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.
37 #include <linux/rhashtable.h>
38 #include <linux/jhash.h>
40 #include "name_table.h"
46 #define SS_LISTENING -1 /* socket is listening */
47 #define SS_READY -2 /* socket is connectionless */
49 #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */
50 #define CONN_PROBING_INTERVAL msecs_to_jiffies(3600000) /* [ms] => 1 h */
51 #define TIPC_FWD_MSG 1
52 #define TIPC_CONN_OK 0
53 #define TIPC_CONN_PROBING 1
54 #define TIPC_MAX_PORT 0xffffffff
55 #define TIPC_MIN_PORT 1
58 * struct tipc_sock - TIPC socket structure
59 * @sk: socket - interacts with 'port' and with user via the socket API
60 * @connected: non-zero if port is currently connected to a peer port
61 * @conn_type: TIPC type used when connection was established
62 * @conn_instance: TIPC instance used when connection was established
63 * @published: non-zero if port has one or more associated names
64 * @max_pkt: maximum packet size "hint" used when building messages sent by port
65 * @portid: unique port identity in TIPC socket hash table
66 * @phdr: preformatted message header used when sending messages
67 * @port_list: adjacent ports in TIPC's global list of ports
68 * @publications: list of publications for port
69 * @pub_count: total # of publications port has made during its lifetime
73 * @port: port - interacts with 'sk' and with the rest of the TIPC stack
74 * @peer_name: the peer of the connection, if any
75 * @conn_timeout: the time we can wait for an unresponded setup request
76 * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue
77 * @link_cong: non-zero if owner must sleep because of link congestion
78 * @sent_unacked: # messages sent by socket, and not yet acked by peer
79 * @rcv_unacked: # messages read by user, but not yet acked back to peer
80 * @node: hash table node
81 * @rcu: rcu struct for tipc_sock
92 struct list_head sock_list
;
93 struct list_head publications
;
96 unsigned long probing_intv
;
97 struct timer_list timer
;
103 struct rhash_head node
;
107 static int tipc_backlog_rcv(struct sock
*sk
, struct sk_buff
*skb
);
108 static void tipc_data_ready(struct sock
*sk
);
109 static void tipc_write_space(struct sock
*sk
);
110 static int tipc_release(struct socket
*sock
);
111 static int tipc_accept(struct socket
*sock
, struct socket
*new_sock
, int flags
);
112 static int tipc_wait_for_sndmsg(struct socket
*sock
, long *timeo_p
);
113 static void tipc_sk_timeout(unsigned long data
);
114 static int tipc_sk_publish(struct tipc_sock
*tsk
, uint scope
,
115 struct tipc_name_seq
const *seq
);
116 static int tipc_sk_withdraw(struct tipc_sock
*tsk
, uint scope
,
117 struct tipc_name_seq
const *seq
);
118 static struct tipc_sock
*tipc_sk_lookup(u32 portid
);
119 static int tipc_sk_insert(struct tipc_sock
*tsk
);
120 static void tipc_sk_remove(struct tipc_sock
*tsk
);
122 static const struct proto_ops packet_ops
;
123 static const struct proto_ops stream_ops
;
124 static const struct proto_ops msg_ops
;
126 static struct proto tipc_proto
;
127 static struct proto tipc_proto_kern
;
129 static const struct nla_policy tipc_nl_sock_policy
[TIPC_NLA_SOCK_MAX
+ 1] = {
130 [TIPC_NLA_SOCK_UNSPEC
] = { .type
= NLA_UNSPEC
},
131 [TIPC_NLA_SOCK_ADDR
] = { .type
= NLA_U32
},
132 [TIPC_NLA_SOCK_REF
] = { .type
= NLA_U32
},
133 [TIPC_NLA_SOCK_CON
] = { .type
= NLA_NESTED
},
134 [TIPC_NLA_SOCK_HAS_PUBL
] = { .type
= NLA_FLAG
}
138 * Revised TIPC socket locking policy:
140 * Most socket operations take the standard socket lock when they start
141 * and hold it until they finish (or until they need to sleep). Acquiring
142 * this lock grants the owner exclusive access to the fields of the socket
143 * data structures, with the exception of the backlog queue. A few socket
144 * operations can be done without taking the socket lock because they only
145 * read socket information that never changes during the life of the socket.
