 			 * subtree. If not, it has lower > end, then any next
189 			 * successor (- an ancestor) of this node cannot
190 			 * satisfy the condition either.
191 			 */
192 			return service_range_match_first(r, start, end);
193 
194 		/* No one in the right subtree can match, go up to find an
195 		 * ancestor of this node which is parent of a left-hand child.
196 		 */
197 		while ((p = rb_parent(n)) && n == p->rb_right)
198 			n = p;
199 		if (!p)
200 			break;
201 
202 		/* Return if this ancestor is an overlap */
203 		sr = service_range_entry(p);
204 		if (service_range_overlap(sr, start, end))
205 			return sr;
206 
207 		/* Ok, try to lookup more from this ancestor */
208 		if (sr->lower <= end) {
209 			n = p;
210 			continue;
211 		}
212 		break;
213 	}
214 
215 	return NULL;
216 }
217 
218 static int hash(int x)
219 {
220 	return x & (TIPC_NAMETBL_SIZE - 1);
221 }
222 
223 /**
224  * tipc_publ_create - create a publication structure
225  * @ua: the service range the user is binding to
226  * @sk: the address of the socket that is bound
227  * @key: publication key
228  */
229 static struct publication *tipc_publ_create(struct tipc_uaddr *ua,
230 					    struct tipc_socket_addr *sk,
231 					    u32 key)
232 {
233 	struct publication *p = kzalloc(sizeof(*p), GFP_ATOMIC);
234 
235 	if (!p)
236 		return NULL;
237 
238 	p->sr = ua->sr;
239 	p->sk = *sk;
240 	p->scope = ua->scope;
241 	p->key = key;
242 	INIT_LIST_HEAD(&p->binding_sock);
243 	INIT_LIST_HEAD(&p->binding_node);
244 	INIT_LIST_HEAD(&p->local_publ);
245 	INIT_LIST_HEAD(&p->all_publ);
246 	INIT_LIST_HEAD(&p->list);
247 	return p;
248 }
249 
250 /**
251  * tipc_service_create - create a service structure for the specified 'type'
252  * @net: network namespace
253  * @ua: address representing the service to be bound
254  *
255  * Allocates a single range structure and sets it to all 0's.
256  */
257 static struct tipc_service *tipc_service_create(struct net *net,
258 						struct tipc_uaddr *ua)
259 {
260 	struct name_table *nt = tipc_name_table(net);
261 	struct tipc_service *service;
262 	struct hlist_head *hd;
263 
264 	service = kzalloc(sizeof(*service), GFP_ATOMIC);
265 	if (!service) {
266 		pr_warn("Service creation failed, no memory\n");
267 		return NULL;
268 	}
269 
270 	spin_lock_init(&service->lock);
271 	service->type = ua->sr.type;
272 	service->ranges = RB_ROOT;
273 	INIT_HLIST_NODE(&service->service_list);
274 	INIT_LIST_HEAD(&service->subscriptions);
275 	hd = &nt->services[hash(ua->sr.type)];
276 	hlist_add_head_rcu(&service->service_list, hd);
277 	return service;
278 }
279 
280 /*  tipc_service_find_range - find service range matching publication parameters
281  */
282 static struct service_range *tipc_service_find_range(struct tipc_service *sc,
283 						     struct tipc_uaddr *ua)
284 {
285 	struct service_range *sr;
286 
287 	service_range_foreach_match(sr, sc, ua->sr.lower, ua->sr.upper) {
288 		/* Look for exact match */
289 		if (sr->lower == ua->sr.lower && sr->upper == ua->sr.upper)
290 			return sr;
291 	}
292 
293 	return NULL;
294 }
295 
296 static struct service_range *tipc_service_create_range(struct tipc_service *sc,
297 						       struct publication *p)
298 {
299 	struct rb_node **n, *parent = NULL;
300 	struct service_range *sr;
301 	u32 lower = p->sr.lower;
302 	u32 upper = p->sr.upper;
303 
304 	n = &sc->ranges.rb_node;
305 	while (*n) {
306 		parent = *n;
307 		sr = service_range_entry(parent);
308 		if (lower == sr->lower && upper == sr->upper)
309 			return sr;
310 		if (sr->max < upper)
311 			sr->max = upper;
312 		if (lower <= sr->lower)
313 			n = &parent->rb_left;
314 		else
315 			n = &parent->rb_right;
316 	}
317 	sr = kzalloc(sizeof(*sr), GFP_ATOMIC);
318 	if (!sr)
