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>
44 #define SS_LISTENING -1 /* socket is listening */
45 #define SS_READY -2 /* socket is connectionless */
47 #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */
49 static int tipc_backlog_rcv(struct sock
*sk
, struct sk_buff
*skb
);
50 static void tipc_data_ready(struct sock
*sk
);
51 static void tipc_write_space(struct sock
*sk
);
52 static int tipc_release(struct socket
*sock
);
53 static int tipc_accept(struct socket
*sock
, struct socket
*new_sock
, int flags
);
55 static const struct proto_ops packet_ops
;
56 static const struct proto_ops stream_ops
;
57 static const struct proto_ops msg_ops
;
59 static struct proto tipc_proto
;
60 static struct proto tipc_proto_kern
;
63 * Revised TIPC socket locking policy:
65 * Most socket operations take the standard socket lock when they start
66 * and hold it until they finish (or until they need to sleep). Acquiring
67 * this lock grants the owner exclusive access to the fields of the socket
68 * data structures, with the exception of the backlog queue. A few socket
69 * operations can be done without taking the socket lock because they only
70 * read socket information that never changes during the life of the socket.
72 * Socket operations may acquire the lock for the associated TIPC port if they
73 * need to perform an operation on the port. If any routine needs to acquire
74 * both the socket lock and the port lock it must take the socket lock first
75 * to avoid the risk of deadlock.
77 * The dispatcher handling incoming messages cannot grab the socket lock in
78 * the standard fashion, since invoked it runs at the BH level and cannot block.
79 * Instead, it checks to see if the socket lock is currently owned by someone,
80 * and either handles the message itself or adds it to the socket's backlog
81 * queue; in the latter case the queued message is processed once the process
82 * owning the socket lock releases it.
84 * NOTE: Releasing the socket lock while an operation is sleeping overcomes
85 * the problem of a blocked socket operation preventing any other operations
86 * from occurring. However, applications must be careful if they have
87 * multiple threads trying to send (or receive) on the same socket, as these
88 * operations might interfere with each other. For example, doing a connect
89 * and a receive at the same time might allow the receive to consume the
90 * ACK message meant for the connect. While additional work could be done
91 * to try and overcome this, it doesn't seem to be worthwhile at the present.
93 * NOTE: Releasing the socket lock while an operation is sleeping also ensures
94 * that another operation that must be performed in a non-blocking manner is
95 * not delayed for very long because the lock has already been taken.
97 * NOTE: This code assumes that certain fields of a port/socket pair are
98 * constant over its lifetime; such fields can be examined without taking
99 * the socket lock and/or port lock, and do not need to be re-read even
100 * after resuming processing after waiting. These fields include:
102 * - pointer to socket sk structure (aka tipc_sock structure)
103 * - pointer to port structure
110 * advance_rx_queue - discard first buffer in socket receive queue
112 * Caller must hold socket lock
114 static void advance_rx_queue(struct sock
*sk
)
116 kfree_skb(__skb_dequeue(&sk
->sk_receive_queue
));
120 * reject_rx_queue - reject all buffers in socket receive queue
122 * Caller must hold socket lock
124 static void reject_rx_queue(struct sock
*sk
)
129 while ((buf
= __skb_dequeue(&sk
->sk_receive_queue
))) {
130 if (tipc_msg_reverse(buf
, &dnode
, TIPC_ERR_NO_PORT
))
131 tipc_link_xmit2(buf
, dnode
, 0);
136 * tipc_sk_create - create a TIPC socket
137 * @net: network namespace (must be default network)
138 * @sock: pre-allocated socket structure
139 * @protocol: protocol indicator (must be 0)
140 * @kern: caused by kernel or by userspace?
142 * This routine creates additional data structures used by the TIPC socket,
143 * initializes them, and links them together.
145 * Returns 0 on success, errno otherwise
147 static int tipc_sk_create(struct net
*net
, struct socket
*sock
,
148 int protocol
, int kern
)
150 const struct proto_ops
*ops
;
153 struct tipc_sock
*tsk
;
154 struct tipc_port
*port
;
157 /* Validate arguments */
158 if (unlikely(protocol
!= 0))
159 return -EPROTONOSUPPORT
;
161 switch (sock
->type
) {
164 state
= SS_UNCONNECTED
;
168 state
= SS_UNCONNECTED
;
179 /* Allocate socket's protocol area */
181 sk
= sk_alloc(net
, AF_TIPC
, GFP_KERNEL
, &tipc_proto
);
183 sk
= sk_alloc(net
, AF_TIPC
, GFP_KERNEL
, &tipc_proto_kern
);
191 ref
= tipc_port_init(port
, TIPC_LOW_IMPORTANCE
);
193 pr_warn("Socket registration failed, ref. table exhausted\n");
198 /* Finish initializing socket data structures */
202 sock_init_data(sock
, sk
);
203 sk
->sk_backlog_rcv
= tipc_backlog_rcv
;
204 sk
->sk_rcvbuf
= sysctl_tipc_rmem
[1];
205 sk
->sk_data_ready
= tipc_data_ready
;
206 sk
->sk_write_space
= tipc_write_space
;
207 tsk
->conn_timeout
= CONN_TIMEOUT_DEFAULT
;
208 atomic_set(&tsk
->dupl_rcvcnt
, 0);
209 tipc_port_unlock(port
);
211 if (sock
->state
== SS_READY
) {
212 tipc_port_set_unreturnable(port
, true);
213 if (sock
->type
== SOCK_DGRAM
)
214 tipc_port_set_unreliable(port
, true);
220 * tipc_sock_create_local - create TIPC socket from inside TIPC module
221 * @type: socket type - SOCK_RDM or SOCK_SEQPACKET
223 * We cannot use sock_creat_kern here because it bumps module user count.
224 * Since socket owner and creator is the same module we must make sure
225 * that module count remains zero for module local sockets, otherwise
226 * we cannot do rmmod.
228 * Returns 0 on success, errno otherwise
230 int tipc_sock_create_local(int type
, struct socket
**res
)
234 rc
= sock_create_lite(AF_TIPC
, type
, 0, res
);
236 pr_err("Failed to create kernel socket\n");
239 tipc_sk_create(&init_net
, *res
, 0, 1);
245 * tipc_sock_release_local - release socket created by tipc_sock_create_local
246 * @sock: the socket to be released.
248 * Module reference count is not incremented when such sockets are created,
249 * so we must keep it from being decremented when they are released.
251 void tipc_sock_release_local(struct socket
*sock
)
259 * tipc_sock_accept_local - accept a connection on a socket created
260 * with tipc_sock_create_local. Use this function to avoid that
261 * module reference count is inadvertently incremented.
263 * @sock: the accepting socket
264 * @newsock: reference to the new socket to be created
265 * @flags: socket flags
268 int tipc_sock_accept_local(struct socket
*sock
, struct socket
**newsock
,
271 struct sock
*sk
= sock
->sk
;
274 ret
= sock_create_lite(sk
->sk_family
, sk
->sk_type
,
275 sk
->sk_protocol
, newsock
);
279 ret
= tipc_accept(sock
, *newsock
, flags
);
281 sock_release(*newsock
);
284 (*newsock
)->ops
= sock
->ops
;
289 * tipc_release - destroy a TIPC socket
290 * @sock: socket to destroy
292 * This routine cleans up any messages that are still queued on the socket.
