2 * net/tipc/socket.c: TIPC socket API
4 * Copyright (c) 2001-2007, 2012 Ericsson AB
5 * Copyright (c) 2004-2008, 2010-2012, Wind River Systems
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the names of the copyright holders nor the names of its
17 * contributors may be used to endorse or promote products derived from
18 * this software without specific prior written permission.
20 * Alternatively, this software may be distributed under the terms of the
21 * GNU General Public License ("GPL") version 2 as published by the Free
22 * Software Foundation.
24 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
25 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
28 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
29 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
30 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
31 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
32 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
33 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
34 * POSSIBILITY OF SUCH DAMAGE.
40 #include <linux/export.h>
43 #define SS_LISTENING -1 /* socket is listening */
44 #define SS_READY -2 /* socket is connectionless */
46 #define CONN_OVERLOAD_LIMIT ((TIPC_FLOW_CONTROL_WIN * 2 + 1) * \
47 SKB_TRUESIZE(TIPC_MAX_USER_MSG_SIZE))
48 #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */
53 struct tipc_portid peer_name;
54 unsigned int conn_timeout;
57 #define tipc_sk(sk) ((struct tipc_sock *)(sk))
58 #define tipc_sk_port(sk) (tipc_sk(sk)->p)
60 #define tipc_rx_ready(sock) (!skb_queue_empty(&sock->sk->sk_receive_queue) || \
61 (sock->state == SS_DISCONNECTING))
63 static int backlog_rcv(struct sock *sk, struct sk_buff *skb);
64 static u32 dispatch(struct tipc_port *tport, struct sk_buff *buf);
65 static void wakeupdispatch(struct tipc_port *tport);
66 static void tipc_data_ready(struct sock *sk, int len);
67 static void tipc_write_space(struct sock *sk);
69 static const struct proto_ops packet_ops;
70 static const struct proto_ops stream_ops;
71 static const struct proto_ops msg_ops;
73 static struct proto tipc_proto;
75 static int sockets_enabled;
78 * Revised TIPC socket locking policy:
80 * Most socket operations take the standard socket lock when they start
81 * and hold it until they finish (or until they need to sleep). Acquiring
82 * this lock grants the owner exclusive access to the fields of the socket
83 * data structures, with the exception of the backlog queue. A few socket
84 * operations can be done without taking the socket lock because they only
85 * read socket information that never changes during the life of the socket.
87 * Socket operations may acquire the lock for the associated TIPC port if they
88 * need to perform an operation on the port. If any routine needs to acquire
89 * both the socket lock and the port lock it must take the socket lock first
90 * to avoid the risk of deadlock.
92 * The dispatcher handling incoming messages cannot grab the socket lock in
93 * the standard fashion, since invoked it runs at the BH level and cannot block.
94 * Instead, it checks to see if the socket lock is currently owned by someone,
95 * and either handles the message itself or adds it to the socket's backlog
96 * queue; in the latter case the queued message is processed once the process
97 * owning the socket lock releases it.
99 * NOTE: Releasing the socket lock while an operation is sleeping overcomes
100 * the problem of a blocked socket operation preventing any other operations
101 * from occurring. However, applications must be careful if they have
102 * multiple threads trying to send (or receive) on the same socket, as these
103 * operations might interfere with each other. For example, doing a connect
104 * and a receive at the same time might allow the receive to consume the
105 * ACK message meant for the connect. While additional work could be done
106 * to try and overcome this, it doesn't seem to be worthwhile at the present.
108 * NOTE: Releasing the socket lock while an operation is sleeping also ensures
109 * that another operation that must be performed in a non-blocking manner is
110 * not delayed for very long because the lock has already been taken.
112 * NOTE: This code assumes that certain fields of a port/socket pair are
113 * constant over its lifetime; such fields can be examined without taking
114 * the socket lock and/or port lock, and do not need to be re-read even
115 * after resuming processing after waiting. These fields include:
117 * - pointer to socket sk structure (aka tipc_sock structure)
118 * - pointer to port structure
123 * advance_rx_queue - discard first buffer in socket receive queue
125 * Caller must hold socket lock
127 static void advance_rx_queue(struct sock *sk)
129 kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
133 * reject_rx_queue - reject all buffers in socket receive queue
135 * Caller must hold socket lock
137 static void reject_rx_queue(struct sock *sk)
141 while ((buf = __skb_dequeue(&sk->sk_receive_queue)))
142 tipc_reject_msg(buf, TIPC_ERR_NO_PORT);
146 * tipc_create - create a TIPC socket
147 * @net: network namespace (must be default network)
148 * @sock: pre-allocated socket structure
149 * @protocol: protocol indicator (must be 0)
150 * @kern: caused by kernel or by userspace?
152 * This routine creates additional data structures used by the TIPC socket,
153 * initializes them, and links them together.
155 * Returns 0 on success, errno otherwise
157 static int tipc_create(struct net *net, struct socket *sock, int protocol,
160 const struct proto_ops *ops;
163 struct tipc_port *tp_ptr;
165 /* Validate arguments */
166 if (unlikely(protocol != 0))
167 return -EPROTONOSUPPORT;
169 switch (sock->type) {
172 state = SS_UNCONNECTED;
176 state = SS_UNCONNECTED;
187 /* Allocate socket's protocol area */
188 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto);
192 /* Allocate TIPC port for socket to use */
193 tp_ptr = tipc_createport_raw(sk, &dispatch, &wakeupdispatch,
194 TIPC_LOW_IMPORTANCE);
195 if (unlikely(!tp_ptr)) {
200 /* Finish initializing socket data structures */
204 sock_init_data(sock, sk);
205 sk->sk_backlog_rcv = backlog_rcv;
206 sk->sk_data_ready = tipc_data_ready;
207 sk->sk_write_space = tipc_write_space;
208 tipc_sk(sk)->p = tp_ptr;
209 tipc_sk(sk)->conn_timeout = CONN_TIMEOUT_DEFAULT;
211 spin_unlock_bh(tp_ptr->lock);
213 if (sock->state == SS_READY) {
214 tipc_set_portunreturnable(tp_ptr->ref, 1);
215 if (sock->type == SOCK_DGRAM)
216 tipc_set_portunreliable(tp_ptr->ref, 1);
223 * release - destroy a TIPC socket
224 * @sock: socket to destroy
226 * This routine cleans up any messages that are still queued on the socket.