319 		return NULL;
320 	sr->lower = lower;
321 	sr->upper = upper;
322 	sr->max = upper;
323 	INIT_LIST_HEAD(&sr->local_publ);
324 	INIT_LIST_HEAD(&sr->all_publ);
325 	rb_link_node(&sr->tree_node, parent, n);
326 	rb_insert_augmented(&sr->tree_node, &sc->ranges, &sr_callbacks);
327 	return sr;
328 }
329 
330 static bool tipc_service_insert_publ(struct net *net,
331 				     struct tipc_service *sc,
332 				     struct publication *p)
333 {
334 	struct tipc_subscription *sub, *tmp;
335 	struct service_range *sr;
336 	struct publication *_p;
337 	u32 node = p->sk.node;
338 	bool first = false;
339 	bool res = false;
340 	u32 key = p->key;
341 
342 	spin_lock_bh(&sc->lock);
343 	sr = tipc_service_create_range(sc, p);
344 	if (!sr)
345 		goto  exit;
346 
347 	first = list_empty(&sr->all_publ);
348 
349 	/* Return if the publication already exists */
350 	list_for_each_entry(_p, &sr->all_publ, all_publ) {
351 		if (_p->key == key && (!_p->sk.node || _p->sk.node == node)) {
352 			pr_debug("Failed to bind duplicate %u,%u,%u/%u:%u/%u\n",
353 				 p->sr.type, p->sr.lower, p->sr.upper,
354 				 node, p->sk.ref, key);
355 			goto exit;
356 		}
357 	}
358 
359 	if (in_own_node(net, p->sk.node))
360 		list_add(&p->local_publ, &sr->local_publ);
361 	list_add(&p->all_publ, &sr->all_publ);
362 	p->id = sc->publ_cnt++;
363 
364 	/* Any subscriptions waiting for notification?  */
365 	list_for_each_entry_safe(sub, tmp, &sc->subscriptions, service_list) {
366 		tipc_sub_report_overlap(sub, p, TIPC_PUBLISHED, first);
367 	}
368 	res = true;
369 exit:
370 	if (!res)
371 		pr_warn("Failed to bind to %u,%u,%u\n",
372 			p->sr.type, p->sr.lower, p->sr.upper);
373 	spin_unlock_bh(&sc->lock);
374 	return res;
375 }
376 
377 /**
378  * tipc_service_remove_publ - remove a publication from a service
379  * @r: service_range to remove publication from
380  * @sk: address publishing socket
381  * @key: target publication key
382  */
383 static struct publication *tipc_service_remove_publ(struct service_range *r,
384 						    struct tipc_socket_addr *sk,
385 						    u32 key)
386 {
387 	struct publication *p;
388 	u32 node = sk->node;
389 
390 	list_for_each_entry(p, &r->all_publ, all_publ) {
391 		if (p->key != key || (node && node != p->sk.node))
392 			continue;
393 		list_del(&p->all_publ);
394 		list_del(&p->local_publ);
395 		return p;
396 	}
397 	return NULL;
398 }
399 
400 /*
401  * Code reused: time_after32() for the same purpose
402  */
403 #define publication_after(pa, pb) time_after32((pa)->id, (pb)->id)
404 static int tipc_publ_sort(void *priv, const struct list_head *a,
405 			  const struct list_head *b)
406 {
407 	struct publication *pa, *pb;
408 
409 	pa = container_of(a, struct publication, list);
410 	pb = container_of(b, struct publication, list);
411 	return publication_after(pa, pb);
412 }
413 
414 /**
415  * tipc_service_subscribe - attach a subscription, and optionally
416  * issue the prescribed number of events if there is any service
417  * range overlapping with the requested range
418  * @service: the tipc_service to attach the @sub to
419  * @sub: the subscription to attach
420  */
421 static void tipc_service_subscribe(struct tipc_service *service,
422 				   struct tipc_subscription *sub)
423 {
424 	struct publication *p, *first, *tmp;
425 	struct list_head publ_list;
426 	struct service_range *sr;
427 	u32 filter, lower, upper;
428 
429 	filter = sub->s.filter;
430 	lower = sub->s.seq.lower;
431 	upper = sub->s.seq.upper;
432 
433 	tipc_sub_get(sub);
434 	list_add(&sub->service_list, &service->subscriptions);
435 
436 	if (filter & TIPC_SUB_NO_STATUS)
437 		return;
438 
439 	INIT_LIST_HEAD(&publ_list);
440 	service_range_foreach_match(sr, service, lower, upper) {
441 		first = NULL;