293 * For DGRAM and RDM socket types, all queued messages are rejected.
294 * For SEQPACKET and STREAM socket types, the first message is rejected
295 * and any others are discarded. (If the first message on a STREAM socket
296 * is partially-read, it is discarded and the next one is rejected instead.)
298 * NOTE: Rejected messages are not necessarily returned to the sender! They
299 * are returned or discarded according to the "destination droppable" setting
300 * specified for the message by the sender.
302 * Returns 0 on success, errno otherwise
304 static int tipc_release(struct socket
*sock
)
306 struct sock
*sk
= sock
->sk
;
307 struct tipc_sock
*tsk
;
308 struct tipc_port
*port
;
313 * Exit if socket isn't fully initialized (occurs when a failed accept()
314 * releases a pre-allocated child socket that was never used)
324 * Reject all unreceived messages, except on an active connection
325 * (which disconnects locally & sends a 'FIN+' to peer)
327 while (sock
->state
!= SS_DISCONNECTING
) {
328 buf
= __skb_dequeue(&sk
->sk_receive_queue
);
331 if (TIPC_SKB_CB(buf
)->handle
!= NULL
)
334 if ((sock
->state
== SS_CONNECTING
) ||
335 (sock
->state
== SS_CONNECTED
)) {
336 sock
->state
= SS_DISCONNECTING
;
337 tipc_port_disconnect(port
->ref
);
339 if (tipc_msg_reverse(buf
, &dnode
, TIPC_ERR_NO_PORT
))
340 tipc_link_xmit2(buf
, dnode
, 0);
344 /* Destroy TIPC port; also disconnects an active connection and
345 * sends a 'FIN-' to peer.
347 tipc_port_destroy(port
);
349 /* Discard any remaining (connection-based) messages in receive queue */
350 __skb_queue_purge(&sk
->sk_receive_queue
);
352 /* Reject any messages that accumulated in backlog queue */
353 sock
->state
= SS_DISCONNECTING
;
363 * tipc_bind - associate or disassocate TIPC name(s) with a socket
364 * @sock: socket structure
365 * @uaddr: socket address describing name(s) and desired operation
366 * @uaddr_len: size of socket address data structure
368 * Name and name sequence binding is indicated using a positive scope value;
369 * a negative scope value unbinds the specified name. Specifying no name
370 * (i.e. a socket address length of 0) unbinds all names from the socket.
372 * Returns 0 on success, errno otherwise
374 * NOTE: This routine doesn't need to take the socket lock since it doesn't
375 * access any non-constant socket information.
377 static int tipc_bind(struct socket
*sock
, struct sockaddr
*uaddr
,
380 struct sock
*sk
= sock
->sk
;
381 struct sockaddr_tipc
*addr
= (struct sockaddr_tipc
*)uaddr
;
382 struct tipc_sock
*tsk
= tipc_sk(sk
);
386 if (unlikely(!uaddr_len
)) {
387 res
= tipc_withdraw(&tsk
->port
, 0, NULL
);
391 if (uaddr_len
< sizeof(struct sockaddr_tipc
)) {
395 if (addr
->family
!= AF_TIPC
) {
400 if (addr
->addrtype
== TIPC_ADDR_NAME
)
401 addr
->addr
.nameseq
.upper
= addr
->addr
.nameseq
.lower
;
402 else if (addr
->addrtype
!= TIPC_ADDR_NAMESEQ
) {
407 if ((addr
->addr
.nameseq
.type
< TIPC_RESERVED_TYPES
) &&
408 (addr
->addr
.nameseq
.type
!= TIPC_TOP_SRV
) &&
409 (addr
->addr
.nameseq
.type
!= TIPC_CFG_SRV
)) {
414 res
= (addr
->scope
> 0) ?
415 tipc_publish(&tsk
->port
, addr
->scope
, &addr
->addr
.nameseq
) :
416 tipc_withdraw(&tsk
->port
, -addr
->scope
, &addr
->addr
.nameseq
);
423 * tipc_getname - get port ID of socket or peer socket
424 * @sock: socket structure
425 * @uaddr: area for returned socket address
426 * @uaddr_len: area for returned length of socket address
427 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
429 * Returns 0 on success, errno otherwise
431 * NOTE: This routine doesn't need to take the socket lock since it only
432 * accesses socket information that is unchanging (or which changes in
433 * a completely predictable manner).
435 static int tipc_getname(struct socket
*sock
, struct sockaddr
*uaddr
,
436 int *uaddr_len
, int peer
)
438 struct sockaddr_tipc
*addr
= (struct sockaddr_tipc
*)uaddr
;
439 struct tipc_sock
*tsk
= tipc_sk(sock
->sk
);
441 memset(addr
, 0, sizeof(*addr
));
443 if ((sock
->state
!= SS_CONNECTED
) &&
444 ((peer
!= 2) || (sock
->state
!= SS_DISCONNECTING
)))
446 addr
->addr
.id
.ref
= tipc_port_peerport(&tsk
->port
);
447 addr
->addr
.id
.node
= tipc_port_peernode(&tsk
->port
);
449 addr
->addr
.id
.ref
= tsk
->port
.ref
;
450 addr
->addr
.id
.node
= tipc_own_addr
;
453 *uaddr_len
= sizeof(*addr
);
454 addr
->addrtype
= TIPC_ADDR_ID
;
455 addr
->family
= AF_TIPC
;
457 addr
->addr
.name
.domain
= 0;
463 * tipc_poll - read and possibly block on pollmask
464 * @file: file structure associated with the socket
465 * @sock: socket for which to calculate the poll bits
468 * Returns pollmask value
471 * It appears that the usual socket locking mechanisms are not useful here
472 * since the pollmask info is potentially out-of-date the moment this routine
473 * exits. TCP and other protocols seem to rely on higher level poll routines
474 * to handle any preventable race conditions, so TIPC will do the same ...
476 * TIPC sets the returned events as follows:
478 * socket state flags set
479 * ------------ ---------
480 * unconnected no read flags
481 * POLLOUT if port is not congested
483 * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue
486 * connected POLLIN/POLLRDNORM if data in rx queue
487 * POLLOUT if port is not congested
489 * disconnecting POLLIN/POLLRDNORM/POLLHUP
492 * listening POLLIN if SYN in rx queue
495 * ready POLLIN/POLLRDNORM if data in rx queue
496 * [connectionless] POLLOUT (since port cannot be congested)
498 * IMPORTANT: The fact that a read or write operation is indicated does NOT
499 * imply that the operation will succeed, merely that it should be performed
500 * and will not block.
502 static unsigned int tipc_poll(struct file
*file
, struct socket
*sock
,
505 struct sock
*sk
= sock
->sk
;
506 struct tipc_sock
*tsk
= tipc_sk(sk
);
509 sock_poll_wait(file
, sk_sleep(sk
), wait
);
511 switch ((int)sock
->state
) {
513 if (!tsk
->port
.congested
)
518 if (!tsk
->port
.congested
)
523 if (!skb_queue_empty(&sk
->sk_receive_queue
))
524 mask
|= (POLLIN
| POLLRDNORM
);
526 case SS_DISCONNECTING
:
527 mask
= (POLLIN
| POLLRDNORM
| POLLHUP
);
535 * dest_name_check - verify user is permitted to send to specified port name
536 * @dest: destination address
537 * @m: descriptor for message to be sent
539 * Prevents restricted configuration commands from being issued by
540 * unauthorized users.
