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.
41 #include <linux/export.h>
43 #define SS_LISTENING -1 /* socket is listening */
44 #define SS_READY -2 /* socket is connectionless */
46 #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */
48 static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *skb);
49 static void tipc_data_ready(struct sock *sk);
50 static void tipc_write_space(struct sock *sk);
51 static int tipc_release(struct socket *sock);
52 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags);
54 static const struct proto_ops packet_ops;
55 static const struct proto_ops stream_ops;
56 static const struct proto_ops msg_ops;
58 static struct proto tipc_proto;
59 static struct proto tipc_proto_kern;
62 * Revised TIPC socket locking policy:
64 * Most socket operations take the standard socket lock when they start
65 * and hold it until they finish (or until they need to sleep). Acquiring
66 * this lock grants the owner exclusive access to the fields of the socket
67 * data structures, with the exception of the backlog queue. A few socket
68 * operations can be done without taking the socket lock because they only
69 * read socket information that never changes during the life of the socket.
71 * Socket operations may acquire the lock for the associated TIPC port if they
72 * need to perform an operation on the port. If any routine needs to acquire
73 * both the socket lock and the port lock it must take the socket lock first
74 * to avoid the risk of deadlock.
76 * The dispatcher handling incoming messages cannot grab the socket lock in
77 * the standard fashion, since invoked it runs at the BH level and cannot block.
78 * Instead, it checks to see if the socket lock is currently owned by someone,
79 * and either handles the message itself or adds it to the socket's backlog
80 * queue; in the latter case the queued message is processed once the process
81 * owning the socket lock releases it.
83 * NOTE: Releasing the socket lock while an operation is sleeping overcomes
84 * the problem of a blocked socket operation preventing any other operations
85 * from occurring. However, applications must be careful if they have
86 * multiple threads trying to send (or receive) on the same socket, as these
87 * operations might interfere with each other. For example, doing a connect
88 * and a receive at the same time might allow the receive to consume the
89 * ACK message meant for the connect. While additional work could be done
90 * to try and overcome this, it doesn't seem to be worthwhile at the present.
92 * NOTE: Releasing the socket lock while an operation is sleeping also ensures
93 * that another operation that must be performed in a non-blocking manner is
94 * not delayed for very long because the lock has already been taken.
96 * NOTE: This code assumes that certain fields of a port/socket pair are
97 * constant over its lifetime; such fields can be examined without taking
98 * the socket lock and/or port lock, and do not need to be re-read even
99 * after resuming processing after waiting. These fields include:
101 * - pointer to socket sk structure (aka tipc_sock structure)
102 * - pointer to port structure
109 * advance_rx_queue - discard first buffer in socket receive queue
111 * Caller must hold socket lock
113 static void advance_rx_queue(struct sock *sk)
115 kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
119 * reject_rx_queue - reject all buffers in socket receive queue
121 * Caller must hold socket lock
123 static void reject_rx_queue(struct sock *sk)
127 while ((buf = __skb_dequeue(&sk->sk_receive_queue)))
128 tipc_reject_msg(buf, TIPC_ERR_NO_PORT);
132 * tipc_sk_create - create a TIPC socket
133 * @net: network namespace (must be default network)
134 * @sock: pre-allocated socket structure
135 * @protocol: protocol indicator (must be 0)
136 * @kern: caused by kernel or by userspace?
138 * This routine creates additional data structures used by the TIPC socket,
139 * initializes them, and links them together.
141 * Returns 0 on success, errno otherwise
143 static int tipc_sk_create(struct net *net, struct socket *sock,
144 int protocol, int kern)
146 const struct proto_ops *ops;
149 struct tipc_sock *tsk;
150 struct tipc_port *port;
153 /* Validate arguments */
154 if (unlikely(protocol != 0))
155 return -EPROTONOSUPPORT;
157 switch (sock->type) {
160 state = SS_UNCONNECTED;
164 state = SS_UNCONNECTED;
175 /* Allocate socket's protocol area */
177 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto);
179 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto_kern);
187 ref = tipc_port_init(port, TIPC_LOW_IMPORTANCE);
189 pr_warn("Socket registration failed, ref. table exhausted\n");
194 /* Finish initializing socket data structures */
198 sock_init_data(sock, sk);
199 sk->sk_backlog_rcv = tipc_backlog_rcv;
200 sk->sk_rcvbuf = sysctl_tipc_rmem[1];
201 sk->sk_data_ready = tipc_data_ready;
202 sk->sk_write_space = tipc_write_space;
203 tsk->conn_timeout = CONN_TIMEOUT_DEFAULT;
204 atomic_set(&tsk->dupl_rcvcnt, 0);
205 tipc_port_unlock(port);
207 if (sock->state == SS_READY) {
208 tipc_port_set_unreturnable(port, true);
209 if (sock->type == SOCK_DGRAM)
210 tipc_port_set_unreliable(port, true);
216 * tipc_sock_create_local - create TIPC socket from inside TIPC module
217 * @type: socket type - SOCK_RDM or SOCK_SEQPACKET
219 * We cannot use sock_creat_kern here because it bumps module user count.
220 * Since socket owner and creator is the same module we must make sure
221 * that module count remains zero for module local sockets, otherwise
222 * we cannot do rmmod.
224 * Returns 0 on success, errno otherwise
226 int tipc_sock_create_local(int type, struct socket **res)
230 rc = sock_create_lite(AF_TIPC, type, 0, res);
232 pr_err("Failed to create kernel socket\n");
235 tipc_sk_create(&init_net, *res, 0, 1);
241 * tipc_sock_release_local - release socket created by tipc_sock_create_local
242 * @sock: the socket to be released.
244 * Module reference count is not incremented when such sockets are created,
245 * so we must keep it from being decremented when they are released.
247 void tipc_sock_release_local(struct socket *sock)
255 * tipc_sock_accept_local - accept a connection on a socket created
256 * with tipc_sock_create_local. Use this function to avoid that
257 * module reference count is inadvertently incremented.
