2 * net/tipc/socket.c: TIPC socket API
4 * Copyright (c) 2001-2007, 2012-2014, Ericsson AB
5 * Copyright (c) 2004-2008, 2010-2013, 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.
38 #include "name_table.h"
41 #include <linux/export.h>
45 #define SS_LISTENING -1 /* socket is listening */
46 #define SS_READY -2 /* socket is connectionless */
48 #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */
49 #define CONN_PROBING_INTERVAL 3600000 /* [ms] => 1 h */
50 #define TIPC_FWD_MSG 1
51 #define TIPC_CONN_OK 0
52 #define TIPC_CONN_PROBING 1
55 * struct tipc_sock - TIPC socket structure
56 * @sk: socket - interacts with 'port' and with user via the socket API
57 * @connected: non-zero if port is currently connected to a peer port
58 * @conn_type: TIPC type used when connection was established
59 * @conn_instance: TIPC instance used when connection was established
60 * @published: non-zero if port has one or more associated names
61 * @max_pkt: maximum packet size "hint" used when building messages sent by port
62 * @ref: unique reference to port in TIPC object registry
63 * @phdr: preformatted message header used when sending messages
64 * @port_list: adjacent ports in TIPC's global list of ports
65 * @publications: list of publications for port
66 * @pub_count: total # of publications port has made during its lifetime
70 * @port: port - interacts with 'sk' and with the rest of the TIPC stack
71 * @peer_name: the peer of the connection, if any
72 * @conn_timeout: the time we can wait for an unresponded setup request
73 * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue
74 * @link_cong: non-zero if owner must sleep because of link congestion
75 * @sent_unacked: # messages sent by socket, and not yet acked by peer
76 * @rcv_unacked: # messages read by user, but not yet acked back to peer
87 struct list_head sock_list;
88 struct list_head publications;
92 struct timer_list timer;
100 static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *skb);
101 static void tipc_data_ready(struct sock *sk);
102 static void tipc_write_space(struct sock *sk);
103 static int tipc_release(struct socket *sock);
104 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags);
105 static int tipc_wait_for_sndmsg(struct socket *sock, long *timeo_p);
106 static void tipc_sk_timeout(unsigned long ref);
107 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
108 struct tipc_name_seq const *seq);
109 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
110 struct tipc_name_seq const *seq);
111 static u32 tipc_sk_ref_acquire(struct tipc_sock *tsk);
112 static void tipc_sk_ref_discard(u32 ref);
113 static struct tipc_sock *tipc_sk_get(u32 ref);
114 static struct tipc_sock *tipc_sk_get_next(u32 *ref);
115 static void tipc_sk_put(struct tipc_sock *tsk);
117 static const struct proto_ops packet_ops;
118 static const struct proto_ops stream_ops;
119 static const struct proto_ops msg_ops;
121 static struct proto tipc_proto;
122 static struct proto tipc_proto_kern;
124 static const struct nla_policy tipc_nl_sock_policy[TIPC_NLA_SOCK_MAX + 1] = {
125 [TIPC_NLA_SOCK_UNSPEC] = { .type = NLA_UNSPEC },
126 [TIPC_NLA_SOCK_ADDR] = { .type = NLA_U32 },
127 [TIPC_NLA_SOCK_REF] = { .type = NLA_U32 },
128 [TIPC_NLA_SOCK_CON] = { .type = NLA_NESTED },
129 [TIPC_NLA_SOCK_HAS_PUBL] = { .type = NLA_FLAG }
133 * Revised TIPC socket locking policy:
135 * Most socket operations take the standard socket lock when they start
136 * and hold it until they finish (or until they need to sleep). Acquiring
137 * this lock grants the owner exclusive access to the fields of the socket
138 * data structures, with the exception of the backlog queue. A few socket
139 * operations can be done without taking the socket lock because they only
140 * read socket information that never changes during the life of the socket.
142 * Socket operations may acquire the lock for the associated TIPC port if they
143 * need to perform an operation on the port. If any routine needs to acquire
144 * both the socket lock and the port lock it must take the socket lock first
145 * to avoid the risk of deadlock.
147 * The dispatcher handling incoming messages cannot grab the socket lock in
148 * the standard fashion, since invoked it runs at the BH level and cannot block.
149 * Instead, it checks to see if the socket lock is currently owned by someone,
150 * and either handles the message itself or adds it to the socket's backlog
151 * queue; in the latter case the queued message is processed once the process
152 * owning the socket lock releases it.
154 * NOTE: Releasing the socket lock while an operation is sleeping overcomes
155 * the problem of a blocked socket operation preventing any other operations
156 * from occurring. However, applications must be careful if they have
157 * multiple threads trying to send (or receive) on the same socket, as these
158 * operations might interfere with each other. For example, doing a connect
159 * and a receive at the same time might allow the receive to consume the
160 * ACK message meant for the connect. While additional work could be done
161 * to try and overcome this, it doesn't seem to be worthwhile at the present.
163 * NOTE: Releasing the socket lock while an operation is sleeping also ensures
164 * that another operation that must be performed in a non-blocking manner is
165 * not delayed for very long because the lock has already been taken.
167 * NOTE: This code assumes that certain fields of a port/socket pair are
168 * constant over its lifetime; such fields can be examined without taking
169 * the socket lock and/or port lock, and do not need to be re-read even
170 * after resuming processing after waiting. These fields include:
172 * - pointer to socket sk structure (aka tipc_sock structure)
173 * - pointer to port structure
177 static u32 tsk_peer_node(struct tipc_sock *tsk)
179 return msg_destnode(&tsk->phdr);
182 static u32 tsk_peer_port(struct tipc_sock *tsk)
184 return msg_destport(&tsk->phdr);
187 static bool tsk_unreliable(struct tipc_sock *tsk)
189 return msg_src_droppable(&tsk->phdr) != 0;
192 static void tsk_set_unreliable(struct tipc_sock *tsk, bool unreliable)
194 msg_set_src_droppable(&tsk->phdr, unreliable ? 1 : 0);
197 static bool tsk_unreturnable(struct tipc_sock *tsk)
199 return msg_dest_droppable(&tsk->phdr) != 0;
202 static void tsk_set_unreturnable(struct tipc_sock *tsk, bool unreturnable)
204 msg_set_dest_droppable(&tsk->phdr, unreturnable ? 1 : 0);
207 static int tsk_importance(struct tipc_sock *tsk)
209 return msg_importance(&tsk->phdr);
212 static int tsk_set_importance(struct tipc_sock *tsk, int imp)
214 if (imp > TIPC_CRITICAL_IMPORTANCE)
216 msg_set_importance(&tsk->phdr, (u32)imp);
220 static struct tipc_sock *tipc_sk(const struct sock *sk)
222 return container_of(sk, struct tipc_sock, sk);
225 static int tsk_conn_cong(struct tipc_sock *tsk)
227 return tsk->sent_unacked >= TIPC_FLOWCTRL_WIN;
231 * tsk_advance_rx_queue - discard first buffer in socket receive queue
233 * Caller must hold socket lock
235 static void tsk_advance_rx_queue(struct sock *sk)
237 kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
241 * tsk_rej_rx_queue - reject all buffers in socket receive queue
243 * Caller must hold socket lock
245 static void tsk_rej_rx_queue(struct sock *sk)
250 while ((buf = __skb_dequeue(&sk->sk_receive_queue))) {
251 if (tipc_msg_reverse(buf, &dnode, TIPC_ERR_NO_PORT))
252 tipc_link_xmit(buf, dnode, 0);
256 /* tsk_peer_msg - verify if message was sent by connected port's peer
258 * Handles cases where the node's network address has changed from
259 * the default of <0.0.0> to its configured setting.
261 static bool tsk_peer_msg(struct tipc_sock *tsk, struct tipc_msg *msg)
263 u32 peer_port = tsk_peer_port(tsk);
267 if (unlikely(!tsk->connected))
270 if (unlikely(msg_origport(msg) != peer_port))
273 orig_node = msg_orignode(msg);
274 peer_node = tsk_peer_node(tsk);
276 if (likely(orig_node == peer_node))
279 if (!orig_node && (peer_node == tipc_own_addr))
282 if (!peer_node && (orig_node == tipc_own_addr))
289 * tipc_sk_create - create a TIPC socket
290 * @net: network namespace (must be default network)
291 * @sock: pre-allocated socket structure
292 * @protocol: protocol indicator (must be 0)
293 * @kern: caused by kernel or by userspace?
295 * This routine creates additional data structures used by the TIPC socket,
296 * initializes them, and links them together.
298 * Returns 0 on success, errno otherwise
300 static int tipc_sk_create(struct net *net, struct socket *sock,
301 int protocol, int kern)
303 const struct proto_ops *ops;
306 struct tipc_sock *tsk;
307 struct tipc_msg *msg;
310 /* Validate arguments */
311 if (unlikely(protocol != 0))
312 return -EPROTONOSUPPORT;
314 switch (sock->type) {
317 state = SS_UNCONNECTED;
321 state = SS_UNCONNECTED;
332 /* Allocate socket's protocol area */
334 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto);
336 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto_kern);
342 ref = tipc_sk_ref_acquire(tsk);
344 pr_warn("Socket create failed; reference table exhausted\n");
347 tsk->max_pkt = MAX_PKT_DEFAULT;
349 INIT_LIST_HEAD(&tsk->publications);
351 tipc_msg_init(msg, TIPC_LOW_IMPORTANCE, TIPC_NAMED_MSG,
353 msg_set_origport(msg, ref);
355 /* Finish initializing socket data structures */
358 sock_init_data(sock, sk);
359 k_init_timer(&tsk->timer, (Handler)tipc_sk_timeout, ref);
360 sk->sk_backlog_rcv = tipc_backlog_rcv;
361 sk->sk_rcvbuf = sysctl_tipc_rmem[1];
362 sk->sk_data_ready = tipc_data_ready;
363 sk->sk_write_space = tipc_write_space;
364 tsk->conn_timeout = CONN_TIMEOUT_DEFAULT;
365 tsk->sent_unacked = 0;
366 atomic_set(&tsk->dupl_rcvcnt, 0);
368 if (sock->state == SS_READY) {
369 tsk_set_unreturnable(tsk, true);
370 if (sock->type == SOCK_DGRAM)
371 tsk_set_unreliable(tsk, true);
377 * tipc_sock_create_local - create TIPC socket from inside TIPC module
378 * @type: socket type - SOCK_RDM or SOCK_SEQPACKET
380 * We cannot use sock_creat_kern here because it bumps module user count.
