blkcg: misc preparations for unified hierarchy interface
[linux-2.6/btrfs-unstable.git] / net / tipc / node.c
blob0b1d61a5f85334b3553780e8c0dd64c3f0549aa3
1 /*
2 * net/tipc/node.c: TIPC node management routines
4 * Copyright (c) 2000-2006, 2012-2015, Ericsson AB
5 * Copyright (c) 2005-2006, 2010-2014, Wind River Systems
6 * All rights reserved.
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions are met:
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.
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.
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.
37 #include "core.h"
38 #include "link.h"
39 #include "node.h"
40 #include "name_distr.h"
41 #include "socket.h"
42 #include "bcast.h"
44 static void node_lost_contact(struct tipc_node *n_ptr);
45 static void node_established_contact(struct tipc_node *n_ptr);
46 static void tipc_node_delete(struct tipc_node *node);
48 struct tipc_sock_conn {
49 u32 port;
50 u32 peer_port;
51 u32 peer_node;
52 struct list_head list;
55 static const struct nla_policy tipc_nl_node_policy[TIPC_NLA_NODE_MAX + 1] = {
56 [TIPC_NLA_NODE_UNSPEC] = { .type = NLA_UNSPEC },
57 [TIPC_NLA_NODE_ADDR] = { .type = NLA_U32 },
58 [TIPC_NLA_NODE_UP] = { .type = NLA_FLAG }
62 * A trivial power-of-two bitmask technique is used for speed, since this
63 * operation is done for every incoming TIPC packet. The number of hash table
64 * entries has been chosen so that no hash chain exceeds 8 nodes and will
65 * usually be much smaller (typically only a single node).
67 static unsigned int tipc_hashfn(u32 addr)
69 return addr & (NODE_HTABLE_SIZE - 1);
72 static void tipc_node_kref_release(struct kref *kref)
74 struct tipc_node *node = container_of(kref, struct tipc_node, kref);
76 tipc_node_delete(node);
79 void tipc_node_put(struct tipc_node *node)
81 kref_put(&node->kref, tipc_node_kref_release);
84 static void tipc_node_get(struct tipc_node *node)
86 kref_get(&node->kref);
90 * tipc_node_find - locate specified node object, if it exists
92 struct tipc_node *tipc_node_find(struct net *net, u32 addr)
94 struct tipc_net *tn = net_generic(net, tipc_net_id);
95 struct tipc_node *node;
97 if (unlikely(!in_own_cluster_exact(net, addr)))
98 return NULL;
100 rcu_read_lock();
101 hlist_for_each_entry_rcu(node, &tn->node_htable[tipc_hashfn(addr)],
102 hash) {
103 if (node->addr == addr) {
104 tipc_node_get(node);
105 rcu_read_unlock();
106 return node;
109 rcu_read_unlock();
110 return NULL;
113 struct tipc_node *tipc_node_create(struct net *net, u32 addr)
115 struct tipc_net *tn = net_generic(net, tipc_net_id);
116 struct tipc_node *n_ptr, *temp_node;
118 spin_lock_bh(&tn->node_list_lock);
119 n_ptr = tipc_node_find(net, addr);
120 if (n_ptr)
121 goto exit;
122 n_ptr = kzalloc(sizeof(*n_ptr), GFP_ATOMIC);
123 if (!n_ptr) {
124 pr_warn("Node creation failed, no memory\n");
125 goto exit;
127 n_ptr->addr = addr;
128 n_ptr->net = net;
129 kref_init(&n_ptr->kref);
130 spin_lock_init(&n_ptr->lock);
131 INIT_HLIST_NODE(&n_ptr->hash);
132 INIT_LIST_HEAD(&n_ptr->list);
133 INIT_LIST_HEAD(&n_ptr->publ_list);
134 INIT_LIST_HEAD(&n_ptr->conn_sks);