259 * @sock: the accepting socket
260 * @newsock: reference to the new socket to be created
261 * @flags: socket flags
264 int tipc_sock_accept_local(struct socket *sock, struct socket **newsock,
267 struct sock *sk = sock->sk;
270 ret = sock_create_lite(sk->sk_family, sk->sk_type,
271 sk->sk_protocol, newsock);
275 ret = tipc_accept(sock, *newsock, flags);
277 sock_release(*newsock);
280 (*newsock)->ops = sock->ops;
285 * tipc_release - destroy a TIPC socket
286 * @sock: socket to destroy
288 * This routine cleans up any messages that are still queued on the socket.
289 * For DGRAM and RDM socket types, all queued messages are rejected.
290 * For SEQPACKET and STREAM socket types, the first message is rejected
291 * and any others are discarded. (If the first message on a STREAM socket
292 * is partially-read, it is discarded and the next one is rejected instead.)
294 * NOTE: Rejected messages are not necessarily returned to the sender! They
295 * are returned or discarded according to the "destination droppable" setting
296 * specified for the message by the sender.
298 * Returns 0 on success, errno otherwise
300 static int tipc_release(struct socket *sock)
302 struct sock *sk = sock->sk;
303 struct tipc_sock *tsk;
304 struct tipc_port *port;
308 * Exit if socket isn't fully initialized (occurs when a failed accept()
309 * releases a pre-allocated child socket that was never used)
319 * Reject all unreceived messages, except on an active connection
320 * (which disconnects locally & sends a 'FIN+' to peer)
322 while (sock->state != SS_DISCONNECTING) {
323 buf = __skb_dequeue(&sk->sk_receive_queue);
326 if (TIPC_SKB_CB(buf)->handle != NULL)
329 if ((sock->state == SS_CONNECTING) ||
330 (sock->state == SS_CONNECTED)) {
331 sock->state = SS_DISCONNECTING;
332 tipc_port_disconnect(port->ref);
334 tipc_reject_msg(buf, TIPC_ERR_NO_PORT);
338 /* Destroy TIPC port; also disconnects an active connection and
339 * sends a 'FIN-' to peer.
341 tipc_port_destroy(port);
343 /* Discard any remaining (connection-based) messages in receive queue */
344 __skb_queue_purge(&sk->sk_receive_queue);
346 /* Reject any messages that accumulated in backlog queue */
347 sock->state = SS_DISCONNECTING;
357 * tipc_bind - associate or disassocate TIPC name(s) with a socket
358 * @sock: socket structure
359 * @uaddr: socket address describing name(s) and desired operation
360 * @uaddr_len: size of socket address data structure
362 * Name and name sequence binding is indicated using a positive scope value;
363 * a negative scope value unbinds the specified name. Specifying no name
364 * (i.e. a socket address length of 0) unbinds all names from the socket.
366 * Returns 0 on success, errno otherwise
368 * NOTE: This routine doesn't need to take the socket lock since it doesn't
369 * access any non-constant socket information.
371 static int tipc_bind(struct socket *sock, struct sockaddr *uaddr,
374 struct sock *sk = sock->sk;
375 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
376 struct tipc_sock *tsk = tipc_sk(sk);
380 if (unlikely(!uaddr_len)) {
381 res = tipc_withdraw(&tsk->port, 0, NULL);
385 if (uaddr_len < sizeof(struct sockaddr_tipc)) {
389 if (addr->family != AF_TIPC) {
394 if (addr->addrtype == TIPC_ADDR_NAME)
395 addr->addr.nameseq.upper = addr->addr.nameseq.lower;
396 else if (addr->addrtype != TIPC_ADDR_NAMESEQ) {
401 if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) &&
402 (addr->addr.nameseq.type != TIPC_TOP_SRV) &&
403 (addr->addr.nameseq.type != TIPC_CFG_SRV)) {
408 res = (addr->scope > 0) ?
409 tipc_publish(&tsk->port, addr->scope, &addr->addr.nameseq) :
410 tipc_withdraw(&tsk->port, -addr->scope, &addr->addr.nameseq);
417 * tipc_getname - get port ID of socket or peer socket
418 * @sock: socket structure
419 * @uaddr: area for returned socket address
420 * @uaddr_len: area for returned length of socket address
421 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
423 * Returns 0 on success, errno otherwise
425 * NOTE: This routine doesn't need to take the socket lock since it only
426 * accesses socket information that is unchanging (or which changes in
427 * a completely predictable manner).
429 static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
430 int *uaddr_len, int peer)
432 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
433 struct tipc_sock *tsk = tipc_sk(sock->sk);
435 memset(addr, 0, sizeof(*addr));
437 if ((sock->state != SS_CONNECTED) &&
438 ((peer != 2) || (sock->state != SS_DISCONNECTING)))
440 addr->addr.id.ref = tipc_port_peerport(&tsk->port);
441 addr->addr.id.node = tipc_port_peernode(&tsk->port);
443 addr->addr.id.ref = tsk->port.ref;
444 addr->addr.id.node = tipc_own_addr;
447 *uaddr_len = sizeof(*addr);
448 addr->addrtype = TIPC_ADDR_ID;
449 addr->family = AF_TIPC;
451 addr->addr.name.domain = 0;
457 * tipc_poll - read and possibly block on pollmask
458 * @file: file structure associated with the socket
459 * @sock: socket for which to calculate the poll bits
462 * Returns pollmask value
465 * It appears that the usual socket locking mechanisms are not useful here
466 * since the pollmask info is potentially out-of-date the moment this routine
467 * exits. TCP and other protocols seem to rely on higher level poll routines
468 * to handle any preventable race conditions, so TIPC will do the same ...
470 * TIPC sets the returned events as follows:
472 * socket state flags set
473 * ------------ ---------
474 * unconnected no read flags
475 * POLLOUT if port is not congested
477 * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue
480 * connected POLLIN/POLLRDNORM if data in rx queue
481 * POLLOUT if port is not congested
483 * disconnecting POLLIN/POLLRDNORM/POLLHUP
486 * listening POLLIN if SYN in rx queue
489 * ready POLLIN/POLLRDNORM if data in rx queue
490 * [connectionless] POLLOUT (since port cannot be congested)
492 * IMPORTANT: The fact that a read or write operation is indicated does NOT
493 * imply that the operation will succeed, merely that it should be performed
494 * and will not block.