381 * Since socket owner and creator is the same module we must make sure
382 * that module count remains zero for module local sockets, otherwise
383 * we cannot do rmmod.
385 * Returns 0 on success, errno otherwise
387 int tipc_sock_create_local(int type, struct socket **res)
391 rc = sock_create_lite(AF_TIPC, type, 0, res);
393 pr_err("Failed to create kernel socket\n");
396 tipc_sk_create(&init_net, *res, 0, 1);
402 * tipc_sock_release_local - release socket created by tipc_sock_create_local
403 * @sock: the socket to be released.
405 * Module reference count is not incremented when such sockets are created,
406 * so we must keep it from being decremented when they are released.
408 void tipc_sock_release_local(struct socket *sock)
416 * tipc_sock_accept_local - accept a connection on a socket created
417 * with tipc_sock_create_local. Use this function to avoid that
418 * module reference count is inadvertently incremented.
420 * @sock: the accepting socket
421 * @newsock: reference to the new socket to be created
422 * @flags: socket flags
425 int tipc_sock_accept_local(struct socket *sock, struct socket **newsock,
428 struct sock *sk = sock->sk;
431 ret = sock_create_lite(sk->sk_family, sk->sk_type,
432 sk->sk_protocol, newsock);
436 ret = tipc_accept(sock, *newsock, flags);
438 sock_release(*newsock);
441 (*newsock)->ops = sock->ops;
446 * tipc_release - destroy a TIPC socket
447 * @sock: socket to destroy
449 * This routine cleans up any messages that are still queued on the socket.
450 * For DGRAM and RDM socket types, all queued messages are rejected.
451 * For SEQPACKET and STREAM socket types, the first message is rejected
452 * and any others are discarded. (If the first message on a STREAM socket
453 * is partially-read, it is discarded and the next one is rejected instead.)
455 * NOTE: Rejected messages are not necessarily returned to the sender! They
456 * are returned or discarded according to the "destination droppable" setting
457 * specified for the message by the sender.
459 * Returns 0 on success, errno otherwise
461 static int tipc_release(struct socket *sock)
463 struct sock *sk = sock->sk;
464 struct tipc_sock *tsk;
469 * Exit if socket isn't fully initialized (occurs when a failed accept()
470 * releases a pre-allocated child socket that was never used)
479 * Reject all unreceived messages, except on an active connection
480 * (which disconnects locally & sends a 'FIN+' to peer)
482 dnode = tsk_peer_node(tsk);
483 while (sock->state != SS_DISCONNECTING) {
484 buf = __skb_dequeue(&sk->sk_receive_queue);
487 if (TIPC_SKB_CB(buf)->handle != NULL)
490 if ((sock->state == SS_CONNECTING) ||
491 (sock->state == SS_CONNECTED)) {
492 sock->state = SS_DISCONNECTING;
494 tipc_node_remove_conn(dnode, tsk->ref);
496 if (tipc_msg_reverse(buf, &dnode, TIPC_ERR_NO_PORT))
497 tipc_link_xmit(buf, dnode, 0);
501 tipc_sk_withdraw(tsk, 0, NULL);
502 tipc_sk_ref_discard(tsk->ref);
503 k_cancel_timer(&tsk->timer);
504 if (tsk->connected) {
505 buf = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE, TIPC_CONN_MSG,
506 SHORT_H_SIZE, 0, dnode, tipc_own_addr,
508 tsk->ref, TIPC_ERR_NO_PORT);
510 tipc_link_xmit(buf, dnode, tsk->ref);
511 tipc_node_remove_conn(dnode, tsk->ref);
513 k_term_timer(&tsk->timer);
515 /* Discard any remaining (connection-based) messages in receive queue */
516 __skb_queue_purge(&sk->sk_receive_queue);
518 /* Reject any messages that accumulated in backlog queue */
519 sock->state = SS_DISCONNECTING;
528 * tipc_bind - associate or disassocate TIPC name(s) with a socket
529 * @sock: socket structure
530 * @uaddr: socket address describing name(s) and desired operation
531 * @uaddr_len: size of socket address data structure
533 * Name and name sequence binding is indicated using a positive scope value;
534 * a negative scope value unbinds the specified name. Specifying no name
535 * (i.e. a socket address length of 0) unbinds all names from the socket.
537 * Returns 0 on success, errno otherwise
539 * NOTE: This routine doesn't need to take the socket lock since it doesn't
540 * access any non-constant socket information.
542 static int tipc_bind(struct socket *sock, struct sockaddr *uaddr,
545 struct sock *sk = sock->sk;
546 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
547 struct tipc_sock *tsk = tipc_sk(sk);
551 if (unlikely(!uaddr_len)) {
552 res = tipc_sk_withdraw(tsk, 0, NULL);
556 if (uaddr_len < sizeof(struct sockaddr_tipc)) {
560 if (addr->family != AF_TIPC) {
565 if (addr->addrtype == TIPC_ADDR_NAME)
566 addr->addr.nameseq.upper = addr->addr.nameseq.lower;
567 else if (addr->addrtype != TIPC_ADDR_NAMESEQ) {
572 if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) &&
573 (addr->addr.nameseq.type != TIPC_TOP_SRV) &&
574 (addr->addr.nameseq.type != TIPC_CFG_SRV)) {
579 res = (addr->scope > 0) ?
580 tipc_sk_publish(tsk, addr->scope, &addr->addr.nameseq) :
581 tipc_sk_withdraw(tsk, -addr->scope, &addr->addr.nameseq);
588 * tipc_getname - get port ID of socket or peer socket
589 * @sock: socket structure
590 * @uaddr: area for returned socket address
591 * @uaddr_len: area for returned length of socket address
592 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
594 * Returns 0 on success, errno otherwise
596 * NOTE: This routine doesn't need to take the socket lock since it only
597 * accesses socket information that is unchanging (or which changes in
598 * a completely predictable manner).
600 static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
601 int *uaddr_len, int peer)
603 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
604 struct tipc_sock *tsk = tipc_sk(sock->sk);
606 memset(addr, 0, sizeof(*addr));
608 if ((sock->state != SS_CONNECTED) &&
609 ((peer != 2) || (sock->state != SS_DISCONNECTING)))
611 addr->addr.id.ref = tsk_peer_port(tsk);
612 addr->addr.id.node = tsk_peer_node(tsk);
614 addr->addr.id.ref = tsk->ref;
615 addr->addr.id.node = tipc_own_addr;
618 *uaddr_len = sizeof(*addr);
619 addr->addrtype = TIPC_ADDR_ID;
620 addr->family = AF_TIPC;
622 addr->addr.name.domain = 0;
628 * tipc_poll - read and possibly block on pollmask
629 * @file: file structure associated with the socket
630 * @sock: socket for which to calculate the poll bits
633 * Returns pollmask value
636 * It appears that the usual socket locking mechanisms are not useful here
637 * since the pollmask info is potentially out-of-date the moment this routine
638 * exits. TCP and other protocols seem to rely on higher level poll routines
639 * to handle any preventable race conditions, so TIPC will do the same ...
641 * TIPC sets the returned events as follows:
643 * socket state flags set
644 * ------------ ---------
645 * unconnected no read flags
646 * POLLOUT if port is not congested
648 * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue
651 * connected POLLIN/POLLRDNORM if data in rx queue
652 * POLLOUT if port is not congested
654 * disconnecting POLLIN/POLLRDNORM/POLLHUP
657 * listening POLLIN if SYN in rx queue
660 * ready POLLIN/POLLRDNORM if data in rx queue
661 * [connectionless] POLLOUT (since port cannot be congested)
663 * IMPORTANT: The fact that a read or write operation is indicated does NOT
664 * imply that the operation will succeed, merely that it should be performed
665 * and will not block.