227 * For DGRAM and RDM socket types, all queued messages are rejected.
228 * For SEQPACKET and STREAM socket types, the first message is rejected
229 * and any others are discarded. (If the first message on a STREAM socket
230 * is partially-read, it is discarded and the next one is rejected instead.)
232 * NOTE: Rejected messages are not necessarily returned to the sender! They
233 * are returned or discarded according to the "destination droppable" setting
234 * specified for the message by the sender.
236 * Returns 0 on success, errno otherwise
238 static int release(struct socket *sock)
240 struct sock *sk = sock->sk;
241 struct tipc_port *tport;
246 * Exit if socket isn't fully initialized (occurs when a failed accept()
247 * releases a pre-allocated child socket that was never used)
252 tport = tipc_sk_port(sk);
256 * Reject all unreceived messages, except on an active connection
257 * (which disconnects locally & sends a 'FIN+' to peer)
259 while (sock->state != SS_DISCONNECTING) {
260 buf = __skb_dequeue(&sk->sk_receive_queue);
263 if (TIPC_SKB_CB(buf)->handle != 0)
266 if ((sock->state == SS_CONNECTING) ||
267 (sock->state == SS_CONNECTED)) {
268 sock->state = SS_DISCONNECTING;
269 tipc_disconnect(tport->ref);
271 tipc_reject_msg(buf, TIPC_ERR_NO_PORT);
276 * Delete TIPC port; this ensures no more messages are queued
277 * (also disconnects an active connection & sends a 'FIN-' to peer)
279 res = tipc_deleteport(tport->ref);
281 /* Discard any remaining (connection-based) messages in receive queue */
282 __skb_queue_purge(&sk->sk_receive_queue);
284 /* Reject any messages that accumulated in backlog queue */
285 sock->state = SS_DISCONNECTING;
295 * bind - associate or disassocate TIPC name(s) with a socket
296 * @sock: socket structure
297 * @uaddr: socket address describing name(s) and desired operation
298 * @uaddr_len: size of socket address data structure
300 * Name and name sequence binding is indicated using a positive scope value;
301 * a negative scope value unbinds the specified name. Specifying no name
302 * (i.e. a socket address length of 0) unbinds all names from the socket.
304 * Returns 0 on success, errno otherwise
306 * NOTE: This routine doesn't need to take the socket lock since it doesn't
307 * access any non-constant socket information.
309 static int bind(struct socket *sock, struct sockaddr *uaddr, int uaddr_len)
311 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
312 u32 portref = tipc_sk_port(sock->sk)->ref;
314 if (unlikely(!uaddr_len))
315 return tipc_withdraw(portref, 0, NULL);
317 if (uaddr_len < sizeof(struct sockaddr_tipc))
319 if (addr->family != AF_TIPC)
320 return -EAFNOSUPPORT;
322 if (addr->addrtype == TIPC_ADDR_NAME)
323 addr->addr.nameseq.upper = addr->addr.nameseq.lower;
324 else if (addr->addrtype != TIPC_ADDR_NAMESEQ)
325 return -EAFNOSUPPORT;
327 if (addr->addr.nameseq.type < TIPC_RESERVED_TYPES)
330 return (addr->scope > 0) ?
331 tipc_publish(portref, addr->scope, &addr->addr.nameseq) :
332 tipc_withdraw(portref, -addr->scope, &addr->addr.nameseq);
336 * get_name - get port ID of socket or peer socket
337 * @sock: socket structure
338 * @uaddr: area for returned socket address
339 * @uaddr_len: area for returned length of socket address
340 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
342 * Returns 0 on success, errno otherwise
344 * NOTE: This routine doesn't need to take the socket lock since it only
345 * accesses socket information that is unchanging (or which changes in
346 * a completely predictable manner).
348 static int get_name(struct socket *sock, struct sockaddr *uaddr,
349 int *uaddr_len, int peer)
351 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
352 struct tipc_sock *tsock = tipc_sk(sock->sk);
354 memset(addr, 0, sizeof(*addr));
356 if ((sock->state != SS_CONNECTED) &&
357 ((peer != 2) || (sock->state != SS_DISCONNECTING)))
359 addr->addr.id.ref = tsock->peer_name.ref;
360 addr->addr.id.node = tsock->peer_name.node;
362 addr->addr.id.ref = tsock->p->ref;
363 addr->addr.id.node = tipc_own_addr;
366 *uaddr_len = sizeof(*addr);
367 addr->addrtype = TIPC_ADDR_ID;
368 addr->family = AF_TIPC;
370 addr->addr.name.domain = 0;
376 * poll - read and possibly block on pollmask
377 * @file: file structure associated with the socket
378 * @sock: socket for which to calculate the poll bits
381 * Returns pollmask value
384 * It appears that the usual socket locking mechanisms are not useful here
385 * since the pollmask info is potentially out-of-date the moment this routine
386 * exits. TCP and other protocols seem to rely on higher level poll routines
387 * to handle any preventable race conditions, so TIPC will do the same ...