496 static unsigned int tipc_poll(struct file *file, struct socket *sock,
499 struct sock *sk = sock->sk;
500 struct tipc_sock *tsk = tipc_sk(sk);
503 sock_poll_wait(file, sk_sleep(sk), wait);
505 switch ((int)sock->state) {
507 if (!tsk->port.congested)
512 if (!tsk->port.congested)
517 if (!skb_queue_empty(&sk->sk_receive_queue))
518 mask |= (POLLIN | POLLRDNORM);
520 case SS_DISCONNECTING:
521 mask = (POLLIN | POLLRDNORM | POLLHUP);
529 * dest_name_check - verify user is permitted to send to specified port name
530 * @dest: destination address
531 * @m: descriptor for message to be sent
533 * Prevents restricted configuration commands from being issued by
534 * unauthorized users.
536 * Returns 0 if permission is granted, otherwise errno
538 static int dest_name_check(struct sockaddr_tipc *dest, struct msghdr *m)
540 struct tipc_cfg_msg_hdr hdr;
542 if (likely(dest->addr.name.name.type >= TIPC_RESERVED_TYPES))
544 if (likely(dest->addr.name.name.type == TIPC_TOP_SRV))
546 if (likely(dest->addr.name.name.type != TIPC_CFG_SRV))
549 if (!m->msg_iovlen || (m->msg_iov[0].iov_len < sizeof(hdr)))
551 if (copy_from_user(&hdr, m->msg_iov[0].iov_base, sizeof(hdr)))
553 if ((ntohs(hdr.tcm_type) & 0xC000) && (!capable(CAP_NET_ADMIN)))
559 static int tipc_wait_for_sndmsg(struct socket *sock, long *timeo_p)
561 struct sock *sk = sock->sk;
562 struct tipc_sock *tsk = tipc_sk(sk);
567 int err = sock_error(sk);
570 if (sock->state == SS_DISCONNECTING)
574 if (signal_pending(current))
575 return sock_intr_errno(*timeo_p);
577 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
578 done = sk_wait_event(sk, timeo_p, !tsk->port.congested);
579 finish_wait(sk_sleep(sk), &wait);
586 * tipc_sendmsg - send message in connectionless manner
587 * @iocb: if NULL, indicates that socket lock is already held
588 * @sock: socket structure
589 * @m: message to send
590 * @total_len: length of message
592 * Message must have an destination specified explicitly.
593 * Used for SOCK_RDM and SOCK_DGRAM messages,
594 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
595 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
597 * Returns the number of bytes sent on success, or errno otherwise
599 static int tipc_sendmsg(struct kiocb *iocb, struct socket *sock,
600 struct msghdr *m, size_t total_len)
602 struct sock *sk = sock->sk;
603 struct tipc_sock *tsk = tipc_sk(sk);
604 struct tipc_port *port = &tsk->port;
605 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
611 return -EDESTADDRREQ;
612 if (unlikely((m->msg_namelen < sizeof(*dest)) ||
613 (dest->family != AF_TIPC)))
615 if (total_len > TIPC_MAX_USER_MSG_SIZE)
621 needs_conn = (sock->state != SS_READY);
622 if (unlikely(needs_conn)) {
623 if (sock->state == SS_LISTENING) {
627 if (sock->state != SS_UNCONNECTED) {
631 if (tsk->port.published) {
635 if (dest->addrtype == TIPC_ADDR_NAME) {
636 tsk->port.conn_type = dest->addr.name.name.type;
637 tsk->port.conn_instance = dest->addr.name.name.instance;
640 /* Abort any pending connection attempts (very unlikely) */
644 timeo = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
646 if (dest->addrtype == TIPC_ADDR_NAME) {
647 res = dest_name_check(dest, m);
650 res = tipc_send2name(port,
651 &dest->addr.name.name,
652 dest->addr.name.domain,
655 } else if (dest->addrtype == TIPC_ADDR_ID) {
656 res = tipc_send2port(port,
660 } else if (dest->addrtype == TIPC_ADDR_MCAST) {
665 res = dest_name_check(dest, m);
668 res = tipc_port_mcast_xmit(port,
673 if (likely(res != -ELINKCONG)) {
674 if (needs_conn && (res >= 0))
675 sock->state = SS_CONNECTING;
678 res = tipc_wait_for_sndmsg(sock, &timeo);
689 static int tipc_wait_for_sndpkt(struct socket *sock, long *timeo_p)
691 struct sock *sk = sock->sk;
692 struct tipc_sock *tsk = tipc_sk(sk);
693 struct tipc_port *port = &tsk->port;
698 int err = sock_error(sk);
701 if (sock->state == SS_DISCONNECTING)
703 else if (sock->state != SS_CONNECTED)
707 if (signal_pending(current))
708 return sock_intr_errno(*timeo_p);
710 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
711 done = sk_wait_event(sk, timeo_p,
712 (!port->congested || !port->connected));
713 finish_wait(sk_sleep(sk), &wait);
719 * tipc_send_packet - send a connection-oriented message
720 * @iocb: if NULL, indicates that socket lock is already held
721 * @sock: socket structure
722 * @m: message to send
723 * @total_len: length of message
725 * Used for SOCK_SEQPACKET messages and SOCK_STREAM data.
727 * Returns the number of bytes sent on success, or errno otherwise
729 static int tipc_send_packet(struct kiocb *iocb, struct socket *sock,
730 struct msghdr *m, size_t total_len)
732 struct sock *sk = sock->sk;
733 struct tipc_sock *tsk = tipc_sk(sk);
734 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
738 /* Handle implied connection establishment */
740 return tipc_sendmsg(iocb, sock, m, total_len);
742 if (total_len > TIPC_MAX_USER_MSG_SIZE)
748 if (unlikely(sock->state != SS_CONNECTED)) {
749 if (sock->state == SS_DISCONNECTING)
756 timeo = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
758 res = tipc_send(&tsk->port, m->msg_iov, total_len);
759 if (likely(res != -ELINKCONG))
761 res = tipc_wait_for_sndpkt(sock, &timeo);
772 * tipc_send_stream - send stream-oriented data
774 * @sock: socket structure
776 * @total_len: total length of data to be sent
778 * Used for SOCK_STREAM data.