135 __skb_queue_head_init(&n_ptr->bclink.deferdq);
136 hlist_add_head_rcu(&n_ptr->hash, &tn->node_htable[tipc_hashfn(addr)]);
137 list_for_each_entry_rcu(temp_node, &tn->node_list, list) {
138 if (n_ptr->addr < temp_node->addr)
139 break;
141 list_add_tail_rcu(&n_ptr->list, &temp_node->list);
142 n_ptr->action_flags = TIPC_WAIT_PEER_LINKS_DOWN;
143 n_ptr->signature = INVALID_NODE_SIG;
144 tipc_node_get(n_ptr);
145 exit:
146 spin_unlock_bh(&tn->node_list_lock);
147 return n_ptr;
150 static void tipc_node_delete(struct tipc_node *node)
152 list_del_rcu(&node->list);
153 hlist_del_rcu(&node->hash);
154 kfree_rcu(node, rcu);
157 void tipc_node_stop(struct net *net)
159 struct tipc_net *tn = net_generic(net, tipc_net_id);
160 struct tipc_node *node, *t_node;
162 spin_lock_bh(&tn->node_list_lock);
163 list_for_each_entry_safe(node, t_node, &tn->node_list, list)
164 tipc_node_put(node);
165 spin_unlock_bh(&tn->node_list_lock);
168 int tipc_node_add_conn(struct net *net, u32 dnode, u32 port, u32 peer_port)
170 struct tipc_node *node;
171 struct tipc_sock_conn *conn;
172 int err = 0;
174 if (in_own_node(net, dnode))
175 return 0;
177 node = tipc_node_find(net, dnode);
178 if (!node) {
179 pr_warn("Connecting sock to node 0x%x failed\n", dnode);
180 return -EHOSTUNREACH;
182 conn = kmalloc(sizeof(*conn), GFP_ATOMIC);
183 if (!conn) {
184 err = -EHOSTUNREACH;
185 goto exit;
187 conn->peer_node = dnode;
188 conn->port = port;
189 conn->peer_port = peer_port;
191 tipc_node_lock(node);
192 list_add_tail(&conn->list, &node->conn_sks);
193 tipc_node_unlock(node);
194 exit:
195 tipc_node_put(node);
196 return err;
199 void tipc_node_remove_conn(struct net *net, u32 dnode, u32 port)
201 struct tipc_node *node;
202 struct tipc_sock_conn *conn, *safe;
204 if (in_own_node(net, dnode))
205 return;
207 node = tipc_node_find(net, dnode);
208 if (!node)
209 return;
211 tipc_node_lock(node);
212 list_for_each_entry_safe(conn, safe, &node->conn_sks, list) {
213 if (port != conn->port)
214 continue;
215 list_del(&conn->list);
216 kfree(conn);
218 tipc_node_unlock(node);
219 tipc_node_put(node);
223 * tipc_node_link_up - handle addition of link
225 * Link becomes active (alone or shared) or standby, depending on its priority.
227 void tipc_node_link_up(struct tipc_node *n_ptr, struct tipc_link *l_ptr)
229 struct tipc_link **active = &n_ptr->active_links[0];
231 n_ptr->working_links++;
232 n_ptr->action_flags |= TIPC_NOTIFY_LINK_UP;
233 n_ptr->link_id = l_ptr->peer_bearer_id << 16 | l_ptr->bearer_id;
235 pr_debug("Established link <%s> on network plane %c\n",
236 l_ptr->name, l_ptr->net_plane);
238 if (!active[0]) {
239 active[0] = active[1] = l_ptr;
240 node_established_contact(n_ptr);
241 goto exit;
243 if (l_ptr->priority < active[0]->priority) {
244 pr_debug("New link <%s> becomes standby\n", l_ptr->name);
245 goto exit;
247 tipc_link_dup_queue_xmit(active[0], l_ptr);
248 if (l_ptr->priority == active[0]->priority) {
249 active[0] = l_ptr;
250 goto exit;
252 pr_debug("Old link <%s> becomes standby\n", active[0]->name);
253 if (active[1] != active[0])