442 		list_for_each_entry(p, &sr->all_publ, all_publ) {
443 			if (filter & TIPC_SUB_PORTS)
444 				list_add_tail(&p->list, &publ_list);
445 			else if (!first || publication_after(first, p))
446 				/* Pick this range's *first* publication */
447 				first = p;
448 		}
449 		if (first)
450 			list_add_tail(&first->list, &publ_list);
451 	}
452 
453 	/* Sort the publications before reporting */
454 	list_sort(NULL, &publ_list, tipc_publ_sort);
455 	list_for_each_entry_safe(p, tmp, &publ_list, list) {
456 		tipc_sub_report_overlap(sub, p, TIPC_PUBLISHED, true);
457 		list_del_init(&p->list);
458 	}
459 }
460 
461 static struct tipc_service *tipc_service_find(struct net *net,
462 					      struct tipc_uaddr *ua)
463 {
464 	struct name_table *nt = tipc_name_table(net);
465 	struct hlist_head *service_head;
466 	struct tipc_service *service;
467 
468 	service_head = &nt->services[hash(ua->sr.type)];
469 	hlist_for_each_entry_rcu(service, service_head, service_list) {
470 		if (service->type == ua->sr.type)
471 			return service;
472 	}
473 	return NULL;
474 };
475 
476 struct publication *tipc_nametbl_insert_publ(struct net *net,
477 					     struct tipc_uaddr *ua,
478 					     struct tipc_socket_addr *sk,
479 					     u32 key)
480 {
481 	struct tipc_service *sc;
482 	struct publication *p;
483 
484 	p = tipc_publ_create(ua, sk, key);
485 	if (!p)
486 		return NULL;
487 
488 	sc = tipc_service_find(net, ua);
489 	if (!sc)
490 		sc = tipc_service_create(net, ua);
491 	if (sc && tipc_service_insert_publ(net, sc, p))
492 		return p;
493 	kfree(p);
494 	return NULL;
495 }
496 
497 struct publication *tipc_nametbl_remove_publ(struct net *net,
498 					     struct tipc_uaddr *ua,
499 					     struct tipc_socket_addr *sk,
500 					     u32 key)
501 {
502 	struct tipc_subscription *sub, *tmp;
503 	struct publication *p = NULL;
504 	struct service_range *sr;
505 	struct tipc_service *sc;
506 	bool last;
507 
508 	sc = tipc_service_find(net, ua);
509 	if (!sc)
510 		goto exit;
511 
512 	spin_lock_bh(&sc->lock);
513 	sr = tipc_service_find_range(sc, ua);
514 	if (!sr)
515 		goto unlock;
516 	p = tipc_service_remove_publ(sr, sk, key);
517 	if (!p)
518 		goto unlock;
519 
520 	/* Notify any waiting subscriptions */
521 	last = list_empty(&sr->all_publ);
522 	list_for_each_entry_safe(sub, tmp, &sc->subscriptions, service_list) {
523 		tipc_sub_report_overlap(sub, p, TIPC_WITHDRAWN, last);
524 	}
525 
526 	/* Remove service range item if this was its last publication */
527 	if (list_empty(&sr->all_publ)) {
528 		rb_erase_augmented(&sr->tree_node, &sc->ranges, &sr_callbacks);
529 		kfree(sr);
530 	}
531 
532 	/* Delete service item if no more publications and subscriptions */
533 	if (RB_EMPTY_ROOT(&sc->ranges) && list_empty(&sc->subscriptions)) {
534 		hlist_del_init_rcu(&sc->service_list);
535 		kfree_rcu(sc, rcu);
536 	}
537 unlock:
538 	spin_unlock_bh(&sc->lock);
539 exit:
540 	if (!p) {
541 		pr_err("Failed to remove unknown binding: %u,%u,%u/%u:%u/%u\n",
542 		       ua->sr.type, ua->sr.lower, ua->sr.upper,
543 		       sk->node, sk->ref, key);
544 	}
545 	return p;
546 }
547 
548 /**
549  * tipc_nametbl_lookup_anycast - perform service instance to socket translation
550  * @net: network namespace
551  * @ua: service address to look up
552  * @sk: address to socket we want to find
553  *
554  * On entry, a non-zero 'sk->node' indicates the node where we want lookup to be
555  * performed, which may not be this one.
556  *
557  * On exit:
558  *
559  * - If lookup is deferred to another node, leave 'sk->node' unchanged and
560  *   return 'true'.
561  * - If lookup is successful, set the 'sk->node' and 'sk->ref' (== portid) which
562  *   represent the bound socket and return 'true'.