389 * TIPC sets the returned events as follows:
391 * socket state flags set
392 * ------------ ---------
393 * unconnected no read flags
394 * POLLOUT if port is not congested
396 * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue
399 * connected POLLIN/POLLRDNORM if data in rx queue
400 * POLLOUT if port is not congested
402 * disconnecting POLLIN/POLLRDNORM/POLLHUP
405 * listening POLLIN if SYN in rx queue
408 * ready POLLIN/POLLRDNORM if data in rx queue
409 * [connectionless] POLLOUT (since port cannot be congested)
411 * IMPORTANT: The fact that a read or write operation is indicated does NOT
412 * imply that the operation will succeed, merely that it should be performed
413 * and will not block.
415 static unsigned int poll(struct file *file, struct socket *sock,
418 struct sock *sk = sock->sk;
421 sock_poll_wait(file, sk_sleep(sk), wait);
423 switch ((int)sock->state) {
425 if (!tipc_sk_port(sk)->congested)
430 if (!tipc_sk_port(sk)->congested)
435 if (!skb_queue_empty(&sk->sk_receive_queue))
436 mask |= (POLLIN | POLLRDNORM);
438 case SS_DISCONNECTING:
439 mask = (POLLIN | POLLRDNORM | POLLHUP);
447 * dest_name_check - verify user is permitted to send to specified port name
448 * @dest: destination address
449 * @m: descriptor for message to be sent
451 * Prevents restricted configuration commands from being issued by
452 * unauthorized users.
454 * Returns 0 if permission is granted, otherwise errno
456 static int dest_name_check(struct sockaddr_tipc *dest, struct msghdr *m)
458 struct tipc_cfg_msg_hdr hdr;
460 if (likely(dest->addr.name.name.type >= TIPC_RESERVED_TYPES))
462 if (likely(dest->addr.name.name.type == TIPC_TOP_SRV))
464 if (likely(dest->addr.name.name.type != TIPC_CFG_SRV))
467 if (!m->msg_iovlen || (m->msg_iov[0].iov_len < sizeof(hdr)))
469 if (copy_from_user(&hdr, m->msg_iov[0].iov_base, sizeof(hdr)))
471 if ((ntohs(hdr.tcm_type) & 0xC000) && (!capable(CAP_NET_ADMIN)))
478 * send_msg - send message in connectionless manner
479 * @iocb: if NULL, indicates that socket lock is already held
480 * @sock: socket structure
481 * @m: message to send
482 * @total_len: length of message
484 * Message must have an destination specified explicitly.
485 * Used for SOCK_RDM and SOCK_DGRAM messages,
486 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
487 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
489 * Returns the number of bytes sent on success, or errno otherwise
491 static int send_msg(struct kiocb *iocb, struct socket *sock,
492 struct msghdr *m, size_t total_len)
494 struct sock *sk = sock->sk;
495 struct tipc_port *tport = tipc_sk_port(sk);
496 struct sockaddr_tipc *dest = (struct sockaddr_tipc *)m->msg_name;
502 return -EDESTADDRREQ;
503 if (unlikely((m->msg_namelen < sizeof(*dest)) ||
504 (dest->family != AF_TIPC)))
506 if (total_len > TIPC_MAX_USER_MSG_SIZE)
512 needs_conn = (sock->state != SS_READY);
513 if (unlikely(needs_conn)) {
514 if (sock->state == SS_LISTENING) {
518 if (sock->state != SS_UNCONNECTED) {
522 if ((tport->published) ||
523 ((sock->type == SOCK_STREAM) && (total_len != 0))) {
527 if (dest->addrtype == TIPC_ADDR_NAME) {
528 tport->conn_type = dest->addr.name.name.type;
529 tport->conn_instance = dest->addr.name.name.instance;
532 /* Abort any pending connection attempts (very unlikely) */
536 timeout_val = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
539 if (dest->addrtype == TIPC_ADDR_NAME) {
540 res = dest_name_check(dest, m);
543 res = tipc_send2name(tport->ref,
544 &dest->addr.name.name,
545 dest->addr.name.domain,
549 } else if (dest->addrtype == TIPC_ADDR_ID) {
550 res = tipc_send2port(tport->ref,
555 } else if (dest->addrtype == TIPC_ADDR_MCAST) {
560 res = dest_name_check(dest, m);
563 res = tipc_multicast(tport->ref,
569 if (likely(res != -ELINKCONG)) {
570 if (needs_conn && (res >= 0))
571 sock->state = SS_CONNECTING;
574 if (timeout_val <= 0L) {
575 res = timeout_val ? timeout_val : -EWOULDBLOCK;
579 timeout_val = wait_event_interruptible_timeout(*sk_sleep(sk),
580 !tport->congested, timeout_val);
591 * send_packet - send a connection-oriented message
592 * @iocb: if NULL, indicates that socket lock is already held
593 * @sock: socket structure
594 * @m: message to send
595 * @total_len: length of message
597 * Used for SOCK_SEQPACKET messages and SOCK_STREAM data.
599 * Returns the number of bytes sent on success, or errno otherwise
601 static int send_packet(struct kiocb *iocb, struct socket *sock,
602 struct msghdr *m, size_t total_len)
604 struct sock *sk = sock->sk;
605 struct tipc_port *tport = tipc_sk_port(sk);
606 struct sockaddr_tipc *dest = (struct sockaddr_tipc *)m->msg_name;
610 /* Handle implied connection establishment */
612 return send_msg(iocb, sock, m, total_len);
614 if (total_len > TIPC_MAX_USER_MSG_SIZE)
620 timeout_val = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
623 if (unlikely(sock->state != SS_CONNECTED)) {
624 if (sock->state == SS_DISCONNECTING)
631 res = tipc_send(tport->ref, m->msg_iovlen, m->msg_iov,
633 if (likely(res != -ELINKCONG))
635 if (timeout_val <= 0L) {
636 res = timeout_val ? timeout_val : -EWOULDBLOCK;
640 timeout_val = wait_event_interruptible_timeout(*sk_sleep(sk),
641 (!tport->congested || !tport->connected), timeout_val);
651 * send_stream - send stream-oriented data
653 * @sock: socket structure
655 * @total_len: total length of data to be sent
657 * Used for SOCK_STREAM data.