254 pr_debug("Old link <%s> becomes standby\n", active[1]->name);
255 active[0] = active[1] = l_ptr;
256 exit:
257 /* Leave room for changeover header when returning 'mtu' to users: */
258 n_ptr->act_mtus[0] = active[0]->mtu - INT_H_SIZE;
259 n_ptr->act_mtus[1] = active[1]->mtu - INT_H_SIZE;
263 * node_select_active_links - select active link
265 static void node_select_active_links(struct tipc_node *n_ptr)
267 struct tipc_link **active = &n_ptr->active_links[0];
268 u32 i;
269 u32 highest_prio = 0;
271 active[0] = active[1] = NULL;
273 for (i = 0; i < MAX_BEARERS; i++) {
274 struct tipc_link *l_ptr = n_ptr->links[i];
276 if (!l_ptr || !tipc_link_is_up(l_ptr) ||
277 (l_ptr->priority < highest_prio))
278 continue;
280 if (l_ptr->priority > highest_prio) {
281 highest_prio = l_ptr->priority;
282 active[0] = active[1] = l_ptr;
283 } else {
284 active[1] = l_ptr;
290 * tipc_node_link_down - handle loss of link
292 void tipc_node_link_down(struct tipc_node *n_ptr, struct tipc_link *l_ptr)
294 struct tipc_net *tn = net_generic(n_ptr->net, tipc_net_id);
295 struct tipc_link **active;
297 n_ptr->working_links--;
298 n_ptr->action_flags |= TIPC_NOTIFY_LINK_DOWN;
299 n_ptr->link_id = l_ptr->peer_bearer_id << 16 | l_ptr->bearer_id;
301 if (!tipc_link_is_active(l_ptr)) {
302 pr_debug("Lost standby link <%s> on network plane %c\n",
303 l_ptr->name, l_ptr->net_plane);
304 return;
306 pr_debug("Lost link <%s> on network plane %c\n",
307 l_ptr->name, l_ptr->net_plane);
309 active = &n_ptr->active_links[0];
310 if (active[0] == l_ptr)
311 active[0] = active[1];
312 if (active[1] == l_ptr)
313 active[1] = active[0];
314 if (active[0] == l_ptr)
315 node_select_active_links(n_ptr);
316 if (tipc_node_is_up(n_ptr))
317 tipc_link_failover_send_queue(l_ptr);
318 else
319 node_lost_contact(n_ptr);
321 /* Leave room for changeover header when returning 'mtu' to users: */
322 if (active[0]) {
323 n_ptr->act_mtus[0] = active[0]->mtu - INT_H_SIZE;
324 n_ptr->act_mtus[1] = active[1]->mtu - INT_H_SIZE;
325 return;
327 /* Loopback link went down? No fragmentation needed from now on. */
328 if (n_ptr->addr == tn->own_addr) {
329 n_ptr->act_mtus[0] = MAX_MSG_SIZE;
330 n_ptr->act_mtus[1] = MAX_MSG_SIZE;
334 int tipc_node_active_links(struct tipc_node *n_ptr)
336 return n_ptr->active_links[0] != NULL;
339 int tipc_node_is_up(struct tipc_node *n_ptr)
341 return tipc_node_active_links(n_ptr);
344 void tipc_node_attach_link(struct tipc_node *n_ptr, struct tipc_link *l_ptr)
346 n_ptr->links[l_ptr->bearer_id] = l_ptr;
347 n_ptr->link_cnt++;
350 void tipc_node_detach_link(struct tipc_node *n_ptr, struct tipc_link *l_ptr)
352 int i;
354 for (i = 0; i < MAX_BEARERS; i++) {
355 if (l_ptr != n_ptr->links[i])
356 continue;
357 n_ptr->links[i] = NULL;
358 n_ptr->link_cnt--;
362 static void node_established_contact(struct tipc_node *n_ptr)
364 n_ptr->action_flags |= TIPC_NOTIFY_NODE_UP;
365 n_ptr->bclink.oos_state = 0;
366 n_ptr->bclink.acked = tipc_bclink_get_last_sent(n_ptr->net);
367 tipc_bclink_add_node(n_ptr->net, n_ptr->addr);
370 static void node_lost_contact(struct tipc_node *n_ptr)
372 char addr_string[16];