563  * - If lookup fails, return 'false'
564  *
565  * Note that for legacy users (node configured with Z.C.N address format) the
566  * 'closest-first' lookup algorithm must be maintained, i.e., if sk.node is 0
567  * we must look in the local binding list first
568  */
569 bool tipc_nametbl_lookup_anycast(struct net *net,
570 				 struct tipc_uaddr *ua,
571 				 struct tipc_socket_addr *sk)
572 {
573 	struct tipc_net *tn = tipc_net(net);
574 	bool legacy = tn->legacy_addr_format;
575 	u32 self = tipc_own_addr(net);
576 	u32 inst = ua->sa.instance;
577 	struct service_range *r;
578 	struct tipc_service *sc;
579 	struct publication *p;
580 	struct list_head *l;
581 	bool res = false;
582 
583 	if (!tipc_in_scope(legacy, sk->node, self))
584 		return true;
585 
586 	rcu_read_lock();
587 	sc = tipc_service_find(net, ua);
588 	if (unlikely(!sc))
589 		goto exit;
590 
591 	spin_lock_bh(&sc->lock);
592 	service_range_foreach_match(r, sc, inst, inst) {
593 		/* Select lookup algo: local, closest-first or round-robin */
594 		if (sk->node == self) {
595 			l = &r->local_publ;
596 			if (list_empty(l))
597 				continue;
598 			p = list_first_entry(l, struct publication, local_publ);
599 			list_move_tail(&p->local_publ, &r->local_publ);
600 		} else if (legacy && !sk->node && !list_empty(&r->local_publ)) {
601 			l = &r->local_publ;
602 			p = list_first_entry(l, struct publication, local_publ);
603 			list_move_tail(&p->local_publ, &r->local_publ);
604 		} else {
605 			l = &r->all_publ;
606 			p = list_first_entry(l, struct publication, all_publ);
607 			list_move_tail(&p->all_publ, &r->all_publ);
608 		}
609 		*sk = p->sk;
610 		res = true;
611 		/* Todo: as for legacy, pick the first matching range only, a
612 		 * "true" round-robin will be performed as needed.
613 		 */
614 		break;
615 	}
616 	spin_unlock_bh(&sc->lock);
617 
618 exit:
619 	rcu_read_unlock();
620 	return res;
621 }
622 
623 /* tipc_nametbl_lookup_group(): lookup destinaton(s) in a communication group
624  * Returns a list of one (== group anycast) or more (== group multicast)
625  * destination socket/node pairs matching the given address.
626  * The requester may or may not want to exclude himself from the list.
627  */
628 bool tipc_nametbl_lookup_group(struct net *net, struct tipc_uaddr *ua,
629 			       struct list_head *dsts, int *dstcnt,
630 			       u32 exclude, bool mcast)
631 {
632 	u32 self = tipc_own_addr(net);
633 	u32 inst = ua->sa.instance;
634 	struct service_range *sr;
635 	struct tipc_service *sc;
636 	struct publication *p;
637 
638 	*dstcnt = 0;
639 	rcu_read_lock();
640 	sc = tipc_service_find(net, ua);
641 	if (unlikely(!sc))
642 		goto exit;
643 
644 	spin_lock_bh(&sc->lock);
645 
646 	/* Todo: a full search i.e. service_range_foreach_match() instead? */
647 	sr = service_range_match_first(sc->ranges.rb_node, inst, inst);
648 	if (!sr)
649 		goto no_match;
650 
651 	list_for_each_entry(p, &sr->all_publ, all_publ) {
652 		if (p->scope != ua->scope)
653 			continue;
654 		if (p->sk.ref == exclude && p->sk.node == self)
655 			continue;
656 		tipc_dest_push(dsts, p->sk.node, p->sk.ref);
657 		(*dstcnt)++;
658 		if (mcast)
659 			continue;
660 		list_move_tail(&p->all_publ, &sr->all_publ);
661 		break;
662 	}
663 no_match:
664 	spin_unlock_bh(&sc->lock);
665 exit:
666 	rcu_read_unlock();
667 	return !list_empty(dsts);
668 }
669 
670 /* tipc_nametbl_lookup_mcast_sockets(): look up node local destinaton sockets
671  *                                      matching the given address
672  * Used on nodes which have received a multicast/broadcast message
673  * Returns a list of local sockets
674  */
675 void tipc_nametbl_lookup_mcast_sockets(struct net *net, struct tipc_uaddr *ua,
676 				       struct list_head *dports)
677 {
678 	struct service_range *sr;
679 	struct tipc_service *sc;
680 	struct publication *p;
681 	u8 scope = ua->scope;
682 
683 	rcu_read_lock();
684 	sc = tipc_service_find(net, ua);
685 	if (!sc)
686 		goto exit;
687 
688 	spin_lock_bh(&sc->lock);
689 	service_range_foreach_match(sr, sc, ua->sr.lower, ua->sr.upper) {
690 		list_for_each_entry(p, &sr->local_publ, local_publ) {
691 			if (scope == p->scope || scope == TIPC_ANY_SCOPE)
692 				tipc_dest_push(dports, 0, p->sk.ref);
693 		}
694 	}
695 	spin_unlock_bh(&sc->lock);
696 exit:
697 	rcu_read_unlock();
698 }
699 
700 /* tipc_nametbl_lookup_mcast_nodes(): look up all destination nodes matching
701  *                                    the given address. Used in sending node.