667 static unsigned int tipc_poll(struct file *file, struct socket *sock,
670 struct sock *sk = sock->sk;
671 struct tipc_sock *tsk = tipc_sk(sk);
674 sock_poll_wait(file, sk_sleep(sk), wait);
676 switch ((int)sock->state) {
683 if (!tsk->link_cong && !tsk_conn_cong(tsk))
688 if (!skb_queue_empty(&sk->sk_receive_queue))
689 mask |= (POLLIN | POLLRDNORM);
691 case SS_DISCONNECTING:
692 mask = (POLLIN | POLLRDNORM | POLLHUP);
700 * tipc_sendmcast - send multicast message
701 * @sock: socket structure
702 * @seq: destination address
703 * @iov: message data to send
704 * @dsz: total length of message data
705 * @timeo: timeout to wait for wakeup
707 * Called from function tipc_sendmsg(), which has done all sanity checks
708 * Returns the number of bytes sent on success, or errno
710 static int tipc_sendmcast(struct socket *sock, struct tipc_name_seq *seq,
711 struct iovec *iov, size_t dsz, long timeo)
713 struct sock *sk = sock->sk;
714 struct tipc_msg *mhdr = &tipc_sk(sk)->phdr;
719 msg_set_type(mhdr, TIPC_MCAST_MSG);
720 msg_set_lookup_scope(mhdr, TIPC_CLUSTER_SCOPE);
721 msg_set_destport(mhdr, 0);
722 msg_set_destnode(mhdr, 0);
723 msg_set_nametype(mhdr, seq->type);
724 msg_set_namelower(mhdr, seq->lower);
725 msg_set_nameupper(mhdr, seq->upper);
726 msg_set_hdr_sz(mhdr, MCAST_H_SIZE);
729 mtu = tipc_bclink_get_mtu();
730 rc = tipc_msg_build(mhdr, iov, 0, dsz, mtu, &buf);
731 if (unlikely(rc < 0))
735 rc = tipc_bclink_xmit(buf);
736 if (likely(rc >= 0)) {
742 if (rc != -ELINKCONG)
744 tipc_sk(sk)->link_cong = 1;
745 rc = tipc_wait_for_sndmsg(sock, &timeo);
752 /* tipc_sk_mcast_rcv - Deliver multicast message to all destination sockets
754 void tipc_sk_mcast_rcv(struct sk_buff *buf)
756 struct tipc_msg *msg = buf_msg(buf);
757 struct tipc_port_list dports = {0, NULL, };
758 struct tipc_port_list *item;
760 uint i, last, dst = 0;
761 u32 scope = TIPC_CLUSTER_SCOPE;
763 if (in_own_node(msg_orignode(msg)))
764 scope = TIPC_NODE_SCOPE;
766 /* Create destination port list: */
767 tipc_nametbl_mc_translate(msg_nametype(msg),
778 for (item = &dports; item; item = item->next) {
779 for (i = 0; i < PLSIZE && ++dst <= last; i++) {
780 b = (dst != last) ? skb_clone(buf, GFP_ATOMIC) : buf;
782 pr_warn("Failed do clone mcast rcv buffer\n");
785 msg_set_destport(msg, item->ports[i]);
789 tipc_port_list_free(&dports);
793 * tipc_sk_proto_rcv - receive a connection mng protocol message
794 * @tsk: receiving socket
795 * @dnode: node to send response message to, if any
796 * @buf: buffer containing protocol message
797 * Returns 0 (TIPC_OK) if message was consumed, 1 (TIPC_FWD_MSG) if
798 * (CONN_PROBE_REPLY) message should be forwarded.
800 static int tipc_sk_proto_rcv(struct tipc_sock *tsk, u32 *dnode,
803 struct tipc_msg *msg = buf_msg(buf);
806 /* Ignore if connection cannot be validated: */
807 if (!tsk_peer_msg(tsk, msg))
810 tsk->probing_state = TIPC_CONN_OK;
812 if (msg_type(msg) == CONN_ACK) {
813 conn_cong = tsk_conn_cong(tsk);
814 tsk->sent_unacked -= msg_msgcnt(msg);
816 tsk->sk.sk_write_space(&tsk->sk);
817 } else if (msg_type(msg) == CONN_PROBE) {
818 if (!tipc_msg_reverse(buf, dnode, TIPC_OK))
820 msg_set_type(msg, CONN_PROBE_REPLY);
823 /* Do nothing if msg_type() == CONN_PROBE_REPLY */
830 * dest_name_check - verify user is permitted to send to specified port name
831 * @dest: destination address
832 * @m: descriptor for message to be sent
834 * Prevents restricted configuration commands from being issued by
835 * unauthorized users.
837 * Returns 0 if permission is granted, otherwise errno
839 static int dest_name_check(struct sockaddr_tipc *dest, struct msghdr *m)
841 struct tipc_cfg_msg_hdr hdr;
843 if (unlikely(dest->addrtype == TIPC_ADDR_ID))
845 if (likely(dest->addr.name.name.type >= TIPC_RESERVED_TYPES))
847 if (likely(dest->addr.name.name.type == TIPC_TOP_SRV))
849 if (likely(dest->addr.name.name.type != TIPC_CFG_SRV))
852 if (!m->msg_iovlen || (m->msg_iov[0].iov_len < sizeof(hdr)))
854 if (copy_from_user(&hdr, m->msg_iov[0].iov_base, sizeof(hdr)))
856 if ((ntohs(hdr.tcm_type) & 0xC000) && (!capable(CAP_NET_ADMIN)))
862 static int tipc_wait_for_sndmsg(struct socket *sock, long *timeo_p)
864 struct sock *sk = sock->sk;
865 struct tipc_sock *tsk = tipc_sk(sk);
870 int err = sock_error(sk);
873 if (sock->state == SS_DISCONNECTING)
877 if (signal_pending(current))
878 return sock_intr_errno(*timeo_p);
880 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
881 done = sk_wait_event(sk, timeo_p, !tsk->link_cong);
882 finish_wait(sk_sleep(sk), &wait);
888 * tipc_sendmsg - send message in connectionless manner
889 * @iocb: if NULL, indicates that socket lock is already held
890 * @sock: socket structure
891 * @m: message to send
892 * @dsz: amount of user data to be sent
894 * Message must have an destination specified explicitly.
895 * Used for SOCK_RDM and SOCK_DGRAM messages,
896 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
897 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
899 * Returns the number of bytes sent on success, or errno otherwise
901 static int tipc_sendmsg(struct kiocb *iocb, struct socket *sock,
902 struct msghdr *m, size_t dsz)
904 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
905 struct sock *sk = sock->sk;
906 struct tipc_sock *tsk = tipc_sk(sk);
907 struct tipc_msg *mhdr = &tsk->phdr;
908 struct iovec *iov = m->msg_iov;
911 struct tipc_name_seq *seq = &dest->addr.nameseq;
917 return -EDESTADDRREQ;
919 if (unlikely((m->msg_namelen < sizeof(*dest)) ||
920 (dest->family != AF_TIPC)))
923 if (dsz > TIPC_MAX_USER_MSG_SIZE)
929 if (unlikely(sock->state != SS_READY)) {
930 if (sock->state == SS_LISTENING) {
934 if (sock->state != SS_UNCONNECTED) {
938 if (tsk->published) {
942 if (dest->addrtype == TIPC_ADDR_NAME) {
943 tsk->conn_type = dest->addr.name.name.type;
944 tsk->conn_instance = dest->addr.name.name.instance;
947 rc = dest_name_check(dest, m);
951 timeo = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
953 if (dest->addrtype == TIPC_ADDR_MCAST) {
954 rc = tipc_sendmcast(sock, seq, iov, dsz, timeo);
956 } else if (dest->addrtype == TIPC_ADDR_NAME) {
957 u32 type = dest->addr.name.name.type;
958 u32 inst = dest->addr.name.name.instance;
959 u32 domain = dest->addr.name.domain;
962 msg_set_type(mhdr, TIPC_NAMED_MSG);
963 msg_set_hdr_sz(mhdr, NAMED_H_SIZE);
964 msg_set_nametype(mhdr, type);
965 msg_set_nameinst(mhdr, inst);
966 msg_set_lookup_scope(mhdr, tipc_addr_scope(domain));
967 dport = tipc_nametbl_translate(type, inst, &dnode);
968 msg_set_destnode(mhdr, dnode);
969 msg_set_destport(mhdr, dport);
970 if (unlikely(!dport && !dnode)) {
974 } else if (dest->addrtype == TIPC_ADDR_ID) {
975 dnode = dest->addr.id.node;
976 msg_set_type(mhdr, TIPC_DIRECT_MSG);
977 msg_set_lookup_scope(mhdr, 0);
978 msg_set_destnode(mhdr, dnode);
979 msg_set_destport(mhdr, dest->addr.id.ref);
980 msg_set_hdr_sz(mhdr, BASIC_H_SIZE);
984 mtu = tipc_node_get_mtu(dnode, tsk->ref);
985 rc = tipc_msg_build(mhdr, iov, 0, dsz, mtu, &buf);
990 TIPC_SKB_CB(buf)->wakeup_pending = tsk->link_cong;
991 rc = tipc_link_xmit(buf, dnode, tsk->ref);
992 if (likely(rc >= 0)) {
993 if (sock->state != SS_READY)
994 sock->state = SS_CONNECTING;
1000 if (rc != -ELINKCONG)
1003 rc = tipc_wait_for_sndmsg(sock, &timeo);
1005 kfree_skb_list(buf);
1014 static int tipc_wait_for_sndpkt(struct socket *sock, long *timeo_p)
1016 struct sock *sk = sock->sk;
1017 struct tipc_sock *tsk = tipc_sk(sk);
1022 int err = sock_error(sk);
1025 if (sock->state == SS_DISCONNECTING)
1027 else if (sock->state != SS_CONNECTED)
1031 if (signal_pending(current))
1032 return sock_intr_errno(*timeo_p);
1034 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1035 done = sk_wait_event(sk, timeo_p,
1037 !tsk_conn_cong(tsk)) ||
1039 finish_wait(sk_sleep(sk), &wait);
1045 * tipc_send_stream - send stream-oriented data
1047 * @sock: socket structure
1049 * @dsz: total length of data to be transmitted
1051 * Used for SOCK_STREAM data.
1053 * Returns the number of bytes sent on success (or partial success),
1054 * or errno if no data sent
1056 static int tipc_send_stream(struct kiocb *iocb, struct socket *sock,
1057 struct msghdr *m, size_t dsz)
1059 struct sock *sk = sock->sk;
1060 struct tipc_sock *tsk = tipc_sk(sk);
1061 struct tipc_msg *mhdr = &tsk->phdr;
1062 struct sk_buff *buf;
1063 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1068 uint mtu, send, sent = 0;
1070 /* Handle implied connection establishment */
1071 if (unlikely(dest)) {
1072 rc = tipc_sendmsg(iocb, sock, m, dsz);
1073 if (dsz && (dsz == rc))
1074 tsk->sent_unacked = 1;
1077 if (dsz > (uint)INT_MAX)
1083 if (unlikely(sock->state != SS_CONNECTED)) {
1084 if (sock->state == SS_DISCONNECTING)
1091 timeo = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1092 dnode = tsk_peer_node(tsk);
1096 send = min_t(uint, dsz - sent, TIPC_MAX_USER_MSG_SIZE);
1097 rc = tipc_msg_build(mhdr, m->msg_iov, sent, send, mtu, &buf);
1098 if (unlikely(rc < 0))
1101 if (likely(!tsk_conn_cong(tsk))) {
1102 rc = tipc_link_xmit(buf, dnode, ref);
1104 tsk->sent_unacked++;
1110 if (rc == -EMSGSIZE) {
1111 tsk->max_pkt = tipc_node_get_mtu(dnode, ref);
1114 if (rc != -ELINKCONG)
1118 rc = tipc_wait_for_sndpkt(sock, &timeo);
1120 kfree_skb_list(buf);
1125 return sent ? sent : rc;
1129 * tipc_send_packet - send a connection-oriented message
1130 * @iocb: if NULL, indicates that socket lock is already held
1131 * @sock: socket structure
1132 * @m: message to send
1133 * @dsz: length of data to be transmitted
1135 * Used for SOCK_SEQPACKET messages.