373 struct tipc_sock_conn *conn, *safe;
374 struct list_head *conns = &n_ptr->conn_sks;
375 struct sk_buff *skb;
376 struct tipc_net *tn = net_generic(n_ptr->net, tipc_net_id);
377 uint i;
379 pr_debug("Lost contact with %s\n",
380 tipc_addr_string_fill(addr_string, n_ptr->addr));
382 /* Flush broadcast link info associated with lost node */
383 if (n_ptr->bclink.recv_permitted) {
384 __skb_queue_purge(&n_ptr->bclink.deferdq);
386 if (n_ptr->bclink.reasm_buf) {
387 kfree_skb(n_ptr->bclink.reasm_buf);
388 n_ptr->bclink.reasm_buf = NULL;
391 tipc_bclink_remove_node(n_ptr->net, n_ptr->addr);
392 tipc_bclink_acknowledge(n_ptr, INVALID_LINK_SEQ);
394 n_ptr->bclink.recv_permitted = false;
397 /* Abort any ongoing link failover */
398 for (i = 0; i < MAX_BEARERS; i++) {
399 struct tipc_link *l_ptr = n_ptr->links[i];
400 if (!l_ptr)
401 continue;
402 l_ptr->flags &= ~LINK_FAILINGOVER;
403 l_ptr->failover_checkpt = 0;
404 l_ptr->failover_pkts = 0;
405 kfree_skb(l_ptr->failover_skb);
406 l_ptr->failover_skb = NULL;
407 tipc_link_reset_fragments(l_ptr);
410 n_ptr->action_flags &= ~TIPC_WAIT_OWN_LINKS_DOWN;
412 /* Prevent re-contact with node until cleanup is done */
413 n_ptr->action_flags |= TIPC_WAIT_PEER_LINKS_DOWN;
415 /* Notify publications from this node */
416 n_ptr->action_flags |= TIPC_NOTIFY_NODE_DOWN;
418 /* Notify sockets connected to node */
419 list_for_each_entry_safe(conn, safe, conns, list) {
420 skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE, TIPC_CONN_MSG,
421 SHORT_H_SIZE, 0, tn->own_addr,
422 conn->peer_node, conn->port,
423 conn->peer_port, TIPC_ERR_NO_NODE);
424 if (likely(skb)) {
425 skb_queue_tail(n_ptr->inputq, skb);
426 n_ptr->action_flags |= TIPC_MSG_EVT;
428 list_del(&conn->list);
429 kfree(conn);
434 * tipc_node_get_linkname - get the name of a link
436 * @bearer_id: id of the bearer
437 * @node: peer node address
438 * @linkname: link name output buffer
440 * Returns 0 on success
442 int tipc_node_get_linkname(struct net *net, u32 bearer_id, u32 addr,
443 char *linkname, size_t len)
445 struct tipc_link *link;
446 int err = -EINVAL;
447 struct tipc_node *node = tipc_node_find(net, addr);
449 if (!node)
450 return err;
452 if (bearer_id >= MAX_BEARERS)
453 goto exit;
455 tipc_node_lock(node);
456 link = node->links[bearer_id];
457 if (link) {
458 strncpy(linkname, link->name, len);
459 err = 0;
461 exit:
462 tipc_node_unlock(node);
463 tipc_node_put(node);
464 return err;
467 void tipc_node_unlock(struct tipc_node *node)
469 struct net *net = node->net;
470 u32 addr = 0;
471 u32 flags = node->action_flags;
472 u32 link_id = 0;
473 struct list_head *publ_list;
474 struct sk_buff_head *inputq = node->inputq;
475 struct sk_buff_head *namedq;
477 if (likely(!flags || (flags == TIPC_MSG_EVT))) {
478 node->action_flags = 0;
479 spin_unlock_bh(&node->lock);
480 if (flags == TIPC_MSG_EVT)
481 tipc_sk_rcv(net, inputq);
482 return;
485 addr = node->addr;
486 link_id = node->link_id;
487 namedq = node->namedq;
488 publ_list = &node->publ_list;
490 node->action_flags &= ~(TIPC_MSG_EVT |
491 TIPC_NOTIFY_NODE_DOWN | TIPC_NOTIFY_NODE_UP |