780 * Returns the number of bytes sent on success (or partial success),
781 * or errno if no data sent
783 static int tipc_send_stream(struct kiocb *iocb, struct socket *sock,
784 struct msghdr *m, size_t total_len)
786 struct sock *sk = sock->sk;
787 struct tipc_sock *tsk = tipc_sk(sk);
788 struct msghdr my_msg;
790 struct iovec *curr_iov;
792 char __user *curr_start;
801 /* Handle special cases where there is no connection */
802 if (unlikely(sock->state != SS_CONNECTED)) {
803 if (sock->state == SS_UNCONNECTED)
804 res = tipc_send_packet(NULL, sock, m, total_len);
806 res = sock->state == SS_DISCONNECTING ? -EPIPE : -ENOTCONN;
810 if (unlikely(m->msg_name)) {
815 if (total_len > (unsigned int)INT_MAX) {
821 * Send each iovec entry using one or more messages
823 * Note: This algorithm is good for the most likely case
824 * (i.e. one large iovec entry), but could be improved to pass sets
825 * of small iovec entries into send_packet().
827 curr_iov = m->msg_iov;
828 curr_iovlen = m->msg_iovlen;
829 my_msg.msg_iov = &my_iov;
830 my_msg.msg_iovlen = 1;
831 my_msg.msg_flags = m->msg_flags;
832 my_msg.msg_name = NULL;
835 hdr_size = msg_hdr_sz(&tsk->port.phdr);
837 while (curr_iovlen--) {
838 curr_start = curr_iov->iov_base;
839 curr_left = curr_iov->iov_len;
842 bytes_to_send = tsk->port.max_pkt - hdr_size;
843 if (bytes_to_send > TIPC_MAX_USER_MSG_SIZE)
844 bytes_to_send = TIPC_MAX_USER_MSG_SIZE;
845 if (curr_left < bytes_to_send)
846 bytes_to_send = curr_left;
847 my_iov.iov_base = curr_start;
848 my_iov.iov_len = bytes_to_send;
849 res = tipc_send_packet(NULL, sock, &my_msg,
856 curr_left -= bytes_to_send;
857 curr_start += bytes_to_send;
858 bytes_sent += bytes_to_send;
870 * auto_connect - complete connection setup to a remote port
871 * @tsk: tipc socket structure
872 * @msg: peer's response message
874 * Returns 0 on success, errno otherwise
876 static int auto_connect(struct tipc_sock *tsk, struct tipc_msg *msg)
878 struct tipc_port *port = &tsk->port;
879 struct socket *sock = tsk->sk.sk_socket;
880 struct tipc_portid peer;
882 peer.ref = msg_origport(msg);
883 peer.node = msg_orignode(msg);
885 __tipc_port_connect(port->ref, port, &peer);
887 if (msg_importance(msg) > TIPC_CRITICAL_IMPORTANCE)
889 msg_set_importance(&port->phdr, (u32)msg_importance(msg));
890 sock->state = SS_CONNECTED;
895 * set_orig_addr - capture sender's address for received message
896 * @m: descriptor for message info
897 * @msg: received message header
899 * Note: Address is not captured if not requested by receiver.
901 static void set_orig_addr(struct msghdr *m, struct tipc_msg *msg)
903 DECLARE_SOCKADDR(struct sockaddr_tipc *, addr, m->msg_name);
906 addr->family = AF_TIPC;
907 addr->addrtype = TIPC_ADDR_ID;
908 memset(&addr->addr, 0, sizeof(addr->addr));
909 addr->addr.id.ref = msg_origport(msg);
910 addr->addr.id.node = msg_orignode(msg);
911 addr->addr.name.domain = 0; /* could leave uninitialized */
912 addr->scope = 0; /* could leave uninitialized */
913 m->msg_namelen = sizeof(struct sockaddr_tipc);
918 * anc_data_recv - optionally capture ancillary data for received message
919 * @m: descriptor for message info
920 * @msg: received message header
921 * @tport: TIPC port associated with message
923 * Note: Ancillary data is not captured if not requested by receiver.
925 * Returns 0 if successful, otherwise errno
927 static int anc_data_recv(struct msghdr *m, struct tipc_msg *msg,
928 struct tipc_port *tport)
936 if (likely(m->msg_controllen == 0))
939 /* Optionally capture errored message object(s) */
940 err = msg ? msg_errcode(msg) : 0;
943 anc_data[1] = msg_data_sz(msg);
944 res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
948 res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
955 /* Optionally capture message destination object */
956 dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
960 anc_data[0] = msg_nametype(msg);
961 anc_data[1] = msg_namelower(msg);
962 anc_data[2] = msg_namelower(msg);
966 anc_data[0] = msg_nametype(msg);
967 anc_data[1] = msg_namelower(msg);
968 anc_data[2] = msg_nameupper(msg);
971 has_name = (tport->conn_type != 0);
972 anc_data[0] = tport->conn_type;
973 anc_data[1] = tport->conn_instance;
974 anc_data[2] = tport->conn_instance;
980 res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
988 static int tipc_wait_for_rcvmsg(struct socket *sock, long timeo)
990 struct sock *sk = sock->sk;
995 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
996 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
997 if (sock->state == SS_DISCONNECTING) {
1002 timeo = schedule_timeout(timeo);
1006 if (!skb_queue_empty(&sk->sk_receive_queue))
1008 err = sock_intr_errno(timeo);
1009 if (signal_pending(current))
1015 finish_wait(sk_sleep(sk), &wait);
1020 * tipc_recvmsg - receive packet-oriented message
1022 * @m: descriptor for message info
1023 * @buf_len: total size of user buffer area
1024 * @flags: receive flags
1026 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1027 * If the complete message doesn't fit in user area, truncate it.