1137 * Returns the number of bytes sent on success, or errno otherwise
1139 static int tipc_send_packet(struct kiocb *iocb, struct socket *sock,
1140 struct msghdr *m, size_t dsz)
1142 if (dsz > TIPC_MAX_USER_MSG_SIZE)
1145 return tipc_send_stream(iocb, sock, m, dsz);
1148 /* tipc_sk_finish_conn - complete the setup of a connection
1150 static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port,
1153 struct tipc_msg *msg = &tsk->phdr;
1155 msg_set_destnode(msg, peer_node);
1156 msg_set_destport(msg, peer_port);
1157 msg_set_type(msg, TIPC_CONN_MSG);
1158 msg_set_lookup_scope(msg, 0);
1159 msg_set_hdr_sz(msg, SHORT_H_SIZE);
1161 tsk->probing_interval = CONN_PROBING_INTERVAL;
1162 tsk->probing_state = TIPC_CONN_OK;
1164 k_start_timer(&tsk->timer, tsk->probing_interval);
1165 tipc_node_add_conn(peer_node, tsk->ref, peer_port);
1166 tsk->max_pkt = tipc_node_get_mtu(peer_node, tsk->ref);
1170 * set_orig_addr - capture sender's address for received message
1171 * @m: descriptor for message info
1172 * @msg: received message header
1174 * Note: Address is not captured if not requested by receiver.
1176 static void set_orig_addr(struct msghdr *m, struct tipc_msg *msg)
1178 DECLARE_SOCKADDR(struct sockaddr_tipc *, addr, m->msg_name);
1181 addr->family = AF_TIPC;
1182 addr->addrtype = TIPC_ADDR_ID;
1183 memset(&addr->addr, 0, sizeof(addr->addr));
1184 addr->addr.id.ref = msg_origport(msg);
1185 addr->addr.id.node = msg_orignode(msg);
1186 addr->addr.name.domain = 0; /* could leave uninitialized */
1187 addr->scope = 0; /* could leave uninitialized */
1188 m->msg_namelen = sizeof(struct sockaddr_tipc);
1193 * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
1194 * @m: descriptor for message info
1195 * @msg: received message header
1196 * @tsk: TIPC port associated with message
1198 * Note: Ancillary data is not captured if not requested by receiver.
1200 * Returns 0 if successful, otherwise errno
1202 static int tipc_sk_anc_data_recv(struct msghdr *m, struct tipc_msg *msg,
1203 struct tipc_sock *tsk)
1211 if (likely(m->msg_controllen == 0))
1214 /* Optionally capture errored message object(s) */
1215 err = msg ? msg_errcode(msg) : 0;
1216 if (unlikely(err)) {
1218 anc_data[1] = msg_data_sz(msg);
1219 res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
1223 res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
1230 /* Optionally capture message destination object */
1231 dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
1232 switch (dest_type) {
1233 case TIPC_NAMED_MSG:
1235 anc_data[0] = msg_nametype(msg);
1236 anc_data[1] = msg_namelower(msg);
1237 anc_data[2] = msg_namelower(msg);
1239 case TIPC_MCAST_MSG:
1241 anc_data[0] = msg_nametype(msg);
1242 anc_data[1] = msg_namelower(msg);
1243 anc_data[2] = msg_nameupper(msg);
1246 has_name = (tsk->conn_type != 0);
1247 anc_data[0] = tsk->conn_type;
1248 anc_data[1] = tsk->conn_instance;
1249 anc_data[2] = tsk->conn_instance;
1255 res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
1263 static void tipc_sk_send_ack(struct tipc_sock *tsk, uint ack)
1265 struct sk_buff *buf = NULL;
1266 struct tipc_msg *msg;
1267 u32 peer_port = tsk_peer_port(tsk);
1268 u32 dnode = tsk_peer_node(tsk);
1270 if (!tsk->connected)
1272 buf = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0, dnode,
1273 tipc_own_addr, peer_port, tsk->ref, TIPC_OK);
1277 msg_set_msgcnt(msg, ack);
1278 tipc_link_xmit(buf, dnode, msg_link_selector(msg));
1281 static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop)
1283 struct sock *sk = sock->sk;
1285 long timeo = *timeop;
1289 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1290 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1291 if (sock->state == SS_DISCONNECTING) {
1296 timeo = schedule_timeout(timeo);
1300 if (!skb_queue_empty(&sk->sk_receive_queue))
1302 err = sock_intr_errno(timeo);
1303 if (signal_pending(current))
1309 finish_wait(sk_sleep(sk), &wait);
1315 * tipc_recvmsg - receive packet-oriented message
1317 * @m: descriptor for message info
1318 * @buf_len: total size of user buffer area
1319 * @flags: receive flags
1321 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1322 * If the complete message doesn't fit in user area, truncate it.
1324 * Returns size of returned message data, errno otherwise
1326 static int tipc_recvmsg(struct kiocb *iocb, struct socket *sock,
1327 struct msghdr *m, size_t buf_len, int flags)
1329 struct sock *sk = sock->sk;
1330 struct tipc_sock *tsk = tipc_sk(sk);
1331 struct sk_buff *buf;
1332 struct tipc_msg *msg;
1338 /* Catch invalid receive requests */
1339 if (unlikely(!buf_len))
1344 if (unlikely(sock->state == SS_UNCONNECTED)) {
1349 timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1352 /* Look for a message in receive queue; wait if necessary */
1353 res = tipc_wait_for_rcvmsg(sock, &timeo);
1357 /* Look at first message in receive queue */
1358 buf = skb_peek(&sk->sk_receive_queue);
1360 sz = msg_data_sz(msg);
1361 err = msg_errcode(msg);
1363 /* Discard an empty non-errored message & try again */
1364 if ((!sz) && (!err)) {
1365 tsk_advance_rx_queue(sk);
1369 /* Capture sender's address (optional) */
1370 set_orig_addr(m, msg);
1372 /* Capture ancillary data (optional) */
1373 res = tipc_sk_anc_data_recv(m, msg, tsk);
1377 /* Capture message data (if valid) & compute return value (always) */
1379 if (unlikely(buf_len < sz)) {
1381 m->msg_flags |= MSG_TRUNC;
1383 res = skb_copy_datagram_msg(buf, msg_hdr_sz(msg), m, sz);
1388 if ((sock->state == SS_READY) ||
1389 ((err == TIPC_CONN_SHUTDOWN) || m->msg_control))
1395 /* Consume received message (optional) */
1396 if (likely(!(flags & MSG_PEEK))) {
1397 if ((sock->state != SS_READY) &&
1398 (++tsk->rcv_unacked >= TIPC_CONNACK_INTV)) {
1399 tipc_sk_send_ack(tsk, tsk->rcv_unacked);
1400 tsk->rcv_unacked = 0;
1402 tsk_advance_rx_queue(sk);
1410 * tipc_recv_stream - receive stream-oriented data
1412 * @m: descriptor for message info
1413 * @buf_len: total size of user buffer area
1414 * @flags: receive flags
1416 * Used for SOCK_STREAM messages only. If not enough data is available
1417 * will optionally wait for more; never truncates data.
