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
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.
40 #include "name_distr.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);
47 static void tipc_node_timeout(unsigned long data);
49 struct tipc_sock_conn {
53 struct list_head list;
56 static const struct nla_policy tipc_nl_node_policy[TIPC_NLA_NODE_MAX + 1] = {
57 [TIPC_NLA_NODE_UNSPEC] = { .type = NLA_UNSPEC },
58 [TIPC_NLA_NODE_ADDR] = { .type = NLA_U32 },
59 [TIPC_NLA_NODE_UP] = { .type = NLA_FLAG }
63 * A trivial power-of-two bitmask technique is used for speed, since this
64 * operation is done for every incoming TIPC packet. The number of hash table
65 * entries has been chosen so that no hash chain exceeds 8 nodes and will
66 * usually be much smaller (typically only a single node).
68 static unsigned int tipc_hashfn(u32 addr)
70 return addr & (NODE_HTABLE_SIZE - 1);
73 static void tipc_node_kref_release(struct kref *kref)
75 struct tipc_node *node = container_of(kref, struct tipc_node, kref);
77 tipc_node_delete(node);
80 void tipc_node_put(struct tipc_node *node)
82 kref_put(&node->kref, tipc_node_kref_release);
85 static void tipc_node_get(struct tipc_node *node)
87 kref_get(&node->kref);
91 * tipc_node_find - locate specified node object, if it exists
93 struct tipc_node *tipc_node_find(struct net *net, u32 addr)
95 struct tipc_net *tn = net_generic(net, tipc_net_id);
96 struct tipc_node *node;
98 if (unlikely(!in_own_cluster_exact(net, addr)))
102 hlist_for_each_entry_rcu(node, &tn->node_htable[tipc_hashfn(addr)],
104 if (node->addr == addr) {
114 struct tipc_node *tipc_node_create(struct net *net, u32 addr)
116 struct tipc_net *tn = net_generic(net, tipc_net_id);
117 struct tipc_node *n_ptr, *temp_node;
119 spin_lock_bh(&tn->node_list_lock);
120 n_ptr = tipc_node_find(net, addr);
123 n_ptr = kzalloc(sizeof(*n_ptr), GFP_ATOMIC);
125 pr_warn("Node creation failed, no memory\n");
130 kref_init(&n_ptr->kref);
131 spin_lock_init(&n_ptr->lock);
132 INIT_HLIST_NODE(&n_ptr->hash);
133 INIT_LIST_HEAD(&n_ptr->list);
134 INIT_LIST_HEAD(&n_ptr->publ_list);
135 INIT_LIST_HEAD(&n_ptr->conn_sks);
136 skb_queue_head_init(&n_ptr->bclink.namedq);
137 __skb_queue_head_init(&n_ptr->bclink.deferdq);
138 hlist_add_head_rcu(&n_ptr->hash, &tn->node_htable[tipc_hashfn(addr)]);
139 list_for_each_entry_rcu(temp_node, &tn->node_list, list) {
140 if (n_ptr->addr < temp_node->addr)
143 list_add_tail_rcu(&n_ptr->list, &temp_node->list);
144 n_ptr->action_flags = TIPC_WAIT_PEER_LINKS_DOWN;
145 n_ptr->signature = INVALID_NODE_SIG;
146 n_ptr->active_links[0] = INVALID_BEARER_ID;
147 n_ptr->active_links[1] = INVALID_BEARER_ID;
148 tipc_node_get(n_ptr);
149 setup_timer(&n_ptr->timer, tipc_node_timeout, (unsigned long)n_ptr);
150 n_ptr->keepalive_intv = U32_MAX;
152 spin_unlock_bh(&tn->node_list_lock);
156 static void tipc_node_calculate_timer(struct tipc_node *n, struct tipc_link *l)
158 unsigned long tol = l->tolerance;
159 unsigned long intv = ((tol / 4) > 500) ? 500 : tol / 4;
160 unsigned long keepalive_intv = msecs_to_jiffies(intv);
162 /* Link with lowest tolerance determines timer interval */
163 if (keepalive_intv < n->keepalive_intv)
164 n->keepalive_intv = keepalive_intv;
166 /* Ensure link's abort limit corresponds to current interval */