702  * Used on nodes which are sending out a multicast/broadcast message
703  * Returns a list of nodes, including own node if applicable
704  */
705 void tipc_nametbl_lookup_mcast_nodes(struct net *net, struct tipc_uaddr *ua,
706 				     struct tipc_nlist *nodes)
707 {
708 	struct service_range *sr;
709 	struct tipc_service *sc;
710 	struct publication *p;
711 
712 	rcu_read_lock();
713 	sc = tipc_service_find(net, ua);
714 	if (!sc)
715 		goto exit;
716 
717 	spin_lock_bh(&sc->lock);
718 	service_range_foreach_match(sr, sc, ua->sr.lower, ua->sr.upper) {
719 		list_for_each_entry(p, &sr->all_publ, all_publ) {
720 			tipc_nlist_add(nodes, p->sk.node);
721 		}
722 	}
723 	spin_unlock_bh(&sc->lock);
724 exit:
725 	rcu_read_unlock();
726 }
727 
728 /* tipc_nametbl_build_group - build list of communication group members
729  */
730 void tipc_nametbl_build_group(struct net *net, struct tipc_group *grp,
731 			      struct tipc_uaddr *ua)
732 {
733 	struct service_range *sr;
734 	struct tipc_service *sc;
735 	struct publication *p;
736 	struct rb_node *n;
737 
738 	rcu_read_lock();
739 	sc = tipc_service_find(net, ua);
740 	if (!sc)
741 		goto exit;
742 
743 	spin_lock_bh(&sc->lock);
744 	for (n = rb_first(&sc->ranges); n; n = rb_next(n)) {
745 		sr = container_of(n, struct service_range, tree_node);
746 		list_for_each_entry(p, &sr->all_publ, all_publ) {
747 			if (p->scope != ua->scope)
748 				continue;
749 			tipc_group_add_member(grp, p->sk.node, p->sk.ref,
750 					      p->sr.lower);
751 		}
752 	}
753 	spin_unlock_bh(&sc->lock);
754 exit:
755 	rcu_read_unlock();
756 }
757 
758 /* tipc_nametbl_publish - add service binding to name table
759  */
760 struct publication *tipc_nametbl_publish(struct net *net, struct tipc_uaddr *ua,
761 					 struct tipc_socket_addr *sk, u32 key)
762 {
763 	struct name_table *nt = tipc_name_table(net);
764 	struct tipc_net *tn = tipc_net(net);
765 	struct publication *p = NULL;
766 	struct sk_buff *skb = NULL;
767 	u32 rc_dests;
768 
769 	spin_lock_bh(&tn->nametbl_lock);
770 
771 	if (nt->local_publ_count >= TIPC_MAX_PUBL) {
772 		pr_warn("Bind failed, max limit %u reached\n", TIPC_MAX_PUBL);
773 		goto exit;
774 	}
775 
776 	p = tipc_nametbl_insert_publ(net, ua, sk, key);
777 	if (p) {
778 		nt->local_publ_count++;
779 		skb = tipc_named_publish(net, p);
780 	}
781 	rc_dests = nt->rc_dests;
782 exit:
783 	spin_unlock_bh(&tn->nametbl_lock);
784 
785 	if (skb)
786 		tipc_node_broadcast(net, skb, rc_dests);
787 	return p;
788 
789 }
790 
791 /**
792  * tipc_nametbl_withdraw - withdraw a service binding
793  * @net: network namespace
794  * @ua: service address/range being unbound
795  * @sk: address of the socket being unbound from
796  * @key: target publication key
797  */
798 void tipc_nametbl_withdraw(struct net *net, struct tipc_uaddr *ua,
799 			   struct tipc_socket_addr *sk, u32 key)
800 {
801 	struct name_table *nt = tipc_name_table(net);
802 	struct tipc_net *tn = tipc_net(net);
803 	struct sk_buff *skb = NULL;
804 	struct publication *p;
805 	u32 rc_dests;
806 
807 	spin_lock_bh(&tn->nametbl_lock);
808 
809 	p = tipc_nametbl_remove_publ(net, ua, sk, key);
810 	if (p) {
811 		nt->local_publ_count--;
812 		skb = tipc_named_withdraw(net, p);
813 		list_del_init(&p->binding_sock);
814 		kfree_rcu(p, rcu);
815 	}
816 	rc_dests = nt->rc_dests;
817 	spin_unlock_bh(&tn->nametbl_lock);
818 
819 	if (skb)
820 		tipc_node_broadcast(net, skb, rc_dests);
821 }
822 
823 /**
824  * tipc_nametbl_subscribe - add a subscription object to the name table
825  * @sub: subscription to add