542 * Returns 0 if permission is granted, otherwise errno
544 static int dest_name_check(struct sockaddr_tipc
*dest
, struct msghdr
*m
)
546 struct tipc_cfg_msg_hdr hdr
;
548 if (likely(dest
->addr
.name
.name
.type
>= TIPC_RESERVED_TYPES
))
550 if (likely(dest
->addr
.name
.name
.type
== TIPC_TOP_SRV
))
552 if (likely(dest
->addr
.name
.name
.type
!= TIPC_CFG_SRV
))
555 if (!m
->msg_iovlen
|| (m
->msg_iov
[0].iov_len
< sizeof(hdr
)))
557 if (copy_from_user(&hdr
, m
->msg_iov
[0].iov_base
, sizeof(hdr
)))
559 if ((ntohs(hdr
.tcm_type
) & 0xC000) && (!capable(CAP_NET_ADMIN
)))
565 static int tipc_wait_for_sndmsg(struct socket
*sock
, long *timeo_p
)
567 struct sock
*sk
= sock
->sk
;
568 struct tipc_sock
*tsk
= tipc_sk(sk
);
573 int err
= sock_error(sk
);
576 if (sock
->state
== SS_DISCONNECTING
)
580 if (signal_pending(current
))
581 return sock_intr_errno(*timeo_p
);
583 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
584 done
= sk_wait_event(sk
, timeo_p
, !tsk
->port
.congested
);
585 finish_wait(sk_sleep(sk
), &wait
);
592 * tipc_sendmsg - send message in connectionless manner
593 * @iocb: if NULL, indicates that socket lock is already held
594 * @sock: socket structure
595 * @m: message to send
596 * @total_len: length of message
598 * Message must have an destination specified explicitly.
599 * Used for SOCK_RDM and SOCK_DGRAM messages,
600 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
601 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
603 * Returns the number of bytes sent on success, or errno otherwise
605 static int tipc_sendmsg(struct kiocb
*iocb
, struct socket
*sock
,
606 struct msghdr
*m
, size_t total_len
)
608 struct sock
*sk
= sock
->sk
;
609 struct tipc_sock
*tsk
= tipc_sk(sk
);
610 struct tipc_port
*port
= &tsk
->port
;
611 DECLARE_SOCKADDR(struct sockaddr_tipc
*, dest
, m
->msg_name
);
617 return -EDESTADDRREQ
;
618 if (unlikely((m
->msg_namelen
< sizeof(*dest
)) ||
619 (dest
->family
!= AF_TIPC
)))
621 if (total_len
> TIPC_MAX_USER_MSG_SIZE
)
627 needs_conn
= (sock
->state
!= SS_READY
);
628 if (unlikely(needs_conn
)) {
629 if (sock
->state
== SS_LISTENING
) {
633 if (sock
->state
!= SS_UNCONNECTED
) {
637 if (tsk
->port
.published
) {
641 if (dest
->addrtype
== TIPC_ADDR_NAME
) {
642 tsk
->port
.conn_type
= dest
->addr
.name
.name
.type
;
643 tsk
->port
.conn_instance
= dest
->addr
.name
.name
.instance
;
646 /* Abort any pending connection attempts (very unlikely) */
650 timeo
= sock_sndtimeo(sk
, m
->msg_flags
& MSG_DONTWAIT
);
652 if (dest
->addrtype
== TIPC_ADDR_NAME
) {
653 res
= dest_name_check(dest
, m
);
656 res
= tipc_send2name(port
,
657 &dest
->addr
.name
.name
,
658 dest
->addr
.name
.domain
,
661 } else if (dest
->addrtype
== TIPC_ADDR_ID
) {
662 res
= tipc_send2port(port
,
666 } else if (dest
->addrtype
== TIPC_ADDR_MCAST
) {
671 res
= dest_name_check(dest
, m
);
674 res
= tipc_port_mcast_xmit(port
,
679 if (likely(res
!= -ELINKCONG
)) {
680 if (needs_conn
&& (res
>= 0))
681 sock
->state
= SS_CONNECTING
;
684 res
= tipc_wait_for_sndmsg(sock
, &timeo
);
695 static int tipc_wait_for_sndpkt(struct socket
*sock
, long *timeo_p
)
697 struct sock
*sk
= sock
->sk
;
698 struct tipc_sock
*tsk
= tipc_sk(sk
);
699 struct tipc_port
*port
= &tsk
->port
;
704 int err
= sock_error(sk
);
707 if (sock
->state
== SS_DISCONNECTING
)
709 else if (sock
->state
!= SS_CONNECTED
)
713 if (signal_pending(current
))
714 return sock_intr_errno(*timeo_p
);
716 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
717 done
= sk_wait_event(sk
, timeo_p
,
718 (!port
->congested
|| !port
->connected
));
719 finish_wait(sk_sleep(sk
), &wait
);
725 * tipc_send_packet - send a connection-oriented message
726 * @iocb: if NULL, indicates that socket lock is already held
727 * @sock: socket structure
728 * @m: message to send
729 * @total_len: length of message
731 * Used for SOCK_SEQPACKET messages and SOCK_STREAM data.
733 * Returns the number of bytes sent on success, or errno otherwise
735 static int tipc_send_packet(struct kiocb
*iocb
, struct socket
*sock
,
736 struct msghdr
*m
, size_t total_len
)
738 struct sock
*sk
= sock
->sk
;
739 struct tipc_sock
*tsk
= tipc_sk(sk
);
740 DECLARE_SOCKADDR(struct sockaddr_tipc
*, dest
, m
->msg_name
);
744 /* Handle implied connection establishment */
746 return tipc_sendmsg(iocb
, sock
, m
, total_len
);
748 if (total_len
> TIPC_MAX_USER_MSG_SIZE
)
754 if (unlikely(sock
->state
!= SS_CONNECTED
)) {
755 if (sock
->state
== SS_DISCONNECTING
)
762 timeo
= sock_sndtimeo(sk
, m
->msg_flags
& MSG_DONTWAIT
);
764 res
= tipc_send(&tsk
->port
, m
->msg_iov
, total_len
);
765 if (likely(res
!= -ELINKCONG
))
767 res
= tipc_wait_for_sndpkt(sock
, &timeo
);
778 * tipc_send_stream - send stream-oriented data
780 * @sock: socket structure
782 * @total_len: total length of data to be sent
784 * Used for SOCK_STREAM data.