147 * Socket operations may acquire the lock for the associated TIPC port if they
148 * need to perform an operation on the port. If any routine needs to acquire
149 * both the socket lock and the port lock it must take the socket lock first
150 * to avoid the risk of deadlock.
152 * The dispatcher handling incoming messages cannot grab the socket lock in
153 * the standard fashion, since invoked it runs at the BH level and cannot block.
154 * Instead, it checks to see if the socket lock is currently owned by someone,
155 * and either handles the message itself or adds it to the socket's backlog
156 * queue; in the latter case the queued message is processed once the process
157 * owning the socket lock releases it.
159 * NOTE: Releasing the socket lock while an operation is sleeping overcomes
160 * the problem of a blocked socket operation preventing any other operations
161 * from occurring. However, applications must be careful if they have
162 * multiple threads trying to send (or receive) on the same socket, as these
163 * operations might interfere with each other. For example, doing a connect
164 * and a receive at the same time might allow the receive to consume the
165 * ACK message meant for the connect. While additional work could be done
166 * to try and overcome this, it doesn't seem to be worthwhile at the present.
168 * NOTE: Releasing the socket lock while an operation is sleeping also ensures
169 * that another operation that must be performed in a non-blocking manner is
170 * not delayed for very long because the lock has already been taken.
172 * NOTE: This code assumes that certain fields of a port/socket pair are
173 * constant over its lifetime; such fields can be examined without taking
174 * the socket lock and/or port lock, and do not need to be re-read even
175 * after resuming processing after waiting. These fields include:
177 * - pointer to socket sk structure (aka tipc_sock structure)
178 * - pointer to port structure
182 /* Protects tipc socket hash table mutations */
183 static struct rhashtable tipc_sk_rht
;
185 static u32
tsk_peer_node(struct tipc_sock
*tsk
)
187 return msg_destnode(&tsk
->phdr
);
190 static u32
tsk_peer_port(struct tipc_sock
*tsk
)
192 return msg_destport(&tsk
->phdr
);
195 static bool tsk_unreliable(struct tipc_sock
*tsk
)
197 return msg_src_droppable(&tsk
->phdr
) != 0;
200 static void tsk_set_unreliable(struct tipc_sock
*tsk
, bool unreliable
)
202 msg_set_src_droppable(&tsk
->phdr
, unreliable
? 1 : 0);
205 static bool tsk_unreturnable(struct tipc_sock
*tsk
)
207 return msg_dest_droppable(&tsk
->phdr
) != 0;
210 static void tsk_set_unreturnable(struct tipc_sock
*tsk
, bool unreturnable
)
212 msg_set_dest_droppable(&tsk
->phdr
, unreturnable
? 1 : 0);
215 static int tsk_importance(struct tipc_sock
*tsk
)
217 return msg_importance(&tsk
->phdr
);
220 static int tsk_set_importance(struct tipc_sock
*tsk
, int imp
)
222 if (imp
> TIPC_CRITICAL_IMPORTANCE
)
224 msg_set_importance(&tsk
->phdr
, (u32
)imp
);
228 static struct tipc_sock
*tipc_sk(const struct sock
*sk
)
230 return container_of(sk
, struct tipc_sock
, sk
);
233 static int tsk_conn_cong(struct tipc_sock
*tsk
)
235 return tsk
->sent_unacked
>= TIPC_FLOWCTRL_WIN
;
239 * tsk_advance_rx_queue - discard first buffer in socket receive queue
241 * Caller must hold socket lock
243 static void tsk_advance_rx_queue(struct sock
*sk
)
245 kfree_skb(__skb_dequeue(&sk
->sk_receive_queue
));
249 * tsk_rej_rx_queue - reject all buffers in socket receive queue
251 * Caller must hold socket lock
253 static void tsk_rej_rx_queue(struct sock
*sk
)
258 while ((skb
= __skb_dequeue(&sk
->sk_receive_queue
))) {
259 if (tipc_msg_reverse(skb
, &dnode
, TIPC_ERR_NO_PORT
))
260 tipc_link_xmit_skb(skb
, dnode
, 0);
264 /* tsk_peer_msg - verify if message was sent by connected port's peer
266 * Handles cases where the node's network address has changed from
267 * the default of <0.0.0> to its configured setting.