826  */
827 bool tipc_nametbl_subscribe(struct tipc_subscription *sub)
828 {
829 	struct tipc_net *tn = tipc_net(sub->net);
830 	u32 type = sub->s.seq.type;
831 	struct tipc_service *sc;
832 	struct tipc_uaddr ua;
833 	bool res = true;
834 
835 	tipc_uaddr(&ua, TIPC_SERVICE_RANGE, TIPC_NODE_SCOPE, type,
836 		   sub->s.seq.lower, sub->s.seq.upper);
837 	spin_lock_bh(&tn->nametbl_lock);
838 	sc = tipc_service_find(sub->net, &ua);
839 	if (!sc)
840 		sc = tipc_service_create(sub->net, &ua);
841 	if (sc) {
842 		spin_lock_bh(&sc->lock);
843 		tipc_service_subscribe(sc, sub);
844 		spin_unlock_bh(&sc->lock);
845 	} else {
846 		pr_warn("Failed to subscribe for {%u,%u,%u}\n",
847 			type, sub->s.seq.lower, sub->s.seq.upper);
848 		res = false;
849 	}
850 	spin_unlock_bh(&tn->nametbl_lock);
851 	return res;
852 }
853 
854 /**
855  * tipc_nametbl_unsubscribe - remove a subscription object from name table
856  * @sub: subscription to remove
857  */
858 void tipc_nametbl_unsubscribe(struct tipc_subscription *sub)
859 {
860 	struct tipc_net *tn = tipc_net(sub->net);
861 	struct tipc_service *sc;
862 	struct tipc_uaddr ua;
863 
864 	tipc_uaddr(&ua, TIPC_SERVICE_RANGE, TIPC_NODE_SCOPE,
865 		   sub->s.seq.type, sub->s.seq.lower, sub->s.seq.upper);
866 	spin_lock_bh(&tn->nametbl_lock);
867 	sc = tipc_service_find(sub->net, &ua);
868 	if (!sc)
869 		goto exit;
870 
871 	spin_lock_bh(&sc->lock);
872 	list_del_init(&sub->service_list);
873 	tipc_sub_put(sub);
874 
875 	/* Delete service item if no more publications and subscriptions */
876 	if (RB_EMPTY_ROOT(&sc->ranges) && list_empty(&sc->subscriptions)) {
877 		hlist_del_init_rcu(&sc->service_list);
878 		kfree_rcu(sc, rcu);
879 	}
880 	spin_unlock_bh(&sc->lock);
881 exit:
882 	spin_unlock_bh(&tn->nametbl_lock);
883 }
884 
885 int tipc_nametbl_init(struct net *net)
886 {
887 	struct tipc_net *tn = tipc_net(net);
888 	struct name_table *nt;
889 	int i;
890 
891 	nt = kzalloc(sizeof(*nt), GFP_KERNEL);
892 	if (!nt)
893 		return -ENOMEM;
894 
895 	for (i = 0; i < TIPC_NAMETBL_SIZE; i++)
896 		INIT_HLIST_HEAD(&nt->services[i]);
897 
898 	INIT_LIST_HEAD(&nt->node_scope);
899 	INIT_LIST_HEAD(&nt->cluster_scope);
900 	rwlock_init(&nt->cluster_scope_lock);
901 	tn->nametbl = nt;
902 	spin_lock_init(&tn->nametbl_lock);
903 	return 0;
904 }
905 
906 /**
907  * tipc_service_delete - purge all publications for a service and delete it
908  * @net: the associated network namespace
909  * @sc: tipc_service to delete
910  */
911 static void tipc_service_delete(struct net *net, struct tipc_service *sc)
912 {
913 	struct service_range *sr, *tmpr;
914 	struct publication *p, *tmp;
915 
916 	spin_lock_bh(&sc->lock);
917 	rbtree_postorder_for_each_entry_safe(sr, tmpr, &sc->ranges, tree_node) {
918 		list_for_each_entry_safe(p, tmp, &sr->all_publ, all_publ) {
919 			tipc_service_remove_publ(sr, &p->sk, p->key);
920 			kfree_rcu(p, rcu);
921 		}
922 		rb_erase_augmented(&sr->tree_node, &sc->ranges, &sr_callbacks);
923 		kfree(sr);
924 	}
925 	hlist_del_init_rcu(&sc->service_list);
926 	spin_unlock_bh(&sc->lock);
927 	kfree_rcu(sc, rcu);
928 }
929 
930 void tipc_nametbl_stop(struct net *net)
931 {
932 	struct name_table *nt = tipc_name_table(net);
933 	struct tipc_net *tn = tipc_net(net);
934 	struct hlist_head *service_head;
935 	struct tipc_service *service;
936 	u32 i;
937 
938 	/* Verify name table is empty and purge any lingering
939 	 * publications, then release the name table
940 	 */
941 	spin_lock_bh(&tn->nametbl_lock);
942 	for (i = 0; i < TIPC_NAMETBL_SIZE; i++) {