659 * Returns the number of bytes sent on success (or partial success),
660 * or errno if no data sent
662 static int send_stream(struct kiocb *iocb, struct socket *sock,
663 struct msghdr *m, size_t total_len)
665 struct sock *sk = sock->sk;
666 struct tipc_port *tport = tipc_sk_port(sk);
667 struct msghdr my_msg;
669 struct iovec *curr_iov;
671 char __user *curr_start;
680 /* Handle special cases where there is no connection */
681 if (unlikely(sock->state != SS_CONNECTED)) {
682 if (sock->state == SS_UNCONNECTED) {
683 res = send_packet(NULL, sock, m, total_len);
685 } else if (sock->state == SS_DISCONNECTING) {
694 if (unlikely(m->msg_name)) {
699 if (total_len > (unsigned int)INT_MAX) {
705 * Send each iovec entry using one or more messages
707 * Note: This algorithm is good for the most likely case
708 * (i.e. one large iovec entry), but could be improved to pass sets
709 * of small iovec entries into send_packet().
711 curr_iov = m->msg_iov;
712 curr_iovlen = m->msg_iovlen;
713 my_msg.msg_iov = &my_iov;
714 my_msg.msg_iovlen = 1;
715 my_msg.msg_flags = m->msg_flags;
716 my_msg.msg_name = NULL;
719 hdr_size = msg_hdr_sz(&tport->phdr);
721 while (curr_iovlen--) {
722 curr_start = curr_iov->iov_base;
723 curr_left = curr_iov->iov_len;
726 bytes_to_send = tport->max_pkt - hdr_size;
727 if (bytes_to_send > TIPC_MAX_USER_MSG_SIZE)
728 bytes_to_send = TIPC_MAX_USER_MSG_SIZE;
729 if (curr_left < bytes_to_send)
730 bytes_to_send = curr_left;
731 my_iov.iov_base = curr_start;
732 my_iov.iov_len = bytes_to_send;
733 res = send_packet(NULL, sock, &my_msg, bytes_to_send);
739 curr_left -= bytes_to_send;
740 curr_start += bytes_to_send;
741 bytes_sent += bytes_to_send;
753 * auto_connect - complete connection setup to a remote port
754 * @sock: socket structure
755 * @msg: peer's response message
757 * Returns 0 on success, errno otherwise
759 static int auto_connect(struct socket *sock, struct tipc_msg *msg)
761 struct tipc_sock *tsock = tipc_sk(sock->sk);
762 struct tipc_port *p_ptr;
764 tsock->peer_name.ref = msg_origport(msg);
765 tsock->peer_name.node = msg_orignode(msg);
766 p_ptr = tipc_port_deref(tsock->p->ref);
770 __tipc_connect(tsock->p->ref, p_ptr, &tsock->peer_name);
772 if (msg_importance(msg) > TIPC_CRITICAL_IMPORTANCE)
774 msg_set_importance(&p_ptr->phdr, (u32)msg_importance(msg));
775 sock->state = SS_CONNECTED;
780 * set_orig_addr - capture sender's address for received message
781 * @m: descriptor for message info
782 * @msg: received message header
784 * Note: Address is not captured if not requested by receiver.
786 static void set_orig_addr(struct msghdr *m, struct tipc_msg *msg)
788 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)m->msg_name;
791 addr->family = AF_TIPC;
792 addr->addrtype = TIPC_ADDR_ID;
793 addr->addr.id.ref = msg_origport(msg);
794 addr->addr.id.node = msg_orignode(msg);
795 addr->addr.name.domain = 0; /* could leave uninitialized */
796 addr->scope = 0; /* could leave uninitialized */
797 m->msg_namelen = sizeof(struct sockaddr_tipc);
802 * anc_data_recv - optionally capture ancillary data for received message
803 * @m: descriptor for message info
804 * @msg: received message header
805 * @tport: TIPC port associated with message
807 * Note: Ancillary data is not captured if not requested by receiver.
809 * Returns 0 if successful, otherwise errno
811 static int anc_data_recv(struct msghdr *m, struct tipc_msg *msg,
812 struct tipc_port *tport)
820 if (likely(m->msg_controllen == 0))
823 /* Optionally capture errored message object(s) */
824 err = msg ? msg_errcode(msg) : 0;
827 anc_data[1] = msg_data_sz(msg);
828 res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
832 res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
839 /* Optionally capture message destination object */
840 dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
844 anc_data[0] = msg_nametype(msg);
845 anc_data[1] = msg_namelower(msg);
846 anc_data[2] = msg_namelower(msg);
850 anc_data[0] = msg_nametype(msg);
851 anc_data[1] = msg_namelower(msg);
852 anc_data[2] = msg_nameupper(msg);
855 has_name = (tport->conn_type != 0);
856 anc_data[0] = tport->conn_type;
857 anc_data[1] = tport->conn_instance;
858 anc_data[2] = tport->conn_instance;
864 res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
873 * recv_msg - receive packet-oriented message
875 * @m: descriptor for message info
876 * @buf_len: total size of user buffer area
877 * @flags: receive flags
879 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
880 * If the complete message doesn't fit in user area, truncate it.