1419 * Returns size of returned message data, errno otherwise
1421 static int tipc_recv_stream(struct kiocb *iocb, struct socket *sock,
1422 struct msghdr *m, size_t buf_len, int flags)
1424 struct sock *sk = sock->sk;
1425 struct tipc_sock *tsk = tipc_sk(sk);
1426 struct sk_buff *buf;
1427 struct tipc_msg *msg;
1430 int sz_to_copy, target, needed;
1435 /* Catch invalid receive attempts */
1436 if (unlikely(!buf_len))
1441 if (unlikely(sock->state == SS_UNCONNECTED)) {
1446 target = sock_rcvlowat(sk, flags & MSG_WAITALL, buf_len);
1447 timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1450 /* Look for a message in receive queue; wait if necessary */
1451 res = tipc_wait_for_rcvmsg(sock, &timeo);
1455 /* Look at first message in receive queue */
1456 buf = skb_peek(&sk->sk_receive_queue);
1458 sz = msg_data_sz(msg);
1459 err = msg_errcode(msg);
1461 /* Discard an empty non-errored message & try again */
1462 if ((!sz) && (!err)) {
1463 tsk_advance_rx_queue(sk);
1467 /* Optionally capture sender's address & ancillary data of first msg */
1468 if (sz_copied == 0) {
1469 set_orig_addr(m, msg);
1470 res = tipc_sk_anc_data_recv(m, msg, tsk);
1475 /* Capture message data (if valid) & compute return value (always) */
1477 u32 offset = (u32)(unsigned long)(TIPC_SKB_CB(buf)->handle);
1480 needed = (buf_len - sz_copied);
1481 sz_to_copy = (sz <= needed) ? sz : needed;
1483 res = skb_copy_datagram_msg(buf, msg_hdr_sz(msg) + offset,
1488 sz_copied += sz_to_copy;
1490 if (sz_to_copy < sz) {
1491 if (!(flags & MSG_PEEK))
1492 TIPC_SKB_CB(buf)->handle =
1493 (void *)(unsigned long)(offset + sz_to_copy);
1498 goto exit; /* can't add error msg to valid data */
1500 if ((err == TIPC_CONN_SHUTDOWN) || m->msg_control)
1506 /* Consume received message (optional) */
1507 if (likely(!(flags & MSG_PEEK))) {
1508 if (unlikely(++tsk->rcv_unacked >= TIPC_CONNACK_INTV)) {
1509 tipc_sk_send_ack(tsk, tsk->rcv_unacked);
1510 tsk->rcv_unacked = 0;
1512 tsk_advance_rx_queue(sk);
1515 /* Loop around if more data is required */
1516 if ((sz_copied < buf_len) && /* didn't get all requested data */
1517 (!skb_queue_empty(&sk->sk_receive_queue) ||
1518 (sz_copied < target)) && /* and more is ready or required */
1519 (!(flags & MSG_PEEK)) && /* and aren't just peeking at data */
1520 (!err)) /* and haven't reached a FIN */
1525 return sz_copied ? sz_copied : res;
1529 * tipc_write_space - wake up thread if port congestion is released
1532 static void tipc_write_space(struct sock *sk)
1534 struct socket_wq *wq;
1537 wq = rcu_dereference(sk->sk_wq);
1538 if (wq_has_sleeper(wq))
1539 wake_up_interruptible_sync_poll(&wq->wait, POLLOUT |
1540 POLLWRNORM | POLLWRBAND);
1545 * tipc_data_ready - wake up threads to indicate messages have been received
1547 * @len: the length of messages
1549 static void tipc_data_ready(struct sock *sk)
1551 struct socket_wq *wq;
1554 wq = rcu_dereference(sk->sk_wq);
1555 if (wq_has_sleeper(wq))
1556 wake_up_interruptible_sync_poll(&wq->wait, POLLIN |
1557 POLLRDNORM | POLLRDBAND);
1562 * filter_connect - Handle all incoming messages for a connection-based socket
1566 * Returns 0 (TIPC_OK) if everything ok, -TIPC_ERR_NO_PORT otherwise
1568 static int filter_connect(struct tipc_sock *tsk, struct sk_buff **buf)
1570 struct sock *sk = &tsk->sk;
1571 struct socket *sock = sk->sk_socket;
1572 struct tipc_msg *msg = buf_msg(*buf);
1573 int retval = -TIPC_ERR_NO_PORT;
1578 switch ((int)sock->state) {
1580 /* Accept only connection-based messages sent by peer */
1581 if (tsk_peer_msg(tsk, msg)) {
1582 if (unlikely(msg_errcode(msg))) {
1583 sock->state = SS_DISCONNECTING;
1585 /* let timer expire on it's own */
1586 tipc_node_remove_conn(tsk_peer_node(tsk),
1593 /* Accept only ACK or NACK message */
1595 if (unlikely(!msg_connected(msg)))
1598 if (unlikely(msg_errcode(msg))) {
1599 sock->state = SS_DISCONNECTING;
1600 sk->sk_err = ECONNREFUSED;
1605 if (unlikely(msg_importance(msg) > TIPC_CRITICAL_IMPORTANCE)) {
1606 sock->state = SS_DISCONNECTING;
1607 sk->sk_err = EINVAL;
1612 tipc_sk_finish_conn(tsk, msg_origport(msg), msg_orignode(msg));
1613 msg_set_importance(&tsk->phdr, msg_importance(msg));
1614 sock->state = SS_CONNECTED;
1616 /* If an incoming message is an 'ACK-', it should be
1617 * discarded here because it doesn't contain useful
1618 * data. In addition, we should try to wake up
1619 * connect() routine if sleeping.
1621 if (msg_data_sz(msg) == 0) {
1624 if (waitqueue_active(sk_sleep(sk)))
1625 wake_up_interruptible(sk_sleep(sk));
1630 case SS_UNCONNECTED:
1631 /* Accept only SYN message */
1632 if (!msg_connected(msg) && !(msg_errcode(msg)))
1635 case SS_DISCONNECTING:
1638 pr_err("Unknown socket state %u\n", sock->state);
1644 * rcvbuf_limit - get proper overload limit of socket receive queue
1648 * For all connection oriented messages, irrespective of importance,
1649 * the default overload value (i.e. 67MB) is set as limit.
1651 * For all connectionless messages, by default new queue limits are
1654 * TIPC_LOW_IMPORTANCE (4 MB)
1655 * TIPC_MEDIUM_IMPORTANCE (8 MB)
1656 * TIPC_HIGH_IMPORTANCE (16 MB)
1657 * TIPC_CRITICAL_IMPORTANCE (32 MB)
1659 * Returns overload limit according to corresponding message importance
1661 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *buf)
1663 struct tipc_msg *msg = buf_msg(buf);
1665 if (msg_connected(msg))
1666 return sysctl_tipc_rmem[2];
1668 return sk->sk_rcvbuf >> TIPC_CRITICAL_IMPORTANCE <<
1669 msg_importance(msg);
1673 * filter_rcv - validate incoming message
1677 * Enqueues message on receive queue if acceptable; optionally handles
1678 * disconnect indication for a connected socket.
1680 * Called with socket lock already taken; port lock may also be taken.
1682 * Returns 0 (TIPC_OK) if message was consumed, -TIPC error code if message
1683 * to be rejected, 1 (TIPC_FWD_MSG) if (CONN_MANAGER) message to be forwarded
1685 static int filter_rcv(struct sock *sk, struct sk_buff *buf)
1687 struct socket *sock = sk->sk_socket;
1688 struct tipc_sock *tsk = tipc_sk(sk);
1689 struct tipc_msg *msg = buf_msg(buf);
1690 unsigned int limit = rcvbuf_limit(sk, buf);
1694 if (unlikely(msg_user(msg) == CONN_MANAGER))
1695 return tipc_sk_proto_rcv(tsk, &onode, buf);
1697 if (unlikely(msg_user(msg) == SOCK_WAKEUP)) {
1700 sk->sk_write_space(sk);
1704 /* Reject message if it is wrong sort of message for socket */
1705 if (msg_type(msg) > TIPC_DIRECT_MSG)
1706 return -TIPC_ERR_NO_PORT;
1708 if (sock->state == SS_READY) {
1709 if (msg_connected(msg))
1710 return -TIPC_ERR_NO_PORT;
1712 rc = filter_connect(tsk, &buf);
1713 if (rc != TIPC_OK || buf == NULL)
1717 /* Reject message if there isn't room to queue it */
1718 if (sk_rmem_alloc_get(sk) + buf->truesize >= limit)
1719 return -TIPC_ERR_OVERLOAD;
1721 /* Enqueue message */
1722 TIPC_SKB_CB(buf)->handle = NULL;
1723 __skb_queue_tail(&sk->sk_receive_queue, buf);
1724 skb_set_owner_r(buf, sk);
1726 sk->sk_data_ready(sk);
1731 * tipc_backlog_rcv - handle incoming message from backlog queue
1735 * Caller must hold socket lock, but not port lock.
1739 static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *buf)
1743 struct tipc_sock *tsk = tipc_sk(sk);
1744 uint truesize = buf->truesize;
1746 rc = filter_rcv(sk, buf);
1749 if (atomic_read(&tsk->dupl_rcvcnt) < TIPC_CONN_OVERLOAD_LIMIT)
1750 atomic_add(truesize, &tsk->dupl_rcvcnt);
1754 if ((rc < 0) && !tipc_msg_reverse(buf, &onode, -rc))
1757 tipc_link_xmit(buf, onode, 0);
1763 * tipc_sk_rcv - handle incoming message
1764 * @buf: buffer containing arriving message
1766 * Returns 0 if success, or errno: -EHOSTUNREACH
1768 int tipc_sk_rcv(struct sk_buff *buf)
1770 struct tipc_sock *tsk;
1772 u32 dport = msg_destport(buf_msg(buf));
1777 /* Validate destination and message */
1778 tsk = tipc_sk_get(dport);
1779 if (unlikely(!tsk)) {
1780 rc = tipc_msg_eval(buf, &dnode);
1786 spin_lock_bh(&sk->sk_lock.slock);
1788 if (!sock_owned_by_user(sk)) {
1789 rc = filter_rcv(sk, buf);
1791 if (sk->sk_backlog.len == 0)
1792 atomic_set(&tsk->dupl_rcvcnt, 0);
1793 limit = rcvbuf_limit(sk, buf) + atomic_read(&tsk->dupl_rcvcnt);
1794 if (sk_add_backlog(sk, buf, limit))
1795 rc = -TIPC_ERR_OVERLOAD;
1797 spin_unlock_bh(&sk->sk_lock.slock);
1802 if ((rc < 0) && !tipc_msg_reverse(buf, &dnode, -rc))
1803 return -EHOSTUNREACH;
1805 tipc_link_xmit(buf, dnode, 0);
1806 return (rc < 0) ? -EHOSTUNREACH : 0;
1809 static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
1811 struct sock *sk = sock->sk;
1816 int err = sock_error(sk);
1821 if (signal_pending(current))
1822 return sock_intr_errno(*timeo_p);
1824 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1825 done = sk_wait_event(sk, timeo_p, sock->state != SS_CONNECTING);
1826 finish_wait(sk_sleep(sk), &wait);
1832 * tipc_connect - establish a connection to another TIPC port
1833 * @sock: socket structure
1834 * @dest: socket address for destination port
1835 * @destlen: size of socket address data structure
1836 * @flags: file-related flags associated with socket
1838 * Returns 0 on success, errno otherwise
1840 static int tipc_connect(struct socket *sock, struct sockaddr *dest,
1841 int destlen, int flags)
1843 struct sock *sk = sock->sk;
1844 struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
1845 struct msghdr m = {NULL,};
1846 long timeout = (flags & O_NONBLOCK) ? 0 : tipc_sk(sk)->conn_timeout;
1847 socket_state previous;
1852 /* For now, TIPC does not allow use of connect() with DGRAM/RDM types */
1853 if (sock->state == SS_READY) {
1859 * Reject connection attempt using multicast address
1861 * Note: send_msg() validates the rest of the address fields,
1862 * so there's no need to do it here
1864 if (dst->addrtype == TIPC_ADDR_MCAST) {
1869 previous = sock->state;
1870 switch (sock->state) {
1871 case SS_UNCONNECTED:
1872 /* Send a 'SYN-' to destination */
1874 m.msg_namelen = destlen;
1876 /* If connect is in non-blocking case, set MSG_DONTWAIT to
1877 * indicate send_msg() is never blocked.