786 * Returns the number of bytes sent on success (or partial success),
787 * or errno if no data sent
789 static int tipc_send_stream(struct kiocb
*iocb
, struct socket
*sock
,
790 struct msghdr
*m
, size_t total_len
)
792 struct sock
*sk
= sock
->sk
;
793 struct tipc_sock
*tsk
= tipc_sk(sk
);
794 struct msghdr my_msg
;
796 struct iovec
*curr_iov
;
798 char __user
*curr_start
;
807 /* Handle special cases where there is no connection */
808 if (unlikely(sock
->state
!= SS_CONNECTED
)) {
809 if (sock
->state
== SS_UNCONNECTED
)
810 res
= tipc_send_packet(NULL
, sock
, m
, total_len
);
812 res
= sock
->state
== SS_DISCONNECTING
? -EPIPE
: -ENOTCONN
;
816 if (unlikely(m
->msg_name
)) {
821 if (total_len
> (unsigned int)INT_MAX
) {
827 * Send each iovec entry using one or more messages
829 * Note: This algorithm is good for the most likely case
830 * (i.e. one large iovec entry), but could be improved to pass sets
831 * of small iovec entries into send_packet().
833 curr_iov
= m
->msg_iov
;
834 curr_iovlen
= m
->msg_iovlen
;
835 my_msg
.msg_iov
= &my_iov
;
836 my_msg
.msg_iovlen
= 1;
837 my_msg
.msg_flags
= m
->msg_flags
;
838 my_msg
.msg_name
= NULL
;
841 hdr_size
= msg_hdr_sz(&tsk
->port
.phdr
);
843 while (curr_iovlen
--) {
844 curr_start
= curr_iov
->iov_base
;
845 curr_left
= curr_iov
->iov_len
;
848 bytes_to_send
= tsk
->port
.max_pkt
- hdr_size
;
849 if (bytes_to_send
> TIPC_MAX_USER_MSG_SIZE
)
850 bytes_to_send
= TIPC_MAX_USER_MSG_SIZE
;
851 if (curr_left
< bytes_to_send
)
852 bytes_to_send
= curr_left
;
853 my_iov
.iov_base
= curr_start
;
854 my_iov
.iov_len
= bytes_to_send
;
855 res
= tipc_send_packet(NULL
, sock
, &my_msg
,
862 curr_left
-= bytes_to_send
;
863 curr_start
+= bytes_to_send
;
864 bytes_sent
+= bytes_to_send
;
876 * auto_connect - complete connection setup to a remote port
877 * @tsk: tipc socket structure
878 * @msg: peer's response message
880 * Returns 0 on success, errno otherwise
882 static int auto_connect(struct tipc_sock
*tsk
, struct tipc_msg
*msg
)
884 struct tipc_port
*port
= &tsk
->port
;
885 struct socket
*sock
= tsk
->sk
.sk_socket
;
886 struct tipc_portid peer
;
888 peer
.ref
= msg_origport(msg
);
889 peer
.node
= msg_orignode(msg
);
891 __tipc_port_connect(port
->ref
, port
, &peer
);
893 if (msg_importance(msg
) > TIPC_CRITICAL_IMPORTANCE
)
895 msg_set_importance(&port
->phdr
, (u32
)msg_importance(msg
));
896 sock
->state
= SS_CONNECTED
;
901 * set_orig_addr - capture sender's address for received message
902 * @m: descriptor for message info
903 * @msg: received message header
905 * Note: Address is not captured if not requested by receiver.
907 static void set_orig_addr(struct msghdr
*m
, struct tipc_msg
*msg
)
909 DECLARE_SOCKADDR(struct sockaddr_tipc
*, addr
, m
->msg_name
);
912 addr
->family
= AF_TIPC
;
913 addr
->addrtype
= TIPC_ADDR_ID
;
914 memset(&addr
->addr
, 0, sizeof(addr
->addr
));
915 addr
->addr
.id
.ref
= msg_origport(msg
);
916 addr
->addr
.id
.node
= msg_orignode(msg
);
917 addr
->addr
.name
.domain
= 0; /* could leave uninitialized */
918 addr
->scope
= 0; /* could leave uninitialized */
919 m
->msg_namelen
= sizeof(struct sockaddr_tipc
);
924 * anc_data_recv - optionally capture ancillary data for received message
925 * @m: descriptor for message info
926 * @msg: received message header
927 * @tport: TIPC port associated with message
929 * Note: Ancillary data is not captured if not requested by receiver.
931 * Returns 0 if successful, otherwise errno
933 static int anc_data_recv(struct msghdr
*m
, struct tipc_msg
*msg
,
934 struct tipc_port
*tport
)
942 if (likely(m
->msg_controllen
== 0))
945 /* Optionally capture errored message object(s) */
946 err
= msg
? msg_errcode(msg
) : 0;
949 anc_data
[1] = msg_data_sz(msg
);
950 res
= put_cmsg(m
, SOL_TIPC
, TIPC_ERRINFO
, 8, anc_data
);
954 res
= put_cmsg(m
, SOL_TIPC
, TIPC_RETDATA
, anc_data
[1],
961 /* Optionally capture message destination object */
962 dest_type
= msg
? msg_type(msg
) : TIPC_DIRECT_MSG
;
966 anc_data
[0] = msg_nametype(msg
);
967 anc_data
[1] = msg_namelower(msg
);
968 anc_data
[2] = msg_namelower(msg
);
972 anc_data
[0] = msg_nametype(msg
);
973 anc_data
[1] = msg_namelower(msg
);
974 anc_data
[2] = msg_nameupper(msg
);
977 has_name
= (tport
->conn_type
!= 0);
978 anc_data
[0] = tport
->conn_type
;
979 anc_data
[1] = tport
->conn_instance
;
980 anc_data
[2] = tport
->conn_instance
;
986 res
= put_cmsg(m
, SOL_TIPC
, TIPC_DESTNAME
, 12, anc_data
);
994 static int tipc_wait_for_rcvmsg(struct socket
*sock
, long *timeop
)
996 struct sock
*sk
= sock
->sk
;
998 long timeo
= *timeop
;
1002 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
1003 if (timeo
&& skb_queue_empty(&sk
->sk_receive_queue
)) {
1004 if (sock
->state
== SS_DISCONNECTING
) {
1009 timeo
= schedule_timeout(timeo
);
1013 if (!skb_queue_empty(&sk
->sk_receive_queue
))
1015 err
= sock_intr_errno(timeo
);
1016 if (signal_pending(current
))
1022 finish_wait(sk_sleep(sk
), &wait
);
1028 * tipc_recvmsg - receive packet-oriented message
1030 * @m: descriptor for message info
1031 * @buf_len: total size of user buffer area
1032 * @flags: receive flags
1034 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1035 * If the complete message doesn't fit in user area, truncate it.