1029 * Returns size of returned message data, errno otherwise
1031 static int tipc_recvmsg(struct kiocb *iocb, struct socket *sock,
1032 struct msghdr *m, size_t buf_len, int flags)
1034 struct sock *sk = sock->sk;
1035 struct tipc_sock *tsk = tipc_sk(sk);
1036 struct tipc_port *port = &tsk->port;
1037 struct sk_buff *buf;
1038 struct tipc_msg *msg;
1044 /* Catch invalid receive requests */
1045 if (unlikely(!buf_len))
1050 if (unlikely(sock->state == SS_UNCONNECTED)) {
1055 timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1058 /* Look for a message in receive queue; wait if necessary */
1059 res = tipc_wait_for_rcvmsg(sock, timeo);
1063 /* Look at first message in receive queue */
1064 buf = skb_peek(&sk->sk_receive_queue);
1066 sz = msg_data_sz(msg);
1067 err = msg_errcode(msg);
1069 /* Discard an empty non-errored message & try again */
1070 if ((!sz) && (!err)) {
1071 advance_rx_queue(sk);
1075 /* Capture sender's address (optional) */
1076 set_orig_addr(m, msg);
1078 /* Capture ancillary data (optional) */
1079 res = anc_data_recv(m, msg, port);
1083 /* Capture message data (if valid) & compute return value (always) */
1085 if (unlikely(buf_len < sz)) {
1087 m->msg_flags |= MSG_TRUNC;
1089 res = skb_copy_datagram_iovec(buf, msg_hdr_sz(msg),
1095 if ((sock->state == SS_READY) ||
1096 ((err == TIPC_CONN_SHUTDOWN) || m->msg_control))
1102 /* Consume received message (optional) */
1103 if (likely(!(flags & MSG_PEEK))) {
1104 if ((sock->state != SS_READY) &&
1105 (++port->conn_unacked >= TIPC_CONNACK_INTV))
1106 tipc_acknowledge(port->ref, port->conn_unacked);
1107 advance_rx_queue(sk);
1115 * tipc_recv_stream - receive stream-oriented data
1117 * @m: descriptor for message info
1118 * @buf_len: total size of user buffer area
1119 * @flags: receive flags
1121 * Used for SOCK_STREAM messages only. If not enough data is available
1122 * will optionally wait for more; never truncates data.
1124 * Returns size of returned message data, errno otherwise
1126 static int tipc_recv_stream(struct kiocb *iocb, struct socket *sock,
1127 struct msghdr *m, size_t buf_len, int flags)
1129 struct sock *sk = sock->sk;
1130 struct tipc_sock *tsk = tipc_sk(sk);
1131 struct tipc_port *port = &tsk->port;
1132 struct sk_buff *buf;
1133 struct tipc_msg *msg;
1136 int sz_to_copy, target, needed;
1141 /* Catch invalid receive attempts */
1142 if (unlikely(!buf_len))
1147 if (unlikely(sock->state == SS_UNCONNECTED)) {
1152 target = sock_rcvlowat(sk, flags & MSG_WAITALL, buf_len);
1153 timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1156 /* Look for a message in receive queue; wait if necessary */
1157 res = tipc_wait_for_rcvmsg(sock, timeo);
1161 /* Look at first message in receive queue */
1162 buf = skb_peek(&sk->sk_receive_queue);
1164 sz = msg_data_sz(msg);
1165 err = msg_errcode(msg);
1167 /* Discard an empty non-errored message & try again */
1168 if ((!sz) && (!err)) {
1169 advance_rx_queue(sk);
1173 /* Optionally capture sender's address & ancillary data of first msg */
1174 if (sz_copied == 0) {
1175 set_orig_addr(m, msg);
1176 res = anc_data_recv(m, msg, port);
1181 /* Capture message data (if valid) & compute return value (always) */
1183 u32 offset = (u32)(unsigned long)(TIPC_SKB_CB(buf)->handle);
1186 needed = (buf_len - sz_copied);
1187 sz_to_copy = (sz <= needed) ? sz : needed;
1189 res = skb_copy_datagram_iovec(buf, msg_hdr_sz(msg) + offset,
1190 m->msg_iov, sz_to_copy);
1194 sz_copied += sz_to_copy;
1196 if (sz_to_copy < sz) {
1197 if (!(flags & MSG_PEEK))
1198 TIPC_SKB_CB(buf)->handle =
1199 (void *)(unsigned long)(offset + sz_to_copy);
1204 goto exit; /* can't add error msg to valid data */
1206 if ((err == TIPC_CONN_SHUTDOWN) || m->msg_control)
1212 /* Consume received message (optional) */
1213 if (likely(!(flags & MSG_PEEK))) {
1214 if (unlikely(++port->conn_unacked >= TIPC_CONNACK_INTV))
1215 tipc_acknowledge(port->ref, port->conn_unacked);
1216 advance_rx_queue(sk);
1219 /* Loop around if more data is required */
1220 if ((sz_copied < buf_len) && /* didn't get all requested data */
1221 (!skb_queue_empty(&sk->sk_receive_queue) ||
1222 (sz_copied < target)) && /* and more is ready or required */
1223 (!(flags & MSG_PEEK)) && /* and aren't just peeking at data */
1224 (!err)) /* and haven't reached a FIN */
1229 return sz_copied ? sz_copied : res;
1233 * tipc_write_space - wake up thread if port congestion is released
1236 static void tipc_write_space(struct sock *sk)
1238 struct socket_wq *wq;
1241 wq = rcu_dereference(sk->sk_wq);
1242 if (wq_has_sleeper(wq))
1243 wake_up_interruptible_sync_poll(&wq->wait, POLLOUT |
1244 POLLWRNORM | POLLWRBAND);
1249 * tipc_data_ready - wake up threads to indicate messages have been received
1251 * @len: the length of messages
1253 static void tipc_data_ready(struct sock *sk)
1255 struct socket_wq *wq;
1258 wq = rcu_dereference(sk->sk_wq);
1259 if (wq_has_sleeper(wq))
1260 wake_up_interruptible_sync_poll(&wq->wait, POLLIN |
1261 POLLRDNORM | POLLRDBAND);
1266 * filter_connect - Handle all incoming messages for a connection-based socket
1270 * Returns TIPC error status code and socket error status code
1271 * once it encounters some errors
1273 static u32 filter_connect(struct tipc_sock *tsk, struct sk_buff **buf)
1275 struct sock *sk = &tsk->sk;
1276 struct tipc_port *port = &tsk->port;
1277 struct socket *sock = sk->sk_socket;
1278 struct tipc_msg *msg = buf_msg(*buf);
1280 u32 retval = TIPC_ERR_NO_PORT;
1286 switch ((int)sock->state) {
1288 /* Accept only connection-based messages sent by peer */
1289 if (msg_connected(msg) && tipc_port_peer_msg(port, msg)) {
1290 if (unlikely(msg_errcode(msg))) {
1291 sock->state = SS_DISCONNECTING;
1292 __tipc_port_disconnect(port);
1298 /* Accept only ACK or NACK message */
1299 if (unlikely(msg_errcode(msg))) {
1300 sock->state = SS_DISCONNECTING;
1301 sk->sk_err = ECONNREFUSED;
1306 if (unlikely(!msg_connected(msg)))
1309 res = auto_connect(tsk, msg);
1311 sock->state = SS_DISCONNECTING;
1317 /* If an incoming message is an 'ACK-', it should be
1318 * discarded here because it doesn't contain useful
1319 * data. In addition, we should try to wake up
1320 * connect() routine if sleeping.