167 l->abort_limit = l->tolerance / jiffies_to_msecs(n->keepalive_intv);
170 static void tipc_node_delete(struct tipc_node *node)
172 list_del_rcu(&node->list);
173 hlist_del_rcu(&node->hash);
174 kfree_rcu(node, rcu);
177 void tipc_node_stop(struct net *net)
179 struct tipc_net *tn = net_generic(net, tipc_net_id);
180 struct tipc_node *node, *t_node;
182 spin_lock_bh(&tn->node_list_lock);
183 list_for_each_entry_safe(node, t_node, &tn->node_list, list) {
184 if (del_timer(&node->timer))
188 spin_unlock_bh(&tn->node_list_lock);
191 int tipc_node_add_conn(struct net *net, u32 dnode, u32 port, u32 peer_port)
193 struct tipc_node *node;
194 struct tipc_sock_conn *conn;
197 if (in_own_node(net, dnode))
200 node = tipc_node_find(net, dnode);
202 pr_warn("Connecting sock to node 0x%x failed\n", dnode);
203 return -EHOSTUNREACH;
205 conn = kmalloc(sizeof(*conn), GFP_ATOMIC);
210 conn->peer_node = dnode;
212 conn->peer_port = peer_port;
214 tipc_node_lock(node);
215 list_add_tail(&conn->list, &node->conn_sks);
216 tipc_node_unlock(node);
222 void tipc_node_remove_conn(struct net *net, u32 dnode, u32 port)
224 struct tipc_node *node;
225 struct tipc_sock_conn *conn, *safe;
227 if (in_own_node(net, dnode))
230 node = tipc_node_find(net, dnode);
234 tipc_node_lock(node);
235 list_for_each_entry_safe(conn, safe, &node->conn_sks, list) {
236 if (port != conn->port)
238 list_del(&conn->list);
241 tipc_node_unlock(node);
245 /* tipc_node_timeout - handle expiration of node timer
247 static void tipc_node_timeout(unsigned long data)
249 struct tipc_node *n = (struct tipc_node *)data;
250 struct sk_buff_head xmitq;
252 struct tipc_media_addr *maddr;
256 __skb_queue_head_init(&xmitq);
258 for (bearer_id = 0; bearer_id < MAX_BEARERS; bearer_id++) {
260 l = n->links[bearer_id].link;
262 /* Link tolerance may change asynchronously: */
263 tipc_node_calculate_timer(n, l);
264 rc = tipc_link_timeout(l, &xmitq);
265 if (rc & TIPC_LINK_DOWN_EVT)
269 maddr = &n->links[bearer_id].maddr;
270 tipc_bearer_xmit(n->net, bearer_id, &xmitq, maddr);
272 if (!mod_timer(&n->timer, jiffies + n->keepalive_intv))
278 * tipc_node_link_up - handle addition of link
280 * Link becomes active (alone or shared) or standby, depending on its priority.
282 void tipc_node_link_up(struct tipc_node *n, int bearer_id)
284 int *slot0 = &n->active_links[0];
285 int *slot1 = &n->active_links[1];
286 struct tipc_link_entry *links = n->links;
287 struct tipc_link *l = n->links[bearer_id].link;
289 /* Leave room for tunnel header when returning 'mtu' to users: */
290 links[bearer_id].mtu = l->mtu - INT_H_SIZE;
293 n->action_flags |= TIPC_NOTIFY_LINK_UP;
294 n->link_id = l->peer_bearer_id << 16 | l->bearer_id;
296 pr_debug("Established link <%s> on network plane %c\n",
297 l->name, l->net_plane);
299 /* No active links ? => take both active slots */
303 node_established_contact(n);
307 /* Lower prio than current active ? => no slot */
308 if (l->priority < links[*slot0].link->priority) {
309 pr_debug("New link <%s> becomes standby\n", l->name);
312 tipc_link_dup_queue_xmit(links[*slot0].link, l);
314 /* Same prio as current active ? => take one slot */
315 if (l->priority == links[*slot0].link->priority) {
320 /* Higher prio than current active => take both active slots */