882 * Returns size of returned message data, errno otherwise
884 static int recv_msg(struct kiocb *iocb, struct socket *sock,
885 struct msghdr *m, size_t buf_len, int flags)
887 struct sock *sk = sock->sk;
888 struct tipc_port *tport = tipc_sk_port(sk);
890 struct tipc_msg *msg;
896 /* Catch invalid receive requests */
897 if (unlikely(!buf_len))
902 if (unlikely(sock->state == SS_UNCONNECTED)) {
907 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
910 /* Look for a message in receive queue; wait if necessary */
911 while (skb_queue_empty(&sk->sk_receive_queue)) {
912 if (sock->state == SS_DISCONNECTING) {
917 res = timeout ? timeout : -EWOULDBLOCK;
921 timeout = wait_event_interruptible_timeout(*sk_sleep(sk),
927 /* Look at first message in receive queue */
928 buf = skb_peek(&sk->sk_receive_queue);
930 sz = msg_data_sz(msg);
931 err = msg_errcode(msg);
933 /* Discard an empty non-errored message & try again */
934 if ((!sz) && (!err)) {
935 advance_rx_queue(sk);
939 /* Capture sender's address (optional) */
940 set_orig_addr(m, msg);
942 /* Capture ancillary data (optional) */
943 res = anc_data_recv(m, msg, tport);
947 /* Capture message data (if valid) & compute return value (always) */
949 if (unlikely(buf_len < sz)) {
951 m->msg_flags |= MSG_TRUNC;
953 res = skb_copy_datagram_iovec(buf, msg_hdr_sz(msg),
959 if ((sock->state == SS_READY) ||
960 ((err == TIPC_CONN_SHUTDOWN) || m->msg_control))
966 /* Consume received message (optional) */
967 if (likely(!(flags & MSG_PEEK))) {
968 if ((sock->state != SS_READY) &&
969 (++tport->conn_unacked >= TIPC_FLOW_CONTROL_WIN))
970 tipc_acknowledge(tport->ref, tport->conn_unacked);
971 advance_rx_queue(sk);
979 * recv_stream - receive stream-oriented data
981 * @m: descriptor for message info
982 * @buf_len: total size of user buffer area
983 * @flags: receive flags
985 * Used for SOCK_STREAM messages only. If not enough data is available
986 * will optionally wait for more; never truncates data.
988 * Returns size of returned message data, errno otherwise
990 static int recv_stream(struct kiocb *iocb, struct socket *sock,
991 struct msghdr *m, size_t buf_len, int flags)
993 struct sock *sk = sock->sk;
994 struct tipc_port *tport = tipc_sk_port(sk);
996 struct tipc_msg *msg;
999 int sz_to_copy, target, needed;
1004 /* Catch invalid receive attempts */
1005 if (unlikely(!buf_len))
1010 if (unlikely((sock->state == SS_UNCONNECTED) ||
1011 (sock->state == SS_CONNECTING))) {
1016 target = sock_rcvlowat(sk, flags & MSG_WAITALL, buf_len);
1017 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1020 /* Look for a message in receive queue; wait if necessary */
1021 while (skb_queue_empty(&sk->sk_receive_queue)) {
1022 if (sock->state == SS_DISCONNECTING) {
1026 if (timeout <= 0L) {
1027 res = timeout ? timeout : -EWOULDBLOCK;
1031 timeout = wait_event_interruptible_timeout(*sk_sleep(sk),
1032 tipc_rx_ready(sock),
1037 /* Look at first message in receive queue */
1038 buf = skb_peek(&sk->sk_receive_queue);
1040 sz = msg_data_sz(msg);
1041 err = msg_errcode(msg);
1043 /* Discard an empty non-errored message & try again */
1044 if ((!sz) && (!err)) {
1045 advance_rx_queue(sk);
1049 /* Optionally capture sender's address & ancillary data of first msg */
1050 if (sz_copied == 0) {
1051 set_orig_addr(m, msg);
1052 res = anc_data_recv(m, msg, tport);
1057 /* Capture message data (if valid) & compute return value (always) */
1059 u32 offset = (u32)(unsigned long)(TIPC_SKB_CB(buf)->handle);
1062 needed = (buf_len - sz_copied);
1063 sz_to_copy = (sz <= needed) ? sz : needed;
1065 res = skb_copy_datagram_iovec(buf, msg_hdr_sz(msg) + offset,
1066 m->msg_iov, sz_to_copy);
1070 sz_copied += sz_to_copy;
1072 if (sz_to_copy < sz) {
1073 if (!(flags & MSG_PEEK))
1074 TIPC_SKB_CB(buf)->handle =
1075 (void *)(unsigned long)(offset + sz_to_copy);
1080 goto exit; /* can't add error msg to valid data */
1082 if ((err == TIPC_CONN_SHUTDOWN) || m->msg_control)
1088 /* Consume received message (optional) */
1089 if (likely(!(flags & MSG_PEEK))) {
1090 if (unlikely(++tport->conn_unacked >= TIPC_FLOW_CONTROL_WIN))
1091 tipc_acknowledge(tport->ref, tport->conn_unacked);
1092 advance_rx_queue(sk);
1095 /* Loop around if more data is required */
1096 if ((sz_copied < buf_len) && /* didn't get all requested data */
1097 (!skb_queue_empty(&sk->sk_receive_queue) ||
1098 (sz_copied < target)) && /* and more is ready or required */
1099 (!(flags & MSG_PEEK)) && /* and aren't just peeking at data */
1100 (!err)) /* and haven't reached a FIN */
1105 return sz_copied ? sz_copied : res;
1109 * tipc_write_space - wake up thread if port congestion is released
1112 static void tipc_write_space(struct sock *sk)
1114 struct socket_wq *wq;
1117 wq = rcu_dereference(sk->sk_wq);
1118 if (wq_has_sleeper(wq))
1119 wake_up_interruptible_sync_poll(&wq->wait, POLLOUT |
1120 POLLWRNORM | POLLWRBAND);
1125 * tipc_data_ready - wake up threads to indicate messages have been received
1127 * @len: the length of messages
1129 static void tipc_data_ready(struct sock *sk, int len)
1131 struct socket_wq *wq;
1134 wq = rcu_dereference(sk->sk_wq);
1135 if (wq_has_sleeper(wq))
1136 wake_up_interruptible_sync_poll(&wq->wait, POLLIN |
1137 POLLRDNORM | POLLRDBAND);
1142 * filter_connect - Handle all incoming messages for a connection-based socket
1143 * @tsock: TIPC socket
1146 * Returns TIPC error status code and socket error status code
1147 * once it encounters some errors
1149 static u32 filter_connect(struct tipc_sock *tsock, struct sk_buff **buf)
1151 struct socket *sock = tsock->sk.sk_socket;
1152 struct tipc_msg *msg = buf_msg(*buf);
1153 struct sock *sk = &tsock->sk;
1154 u32 retval = TIPC_ERR_NO_PORT;
1160 switch ((int)sock->state) {
1162 /* Accept only connection-based messages sent by peer */
1163 if (msg_connected(msg) && tipc_port_peer_msg(tsock->p, msg)) {
1164 if (unlikely(msg_errcode(msg))) {
1165 sock->state = SS_DISCONNECTING;
1166 __tipc_disconnect(tsock->p);
1172 /* Accept only ACK or NACK message */
1173 if (unlikely(msg_errcode(msg))) {
1174 sock->state = SS_DISCONNECTING;
1175 sk->sk_err = -ECONNREFUSED;
1180 if (unlikely(!msg_connected(msg)))
1183 res = auto_connect(sock, msg);
1185 sock->state = SS_DISCONNECTING;
1191 /* If an incoming message is an 'ACK-', it should be
1192 * discarded here because it doesn't contain useful
1193 * data. In addition, we should try to wake up
1194 * connect() routine if sleeping.