1880 m.msg_flags = MSG_DONTWAIT;
1882 res = tipc_sendmsg(NULL, sock, &m, 0);
1883 if ((res < 0) && (res != -EWOULDBLOCK))
1886 /* Just entered SS_CONNECTING state; the only
1887 * difference is that return value in non-blocking
1888 * case is EINPROGRESS, rather than EALREADY.
1892 if (previous == SS_CONNECTING)
1896 timeout = msecs_to_jiffies(timeout);
1897 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
1898 res = tipc_wait_for_connect(sock, &timeout);
1913 * tipc_listen - allow socket to listen for incoming connections
1914 * @sock: socket structure
1917 * Returns 0 on success, errno otherwise
1919 static int tipc_listen(struct socket *sock, int len)
1921 struct sock *sk = sock->sk;
1926 if (sock->state != SS_UNCONNECTED)
1929 sock->state = SS_LISTENING;
1937 static int tipc_wait_for_accept(struct socket *sock, long timeo)
1939 struct sock *sk = sock->sk;
1943 /* True wake-one mechanism for incoming connections: only
1944 * one process gets woken up, not the 'whole herd'.
1945 * Since we do not 'race & poll' for established sockets
1946 * anymore, the common case will execute the loop only once.
1949 prepare_to_wait_exclusive(sk_sleep(sk), &wait,
1950 TASK_INTERRUPTIBLE);
1951 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1953 timeo = schedule_timeout(timeo);
1957 if (!skb_queue_empty(&sk->sk_receive_queue))
1960 if (sock->state != SS_LISTENING)
1962 err = sock_intr_errno(timeo);
1963 if (signal_pending(current))
1969 finish_wait(sk_sleep(sk), &wait);
1974 * tipc_accept - wait for connection request
1975 * @sock: listening socket
1976 * @newsock: new socket that is to be connected
1977 * @flags: file-related flags associated with socket
1979 * Returns 0 on success, errno otherwise
1981 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags)
1983 struct sock *new_sk, *sk = sock->sk;
1984 struct sk_buff *buf;
1985 struct tipc_sock *new_tsock;
1986 struct tipc_msg *msg;
1992 if (sock->state != SS_LISTENING) {
1996 timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
1997 res = tipc_wait_for_accept(sock, timeo);
2001 buf = skb_peek(&sk->sk_receive_queue);
2003 res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, 1);
2007 new_sk = new_sock->sk;
2008 new_tsock = tipc_sk(new_sk);
2011 /* we lock on new_sk; but lockdep sees the lock on sk */
2012 lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
2015 * Reject any stray messages received by new socket
2016 * before the socket lock was taken (very, very unlikely)
2018 tsk_rej_rx_queue(new_sk);
2020 /* Connect new socket to it's peer */
2021 tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg));
2022 new_sock->state = SS_CONNECTED;
2024 tsk_set_importance(new_tsock, msg_importance(msg));
2025 if (msg_named(msg)) {
2026 new_tsock->conn_type = msg_nametype(msg);
2027 new_tsock->conn_instance = msg_nameinst(msg);
2031 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2032 * Respond to 'SYN+' by queuing it on new socket.
2034 if (!msg_data_sz(msg)) {
2035 struct msghdr m = {NULL,};
2037 tsk_advance_rx_queue(sk);
2038 tipc_send_packet(NULL, new_sock, &m, 0);
2040 __skb_dequeue(&sk->sk_receive_queue);
2041 __skb_queue_head(&new_sk->sk_receive_queue, buf);
2042 skb_set_owner_r(buf, new_sk);
2044 release_sock(new_sk);
2051 * tipc_shutdown - shutdown socket connection
2052 * @sock: socket structure
2053 * @how: direction to close (must be SHUT_RDWR)
2055 * Terminates connection (if necessary), then purges socket's receive queue.
2057 * Returns 0 on success, errno otherwise
2059 static int tipc_shutdown(struct socket *sock, int how)
2061 struct sock *sk = sock->sk;
2062 struct tipc_sock *tsk = tipc_sk(sk);
2063 struct sk_buff *buf;
2067 if (how != SHUT_RDWR)
2072 switch (sock->state) {
2077 /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */
2078 buf = __skb_dequeue(&sk->sk_receive_queue);
2080 if (TIPC_SKB_CB(buf)->handle != NULL) {
2084 if (tipc_msg_reverse(buf, &dnode, TIPC_CONN_SHUTDOWN))
2085 tipc_link_xmit(buf, dnode, tsk->ref);
2086 tipc_node_remove_conn(dnode, tsk->ref);
2088 dnode = tsk_peer_node(tsk);
2089 buf = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE,
2090 TIPC_CONN_MSG, SHORT_H_SIZE,
2091 0, dnode, tipc_own_addr,
2093 tsk->ref, TIPC_CONN_SHUTDOWN);
2094 tipc_link_xmit(buf, dnode, tsk->ref);
2097 sock->state = SS_DISCONNECTING;
2098 tipc_node_remove_conn(dnode, tsk->ref);
2101 case SS_DISCONNECTING:
2103 /* Discard any unreceived messages */
2104 __skb_queue_purge(&sk->sk_receive_queue);
2106 /* Wake up anyone sleeping in poll */
2107 sk->sk_state_change(sk);
2119 static void tipc_sk_timeout(unsigned long ref)
2121 struct tipc_sock *tsk;
2123 struct sk_buff *buf = NULL;
2124 u32 peer_port, peer_node;
2126 tsk = tipc_sk_get(ref);
2132 if (!tsk->connected) {
2136 peer_port = tsk_peer_port(tsk);
2137 peer_node = tsk_peer_node(tsk);
2139 if (tsk->probing_state == TIPC_CONN_PROBING) {
2140 /* Previous probe not answered -> self abort */
2141 buf = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE, TIPC_CONN_MSG,
2142 SHORT_H_SIZE, 0, tipc_own_addr,
2143 peer_node, ref, peer_port,
2146 buf = tipc_msg_create(CONN_MANAGER, CONN_PROBE, INT_H_SIZE,
2147 0, peer_node, tipc_own_addr,
2148 peer_port, ref, TIPC_OK);
2149 tsk->probing_state = TIPC_CONN_PROBING;
2150 k_start_timer(&tsk->timer, tsk->probing_interval);
2154 tipc_link_xmit(buf, peer_node, ref);
2159 static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
2160 struct tipc_name_seq const *seq)
2162 struct publication *publ;
2167 key = tsk->ref + tsk->pub_count + 1;
2168 if (key == tsk->ref)
2171 publ = tipc_nametbl_publish(seq->type, seq->lower, seq->upper,
2172 scope, tsk->ref, key);
2173 if (unlikely(!publ))
2176 list_add(&publ->pport_list, &tsk->publications);
2182 static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
2183 struct tipc_name_seq const *seq)
2185 struct publication *publ;
2186 struct publication *safe;
2189 list_for_each_entry_safe(publ, safe, &tsk->publications, pport_list) {
2191 if (publ->scope != scope)
2193 if (publ->type != seq->type)
2195 if (publ->lower != seq->lower)
2197 if (publ->upper != seq->upper)
2199 tipc_nametbl_withdraw(publ->type, publ->lower,
2200 publ->ref, publ->key);
2204 tipc_nametbl_withdraw(publ->type, publ->lower,
2205 publ->ref, publ->key);
2208 if (list_empty(&tsk->publications))
2213 static int tipc_sk_show(struct tipc_sock *tsk, char *buf,
2214 int len, int full_id)
2216 struct publication *publ;
2220 ret = tipc_snprintf(buf, len, "<%u.%u.%u:%u>:",
2221 tipc_zone(tipc_own_addr),
2222 tipc_cluster(tipc_own_addr),
2223 tipc_node(tipc_own_addr), tsk->ref);
2225 ret = tipc_snprintf(buf, len, "%-10u:", tsk->ref);
2227 if (tsk->connected) {
2228 u32 dport = tsk_peer_port(tsk);
2229 u32 destnode = tsk_peer_node(tsk);
2231 ret += tipc_snprintf(buf + ret, len - ret,
2232 " connected to <%u.%u.%u:%u>",
2233 tipc_zone(destnode),
2234 tipc_cluster(destnode),
2235 tipc_node(destnode), dport);
2236 if (tsk->conn_type != 0)
2237 ret += tipc_snprintf(buf + ret, len - ret,
2238 " via {%u,%u}", tsk->conn_type,
2239 tsk->conn_instance);
2240 } else if (tsk->published) {
2241 ret += tipc_snprintf(buf + ret, len - ret, " bound to");
2242 list_for_each_entry(publ, &tsk->publications, pport_list) {
2243 if (publ->lower == publ->upper)
2244 ret += tipc_snprintf(buf + ret, len - ret,
2245 " {%u,%u}", publ->type,
2248 ret += tipc_snprintf(buf + ret, len - ret,
2249 " {%u,%u,%u}", publ->type,
2250 publ->lower, publ->upper);
2253 ret += tipc_snprintf(buf + ret, len - ret, "\n");
2257 struct sk_buff *tipc_sk_socks_show(void)
2259 struct sk_buff *buf;
2260 struct tlv_desc *rep_tlv;
2263 struct tipc_sock *tsk;
2267 buf = tipc_cfg_reply_alloc(TLV_SPACE(ULTRA_STRING_MAX_LEN));
2270 rep_tlv = (struct tlv_desc *)buf->data;
2271 pb = TLV_DATA(rep_tlv);
2272 pb_len = ULTRA_STRING_MAX_LEN;
2274 tsk = tipc_sk_get_next(&ref);
2275 for (; tsk; tsk = tipc_sk_get_next(&ref)) {
2276 lock_sock(&tsk->sk);
2277 str_len += tipc_sk_show(tsk, pb + str_len,
2278 pb_len - str_len, 0);
2279 release_sock(&tsk->sk);
2282 str_len += 1; /* for "\0" */
2283 skb_put(buf, TLV_SPACE(str_len));
2284 TLV_SET(rep_tlv, TIPC_TLV_ULTRA_STRING, NULL, str_len);
2289 /* tipc_sk_reinit: set non-zero address in all existing sockets
2290 * when we go from standalone to network mode.