321 pr_debug("Old link <%s> now standby\n", links[*slot0].link->name);
327 * tipc_node_link_down - handle loss of link
329 void tipc_node_link_down(struct tipc_node *n, int bearer_id)
331 int *slot0 = &n->active_links[0];
332 int *slot1 = &n->active_links[1];
334 struct tipc_link *l, *_l;
336 l = n->links[bearer_id].link;
338 n->action_flags |= TIPC_NOTIFY_LINK_DOWN;
339 n->link_id = l->peer_bearer_id << 16 | l->bearer_id;
341 pr_debug("Lost link <%s> on network plane %c\n",
342 l->name, l->net_plane);
344 /* Select new active link if any available */
345 *slot0 = INVALID_BEARER_ID;
346 *slot1 = INVALID_BEARER_ID;
347 for (i = 0; i < MAX_BEARERS; i++) {
348 _l = n->links[i].link;
349 if (!_l || !tipc_link_is_up(_l))
351 if (_l->priority < highest)
353 if (_l->priority > highest) {
354 highest = _l->priority;
361 if (tipc_node_is_up(n))
362 tipc_link_failover_send_queue(l);
364 node_lost_contact(n);
367 bool tipc_node_is_up(struct tipc_node *n)
369 return n->active_links[0] != INVALID_BEARER_ID;
372 void tipc_node_check_dest(struct tipc_node *n, struct tipc_bearer *b,
373 bool *link_up, bool *addr_match,
374 struct tipc_media_addr *maddr)
376 struct tipc_link *l = n->links[b->identity].link;
377 struct tipc_media_addr *curr = &n->links[b->identity].maddr;
379 *link_up = l && tipc_link_is_up(l);
380 *addr_match = l && !memcmp(curr, maddr, sizeof(*maddr));
383 bool tipc_node_update_dest(struct tipc_node *n, struct tipc_bearer *b,
384 struct tipc_media_addr *maddr)
386 struct tipc_link *l = n->links[b->identity].link;
387 struct tipc_media_addr *curr = &n->links[b->identity].maddr;
388 struct sk_buff_head *inputq = &n->links[b->identity].inputq;
391 l = tipc_link_create(n, b, maddr, inputq, &n->bclink.namedq);
394 tipc_node_calculate_timer(n, l);
395 if (n->link_cnt == 1) {
396 if (!mod_timer(&n->timer, jiffies + n->keepalive_intv))
400 memcpy(&l->media_addr, maddr, sizeof(*maddr));
401 memcpy(curr, maddr, sizeof(*maddr));
406 void tipc_node_attach_link(struct tipc_node *n_ptr, struct tipc_link *l_ptr)
408 n_ptr->links[l_ptr->bearer_id].link = l_ptr;
412 void tipc_node_detach_link(struct tipc_node *n_ptr, struct tipc_link *l_ptr)
416 for (i = 0; i < MAX_BEARERS; i++) {
417 if (l_ptr != n_ptr->links[i].link)
419 n_ptr->links[i].link = NULL;
424 static void node_established_contact(struct tipc_node *n_ptr)
426 n_ptr->action_flags |= TIPC_NOTIFY_NODE_UP;
427 n_ptr->bclink.oos_state = 0;
428 n_ptr->bclink.acked = tipc_bclink_get_last_sent(n_ptr->net);
429 tipc_bclink_add_node(n_ptr->net, n_ptr->addr);
432 static void node_lost_contact(struct tipc_node *n_ptr)
434 char addr_string[16];
435 struct tipc_sock_conn *conn, *safe;
436 struct list_head *conns = &n_ptr->conn_sks;
438 struct tipc_net *tn = net_generic(n_ptr->net, tipc_net_id);
441 pr_debug("Lost contact with %s\n",
442 tipc_addr_string_fill(addr_string, n_ptr->addr));
444 /* Flush broadcast link info associated with lost node */
445 if (n_ptr->bclink.recv_permitted) {
446 __skb_queue_purge(&n_ptr->bclink.deferdq);
448 if (n_ptr->bclink.reasm_buf) {
449 kfree_skb(n_ptr->bclink.reasm_buf);
450 n_ptr->bclink.reasm_buf = NULL;
453 tipc_bclink_remove_node(n_ptr->net, n_ptr->addr);
454 tipc_bclink_acknowledge(n_ptr, INVALID_LINK_SEQ);
456 n_ptr->bclink.recv_permitted = false;