1322 if (msg_data_sz(msg) == 0) {
1325 if (waitqueue_active(sk_sleep(sk)))
1326 wake_up_interruptible(sk_sleep(sk));
1331 case SS_UNCONNECTED:
1332 /* Accept only SYN message */
1333 if (!msg_connected(msg) && !(msg_errcode(msg)))
1336 case SS_DISCONNECTING:
1339 pr_err("Unknown socket state %u\n", sock->state);
1345 * rcvbuf_limit - get proper overload limit of socket receive queue
1349 * For all connection oriented messages, irrespective of importance,
1350 * the default overload value (i.e. 67MB) is set as limit.
1352 * For all connectionless messages, by default new queue limits are
1355 * TIPC_LOW_IMPORTANCE (4 MB)
1356 * TIPC_MEDIUM_IMPORTANCE (8 MB)
1357 * TIPC_HIGH_IMPORTANCE (16 MB)
1358 * TIPC_CRITICAL_IMPORTANCE (32 MB)
1360 * Returns overload limit according to corresponding message importance
1362 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *buf)
1364 struct tipc_msg *msg = buf_msg(buf);
1366 if (msg_connected(msg))
1367 return sysctl_tipc_rmem[2];
1369 return sk->sk_rcvbuf >> TIPC_CRITICAL_IMPORTANCE <<
1370 msg_importance(msg);
1374 * filter_rcv - validate incoming message
1378 * Enqueues message on receive queue if acceptable; optionally handles
1379 * disconnect indication for a connected socket.
1381 * Called with socket lock already taken; port lock may also be taken.
1383 * Returns TIPC error status code (TIPC_OK if message is not to be rejected)
1385 static u32 filter_rcv(struct sock *sk, struct sk_buff *buf)
1387 struct socket *sock = sk->sk_socket;
1388 struct tipc_sock *tsk = tipc_sk(sk);
1389 struct tipc_msg *msg = buf_msg(buf);
1390 unsigned int limit = rcvbuf_limit(sk, buf);
1393 /* Reject message if it is wrong sort of message for socket */
1394 if (msg_type(msg) > TIPC_DIRECT_MSG)
1395 return TIPC_ERR_NO_PORT;
1397 if (sock->state == SS_READY) {
1398 if (msg_connected(msg))
1399 return TIPC_ERR_NO_PORT;
1401 res = filter_connect(tsk, &buf);
1402 if (res != TIPC_OK || buf == NULL)
1406 /* Reject message if there isn't room to queue it */
1407 if (sk_rmem_alloc_get(sk) + buf->truesize >= limit)
1408 return TIPC_ERR_OVERLOAD;
1410 /* Enqueue message */
1411 TIPC_SKB_CB(buf)->handle = NULL;
1412 __skb_queue_tail(&sk->sk_receive_queue, buf);
1413 skb_set_owner_r(buf, sk);
1415 sk->sk_data_ready(sk);
1420 * tipc_backlog_rcv - handle incoming message from backlog queue
1424 * Caller must hold socket lock, but not port lock.
1428 static int tipc_backlog_rcv(struct sock *sk, struct sk_buff *buf)
1431 struct tipc_sock *tsk = tipc_sk(sk);
1433 res = filter_rcv(sk, buf);
1435 tipc_reject_msg(buf, res);
1437 if (atomic_read(&tsk->dupl_rcvcnt) < TIPC_CONN_OVERLOAD_LIMIT)
1438 atomic_add(buf->truesize, &tsk->dupl_rcvcnt);
1444 * tipc_sk_rcv - handle incoming message
1445 * @buf: buffer containing arriving message
1447 * Returns 0 if success, or errno: -EHOSTUNREACH
1449 int tipc_sk_rcv(struct sk_buff *buf)
1451 struct tipc_sock *tsk;
1452 struct tipc_port *port;
1454 u32 dport = msg_destport(buf_msg(buf));
1458 /* Forward unresolved named message */
1459 if (unlikely(!dport)) {
1460 tipc_net_route_msg(buf);
1464 /* Validate destination */
1465 port = tipc_port_lock(dport);
1466 if (unlikely(!port)) {
1467 err = TIPC_ERR_NO_PORT;
1471 tsk = tipc_port_to_sock(port);
1477 if (!sock_owned_by_user(sk)) {
1478 err = filter_rcv(sk, buf);
1480 if (sk->sk_backlog.len == 0)
1481 atomic_set(&tsk->dupl_rcvcnt, 0);
1482 limit = rcvbuf_limit(sk, buf) + atomic_read(&tsk->dupl_rcvcnt);
1483 if (sk_add_backlog(sk, buf, limit))
1484 err = TIPC_ERR_OVERLOAD;
1488 tipc_port_unlock(port);
1493 tipc_reject_msg(buf, err);
1494 return -EHOSTUNREACH;
1497 static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
1499 struct sock *sk = sock->sk;
1504 int err = sock_error(sk);
1509 if (signal_pending(current))
1510 return sock_intr_errno(*timeo_p);
1512 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1513 done = sk_wait_event(sk, timeo_p, sock->state != SS_CONNECTING);
1514 finish_wait(sk_sleep(sk), &wait);
1520 * tipc_connect - establish a connection to another TIPC port
1521 * @sock: socket structure
1522 * @dest: socket address for destination port
1523 * @destlen: size of socket address data structure
1524 * @flags: file-related flags associated with socket
1526 * Returns 0 on success, errno otherwise
1528 static int tipc_connect(struct socket *sock, struct sockaddr *dest,
1529 int destlen, int flags)
1531 struct sock *sk = sock->sk;
1532 struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
1533 struct msghdr m = {NULL,};
1534 long timeout = (flags & O_NONBLOCK) ? 0 : tipc_sk(sk)->conn_timeout;
1535 socket_state previous;
1540 /* For now, TIPC does not allow use of connect() with DGRAM/RDM types */
1541 if (sock->state == SS_READY) {
1547 * Reject connection attempt using multicast address
1549 * Note: send_msg() validates the rest of the address fields,
1550 * so there's no need to do it here
1552 if (dst->addrtype == TIPC_ADDR_MCAST) {
1557 previous = sock->state;
1558 switch (sock->state) {
1559 case SS_UNCONNECTED:
1560 /* Send a 'SYN-' to destination */
1562 m.msg_namelen = destlen;
1564 /* If connect is in non-blocking case, set MSG_DONTWAIT to
1565 * indicate send_msg() is never blocked.