492 TIPC_NOTIFY_LINK_DOWN | TIPC_NOTIFY_LINK_UP |
493 TIPC_WAKEUP_BCAST_USERS | TIPC_BCAST_MSG_EVT |
494 TIPC_NAMED_MSG_EVT | TIPC_BCAST_RESET);
496 spin_unlock_bh(&node->lock);
498 if (flags & TIPC_NOTIFY_NODE_DOWN)
499 tipc_publ_notify(net, publ_list, addr);
501 if (flags & TIPC_WAKEUP_BCAST_USERS)
502 tipc_bclink_wakeup_users(net);
504 if (flags & TIPC_NOTIFY_NODE_UP)
505 tipc_named_node_up(net, addr);
507 if (flags & TIPC_NOTIFY_LINK_UP)
508 tipc_nametbl_publish(net, TIPC_LINK_STATE, addr, addr,
509 TIPC_NODE_SCOPE, link_id, addr);
511 if (flags & TIPC_NOTIFY_LINK_DOWN)
512 tipc_nametbl_withdraw(net, TIPC_LINK_STATE, addr,
513 link_id, addr);
515 if (flags & TIPC_MSG_EVT)
516 tipc_sk_rcv(net, inputq);
518 if (flags & TIPC_NAMED_MSG_EVT)
519 tipc_named_rcv(net, namedq);
521 if (flags & TIPC_BCAST_MSG_EVT)
522 tipc_bclink_input(net);
524 if (flags & TIPC_BCAST_RESET)
525 tipc_link_reset_all(node);
528 /* Caller should hold node lock for the passed node */
529 static int __tipc_nl_add_node(struct tipc_nl_msg *msg, struct tipc_node *node)
531 void *hdr;
532 struct nlattr *attrs;
534 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
535 NLM_F_MULTI, TIPC_NL_NODE_GET);
536 if (!hdr)
537 return -EMSGSIZE;
539 attrs = nla_nest_start(msg->skb, TIPC_NLA_NODE);
540 if (!attrs)
541 goto msg_full;
543 if (nla_put_u32(msg->skb, TIPC_NLA_NODE_ADDR, node->addr))
544 goto attr_msg_full;
545 if (tipc_node_is_up(node))
546 if (nla_put_flag(msg->skb, TIPC_NLA_NODE_UP))
547 goto attr_msg_full;
549 nla_nest_end(msg->skb, attrs);
550 genlmsg_end(msg->skb, hdr);
552 return 0;
554 attr_msg_full:
555 nla_nest_cancel(msg->skb, attrs);
556 msg_full:
557 genlmsg_cancel(msg->skb, hdr);
559 return -EMSGSIZE;
562 int tipc_nl_node_dump(struct sk_buff *skb, struct netlink_callback *cb)
564 int err;
565 struct net *net = sock_net(skb->sk);
566 struct tipc_net *tn = net_generic(net, tipc_net_id);
567 int done = cb->args[0];
568 int last_addr = cb->args[1];
569 struct tipc_node *node;
570 struct tipc_nl_msg msg;
572 if (done)
573 return 0;
575 msg.skb = skb;
576 msg.portid = NETLINK_CB(cb->skb).portid;
577 msg.seq = cb->nlh->nlmsg_seq;
579 rcu_read_lock();
580 if (last_addr) {
581 node = tipc_node_find(net, last_addr);
582 if (!node) {
583 rcu_read_unlock();
584 /* We never set seq or call nl_dump_check_consistent()
585 * this means that setting prev_seq here will cause the
586 * consistence check to fail in the netlink callback
587 * handler. Resulting in the NLMSG_DONE message having
588 * the NLM_F_DUMP_INTR flag set if the node state
589 * changed while we released the lock.
591 cb->prev_seq = 1;
592 return -EPIPE;
594 tipc_node_put(node);
597 list_for_each_entry_rcu(node, &tn->node_list, list) {
598 if (last_addr) {
599 if (node->addr == last_addr)
600 last_addr = 0;
601 else
602 continue;
605 tipc_node_lock(node);
606 err = __tipc_nl_add_node(&msg, node);
607 if (err) {
608 last_addr = node->addr;
609 tipc_node_unlock(node);
610 goto out;
613 tipc_node_unlock(node);
615 done = 1;
616 out:
617 cb->args[0] = done;
618 cb->args[1] = last_addr;
619 rcu_read_unlock();
621 return skb->len;