459 /* Abort any ongoing link failover */
460 for (i = 0; i < MAX_BEARERS; i++) {
461 struct tipc_link *l_ptr = n_ptr->links[i].link;
464 l_ptr->exec_mode = TIPC_LINK_OPEN;
465 l_ptr->failover_checkpt = 0;
466 l_ptr->failover_pkts = 0;
467 kfree_skb(l_ptr->failover_skb);
468 l_ptr->failover_skb = NULL;
469 tipc_link_reset_fragments(l_ptr);
472 n_ptr->action_flags &= ~TIPC_WAIT_OWN_LINKS_DOWN;
474 /* Prevent re-contact with node until cleanup is done */
475 n_ptr->action_flags |= TIPC_WAIT_PEER_LINKS_DOWN;
477 /* Notify publications from this node */
478 n_ptr->action_flags |= TIPC_NOTIFY_NODE_DOWN;
480 /* Notify sockets connected to node */
481 list_for_each_entry_safe(conn, safe, conns, list) {
482 skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE, TIPC_CONN_MSG,
483 SHORT_H_SIZE, 0, tn->own_addr,
484 conn->peer_node, conn->port,
485 conn->peer_port, TIPC_ERR_NO_NODE);
487 skb_queue_tail(n_ptr->inputq, skb);
488 n_ptr->action_flags |= TIPC_MSG_EVT;
490 list_del(&conn->list);
496 * tipc_node_get_linkname - get the name of a link
498 * @bearer_id: id of the bearer
499 * @node: peer node address
500 * @linkname: link name output buffer
502 * Returns 0 on success
504 int tipc_node_get_linkname(struct net *net, u32 bearer_id, u32 addr,
505 char *linkname, size_t len)
507 struct tipc_link *link;
509 struct tipc_node *node = tipc_node_find(net, addr);
514 if (bearer_id >= MAX_BEARERS)
517 tipc_node_lock(node);
518 link = node->links[bearer_id].link;
520 strncpy(linkname, link->name, len);
524 tipc_node_unlock(node);
529 void tipc_node_unlock(struct tipc_node *node)
531 struct net *net = node->net;
533 u32 flags = node->action_flags;
535 struct list_head *publ_list;
536 struct sk_buff_head *inputq = node->inputq;
537 struct sk_buff_head *namedq;
539 if (likely(!flags || (flags == TIPC_MSG_EVT))) {
540 node->action_flags = 0;
541 spin_unlock_bh(&node->lock);
542 if (flags == TIPC_MSG_EVT)
543 tipc_sk_rcv(net, inputq);
548 link_id = node->link_id;
549 namedq = node->namedq;
550 publ_list = &node->publ_list;
552 node->action_flags &= ~(TIPC_MSG_EVT |
553 TIPC_NOTIFY_NODE_DOWN | TIPC_NOTIFY_NODE_UP |
554 TIPC_NOTIFY_LINK_DOWN | TIPC_NOTIFY_LINK_UP |
555 TIPC_WAKEUP_BCAST_USERS | TIPC_BCAST_MSG_EVT |
556 TIPC_NAMED_MSG_EVT | TIPC_BCAST_RESET);
558 spin_unlock_bh(&node->lock);
560 if (flags & TIPC_NOTIFY_NODE_DOWN)
561 tipc_publ_notify(net, publ_list, addr);
563 if (flags & TIPC_WAKEUP_BCAST_USERS)
564 tipc_bclink_wakeup_users(net);
566 if (flags & TIPC_NOTIFY_NODE_UP)
567 tipc_named_node_up(net, addr);
569 if (flags & TIPC_NOTIFY_LINK_UP)
570 tipc_nametbl_publish(net, TIPC_LINK_STATE, addr, addr,
571 TIPC_NODE_SCOPE, link_id, addr);
573 if (flags & TIPC_NOTIFY_LINK_DOWN)
574 tipc_nametbl_withdraw(net, TIPC_LINK_STATE, addr,
577 if (flags & TIPC_MSG_EVT)
578 tipc_sk_rcv(net, inputq);
580 if (flags & TIPC_NAMED_MSG_EVT)
581 tipc_named_rcv(net, namedq);
583 if (flags & TIPC_BCAST_MSG_EVT)
584 tipc_bclink_input(net);
586 if (flags & TIPC_BCAST_RESET)
587 tipc_link_reset_all(node);
590 /* Caller should hold node lock for the passed node */
591 static int __tipc_nl_add_node(struct tipc_nl_msg *msg, struct tipc_node *node)
594 struct nlattr *attrs;
596 hdr = genlmsg_put(msg->skb, msg->portid, msg->seq, &tipc_genl_family,