1037 * Returns size of returned message data, errno otherwise
1039 static int tipc_recvmsg(struct kiocb
*iocb
, struct socket
*sock
,
1040 struct msghdr
*m
, size_t buf_len
, int flags
)
1042 struct sock
*sk
= sock
->sk
;
1043 struct tipc_sock
*tsk
= tipc_sk(sk
);
1044 struct tipc_port
*port
= &tsk
->port
;
1045 struct sk_buff
*buf
;
1046 struct tipc_msg
*msg
;
1052 /* Catch invalid receive requests */
1053 if (unlikely(!buf_len
))
1058 if (unlikely(sock
->state
== SS_UNCONNECTED
)) {
1063 timeo
= sock_rcvtimeo(sk
, flags
& MSG_DONTWAIT
);
1066 /* Look for a message in receive queue; wait if necessary */
1067 res
= tipc_wait_for_rcvmsg(sock
, &timeo
);
1071 /* Look at first message in receive queue */
1072 buf
= skb_peek(&sk
->sk_receive_queue
);
1074 sz
= msg_data_sz(msg
);
1075 err
= msg_errcode(msg
);
1077 /* Discard an empty non-errored message & try again */
1078 if ((!sz
) && (!err
)) {
1079 advance_rx_queue(sk
);
1083 /* Capture sender's address (optional) */
1084 set_orig_addr(m
, msg
);
1086 /* Capture ancillary data (optional) */
1087 res
= anc_data_recv(m
, msg
, port
);
1091 /* Capture message data (if valid) & compute return value (always) */
1093 if (unlikely(buf_len
< sz
)) {
1095 m
->msg_flags
|= MSG_TRUNC
;
1097 res
= skb_copy_datagram_iovec(buf
, msg_hdr_sz(msg
),
1103 if ((sock
->state
== SS_READY
) ||
1104 ((err
== TIPC_CONN_SHUTDOWN
) || m
->msg_control
))
1110 /* Consume received message (optional) */
1111 if (likely(!(flags
& MSG_PEEK
))) {
1112 if ((sock
->state
!= SS_READY
) &&
1113 (++port
->conn_unacked
>= TIPC_CONNACK_INTV
))
1114 tipc_acknowledge(port
->ref
, port
->conn_unacked
);
1115 advance_rx_queue(sk
);
1123 * tipc_recv_stream - receive stream-oriented data
1125 * @m: descriptor for message info
1126 * @buf_len: total size of user buffer area
1127 * @flags: receive flags
1129 * Used for SOCK_STREAM messages only. If not enough data is available
1130 * will optionally wait for more; never truncates data.
1132 * Returns size of returned message data, errno otherwise
1134 static int tipc_recv_stream(struct kiocb
*iocb
, struct socket
*sock
,
1135 struct msghdr
*m
, size_t buf_len
, int flags
)
1137 struct sock
*sk
= sock
->sk
;
1138 struct tipc_sock
*tsk
= tipc_sk(sk
);
1139 struct tipc_port
*port
= &tsk
->port
;
1140 struct sk_buff
*buf
;
1141 struct tipc_msg
*msg
;
1144 int sz_to_copy
, target
, needed
;
1149 /* Catch invalid receive attempts */
1150 if (unlikely(!buf_len
))
1155 if (unlikely(sock
->state
== SS_UNCONNECTED
)) {
1160 target
= sock_rcvlowat(sk
, flags
& MSG_WAITALL
, buf_len
);
1161 timeo
= sock_rcvtimeo(sk
, flags
& MSG_DONTWAIT
);
1164 /* Look for a message in receive queue; wait if necessary */
1165 res
= tipc_wait_for_rcvmsg(sock
, &timeo
);
1169 /* Look at first message in receive queue */
1170 buf
= skb_peek(&sk
->sk_receive_queue
);
1172 sz
= msg_data_sz(msg
);
1173 err
= msg_errcode(msg
);
1175 /* Discard an empty non-errored message & try again */
1176 if ((!sz
) && (!err
)) {
1177 advance_rx_queue(sk
);
1181 /* Optionally capture sender's address & ancillary data of first msg */
1182 if (sz_copied
== 0) {
1183 set_orig_addr(m
, msg
);
1184 res
= anc_data_recv(m
, msg
, port
);
1189 /* Capture message data (if valid) & compute return value (always) */
1191 u32 offset
= (u32
)(unsigned long)(TIPC_SKB_CB(buf
)->handle
);
1194 needed
= (buf_len
- sz_copied
);
1195 sz_to_copy
= (sz
<= needed
) ? sz
: needed
;
1197 res
= skb_copy_datagram_iovec(buf
, msg_hdr_sz(msg
) + offset
,
1198 m
->msg_iov
, sz_to_copy
);
1202 sz_copied
+= sz_to_copy
;
1204 if (sz_to_copy
< sz
) {
1205 if (!(flags
& MSG_PEEK
))
1206 TIPC_SKB_CB(buf
)->handle
=
1207 (void *)(unsigned long)(offset
+ sz_to_copy
);
1212 goto exit
; /* can't add error msg to valid data */
1214 if ((err
== TIPC_CONN_SHUTDOWN
) || m
->msg_control
)
1220 /* Consume received message (optional) */
1221 if (likely(!(flags
& MSG_PEEK
))) {
1222 if (unlikely(++port
->conn_unacked
>= TIPC_CONNACK_INTV
))
1223 tipc_acknowledge(port
->ref
, port
->conn_unacked
);
1224 advance_rx_queue(sk
);
1227 /* Loop around if more data is required */
1228 if ((sz_copied
< buf_len
) && /* didn't get all requested data */
1229 (!skb_queue_empty(&sk
->sk_receive_queue
) ||
1230 (sz_copied
< target
)) && /* and more is ready or required */
1231 (!(flags
& MSG_PEEK
)) && /* and aren't just peeking at data */
1232 (!err
)) /* and haven't reached a FIN */
1237 return sz_copied
? sz_copied
: res
;
1241 * tipc_write_space - wake up thread if port congestion is released
1244 static void tipc_write_space(struct sock
*sk
)
1246 struct socket_wq
*wq
;
1249 wq
= rcu_dereference(sk
->sk_wq
);
1250 if (wq_has_sleeper(wq
))
1251 wake_up_interruptible_sync_poll(&wq
->wait
, POLLOUT
|
1252 POLLWRNORM
| POLLWRBAND
);
1257 * tipc_data_ready - wake up threads to indicate messages have been received
1259 * @len: the length of messages
1261 static void tipc_data_ready(struct sock
*sk
)
1263 struct socket_wq
*wq
;
1266 wq
= rcu_dereference(sk
->sk_wq
);
1267 if (wq_has_sleeper(wq
))
1268 wake_up_interruptible_sync_poll(&wq
->wait
, POLLIN
|
1269 POLLRDNORM
| POLLRDBAND
);
1274 * filter_connect - Handle all incoming messages for a connection-based socket
1278 * Returns 0 (TIPC_OK) if everyting ok, -TIPC_ERR_NO_PORT otherwise
1280 static int filter_connect(struct tipc_sock
*tsk
, struct sk_buff
**buf
)
1282 struct sock
*sk
= &tsk
->sk
;
1283 struct tipc_port
*port
= &tsk
->port
;
1284 struct socket
*sock
= sk
->sk_socket
;
1285 struct tipc_msg
*msg
= buf_msg(*buf
);
1287 int retval
= -TIPC_ERR_NO_PORT
;
1293 switch ((int)sock
->state
) {
1295 /* Accept only connection-based messages sent by peer */
1296 if (msg_connected(msg
) && tipc_port_peer_msg(port
, msg
)) {
1297 if (unlikely(msg_errcode(msg
))) {
1298 sock
->state
= SS_DISCONNECTING
;
1299 __tipc_port_disconnect(port
);
1305 /* Accept only ACK or NACK message */
1306 if (unlikely(msg_errcode(msg
))) {
1307 sock
->state
= SS_DISCONNECTING
;
1308 sk
->sk_err
= ECONNREFUSED
;
1313 if (unlikely(!msg_connected(msg
)))
1316 res
= auto_connect(tsk
, msg
);
1318 sock
->state
= SS_DISCONNECTING
;
1324 /* If an incoming message is an 'ACK-', it should be
1325 * discarded here because it doesn't contain useful
1326 * data. In addition, we should try to wake up
1327 * connect() routine if sleeping.