943 		if (hlist_empty(&nt->services[i]))
944 			continue;
945 		service_head = &nt->services[i];
946 		hlist_for_each_entry_rcu(service, service_head, service_list) {
947 			tipc_service_delete(net, service);
948 		}
949 	}
950 	spin_unlock_bh(&tn->nametbl_lock);
951 
952 	synchronize_net();
953 	kfree(nt);
954 }
955 
956 static int __tipc_nl_add_nametable_publ(struct tipc_nl_msg *msg,
957 					struct tipc_service *service,
958 					struct service_range *sr,
959 					u32 *last_key)
960 {
961 	struct publication *p;
962 	struct nlattr *attrs;
963 	struct nlattr *b;
964 	void *hdr;
965 
966 	if (*last_key) {
967 		list_for_each_entry(p, &sr->all_publ, all_publ)
968 			if (p->key == *last_key)
969 				break;
970 		if (list_entry_is_head(p, &sr->all_publ, all_publ))
971 			return -EPIPE;
972 	} else {
973 		p = list_first_entry(&sr->all_publ,
974 				     struct publication,
975 				     all_publ);
976 	}
977 
978 	list_for_each_entry_from(p, &sr->all_publ, all_publ) {
979 		*last_key = p->key;
980 
981 		hdr = genlmsg_put(msg->skb, msg->portid, msg->seq,
982 				  &tipc_genl_family, NLM_F_MULTI,
983 				  TIPC_NL_NAME_TABLE_GET);
984 		if (!hdr)
985 			return -EMSGSIZE;
986 
987 		attrs = nla_nest_start_noflag(msg->skb, TIPC_NLA_NAME_TABLE);
988 		if (!attrs)
989 			goto msg_full;
990 
991 		b = nla_nest_start_noflag(msg->skb, TIPC_NLA_NAME_TABLE_PUBL);
992 		if (!b)
993 			goto attr_msg_full;
994 
995 		if (nla_put_u32(msg->skb, TIPC_NLA_PUBL_TYPE, service->type))
996 			goto publ_msg_full;
997 		if (nla_put_u32(msg->skb, TIPC_NLA_PUBL_LOWER, sr->lower))
998 			goto publ_msg_full;
999 		if (nla_put_u32(msg->skb, TIPC_NLA_PUBL_UPPER, sr->upper))
1000 			goto publ_msg_full;
1001 		if (nla_put_u32(msg->skb, TIPC_NLA_PUBL_SCOPE, p->scope))
1002 			goto publ_msg_full;
1003 		if (nla_put_u32(msg->skb, TIPC_NLA_PUBL_NODE, p->sk.node))
1004 			goto publ_msg_full;
1005 		if (nla_put_u32(msg->skb, TIPC_NLA_PUBL_REF, p->sk.ref))
1006 			goto publ_msg_full;
1007 		if (nla_put_u32(msg->skb, TIPC_NLA_PUBL_KEY, p->key))
1008 			goto publ_msg_full;
1009 
1010 		nla_nest_end(msg->skb, b);
1011 		nla_nest_end(msg->skb, attrs);
1012 		genlmsg_end(msg->skb, hdr);
1013 	}
1014 	*last_key = 0;
1015 
1016 	return 0;
1017 
1018 publ_msg_full:
1019 	nla_nest_cancel(msg->skb, b);
1020 attr_msg_full:
1021 	nla_nest_cancel(msg->skb, attrs);
1022 msg_full:
1023 	genlmsg_cancel(msg->skb, hdr);
1024 
1025 	return -EMSGSIZE;
1026 }
1027 
1028 static int __tipc_nl_service_range_list(struct tipc_nl_msg *msg,
1029 					struct tipc_service *sc,
1030 					u32 *last_lower, u32 *last_key)
1031 {
1032 	struct service_range *sr;
1033 	struct rb_node *n;
1034 	int err;
1035 
1036 	for (n = rb_first(&sc->ranges); n; n = rb_next(n)) {
1037 		sr = container_of(n, struct service_range, tree_node);
1038 		if (sr->lower < *last_lower)
1039 			continue;
1040 		err = __tipc_nl_add_nametable_publ(msg, sc, sr, last_key);
1041 		if (err) {
1042 			*last_lower = sr->lower;
1043 			return err;
1044 		}
1045 	}
1046 	*last_lower = 0;
1047 	return 0;
1048 }
1049 
1050 static int tipc_nl_service_list(struct net *net, struct tipc_nl_msg *msg,
1051 				u32 *last_type, u32 *last_lower, u32 *last_key)
1052 {
1053 	struct tipc_net *tn = tipc_net(net);
1054 	struct tipc_service *service = NULL;
1055 	struct hlist_head *head;
1056 	struct tipc_uaddr ua;
1057 	int err;
1058 	int i;
1059 
1060 	if (*last_type)
1061 		i = hash(*last_type);
1062 	else
1063 		i = 0;
1064 
1065 	for (; i < TIPC_NAMETBL_SIZE; i++) {
1066 		head = &tn->nametbl->services[i];
1067 
1068 		if (*last_type ||