597 NLM_F_MULTI, TIPC_NL_NODE_GET);
601 attrs = nla_nest_start(msg->skb, TIPC_NLA_NODE);
605 if (nla_put_u32(msg->skb, TIPC_NLA_NODE_ADDR, node->addr))
607 if (tipc_node_is_up(node))
608 if (nla_put_flag(msg->skb, TIPC_NLA_NODE_UP))
611 nla_nest_end(msg->skb, attrs);
612 genlmsg_end(msg->skb, hdr);
617 nla_nest_cancel(msg->skb, attrs);
619 genlmsg_cancel(msg->skb, hdr);
624 static struct tipc_link *tipc_node_select_link(struct tipc_node *n, int sel,
626 struct tipc_media_addr **maddr)
628 int id = n->active_links[sel & 1];
630 if (unlikely(id < 0))
634 *maddr = &n->links[id].maddr;
635 return n->links[id].link;
639 * tipc_node_xmit() is the general link level function for message sending
640 * @net: the applicable net namespace
641 * @list: chain of buffers containing message
642 * @dnode: address of destination node
643 * @selector: a number used for deterministic link selection
644 * Consumes the buffer chain, except when returning -ELINKCONG
645 * Returns 0 if success, otherwise errno: -ELINKCONG,-EHOSTUNREACH,-EMSGSIZE
647 int tipc_node_xmit(struct net *net, struct sk_buff_head *list,
648 u32 dnode, int selector)
650 struct tipc_link *l = NULL;
652 struct sk_buff_head xmitq;
653 struct tipc_media_addr *maddr;
655 int rc = -EHOSTUNREACH;
657 __skb_queue_head_init(&xmitq);
658 n = tipc_node_find(net, dnode);
661 l = tipc_node_select_link(n, selector, &bearer_id, &maddr);
663 rc = tipc_link_xmit(l, list, &xmitq);
664 if (unlikely(rc == -ENOBUFS))
670 tipc_bearer_xmit(net, bearer_id, &xmitq, maddr);
673 if (likely(in_own_node(net, dnode))) {
674 tipc_sk_rcv(net, list);
680 /* tipc_node_xmit_skb(): send single buffer to destination
681 * Buffers sent via this functon are generally TIPC_SYSTEM_IMPORTANCE
682 * messages, which will not be rejected
683 * The only exception is datagram messages rerouted after secondary
684 * lookup, which are rare and safe to dispose of anyway.
685 * TODO: Return real return value, and let callers use
686 * tipc_wait_for_sendpkt() where applicable
688 int tipc_node_xmit_skb(struct net *net, struct sk_buff *skb, u32 dnode,
691 struct sk_buff_head head;
694 skb_queue_head_init(&head);
695 __skb_queue_tail(&head, skb);
696 rc = tipc_node_xmit(net, &head, dnode, selector);
697 if (rc == -ELINKCONG)
702 int tipc_nl_node_dump(struct sk_buff *skb, struct netlink_callback *cb)
705 struct net *net = sock_net(skb->sk);
706 struct tipc_net *tn = net_generic(net, tipc_net_id);
707 int done = cb->args[0];
708 int last_addr = cb->args[1];
709 struct tipc_node *node;
710 struct tipc_nl_msg msg;
716 msg.portid = NETLINK_CB(cb->skb).portid;
717 msg.seq = cb->nlh->nlmsg_seq;
721 node = tipc_node_find(net, last_addr);
724 /* We never set seq or call nl_dump_check_consistent()
725 * this means that setting prev_seq here will cause the
726 * consistence check to fail in the netlink callback
727 * handler. Resulting in the NLMSG_DONE message having
728 * the NLM_F_DUMP_INTR flag set if the node state
729 * changed while we released the lock.
737 list_for_each_entry_rcu(node, &tn->node_list, list) {
739 if (node->addr == last_addr)
745 tipc_node_lock(node);
746 err = __tipc_nl_add_node(&msg, node);
748 last_addr = node->addr;
749 tipc_node_unlock(node);
753 tipc_node_unlock(node);
758 cb->args[1] = last_addr;