1196 if (msg_data_sz(msg) == 0) {
1199 if (waitqueue_active(sk_sleep(sk)))
1200 wake_up_interruptible(sk_sleep(sk));
1205 case SS_UNCONNECTED:
1206 /* Accept only SYN message */
1207 if (!msg_connected(msg) && !(msg_errcode(msg)))
1210 case SS_DISCONNECTING:
1213 pr_err("Unknown socket state %u\n", sock->state);
1219 * rcvbuf_limit - get proper overload limit of socket receive queue
1223 * For all connection oriented messages, irrespective of importance,
1224 * the default overload value (i.e. 67MB) is set as limit.
1226 * For all connectionless messages, by default new queue limits are
1229 * TIPC_LOW_IMPORTANCE (5MB)
1230 * TIPC_MEDIUM_IMPORTANCE (10MB)
1231 * TIPC_HIGH_IMPORTANCE (20MB)
1232 * TIPC_CRITICAL_IMPORTANCE (40MB)
1234 * Returns overload limit according to corresponding message importance
1236 static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *buf)
1238 struct tipc_msg *msg = buf_msg(buf);
1241 if (msg_connected(msg))
1242 limit = CONN_OVERLOAD_LIMIT;
1244 limit = sk->sk_rcvbuf << (msg_importance(msg) + 5);
1249 * filter_rcv - validate incoming message
1253 * Enqueues message on receive queue if acceptable; optionally handles
1254 * disconnect indication for a connected socket.
1256 * Called with socket lock already taken; port lock may also be taken.
1258 * Returns TIPC error status code (TIPC_OK if message is not to be rejected)
1260 static u32 filter_rcv(struct sock *sk, struct sk_buff *buf)
1262 struct socket *sock = sk->sk_socket;
1263 struct tipc_msg *msg = buf_msg(buf);
1264 unsigned int limit = rcvbuf_limit(sk, buf);
1267 /* Reject message if it is wrong sort of message for socket */
1268 if (msg_type(msg) > TIPC_DIRECT_MSG)
1269 return TIPC_ERR_NO_PORT;
1271 if (sock->state == SS_READY) {
1272 if (msg_connected(msg))
1273 return TIPC_ERR_NO_PORT;
1275 res = filter_connect(tipc_sk(sk), &buf);
1276 if (res != TIPC_OK || buf == NULL)
1280 /* Reject message if there isn't room to queue it */
1281 if (sk_rmem_alloc_get(sk) + buf->truesize >= limit)
1282 return TIPC_ERR_OVERLOAD;
1284 /* Enqueue message */
1285 TIPC_SKB_CB(buf)->handle = 0;
1286 __skb_queue_tail(&sk->sk_receive_queue, buf);
1287 skb_set_owner_r(buf, sk);
1289 sk->sk_data_ready(sk, 0);
1294 * backlog_rcv - handle incoming message from backlog queue
1298 * Caller must hold socket lock, but not port lock.
1302 static int backlog_rcv(struct sock *sk, struct sk_buff *buf)
1306 res = filter_rcv(sk, buf);
1308 tipc_reject_msg(buf, res);
1313 * dispatch - handle incoming message
1314 * @tport: TIPC port that received message
1317 * Called with port lock already taken.
1319 * Returns TIPC error status code (TIPC_OK if message is not to be rejected)
1321 static u32 dispatch(struct tipc_port *tport, struct sk_buff *buf)
1323 struct sock *sk = (struct sock *)tport->usr_handle;
1327 * Process message if socket is unlocked; otherwise add to backlog queue
1329 * This code is based on sk_receive_skb(), but must be distinct from it
1330 * since a TIPC-specific filter/reject mechanism is utilized
1333 if (!sock_owned_by_user(sk)) {
1334 res = filter_rcv(sk, buf);
1336 if (sk_add_backlog(sk, buf, rcvbuf_limit(sk, buf)))
1337 res = TIPC_ERR_OVERLOAD;
1347 * wakeupdispatch - wake up port after congestion
1348 * @tport: port to wakeup
1350 * Called with port lock already taken.