1329 if (msg_data_sz(msg
) == 0) {
1332 if (waitqueue_active(sk_sleep(sk
)))
1333 wake_up_interruptible(sk_sleep(sk
));
1338 case SS_UNCONNECTED
:
1339 /* Accept only SYN message */
1340 if (!msg_connected(msg
) && !(msg_errcode(msg
)))
1343 case SS_DISCONNECTING
:
1346 pr_err("Unknown socket state %u\n", sock
->state
);
1352 * rcvbuf_limit - get proper overload limit of socket receive queue
1356 * For all connection oriented messages, irrespective of importance,
1357 * the default overload value (i.e. 67MB) is set as limit.
1359 * For all connectionless messages, by default new queue limits are
1362 * TIPC_LOW_IMPORTANCE (4 MB)
1363 * TIPC_MEDIUM_IMPORTANCE (8 MB)
1364 * TIPC_HIGH_IMPORTANCE (16 MB)
1365 * TIPC_CRITICAL_IMPORTANCE (32 MB)
1367 * Returns overload limit according to corresponding message importance
1369 static unsigned int rcvbuf_limit(struct sock
*sk
, struct sk_buff
*buf
)
1371 struct tipc_msg
*msg
= buf_msg(buf
);
1373 if (msg_connected(msg
))
1374 return sysctl_tipc_rmem
[2];
1376 return sk
->sk_rcvbuf
>> TIPC_CRITICAL_IMPORTANCE
<<
1377 msg_importance(msg
);
1381 * filter_rcv - validate incoming message
1385 * Enqueues message on receive queue if acceptable; optionally handles
1386 * disconnect indication for a connected socket.
1388 * Called with socket lock already taken; port lock may also be taken.
1390 * Returns 0 (TIPC_OK) if message was consumed, -TIPC error code if message
1393 static int filter_rcv(struct sock
*sk
, struct sk_buff
*buf
)
1395 struct socket
*sock
= sk
->sk_socket
;
1396 struct tipc_sock
*tsk
= tipc_sk(sk
);
1397 struct tipc_msg
*msg
= buf_msg(buf
);
1398 unsigned int limit
= rcvbuf_limit(sk
, buf
);
1401 /* Reject message if it is wrong sort of message for socket */
1402 if (msg_type(msg
) > TIPC_DIRECT_MSG
)
1403 return -TIPC_ERR_NO_PORT
;
1405 if (sock
->state
== SS_READY
) {
1406 if (msg_connected(msg
))
1407 return -TIPC_ERR_NO_PORT
;
1409 rc
= filter_connect(tsk
, &buf
);
1410 if (rc
!= TIPC_OK
|| buf
== NULL
)
1414 /* Reject message if there isn't room to queue it */
1415 if (sk_rmem_alloc_get(sk
) + buf
->truesize
>= limit
)
1416 return -TIPC_ERR_OVERLOAD
;
1418 /* Enqueue message */
1419 TIPC_SKB_CB(buf
)->handle
= NULL
;
1420 __skb_queue_tail(&sk
->sk_receive_queue
, buf
);
1421 skb_set_owner_r(buf
, sk
);
1423 sk
->sk_data_ready(sk
);
1428 * tipc_backlog_rcv - handle incoming message from backlog queue
1432 * Caller must hold socket lock, but not port lock.
1436 static int tipc_backlog_rcv(struct sock
*sk
, struct sk_buff
*buf
)
1440 struct tipc_sock
*tsk
= tipc_sk(sk
);
1441 uint truesize
= buf
->truesize
;
1443 rc
= filter_rcv(sk
, buf
);
1445 if (unlikely(rc
&& tipc_msg_reverse(buf
, &onode
, -rc
)))
1446 tipc_link_xmit2(buf
, onode
, 0);
1447 else if (atomic_read(&tsk
->dupl_rcvcnt
) < TIPC_CONN_OVERLOAD_LIMIT
)
1448 atomic_add(truesize
, &tsk
->dupl_rcvcnt
);
1454 * tipc_sk_rcv - handle incoming message
1455 * @buf: buffer containing arriving message
1457 * Returns 0 if success, or errno: -EHOSTUNREACH
1459 int tipc_sk_rcv(struct sk_buff
*buf
)
1461 struct tipc_sock
*tsk
;
1462 struct tipc_port
*port
;
1464 u32 dport
= msg_destport(buf_msg(buf
));
1469 /* Forward unresolved named message */
1470 if (unlikely(!dport
)) {
1471 tipc_net_route_msg(buf
);
1475 /* Validate destination */
1476 port
= tipc_port_lock(dport
);
1477 if (unlikely(!port
)) {
1478 rc
= -TIPC_ERR_NO_PORT
;
1482 tsk
= tipc_port_to_sock(port
);
1488 if (!sock_owned_by_user(sk
)) {
1489 rc
= filter_rcv(sk
, buf
);
1491 if (sk
->sk_backlog
.len
== 0)
1492 atomic_set(&tsk
->dupl_rcvcnt
, 0);
1493 limit
= rcvbuf_limit(sk
, buf
) + atomic_read(&tsk
->dupl_rcvcnt
);
1494 if (sk_add_backlog(sk
, buf
, limit
))
1495 rc
= -TIPC_ERR_OVERLOAD
;
1499 tipc_port_unlock(port
);
1504 if (!tipc_msg_reverse(buf
, &dnode
, -rc
))
1505 return -EHOSTUNREACH
;
1506 tipc_link_xmit2(buf
, dnode
, 0);
1507 return -EHOSTUNREACH
;
1510 static int tipc_wait_for_connect(struct socket
*sock
, long *timeo_p
)
1512 struct sock
*sk
= sock
->sk
;
1517 int err
= sock_error(sk
);
1522 if (signal_pending(current
))
1523 return sock_intr_errno(*timeo_p
);
1525 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
1526 done
= sk_wait_event(sk
, timeo_p
, sock
->state
!= SS_CONNECTING
);
1527 finish_wait(sk_sleep(sk
), &wait
);
1533 * tipc_connect - establish a connection to another TIPC port
1534 * @sock: socket structure
1535 * @dest: socket address for destination port
1536 * @destlen: size of socket address data structure
1537 * @flags: file-related flags associated with socket
1539 * Returns 0 on success, errno otherwise
1541 static int tipc_connect(struct socket
*sock
, struct sockaddr
*dest
,
1542 int destlen
, int flags
)
1544 struct sock
*sk
= sock
->sk
;
1545 struct sockaddr_tipc
*dst
= (struct sockaddr_tipc
*)dest
;
1546 struct msghdr m
= {NULL
,};
1547 long timeout
= (flags
& O_NONBLOCK
) ? 0 : tipc_sk(sk
)->conn_timeout
;
1548 socket_state previous
;
1553 /* For now, TIPC does not allow use of connect() with DGRAM/RDM types */
1554 if (sock
->state
== SS_READY
) {
1560 * Reject connection attempt using multicast address
1562 * Note: send_msg() validates the rest of the address fields,
1563 * so there's no need to do it here
1565 if (dst
->addrtype
== TIPC_ADDR_MCAST
) {
1570 previous
= sock
->state
;
1571 switch (sock
->state
) {
1572 case SS_UNCONNECTED
:
1573 /* Send a 'SYN-' to destination */
1575 m
.msg_namelen
= destlen
;
1577 /* If connect is in non-blocking case, set MSG_DONTWAIT to
1578 * indicate send_msg() is never blocked.