1568 m.msg_flags = MSG_DONTWAIT;
1570 res = tipc_sendmsg(NULL, sock, &m, 0);
1571 if ((res < 0) && (res != -EWOULDBLOCK))
1574 /* Just entered SS_CONNECTING state; the only
1575 * difference is that return value in non-blocking
1576 * case is EINPROGRESS, rather than EALREADY.
1580 if (previous == SS_CONNECTING)
1584 timeout = msecs_to_jiffies(timeout);
1585 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
1586 res = tipc_wait_for_connect(sock, &timeout);
1601 * tipc_listen - allow socket to listen for incoming connections
1602 * @sock: socket structure
1605 * Returns 0 on success, errno otherwise
1607 static int tipc_listen(struct socket *sock, int len)
1609 struct sock *sk = sock->sk;
1614 if (sock->state != SS_UNCONNECTED)
1617 sock->state = SS_LISTENING;
1625 static int tipc_wait_for_accept(struct socket *sock, long timeo)
1627 struct sock *sk = sock->sk;
1631 /* True wake-one mechanism for incoming connections: only
1632 * one process gets woken up, not the 'whole herd'.
1633 * Since we do not 'race & poll' for established sockets
1634 * anymore, the common case will execute the loop only once.
1637 prepare_to_wait_exclusive(sk_sleep(sk), &wait,
1638 TASK_INTERRUPTIBLE);
1639 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
1641 timeo = schedule_timeout(timeo);
1645 if (!skb_queue_empty(&sk->sk_receive_queue))
1648 if (sock->state != SS_LISTENING)
1650 err = sock_intr_errno(timeo);
1651 if (signal_pending(current))
1657 finish_wait(sk_sleep(sk), &wait);
1662 * tipc_accept - wait for connection request
1663 * @sock: listening socket
1664 * @newsock: new socket that is to be connected
1665 * @flags: file-related flags associated with socket
1667 * Returns 0 on success, errno otherwise
1669 static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags)
1671 struct sock *new_sk, *sk = sock->sk;
1672 struct sk_buff *buf;
1673 struct tipc_port *new_port;
1674 struct tipc_msg *msg;
1675 struct tipc_portid peer;
1682 if (sock->state != SS_LISTENING) {
1686 timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
1687 res = tipc_wait_for_accept(sock, timeo);
1691 buf = skb_peek(&sk->sk_receive_queue);
1693 res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, 1);
1697 new_sk = new_sock->sk;
1698 new_port = &tipc_sk(new_sk)->port;
1699 new_ref = new_port->ref;
1702 /* we lock on new_sk; but lockdep sees the lock on sk */
1703 lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
1706 * Reject any stray messages received by new socket
1707 * before the socket lock was taken (very, very unlikely)
1709 reject_rx_queue(new_sk);
1711 /* Connect new socket to it's peer */
1712 peer.ref = msg_origport(msg);
1713 peer.node = msg_orignode(msg);
1714 tipc_port_connect(new_ref, &peer);
1715 new_sock->state = SS_CONNECTED;
1717 tipc_port_set_importance(new_port, msg_importance(msg));
1718 if (msg_named(msg)) {
1719 new_port->conn_type = msg_nametype(msg);
1720 new_port->conn_instance = msg_nameinst(msg);
1724 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
1725 * Respond to 'SYN+' by queuing it on new socket.
1727 if (!msg_data_sz(msg)) {
1728 struct msghdr m = {NULL,};
1730 advance_rx_queue(sk);
1731 tipc_send_packet(NULL, new_sock, &m, 0);
1733 __skb_dequeue(&sk->sk_receive_queue);
1734 __skb_queue_head(&new_sk->sk_receive_queue, buf);
1735 skb_set_owner_r(buf, new_sk);
1737 release_sock(new_sk);
1744 * tipc_shutdown - shutdown socket connection
1745 * @sock: socket structure
1746 * @how: direction to close (must be SHUT_RDWR)
1748 * Terminates connection (if necessary), then purges socket's receive queue.
1750 * Returns 0 on success, errno otherwise
1752 static int tipc_shutdown(struct socket *sock, int how)
1754 struct sock *sk = sock->sk;
1755 struct tipc_sock *tsk = tipc_sk(sk);
1756 struct tipc_port *port = &tsk->port;
1757 struct sk_buff *buf;
1760 if (how != SHUT_RDWR)
1765 switch (sock->state) {
1770 /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */
1771 buf = __skb_dequeue(&sk->sk_receive_queue);
1773 if (TIPC_SKB_CB(buf)->handle != NULL) {
1777 tipc_port_disconnect(port->ref);
1778 tipc_reject_msg(buf, TIPC_CONN_SHUTDOWN);
1780 tipc_port_shutdown(port->ref);
1783 sock->state = SS_DISCONNECTING;
1787 case SS_DISCONNECTING:
1789 /* Discard any unreceived messages */
1790 __skb_queue_purge(&sk->sk_receive_queue);
1792 /* Wake up anyone sleeping in poll */
1793 sk->sk_state_change(sk);
1806 * tipc_setsockopt - set socket option
1807 * @sock: socket structure
1808 * @lvl: option level
1809 * @opt: option identifier
1810 * @ov: pointer to new option value
1811 * @ol: length of option value
1813 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
1814 * (to ease compatibility).