269 static bool tsk_peer_msg(struct tipc_sock
*tsk
, struct tipc_msg
*msg
)
271 u32 peer_port
= tsk_peer_port(tsk
);
275 if (unlikely(!tsk
->connected
))
278 if (unlikely(msg_origport(msg
) != peer_port
))
281 orig_node
= msg_orignode(msg
);
282 peer_node
= tsk_peer_node(tsk
);
284 if (likely(orig_node
== peer_node
))
287 if (!orig_node
&& (peer_node
== tipc_own_addr
))
290 if (!peer_node
&& (orig_node
== tipc_own_addr
))
297 * tipc_sk_create - create a TIPC socket
298 * @net: network namespace (must be default network)
299 * @sock: pre-allocated socket structure
300 * @protocol: protocol indicator (must be 0)
301 * @kern: caused by kernel or by userspace?
303 * This routine creates additional data structures used by the TIPC socket,
304 * initializes them, and links them together.
306 * Returns 0 on success, errno otherwise
308 static int tipc_sk_create(struct net
*net
, struct socket
*sock
,
309 int protocol
, int kern
)
311 const struct proto_ops
*ops
;
314 struct tipc_sock
*tsk
;
315 struct tipc_msg
*msg
;
317 /* Validate arguments */
318 if (unlikely(protocol
!= 0))
319 return -EPROTONOSUPPORT
;
321 switch (sock
->type
) {
324 state
= SS_UNCONNECTED
;
328 state
= SS_UNCONNECTED
;
339 /* Allocate socket's protocol area */
341 sk
= sk_alloc(net
, AF_TIPC
, GFP_KERNEL
, &tipc_proto
);
343 sk
= sk_alloc(net
, AF_TIPC
, GFP_KERNEL
, &tipc_proto_kern
);
349 tsk
->max_pkt
= MAX_PKT_DEFAULT
;
350 INIT_LIST_HEAD(&tsk
->publications
);
352 tipc_msg_init(msg
, TIPC_LOW_IMPORTANCE
, TIPC_NAMED_MSG
,
355 /* Finish initializing socket data structures */
358 sock_init_data(sock
, sk
);
359 if (tipc_sk_insert(tsk
)) {
360 pr_warn("Socket create failed; port numbrer exhausted\n");
363 msg_set_origport(msg
, tsk
->portid
);
364 setup_timer(&tsk
->timer
, tipc_sk_timeout
, (unsigned long)tsk
);
365 sk
->sk_backlog_rcv
= tipc_backlog_rcv
;
366 sk
->sk_rcvbuf
= sysctl_tipc_rmem
[1];
367 sk
->sk_data_ready
= tipc_data_ready
;
368 sk
->sk_write_space
= tipc_write_space
;
369 tsk
->conn_timeout
= CONN_TIMEOUT_DEFAULT
;
370 tsk
->sent_unacked
= 0;
371 atomic_set(&tsk
->dupl_rcvcnt
, 0);
373 if (sock
->state
== SS_READY
) {
374 tsk_set_unreturnable(tsk
, true);
375 if (sock
->type
== SOCK_DGRAM
)
376 tsk_set_unreliable(tsk
, true);
382 * tipc_sock_create_local - create TIPC socket from inside TIPC module
383 * @type: socket type - SOCK_RDM or SOCK_SEQPACKET
385 * We cannot use sock_creat_kern here because it bumps module user count.
386 * Since socket owner and creator is the same module we must make sure
387 * that module count remains zero for module local sockets, otherwise
388 * we cannot do rmmod.
390 * Returns 0 on success, errno otherwise
392 int tipc_sock_create_local(int type
, struct socket
**res
)
396 rc
= sock_create_lite(AF_TIPC
, type
, 0, res
);
398 pr_err("Failed to create kernel socket\n");
401 tipc_sk_create(&init_net
, *res
, 0, 1);
407 * tipc_sock_release_local - release socket created by tipc_sock_create_local
408 * @sock: the socket to be released.
410 * Module reference count is not incremented when such sockets are created,
411 * so we must keep it from being decremented when they are released.
413 void tipc_sock_release_local(struct socket
*sock
)
421 * tipc_sock_accept_local - accept a connection on a socket created
422 * with tipc_sock_create_local. Use this function to avoid that
423 * module reference count is inadvertently incremented.