1069 		    (!i && *last_key && (*last_lower == *last_key))) {
1070 			tipc_uaddr(&ua, TIPC_SERVICE_RANGE, TIPC_NODE_SCOPE,
1071 				   *last_type, *last_lower, *last_lower);
1072 			service = tipc_service_find(net, &ua);
1073 			if (!service)
1074 				return -EPIPE;
1075 		} else {
1076 			hlist_for_each_entry_rcu(service, head, service_list)
1077 				break;
1078 			if (!service)
1079 				continue;
1080 		}
1081 
1082 		hlist_for_each_entry_from_rcu(service, service_list) {
1083 			spin_lock_bh(&service->lock);
1084 			err = __tipc_nl_service_range_list(msg, service,
1085 							   last_lower,
1086 							   last_key);
1087 
1088 			if (err) {
1089 				*last_type = service->type;
1090 				spin_unlock_bh(&service->lock);
1091 				return err;
1092 			}
1093 			spin_unlock_bh(&service->lock);
1094 		}
1095 		*last_type = 0;
1096 	}
1097 	return 0;
1098 }
1099 
1100 int tipc_nl_name_table_dump(struct sk_buff *skb, struct netlink_callback *cb)
1101 {
1102 	struct net *net = sock_net(skb->sk);
1103 	u32 last_type = cb->args[0];
1104 	u32 last_lower = cb->args[1];
1105 	u32 last_key = cb->args[2];
1106 	int done = cb->args[3];
1107 	struct tipc_nl_msg msg;
1108 	int err;
1109 
1110 	if (done)
1111 		return 0;
1112 
1113 	msg.skb = skb;
1114 	msg.portid = NETLINK_CB(cb->skb).portid;
1115 	msg.seq = cb->nlh->nlmsg_seq;
1116 
1117 	rcu_read_lock();
1118 	err = tipc_nl_service_list(net, &msg, &last_type,
1119 				   &last_lower, &last_key);
1120 	if (!err) {
1121 		done = 1;
1122 	} else if (err != -EMSGSIZE) {
1123 		/* We never set seq or call nl_dump_check_consistent() this
1124 		 * means that setting prev_seq here will cause the consistence
1125 		 * check to fail in the netlink callback handler. Resulting in
1126 		 * the NLMSG_DONE message having the NLM_F_DUMP_INTR flag set if
1127 		 * we got an error.
1128 		 */
1129 		cb->prev_seq = 1;
1130 	}
1131 	rcu_read_unlock();
1132 
1133 	cb->args[0] = last_type;
1134 	cb->args[1] = last_lower;
1135 	cb->args[2] = last_key;
1136 	cb->args[3] = done;
1137 
1138 	return skb->len;
1139 }
1140 
1141 struct tipc_dest *tipc_dest_find(struct list_head *l, u32 node, u32 port)
1142 {
1143 	struct tipc_dest *dst;
1144 
1145 	list_for_each_entry(dst, l, list) {
1146 		if (dst->node == node && dst->port == port)
1147 			return dst;
1148 	}
1149 	return NULL;
1150 }
1151 
1152 bool tipc_dest_push(struct list_head *l, u32 node, u32 port)
1153 {
1154 	struct tipc_dest *dst;
1155 
1156 	if (tipc_dest_find(l, node, port))
1157 		return false;
1158 
1159 	dst = kmalloc(sizeof(*dst), GFP_ATOMIC);
1160 	if (unlikely(!dst))
1161 		return false;
1162 	dst->node = node;
1163 	dst->port = port;
1164 	list_add(&dst->list, l);
1165 	return true;
1166 }
1167 
1168 bool tipc_dest_pop(struct list_head *l, u32 *node, u32 *port)
1169 {
1170 	struct tipc_dest *dst;
1171 
1172 	if (list_empty(l))
1173 		return false;
1174 	dst = list_first_entry(l, typeof(*dst), list);
1175 	if (port)
1176 		*port = dst->port;
1177 	if (node)
1178 		*node = dst->node;
1179 	list_del(&dst->list);
1180 	kfree(dst);
1181 	return true;
1182 }
1183 
1184 bool tipc_dest_del(struct list_head *l, u32 node, u32 port)
1185 {
1186 	struct tipc_dest *dst;
1187 
1188 	dst = tipc_dest_find(l, node, port);
1189 	if (!dst)
1190 		return false;
1191 	list_del(&dst->list);
1192 	kfree(dst);
1193 	return true;
1194 }
1195 
1196 void tipc_dest_list_purge(struct list_head *l)
1197 {
1198 	struct tipc_dest *dst, *tmp;
1199 
1200 	list_for_each_entry_safe(dst, tmp, l, list) {
1201 		list_del(&dst->list);
1202 		kfree(dst);
1203 	}
1204 }
1205