1581 m
.msg_flags
= MSG_DONTWAIT
;
1583 res
= tipc_sendmsg(NULL
, sock
, &m
, 0);
1584 if ((res
< 0) && (res
!= -EWOULDBLOCK
))
1587 /* Just entered SS_CONNECTING state; the only
1588 * difference is that return value in non-blocking
1589 * case is EINPROGRESS, rather than EALREADY.
1593 if (previous
== SS_CONNECTING
)
1597 timeout
= msecs_to_jiffies(timeout
);
1598 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
1599 res
= tipc_wait_for_connect(sock
, &timeout
);
1614 * tipc_listen - allow socket to listen for incoming connections
1615 * @sock: socket structure
1618 * Returns 0 on success, errno otherwise
1620 static int tipc_listen(struct socket
*sock
, int len
)
1622 struct sock
*sk
= sock
->sk
;
1627 if (sock
->state
!= SS_UNCONNECTED
)
1630 sock
->state
= SS_LISTENING
;
1638 static int tipc_wait_for_accept(struct socket
*sock
, long timeo
)
1640 struct sock
*sk
= sock
->sk
;
1644 /* True wake-one mechanism for incoming connections: only
1645 * one process gets woken up, not the 'whole herd'.
1646 * Since we do not 'race & poll' for established sockets
1647 * anymore, the common case will execute the loop only once.
1650 prepare_to_wait_exclusive(sk_sleep(sk
), &wait
,
1651 TASK_INTERRUPTIBLE
);
1652 if (timeo
&& skb_queue_empty(&sk
->sk_receive_queue
)) {
1654 timeo
= schedule_timeout(timeo
);
1658 if (!skb_queue_empty(&sk
->sk_receive_queue
))
1661 if (sock
->state
!= SS_LISTENING
)
1663 err
= sock_intr_errno(timeo
);
1664 if (signal_pending(current
))
1670 finish_wait(sk_sleep(sk
), &wait
);
1675 * tipc_accept - wait for connection request
1676 * @sock: listening socket
1677 * @newsock: new socket that is to be connected
1678 * @flags: file-related flags associated with socket
1680 * Returns 0 on success, errno otherwise
1682 static int tipc_accept(struct socket
*sock
, struct socket
*new_sock
, int flags
)
1684 struct sock
*new_sk
, *sk
= sock
->sk
;
1685 struct sk_buff
*buf
;
1686 struct tipc_port
*new_port
;
1687 struct tipc_msg
*msg
;
1688 struct tipc_portid peer
;
1695 if (sock
->state
!= SS_LISTENING
) {
1699 timeo
= sock_rcvtimeo(sk
, flags
& O_NONBLOCK
);
1700 res
= tipc_wait_for_accept(sock
, timeo
);
1704 buf
= skb_peek(&sk
->sk_receive_queue
);
1706 res
= tipc_sk_create(sock_net(sock
->sk
), new_sock
, 0, 1);
1710 new_sk
= new_sock
->sk
;
1711 new_port
= &tipc_sk(new_sk
)->port
;
1712 new_ref
= new_port
->ref
;
1715 /* we lock on new_sk; but lockdep sees the lock on sk */
1716 lock_sock_nested(new_sk
, SINGLE_DEPTH_NESTING
);
1719 * Reject any stray messages received by new socket
1720 * before the socket lock was taken (very, very unlikely)
1722 reject_rx_queue(new_sk
);
1724 /* Connect new socket to it's peer */
1725 peer
.ref
= msg_origport(msg
);
1726 peer
.node
= msg_orignode(msg
);
1727 tipc_port_connect(new_ref
, &peer
);
1728 new_sock
->state
= SS_CONNECTED
;
1730 tipc_port_set_importance(new_port
, msg_importance(msg
));
1731 if (msg_named(msg
)) {
1732 new_port
->conn_type
= msg_nametype(msg
);
1733 new_port
->conn_instance
= msg_nameinst(msg
);
1737 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
1738 * Respond to 'SYN+' by queuing it on new socket.
1740 if (!msg_data_sz(msg
)) {
1741 struct msghdr m
= {NULL
,};
1743 advance_rx_queue(sk
);
1744 tipc_send_packet(NULL
, new_sock
, &m
, 0);
1746 __skb_dequeue(&sk
->sk_receive_queue
);
1747 __skb_queue_head(&new_sk
->sk_receive_queue
, buf
);
1748 skb_set_owner_r(buf
, new_sk
);
1750 release_sock(new_sk
);
1757 * tipc_shutdown - shutdown socket connection
1758 * @sock: socket structure
1759 * @how: direction to close (must be SHUT_RDWR)
1761 * Terminates connection (if necessary), then purges socket's receive queue.
1763 * Returns 0 on success, errno otherwise
1765 static int tipc_shutdown(struct socket
*sock
, int how
)
1767 struct sock
*sk
= sock
->sk
;
1768 struct tipc_sock
*tsk
= tipc_sk(sk
);
1769 struct tipc_port
*port
= &tsk
->port
;
1770 struct sk_buff
*buf
;
1774 if (how
!= SHUT_RDWR
)
1779 switch (sock
->state
) {
1784 /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */
1785 buf
= __skb_dequeue(&sk
->sk_receive_queue
);
1787 if (TIPC_SKB_CB(buf
)->handle
!= NULL
) {
1791 tipc_port_disconnect(port
->ref
);
1792 if (tipc_msg_reverse(buf
, &peer
, TIPC_CONN_SHUTDOWN
))
1793 tipc_link_xmit2(buf
, peer
, 0);
1795 tipc_port_shutdown(port
->ref
);
1798 sock
->state
= SS_DISCONNECTING
;
1802 case SS_DISCONNECTING
:
1804 /* Discard any unreceived messages */
1805 __skb_queue_purge(&sk
->sk_receive_queue
);
1807 /* Wake up anyone sleeping in poll */
1808 sk
->sk_state_change(sk
);
1821 * tipc_setsockopt - set socket option
1822 * @sock: socket structure
1823 * @lvl: option level
1824 * @opt: option identifier
1825 * @ov: pointer to new option value
1826 * @ol: length of option value
1828 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
1829 * (to ease compatibility).