2292 void tipc_sk_reinit(void)
2294 struct tipc_msg *msg;
2296 struct tipc_sock *tsk = tipc_sk_get_next(&ref);
2298 for (; tsk; tsk = tipc_sk_get_next(&ref)) {
2299 lock_sock(&tsk->sk);
2301 msg_set_prevnode(msg, tipc_own_addr);
2302 msg_set_orignode(msg, tipc_own_addr);
2303 release_sock(&tsk->sk);
2309 * struct reference - TIPC socket reference entry
2310 * @tsk: pointer to socket associated with reference entry
2311 * @ref: reference value for socket (combines instance & array index info)
2314 struct tipc_sock *tsk;
2319 * struct tipc_ref_table - table of TIPC socket reference entries
2320 * @entries: pointer to array of reference entries
2321 * @capacity: array index of first unusable entry
2322 * @init_point: array index of first uninitialized entry
2323 * @first_free: array index of first unused socket reference entry
2324 * @last_free: array index of last unused socket reference entry
2325 * @index_mask: bitmask for array index portion of reference values
2326 * @start_mask: initial value for instance value portion of reference values
2329 struct reference *entries;
2338 /* Socket reference table consists of 2**N entries.
2340 * State Socket ptr Reference
2341 * ----- ---------- ---------
2342 * In use non-NULL XXXX|own index
2343 * (XXXX changes each time entry is acquired)
2344 * Free NULL YYYY|next free index
2345 * (YYYY is one more than last used XXXX)
2346 * Uninitialized NULL 0
2348 * Entry 0 is not used; this allows index 0 to denote the end of the free list.
2350 * Note that a reference value of 0 does not necessarily indicate that an
2351 * entry is uninitialized, since the last entry in the free list could also
2352 * have a reference value of 0 (although this is unlikely).
2355 static struct ref_table tipc_ref_table;
2357 static DEFINE_RWLOCK(ref_table_lock);
2360 * tipc_ref_table_init - create reference table for sockets
2362 int tipc_sk_ref_table_init(u32 req_sz, u32 start)
2364 struct reference *table;
2367 /* account for unused entry, then round up size to a power of 2 */
2370 for (actual_sz = 16; actual_sz < req_sz; actual_sz <<= 1) {
2374 /* allocate table & mark all entries as uninitialized */
2375 table = vzalloc(actual_sz * sizeof(struct reference));
2379 tipc_ref_table.entries = table;
2380 tipc_ref_table.capacity = req_sz;
2381 tipc_ref_table.init_point = 1;
2382 tipc_ref_table.first_free = 0;
2383 tipc_ref_table.last_free = 0;
2384 tipc_ref_table.index_mask = actual_sz - 1;
2385 tipc_ref_table.start_mask = start & ~tipc_ref_table.index_mask;
2391 * tipc_ref_table_stop - destroy reference table for sockets
2393 void tipc_sk_ref_table_stop(void)
2395 if (!tipc_ref_table.entries)
2397 vfree(tipc_ref_table.entries);
2398 tipc_ref_table.entries = NULL;
2401 /* tipc_ref_acquire - create reference to a socket
2403 * Register an socket pointer in the reference table.
2404 * Returns a unique reference value that is used from then on to retrieve the
2405 * socket pointer, or to determine if the socket has been deregistered.
2407 u32 tipc_sk_ref_acquire(struct tipc_sock *tsk)
2411 u32 next_plus_upper;
2413 struct reference *entry;
2415 if (unlikely(!tsk)) {
2416 pr_err("Attempt to acquire ref. to non-existent obj\n");
2419 if (unlikely(!tipc_ref_table.entries)) {
2420 pr_err("Ref. table not found in acquisition attempt\n");
2424 /* Take a free entry, if available; otherwise initialize a new one */
2425 write_lock_bh(&ref_table_lock);
2426 index = tipc_ref_table.first_free;
2427 entry = &tipc_ref_table.entries[index];
2429 if (likely(index)) {
2430 index = tipc_ref_table.first_free;
2431 entry = &tipc_ref_table.entries[index];
2432 index_mask = tipc_ref_table.index_mask;
2433 next_plus_upper = entry->ref;
2434 tipc_ref_table.first_free = next_plus_upper & index_mask;
2435 ref = (next_plus_upper & ~index_mask) + index;
2437 } else if (tipc_ref_table.init_point < tipc_ref_table.capacity) {
2438 index = tipc_ref_table.init_point++;
2439 entry = &tipc_ref_table.entries[index];
2440 ref = tipc_ref_table.start_mask + index;
2447 write_unlock_bh(&ref_table_lock);
2451 /* tipc_sk_ref_discard - invalidate reference to an socket
2453 * Disallow future references to an socket and free up the entry for re-use.
2455 void tipc_sk_ref_discard(u32 ref)
2457 struct reference *entry;
2461 if (unlikely(!tipc_ref_table.entries)) {
2462 pr_err("Ref. table not found during discard attempt\n");
2466 index_mask = tipc_ref_table.index_mask;
2467 index = ref & index_mask;
2468 entry = &tipc_ref_table.entries[index];
2470 write_lock_bh(&ref_table_lock);
2472 if (unlikely(!entry->tsk)) {
2473 pr_err("Attempt to discard ref. to non-existent socket\n");
2476 if (unlikely(entry->ref != ref)) {
2477 pr_err("Attempt to discard non-existent reference\n");
2481 /* Mark entry as unused; increment instance part of entry's
2482 * reference to invalidate any subsequent references
2486 entry->ref = (ref & ~index_mask) + (index_mask + 1);
2488 /* Append entry to free entry list */
2489 if (unlikely(tipc_ref_table.first_free == 0))
2490 tipc_ref_table.first_free = index;
2492 tipc_ref_table.entries[tipc_ref_table.last_free].ref |= index;
2493 tipc_ref_table.last_free = index;
2495 write_unlock_bh(&ref_table_lock);
2498 /* tipc_sk_get - find referenced socket and return pointer to it
2500 struct tipc_sock *tipc_sk_get(u32 ref)
2502 struct reference *entry;
2503 struct tipc_sock *tsk;
2505 if (unlikely(!tipc_ref_table.entries))
2507 read_lock_bh(&ref_table_lock);
2508 entry = &tipc_ref_table.entries[ref & tipc_ref_table.index_mask];
2510 if (likely(tsk && (entry->ref == ref)))
2511 sock_hold(&tsk->sk);
2514 read_unlock_bh(&ref_table_lock);
2518 /* tipc_sk_get_next - lock & return next socket after referenced one
2520 struct tipc_sock *tipc_sk_get_next(u32 *ref)
2522 struct reference *entry;
2523 struct tipc_sock *tsk = NULL;
2524 uint index = *ref & tipc_ref_table.index_mask;
2526 read_lock_bh(&ref_table_lock);
2527 while (++index < tipc_ref_table.capacity) {
2528 entry = &tipc_ref_table.entries[index];
2532 sock_hold(&tsk->sk);
2536 read_unlock_bh(&ref_table_lock);
2540 static void tipc_sk_put(struct tipc_sock *tsk)
2546 * tipc_setsockopt - set socket option
2547 * @sock: socket structure
2548 * @lvl: option level
2549 * @opt: option identifier
2550 * @ov: pointer to new option value
2551 * @ol: length of option value
2553 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
2554 * (to ease compatibility).
2556 * Returns 0 on success, errno otherwise
2558 static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
2559 char __user *ov, unsigned int ol)
2561 struct sock *sk = sock->sk;
2562 struct tipc_sock *tsk = tipc_sk(sk);
2566 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
2568 if (lvl != SOL_TIPC)
2569 return -ENOPROTOOPT;
2570 if (ol < sizeof(value))
2572 res = get_user(value, (u32 __user *)ov);
2579 case TIPC_IMPORTANCE:
2580 res = tsk_set_importance(tsk, value);
2582 case TIPC_SRC_DROPPABLE:
2583 if (sock->type != SOCK_STREAM)
2584 tsk_set_unreliable(tsk, value);
2588 case TIPC_DEST_DROPPABLE:
2589 tsk_set_unreturnable(tsk, value);
2591 case TIPC_CONN_TIMEOUT:
2592 tipc_sk(sk)->conn_timeout = value;
2593 /* no need to set "res", since already 0 at this point */
2605 * tipc_getsockopt - get socket option
2606 * @sock: socket structure
2607 * @lvl: option level
2608 * @opt: option identifier
2609 * @ov: receptacle for option value
2610 * @ol: receptacle for length of option value
2612 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
2613 * (to ease compatibility).