1352 static void wakeupdispatch(struct tipc_port *tport)
1354 struct sock *sk = (struct sock *)tport->usr_handle;
1356 sk->sk_write_space(sk);
1360 * connect - establish a connection to another TIPC port
1361 * @sock: socket structure
1362 * @dest: socket address for destination port
1363 * @destlen: size of socket address data structure
1364 * @flags: file-related flags associated with socket
1366 * Returns 0 on success, errno otherwise
1368 static int connect(struct socket *sock, struct sockaddr *dest, int destlen,
1371 struct sock *sk = sock->sk;
1372 struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
1373 struct msghdr m = {NULL,};
1374 unsigned int timeout;
1379 /* For now, TIPC does not allow use of connect() with DGRAM/RDM types */
1380 if (sock->state == SS_READY) {
1386 * Reject connection attempt using multicast address
1388 * Note: send_msg() validates the rest of the address fields,
1389 * so there's no need to do it here
1391 if (dst->addrtype == TIPC_ADDR_MCAST) {
1396 timeout = (flags & O_NONBLOCK) ? 0 : tipc_sk(sk)->conn_timeout;
1398 switch (sock->state) {
1399 case SS_UNCONNECTED:
1400 /* Send a 'SYN-' to destination */
1402 m.msg_namelen = destlen;
1404 /* If connect is in non-blocking case, set MSG_DONTWAIT to
1405 * indicate send_msg() is never blocked.
1408 m.msg_flags = MSG_DONTWAIT;
1410 res = send_msg(NULL, sock, &m, 0);
1411 if ((res < 0) && (res != -EWOULDBLOCK))
1414 /* Just entered SS_CONNECTING state; the only
1415 * difference is that return value in non-blocking
1416 * case is EINPROGRESS, rather than EALREADY.
1431 if (sock->state == SS_CONNECTING) {
1435 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
1437 res = wait_event_interruptible_timeout(*sk_sleep(sk),
1438 sock->state != SS_CONNECTING,
1439 timeout ? (long)msecs_to_jiffies(timeout)
1440 : MAX_SCHEDULE_TIMEOUT);
1446 ; /* leave "res" unchanged */
1451 if (unlikely(sock->state == SS_DISCONNECTING))
1452 res = sock_error(sk);
1462 * listen - allow socket to listen for incoming connections
1463 * @sock: socket structure
1466 * Returns 0 on success, errno otherwise
1468 static int listen(struct socket *sock, int len)
1470 struct sock *sk = sock->sk;
1475 if (sock->state != SS_UNCONNECTED)
1478 sock->state = SS_LISTENING;
1487 * accept - wait for connection request
1488 * @sock: listening socket
1489 * @newsock: new socket that is to be connected
1490 * @flags: file-related flags associated with socket
1492 * Returns 0 on success, errno otherwise
1494 static int accept(struct socket *sock, struct socket *new_sock, int flags)
1496 struct sock *new_sk, *sk = sock->sk;
1497 struct sk_buff *buf;
1498 struct tipc_sock *new_tsock;
1499 struct tipc_port *new_tport;
1500 struct tipc_msg *msg;
1507 if (sock->state != SS_LISTENING) {
1512 while (skb_queue_empty(&sk->sk_receive_queue)) {
1513 if (flags & O_NONBLOCK) {
1518 res = wait_event_interruptible(*sk_sleep(sk),
1519 (!skb_queue_empty(&sk->sk_receive_queue)));
1525 buf = skb_peek(&sk->sk_receive_queue);
1527 res = tipc_create(sock_net(sock->sk), new_sock, 0, 0);
1531 new_sk = new_sock->sk;
1532 new_tsock = tipc_sk(new_sk);
1533 new_tport = new_tsock->p;
1534 new_ref = new_tport->ref;
1537 /* we lock on new_sk; but lockdep sees the lock on sk */
1538 lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
1541 * Reject any stray messages received by new socket
1542 * before the socket lock was taken (very, very unlikely)
1544 reject_rx_queue(new_sk);
1546 /* Connect new socket to it's peer */
1547 new_tsock->peer_name.ref = msg_origport(msg);
1548 new_tsock->peer_name.node = msg_orignode(msg);
1549 tipc_connect(new_ref, &new_tsock->peer_name);
1550 new_sock->state = SS_CONNECTED;
1552 tipc_set_portimportance(new_ref, msg_importance(msg));
1553 if (msg_named(msg)) {
1554 new_tport->conn_type = msg_nametype(msg);
1555 new_tport->conn_instance = msg_nameinst(msg);
1559 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
1560 * Respond to 'SYN+' by queuing it on new socket.
1562 if (!msg_data_sz(msg)) {
1563 struct msghdr m = {NULL,};
1565 advance_rx_queue(sk);
1566 send_packet(NULL, new_sock, &m, 0);
1568 __skb_dequeue(&sk->sk_receive_queue);
1569 __skb_queue_head(&new_sk->sk_receive_queue, buf);
1570 skb_set_owner_r(buf, new_sk);
1572 release_sock(new_sk);
1580 * shutdown - shutdown socket connection
1581 * @sock: socket structure
1582 * @how: direction to close (must be SHUT_RDWR)
1584 * Terminates connection (if necessary), then purges socket's receive queue.
1586 * Returns 0 on success, errno otherwise
1588 static int shutdown(struct socket *sock, int how)
1590 struct sock *sk = sock->sk;
1591 struct tipc_port *tport = tipc_sk_port(sk);
1592 struct sk_buff *buf;
1595 if (how != SHUT_RDWR)
1600 switch (sock->state) {
1605 /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */
1606 buf = __skb_dequeue(&sk->sk_receive_queue);
1608 if (TIPC_SKB_CB(buf)->handle != 0) {
1612 tipc_disconnect(tport->ref);
1613 tipc_reject_msg(buf, TIPC_CONN_SHUTDOWN);
1615 tipc_shutdown(tport->ref);
1618 sock->state = SS_DISCONNECTING;
1622 case SS_DISCONNECTING:
1624 /* Discard any unreceived messages */
1625 __skb_queue_purge(&sk->sk_receive_queue);
1627 /* Wake up anyone sleeping in poll */
1628 sk->sk_state_change(sk);
1641 * setsockopt - set socket option
1642 * @sock: socket structure
1643 * @lvl: option level
1644 * @opt: option identifier
1645 * @ov: pointer to new option value
1646 * @ol: length of option value
1648 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
1649 * (to ease compatibility).