1816 * Returns 0 on success, errno otherwise
1818 static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
1819 char __user *ov, unsigned int ol)
1821 struct sock *sk = sock->sk;
1822 struct tipc_sock *tsk = tipc_sk(sk);
1823 struct tipc_port *port = &tsk->port;
1827 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
1829 if (lvl != SOL_TIPC)
1830 return -ENOPROTOOPT;
1831 if (ol < sizeof(value))
1833 res = get_user(value, (u32 __user *)ov);
1840 case TIPC_IMPORTANCE:
1841 tipc_port_set_importance(port, value);
1843 case TIPC_SRC_DROPPABLE:
1844 if (sock->type != SOCK_STREAM)
1845 tipc_port_set_unreliable(port, value);
1849 case TIPC_DEST_DROPPABLE:
1850 tipc_port_set_unreturnable(port, value);
1852 case TIPC_CONN_TIMEOUT:
1853 tipc_sk(sk)->conn_timeout = value;
1854 /* no need to set "res", since already 0 at this point */
1866 * tipc_getsockopt - get socket option
1867 * @sock: socket structure
1868 * @lvl: option level
1869 * @opt: option identifier
1870 * @ov: receptacle for option value
1871 * @ol: receptacle for length of option value
1873 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
1874 * (to ease compatibility).
1876 * Returns 0 on success, errno otherwise
1878 static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
1879 char __user *ov, int __user *ol)
1881 struct sock *sk = sock->sk;
1882 struct tipc_sock *tsk = tipc_sk(sk);
1883 struct tipc_port *port = &tsk->port;
1888 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
1889 return put_user(0, ol);
1890 if (lvl != SOL_TIPC)
1891 return -ENOPROTOOPT;
1892 res = get_user(len, ol);
1899 case TIPC_IMPORTANCE:
1900 value = tipc_port_importance(port);
1902 case TIPC_SRC_DROPPABLE:
1903 value = tipc_port_unreliable(port);
1905 case TIPC_DEST_DROPPABLE:
1906 value = tipc_port_unreturnable(port);
1908 case TIPC_CONN_TIMEOUT:
1909 value = tipc_sk(sk)->conn_timeout;
1910 /* no need to set "res", since already 0 at this point */
1912 case TIPC_NODE_RECVQ_DEPTH:
1913 value = 0; /* was tipc_queue_size, now obsolete */
1915 case TIPC_SOCK_RECVQ_DEPTH:
1916 value = skb_queue_len(&sk->sk_receive_queue);
1925 return res; /* "get" failed */
1927 if (len < sizeof(value))
1930 if (copy_to_user(ov, &value, sizeof(value)))
1933 return put_user(sizeof(value), ol);
1936 int tipc_ioctl(struct socket *sk, unsigned int cmd, unsigned long arg)
1938 struct tipc_sioc_ln_req lnr;
1939 void __user *argp = (void __user *)arg;
1942 case SIOCGETLINKNAME:
1943 if (copy_from_user(&lnr, argp, sizeof(lnr)))
1945 if (!tipc_node_get_linkname(lnr.bearer_id, lnr.peer,
1946 lnr.linkname, TIPC_MAX_LINK_NAME)) {
1947 if (copy_to_user(argp, &lnr, sizeof(lnr)))
1951 return -EADDRNOTAVAIL;
1954 return -ENOIOCTLCMD;
1958 /* Protocol switches for the various types of TIPC sockets */
1960 static const struct proto_ops msg_ops = {
1961 .owner = THIS_MODULE,
1963 .release = tipc_release,
1965 .connect = tipc_connect,
1966 .socketpair = sock_no_socketpair,
1967 .accept = sock_no_accept,
1968 .getname = tipc_getname,
1970 .ioctl = tipc_ioctl,
1971 .listen = sock_no_listen,
1972 .shutdown = tipc_shutdown,
1973 .setsockopt = tipc_setsockopt,
1974 .getsockopt = tipc_getsockopt,
1975 .sendmsg = tipc_sendmsg,
1976 .recvmsg = tipc_recvmsg,
1977 .mmap = sock_no_mmap,
1978 .sendpage = sock_no_sendpage
1981 static const struct proto_ops packet_ops = {
1982 .owner = THIS_MODULE,
1984 .release = tipc_release,
1986 .connect = tipc_connect,
1987 .socketpair = sock_no_socketpair,
1988 .accept = tipc_accept,
1989 .getname = tipc_getname,
1991 .ioctl = tipc_ioctl,
1992 .listen = tipc_listen,
1993 .shutdown = tipc_shutdown,
1994 .setsockopt = tipc_setsockopt,
1995 .getsockopt = tipc_getsockopt,
1996 .sendmsg = tipc_send_packet,
1997 .recvmsg = tipc_recvmsg,
1998 .mmap = sock_no_mmap,
1999 .sendpage = sock_no_sendpage
2002 static const struct proto_ops stream_ops = {
2003 .owner = THIS_MODULE,
2005 .release = tipc_release,
2007 .connect = tipc_connect,
2008 .socketpair = sock_no_socketpair,
2009 .accept = tipc_accept,
2010 .getname = tipc_getname,
2012 .ioctl = tipc_ioctl,
2013 .listen = tipc_listen,
2014 .shutdown = tipc_shutdown,
2015 .setsockopt = tipc_setsockopt,
2016 .getsockopt = tipc_getsockopt,
2017 .sendmsg = tipc_send_stream,
2018 .recvmsg = tipc_recv_stream,
2019 .mmap = sock_no_mmap,
2020 .sendpage = sock_no_sendpage
2023 static const struct net_proto_family tipc_family_ops = {
2024 .owner = THIS_MODULE,
2026 .create = tipc_sk_create
2029 static struct proto tipc_proto = {
2031 .owner = THIS_MODULE,
2032 .obj_size = sizeof(struct tipc_sock),
2033 .sysctl_rmem = sysctl_tipc_rmem
2036 static struct proto tipc_proto_kern = {
2038 .obj_size = sizeof(struct tipc_sock),
2039 .sysctl_rmem = sysctl_tipc_rmem
2043 * tipc_socket_init - initialize TIPC socket interface
2045 * Returns 0 on success, errno otherwise
2047 int tipc_socket_init(void)
2051 res = proto_register(&tipc_proto, 1);
2053 pr_err("Failed to register TIPC protocol type\n");
2057 res = sock_register(&tipc_family_ops);
2059 pr_err("Failed to register TIPC socket type\n");
2060 proto_unregister(&tipc_proto);
2068 * tipc_socket_stop - stop TIPC socket interface
2070 void tipc_socket_stop(void)
2072 sock_unregister(tipc_family_ops.family);
2073 proto_unregister(&tipc_proto);