2615 * Returns 0 on success, errno otherwise
2617 static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
2618 char __user *ov, int __user *ol)
2620 struct sock *sk = sock->sk;
2621 struct tipc_sock *tsk = tipc_sk(sk);
2626 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
2627 return put_user(0, ol);
2628 if (lvl != SOL_TIPC)
2629 return -ENOPROTOOPT;
2630 res = get_user(len, ol);
2637 case TIPC_IMPORTANCE:
2638 value = tsk_importance(tsk);
2640 case TIPC_SRC_DROPPABLE:
2641 value = tsk_unreliable(tsk);
2643 case TIPC_DEST_DROPPABLE:
2644 value = tsk_unreturnable(tsk);
2646 case TIPC_CONN_TIMEOUT:
2647 value = tsk->conn_timeout;
2648 /* no need to set "res", since already 0 at this point */
2650 case TIPC_NODE_RECVQ_DEPTH:
2651 value = 0; /* was tipc_queue_size, now obsolete */
2653 case TIPC_SOCK_RECVQ_DEPTH:
2654 value = skb_queue_len(&sk->sk_receive_queue);
2663 return res; /* "get" failed */
2665 if (len < sizeof(value))
2668 if (copy_to_user(ov, &value, sizeof(value)))
2671 return put_user(sizeof(value), ol);
2674 static int tipc_ioctl(struct socket *sk, unsigned int cmd, unsigned long arg)
2676 struct tipc_sioc_ln_req lnr;
2677 void __user *argp = (void __user *)arg;
2680 case SIOCGETLINKNAME:
2681 if (copy_from_user(&lnr, argp, sizeof(lnr)))
2683 if (!tipc_node_get_linkname(lnr.bearer_id & 0xffff, lnr.peer,
2684 lnr.linkname, TIPC_MAX_LINK_NAME)) {
2685 if (copy_to_user(argp, &lnr, sizeof(lnr)))
2689 return -EADDRNOTAVAIL;
2691 return -ENOIOCTLCMD;
2695 /* Protocol switches for the various types of TIPC sockets */
2697 static const struct proto_ops msg_ops = {
2698 .owner = THIS_MODULE,
2700 .release = tipc_release,
2702 .connect = tipc_connect,
2703 .socketpair = sock_no_socketpair,
2704 .accept = sock_no_accept,
2705 .getname = tipc_getname,
2707 .ioctl = tipc_ioctl,
2708 .listen = sock_no_listen,
2709 .shutdown = tipc_shutdown,
2710 .setsockopt = tipc_setsockopt,
2711 .getsockopt = tipc_getsockopt,
2712 .sendmsg = tipc_sendmsg,
2713 .recvmsg = tipc_recvmsg,
2714 .mmap = sock_no_mmap,
2715 .sendpage = sock_no_sendpage
2718 static const struct proto_ops packet_ops = {
2719 .owner = THIS_MODULE,
2721 .release = tipc_release,
2723 .connect = tipc_connect,
2724 .socketpair = sock_no_socketpair,
2725 .accept = tipc_accept,
2726 .getname = tipc_getname,
2728 .ioctl = tipc_ioctl,
2729 .listen = tipc_listen,
2730 .shutdown = tipc_shutdown,
2731 .setsockopt = tipc_setsockopt,
2732 .getsockopt = tipc_getsockopt,
2733 .sendmsg = tipc_send_packet,
2734 .recvmsg = tipc_recvmsg,
2735 .mmap = sock_no_mmap,
2736 .sendpage = sock_no_sendpage
2739 static const struct proto_ops stream_ops = {
2740 .owner = THIS_MODULE,
2742 .release = tipc_release,
2744 .connect = tipc_connect,
2745 .socketpair = sock_no_socketpair,
2746 .accept = tipc_accept,
2747 .getname = tipc_getname,
2749 .ioctl = tipc_ioctl,
2750 .listen = tipc_listen,
2751 .shutdown = tipc_shutdown,
2752 .setsockopt = tipc_setsockopt,
2753 .getsockopt = tipc_getsockopt,
2754 .sendmsg = tipc_send_stream,
2755 .recvmsg = tipc_recv_stream,
2756 .mmap = sock_no_mmap,
2757 .sendpage = sock_no_sendpage
2760 static const struct net_proto_family tipc_family_ops = {
2761 .owner = THIS_MODULE,
2763 .create = tipc_sk_create
2766 static struct proto tipc_proto = {
2768 .owner = THIS_MODULE,
2769 .obj_size = sizeof(struct tipc_sock),
2770 .sysctl_rmem = sysctl_tipc_rmem
2773 static struct proto tipc_proto_kern = {
2775 .obj_size = sizeof(struct tipc_sock),
2776 .sysctl_rmem = sysctl_tipc_rmem
2780 * tipc_socket_init - initialize TIPC socket interface
2782 * Returns 0 on success, errno otherwise
2784 int tipc_socket_init(void)
2788 res = proto_register(&tipc_proto, 1);
2790 pr_err("Failed to register TIPC protocol type\n");
2794 res = sock_register(&tipc_family_ops);
2796 pr_err("Failed to register TIPC socket type\n");
2797 proto_unregister(&tipc_proto);
2805 * tipc_socket_stop - stop TIPC socket interface
2807 void tipc_socket_stop(void)
2809 sock_unregister(tipc_family_ops.family);
2810 proto_unregister(&tipc_proto);
2813 /* Caller should hold socket lock for the passed tipc socket. */
2814 int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk)
2818 struct nlattr *nest;
2820 peer_node = tsk_peer_node(tsk);
2821 peer_port = tsk_peer_port(tsk);
2823 nest = nla_nest_start(skb, TIPC_NLA_SOCK_CON);
2825 if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node))
2827 if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port))
2830 if (tsk->conn_type != 0) {
2831 if (nla_put_flag(skb, TIPC_NLA_CON_FLAG))
2833 if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, tsk->conn_type))
2835 if (nla_put_u32(skb, TIPC_NLA_CON_INST, tsk->conn_instance))
2838 nla_nest_end(skb, nest);
2843 nla_nest_cancel(skb, nest);
2848 /* Caller should hold socket lock for the passed tipc socket. */
2849 int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb,
2850 struct tipc_sock *tsk)
2854 struct nlattr *attrs;
2856 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
2857 &tipc_genl_v2_family, NLM_F_MULTI, TIPC_NL_SOCK_GET);
2861 attrs = nla_nest_start(skb, TIPC_NLA_SOCK);
2863 goto genlmsg_cancel;
2864 if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->ref))
2865 goto attr_msg_cancel;
2866 if (nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tipc_own_addr))
2867 goto attr_msg_cancel;
2869 if (tsk->connected) {
2870 err = __tipc_nl_add_sk_con(skb, tsk);
2872 goto attr_msg_cancel;
2873 } else if (!list_empty(&tsk->publications)) {
2874 if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL))
2875 goto attr_msg_cancel;
2877 nla_nest_end(skb, attrs);
2878 genlmsg_end(skb, hdr);
2883 nla_nest_cancel(skb, attrs);
2885 genlmsg_cancel(skb, hdr);
2890 int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb)
2893 struct tipc_sock *tsk;
2894 u32 prev_ref = cb->args[0];
2897 tsk = tipc_sk_get_next(&ref);
2898 for (; tsk; tsk = tipc_sk_get_next(&ref)) {
2899 lock_sock(&tsk->sk);
2900 err = __tipc_nl_add_sk(skb, cb, tsk);
2901 release_sock(&tsk->sk);
2909 cb->args[0] = prev_ref;
2914 /* Caller should hold socket lock for the passed tipc socket. */
2915 int __tipc_nl_add_sk_publ(struct sk_buff *skb, struct netlink_callback *cb,
2916 struct publication *publ)
2919 struct nlattr *attrs;
2921 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
2922 &tipc_genl_v2_family, NLM_F_MULTI, TIPC_NL_PUBL_GET);
2926 attrs = nla_nest_start(skb, TIPC_NLA_PUBL);
2928 goto genlmsg_cancel;
2930 if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key))
2931 goto attr_msg_cancel;
2932 if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->type))
2933 goto attr_msg_cancel;
2934 if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->lower))
2935 goto attr_msg_cancel;
2936 if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->upper))
2937 goto attr_msg_cancel;
2939 nla_nest_end(skb, attrs);
2940 genlmsg_end(skb, hdr);
2945 nla_nest_cancel(skb, attrs);
2947 genlmsg_cancel(skb, hdr);
2952 /* Caller should hold socket lock for the passed tipc socket. */
2953 int __tipc_nl_list_sk_publ(struct sk_buff *skb, struct netlink_callback *cb,
2954 struct tipc_sock *tsk, u32 *last_publ)
2957 struct publication *p;
2960 list_for_each_entry(p, &tsk->publications, pport_list) {
2961 if (p->key == *last_publ)
2964 if (p->key != *last_publ) {
2965 /* We never set seq or call nl_dump_check_consistent()
2966 * this means that setting prev_seq here will cause the
2967 * consistence check to fail in the netlink callback
2968 * handler. Resulting in the last NLMSG_DONE message
2969 * having the NLM_F_DUMP_INTR flag set.
2976 p = list_first_entry(&tsk->publications, struct publication,
2980 list_for_each_entry_from(p, &tsk->publications, pport_list) {
2981 err = __tipc_nl_add_sk_publ(skb, cb, p);
2983 *last_publ = p->key;
2992 int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)
2995 u32 tsk_ref = cb->args[0];
2996 u32 last_publ = cb->args[1];
2997 u32 done = cb->args[2];
2998 struct tipc_sock *tsk;
3001 struct nlattr **attrs;
3002 struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1];
3004 err = tipc_nlmsg_parse(cb->nlh, &attrs);
3008 err = nla_parse_nested(sock, TIPC_NLA_SOCK_MAX,
3009 attrs[TIPC_NLA_SOCK],
3010 tipc_nl_sock_policy);
3014 if (!sock[TIPC_NLA_SOCK_REF])
3017 tsk_ref = nla_get_u32(sock[TIPC_NLA_SOCK_REF]);
3023 tsk = tipc_sk_get(tsk_ref);
3027 lock_sock(&tsk->sk);
3028 err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ);
3031 release_sock(&tsk->sk);
3034 cb->args[0] = tsk_ref;
3035 cb->args[1] = last_publ;