1651 * Returns 0 on success, errno otherwise
1653 static int setsockopt(struct socket *sock,
1654 int lvl, int opt, char __user *ov, unsigned int ol)
1656 struct sock *sk = sock->sk;
1657 struct tipc_port *tport = tipc_sk_port(sk);
1661 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
1663 if (lvl != SOL_TIPC)
1664 return -ENOPROTOOPT;
1665 if (ol < sizeof(value))
1667 res = get_user(value, (u32 __user *)ov);
1674 case TIPC_IMPORTANCE:
1675 res = tipc_set_portimportance(tport->ref, value);
1677 case TIPC_SRC_DROPPABLE:
1678 if (sock->type != SOCK_STREAM)
1679 res = tipc_set_portunreliable(tport->ref, value);
1683 case TIPC_DEST_DROPPABLE:
1684 res = tipc_set_portunreturnable(tport->ref, value);
1686 case TIPC_CONN_TIMEOUT:
1687 tipc_sk(sk)->conn_timeout = value;
1688 /* no need to set "res", since already 0 at this point */
1700 * getsockopt - get socket option
1701 * @sock: socket structure
1702 * @lvl: option level
1703 * @opt: option identifier
1704 * @ov: receptacle for option value
1705 * @ol: receptacle for length of option value
1707 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
1708 * (to ease compatibility).
1710 * Returns 0 on success, errno otherwise
1712 static int getsockopt(struct socket *sock,
1713 int lvl, int opt, char __user *ov, int __user *ol)
1715 struct sock *sk = sock->sk;
1716 struct tipc_port *tport = tipc_sk_port(sk);
1721 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
1722 return put_user(0, ol);
1723 if (lvl != SOL_TIPC)
1724 return -ENOPROTOOPT;
1725 res = get_user(len, ol);
1732 case TIPC_IMPORTANCE:
1733 res = tipc_portimportance(tport->ref, &value);
1735 case TIPC_SRC_DROPPABLE:
1736 res = tipc_portunreliable(tport->ref, &value);
1738 case TIPC_DEST_DROPPABLE:
1739 res = tipc_portunreturnable(tport->ref, &value);
1741 case TIPC_CONN_TIMEOUT:
1742 value = tipc_sk(sk)->conn_timeout;
1743 /* no need to set "res", since already 0 at this point */
1745 case TIPC_NODE_RECVQ_DEPTH:
1746 value = 0; /* was tipc_queue_size, now obsolete */
1748 case TIPC_SOCK_RECVQ_DEPTH:
1749 value = skb_queue_len(&sk->sk_receive_queue);
1758 return res; /* "get" failed */
1760 if (len < sizeof(value))
1763 if (copy_to_user(ov, &value, sizeof(value)))
1766 return put_user(sizeof(value), ol);
1769 /* Protocol switches for the various types of TIPC sockets */
1771 static const struct proto_ops msg_ops = {
1772 .owner = THIS_MODULE,
1777 .socketpair = sock_no_socketpair,
1778 .accept = sock_no_accept,
1779 .getname = get_name,
1781 .ioctl = sock_no_ioctl,
1782 .listen = sock_no_listen,
1783 .shutdown = shutdown,
1784 .setsockopt = setsockopt,
1785 .getsockopt = getsockopt,
1786 .sendmsg = send_msg,
1787 .recvmsg = recv_msg,
1788 .mmap = sock_no_mmap,
1789 .sendpage = sock_no_sendpage
1792 static const struct proto_ops packet_ops = {
1793 .owner = THIS_MODULE,
1798 .socketpair = sock_no_socketpair,
1800 .getname = get_name,
1802 .ioctl = sock_no_ioctl,
1804 .shutdown = shutdown,
1805 .setsockopt = setsockopt,
1806 .getsockopt = getsockopt,
1807 .sendmsg = send_packet,
1808 .recvmsg = recv_msg,
1809 .mmap = sock_no_mmap,
1810 .sendpage = sock_no_sendpage
1813 static const struct proto_ops stream_ops = {
1814 .owner = THIS_MODULE,
1819 .socketpair = sock_no_socketpair,
1821 .getname = get_name,
1823 .ioctl = sock_no_ioctl,
1825 .shutdown = shutdown,
1826 .setsockopt = setsockopt,
1827 .getsockopt = getsockopt,
1828 .sendmsg = send_stream,
1829 .recvmsg = recv_stream,
1830 .mmap = sock_no_mmap,
1831 .sendpage = sock_no_sendpage
1834 static const struct net_proto_family tipc_family_ops = {
1835 .owner = THIS_MODULE,
1837 .create = tipc_create
1840 static struct proto tipc_proto = {
1842 .owner = THIS_MODULE,
1843 .obj_size = sizeof(struct tipc_sock)
1847 * tipc_socket_init - initialize TIPC socket interface
1849 * Returns 0 on success, errno otherwise
1851 int tipc_socket_init(void)
1855 res = proto_register(&tipc_proto, 1);
1857 pr_err("Failed to register TIPC protocol type\n");
1861 res = sock_register(&tipc_family_ops);
1863 pr_err("Failed to register TIPC socket type\n");
1864 proto_unregister(&tipc_proto);
1868 sockets_enabled = 1;
1874 * tipc_socket_stop - stop TIPC socket interface
1876 void tipc_socket_stop(void)
1878 if (!sockets_enabled)
1881 sockets_enabled = 0;
1882 sock_unregister(tipc_family_ops.family);
1883 proto_unregister(&tipc_proto);