1831 * Returns 0 on success, errno otherwise
1833 static int tipc_setsockopt(struct socket
*sock
, int lvl
, int opt
,
1834 char __user
*ov
, unsigned int ol
)
1836 struct sock
*sk
= sock
->sk
;
1837 struct tipc_sock
*tsk
= tipc_sk(sk
);
1838 struct tipc_port
*port
= &tsk
->port
;
1842 if ((lvl
== IPPROTO_TCP
) && (sock
->type
== SOCK_STREAM
))
1844 if (lvl
!= SOL_TIPC
)
1845 return -ENOPROTOOPT
;
1846 if (ol
< sizeof(value
))
1848 res
= get_user(value
, (u32 __user
*)ov
);
1855 case TIPC_IMPORTANCE
:
1856 tipc_port_set_importance(port
, value
);
1858 case TIPC_SRC_DROPPABLE
:
1859 if (sock
->type
!= SOCK_STREAM
)
1860 tipc_port_set_unreliable(port
, value
);
1864 case TIPC_DEST_DROPPABLE
:
1865 tipc_port_set_unreturnable(port
, value
);
1867 case TIPC_CONN_TIMEOUT
:
1868 tipc_sk(sk
)->conn_timeout
= value
;
1869 /* no need to set "res", since already 0 at this point */
1881 * tipc_getsockopt - get socket option
1882 * @sock: socket structure
1883 * @lvl: option level
1884 * @opt: option identifier
1885 * @ov: receptacle for option value
1886 * @ol: receptacle for length of option value
1888 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
1889 * (to ease compatibility).
1891 * Returns 0 on success, errno otherwise
1893 static int tipc_getsockopt(struct socket
*sock
, int lvl
, int opt
,
1894 char __user
*ov
, int __user
*ol
)
1896 struct sock
*sk
= sock
->sk
;
1897 struct tipc_sock
*tsk
= tipc_sk(sk
);
1898 struct tipc_port
*port
= &tsk
->port
;
1903 if ((lvl
== IPPROTO_TCP
) && (sock
->type
== SOCK_STREAM
))
1904 return put_user(0, ol
);
1905 if (lvl
!= SOL_TIPC
)
1906 return -ENOPROTOOPT
;
1907 res
= get_user(len
, ol
);
1914 case TIPC_IMPORTANCE
:
1915 value
= tipc_port_importance(port
);
1917 case TIPC_SRC_DROPPABLE
:
1918 value
= tipc_port_unreliable(port
);
1920 case TIPC_DEST_DROPPABLE
:
1921 value
= tipc_port_unreturnable(port
);
1923 case TIPC_CONN_TIMEOUT
:
1924 value
= tipc_sk(sk
)->conn_timeout
;
1925 /* no need to set "res", since already 0 at this point */
1927 case TIPC_NODE_RECVQ_DEPTH
:
1928 value
= 0; /* was tipc_queue_size, now obsolete */
1930 case TIPC_SOCK_RECVQ_DEPTH
:
1931 value
= skb_queue_len(&sk
->sk_receive_queue
);
1940 return res
; /* "get" failed */
1942 if (len
< sizeof(value
))
1945 if (copy_to_user(ov
, &value
, sizeof(value
)))
1948 return put_user(sizeof(value
), ol
);
1951 int tipc_ioctl(struct socket
*sk
, unsigned int cmd
, unsigned long arg
)
1953 struct tipc_sioc_ln_req lnr
;
1954 void __user
*argp
= (void __user
*)arg
;
1957 case SIOCGETLINKNAME
:
1958 if (copy_from_user(&lnr
, argp
, sizeof(lnr
)))
1960 if (!tipc_node_get_linkname(lnr
.bearer_id
, lnr
.peer
,
1961 lnr
.linkname
, TIPC_MAX_LINK_NAME
)) {
1962 if (copy_to_user(argp
, &lnr
, sizeof(lnr
)))
1966 return -EADDRNOTAVAIL
;
1969 return -ENOIOCTLCMD
;
1973 /* Protocol switches for the various types of TIPC sockets */
1975 static const struct proto_ops msg_ops
= {
1976 .owner
= THIS_MODULE
,
1978 .release
= tipc_release
,
1980 .connect
= tipc_connect
,
1981 .socketpair
= sock_no_socketpair
,
1982 .accept
= sock_no_accept
,
1983 .getname
= tipc_getname
,
1985 .ioctl
= tipc_ioctl
,
1986 .listen
= sock_no_listen
,
1987 .shutdown
= tipc_shutdown
,
1988 .setsockopt
= tipc_setsockopt
,
1989 .getsockopt
= tipc_getsockopt
,
1990 .sendmsg
= tipc_sendmsg
,
1991 .recvmsg
= tipc_recvmsg
,
1992 .mmap
= sock_no_mmap
,
1993 .sendpage
= sock_no_sendpage
1996 static const struct proto_ops packet_ops
= {
1997 .owner
= THIS_MODULE
,
1999 .release
= tipc_release
,
2001 .connect
= tipc_connect
,
2002 .socketpair
= sock_no_socketpair
,
2003 .accept
= tipc_accept
,
2004 .getname
= tipc_getname
,
2006 .ioctl
= tipc_ioctl
,
2007 .listen
= tipc_listen
,
2008 .shutdown
= tipc_shutdown
,
2009 .setsockopt
= tipc_setsockopt
,
2010 .getsockopt
= tipc_getsockopt
,
2011 .sendmsg
= tipc_send_packet
,
2012 .recvmsg
= tipc_recvmsg
,
2013 .mmap
= sock_no_mmap
,
2014 .sendpage
= sock_no_sendpage
2017 static const struct proto_ops stream_ops
= {
2018 .owner
= THIS_MODULE
,
2020 .release
= tipc_release
,
2022 .connect
= tipc_connect
,
2023 .socketpair
= sock_no_socketpair
,
2024 .accept
= tipc_accept
,
2025 .getname
= tipc_getname
,
2027 .ioctl
= tipc_ioctl
,
2028 .listen
= tipc_listen
,
2029 .shutdown
= tipc_shutdown
,
2030 .setsockopt
= tipc_setsockopt
,
2031 .getsockopt
= tipc_getsockopt
,
2032 .sendmsg
= tipc_send_stream
,
2033 .recvmsg
= tipc_recv_stream
,
2034 .mmap
= sock_no_mmap
,
2035 .sendpage
= sock_no_sendpage
2038 static const struct net_proto_family tipc_family_ops
= {
2039 .owner
= THIS_MODULE
,
2041 .create
= tipc_sk_create
2044 static struct proto tipc_proto
= {
2046 .owner
= THIS_MODULE
,
2047 .obj_size
= sizeof(struct tipc_sock
),
2048 .sysctl_rmem
= sysctl_tipc_rmem
2051 static struct proto tipc_proto_kern
= {
2053 .obj_size
= sizeof(struct tipc_sock
),
2054 .sysctl_rmem
= sysctl_tipc_rmem
2058 * tipc_socket_init - initialize TIPC socket interface
2060 * Returns 0 on success, errno otherwise
2062 int tipc_socket_init(void)
2066 res
= proto_register(&tipc_proto
, 1);
2068 pr_err("Failed to register TIPC protocol type\n");
2072 res
= sock_register(&tipc_family_ops
);
2074 pr_err("Failed to register TIPC socket type\n");
2075 proto_unregister(&tipc_proto
);
2083 * tipc_socket_stop - stop TIPC socket interface
2085 void tipc_socket_stop(void)
2087 sock_unregister(tipc_family_ops
.family
);
2088 proto_unregister(&tipc_proto
);