425 * @sock: the accepting socket
426 * @newsock: reference to the new socket to be created
427 * @flags: socket flags
430 int tipc_sock_accept_local(struct socket
*sock
, struct socket
**newsock
,
433 struct sock
*sk
= sock
->sk
;
436 ret
= sock_create_lite(sk
->sk_family
, sk
->sk_type
,
437 sk
->sk_protocol
, newsock
);
441 ret
= tipc_accept(sock
, *newsock
, flags
);
443 sock_release(*newsock
);
446 (*newsock
)->ops
= sock
->ops
;
450 static void tipc_sk_callback(struct rcu_head
*head
)
452 struct tipc_sock
*tsk
= container_of(head
, struct tipc_sock
, rcu
);
458 * tipc_release - destroy a TIPC socket
459 * @sock: socket to destroy
461 * This routine cleans up any messages that are still queued on the socket.
462 * For DGRAM and RDM socket types, all queued messages are rejected.
463 * For SEQPACKET and STREAM socket types, the first message is rejected
464 * and any others are discarded. (If the first message on a STREAM socket
465 * is partially-read, it is discarded and the next one is rejected instead.)
467 * NOTE: Rejected messages are not necessarily returned to the sender! They
468 * are returned or discarded according to the "destination droppable" setting
469 * specified for the message by the sender.
471 * Returns 0 on success, errno otherwise
473 static int tipc_release(struct socket
*sock
)
475 struct sock
*sk
= sock
->sk
;
476 struct tipc_sock
*tsk
;
478 u32 dnode
, probing_state
;
481 * Exit if socket isn't fully initialized (occurs when a failed accept()
482 * releases a pre-allocated child socket that was never used)
491 * Reject all unreceived messages, except on an active connection
492 * (which disconnects locally & sends a 'FIN+' to peer)
494 dnode
= tsk_peer_node(tsk
);
495 while (sock
->state
!= SS_DISCONNECTING
) {
496 skb
= __skb_dequeue(&sk
->sk_receive_queue
);
499 if (TIPC_SKB_CB(skb
)->handle
!= NULL
)
502 if ((sock
->state
== SS_CONNECTING
) ||
503 (sock
->state
== SS_CONNECTED
)) {
504 sock
->state
= SS_DISCONNECTING
;
506 tipc_node_remove_conn(dnode
, tsk
->portid
);
508 if (tipc_msg_reverse(skb
, &dnode
, TIPC_ERR_NO_PORT
))
509 tipc_link_xmit_skb(skb
, dnode
, 0);
513 tipc_sk_withdraw(tsk
, 0, NULL
);
514 probing_state
= tsk
->probing_state
;
515 if (del_timer_sync(&tsk
->timer
) && probing_state
!= TIPC_CONN_PROBING
)
518 if (tsk
->connected
) {
519 skb
= tipc_msg_create(TIPC_CRITICAL_IMPORTANCE
, TIPC_CONN_MSG
,
520 SHORT_H_SIZE
, 0, dnode
, tipc_own_addr
,
522 tsk
->portid
, TIPC_ERR_NO_PORT
);
524 tipc_link_xmit_skb(skb
, dnode
, tsk
->portid
);
525 tipc_node_remove_conn(dnode
, tsk
->portid
);
528 /* Discard any remaining (connection-based) messages in receive queue */
529 __skb_queue_purge(&sk
->sk_receive_queue
);
531 /* Reject any messages that accumulated in backlog queue */
532 sock
->state
= SS_DISCONNECTING
;
535 call_rcu(&tsk
->rcu
, tipc_sk_callback
);
542 * tipc_bind - associate or disassocate TIPC name(s) with a socket
543 * @sock: socket structure
544 * @uaddr: socket address describing name(s) and desired operation
545 * @uaddr_len: size of socket address data structure
547 * Name and name sequence binding is indicated using a positive scope value;
548 * a negative scope value unbinds the specified name. Specifying no name
549 * (i.e. a socket address length of 0) unbinds all names from the socket.
551 * Returns 0 on success, errno otherwise
553 * NOTE: This routine doesn't need to take the socket lock since it doesn't
554 * access any non-constant socket information.
556 static int tipc_bind(struct socket
*sock
, struct sockaddr
*uaddr
,
559 struct sock
*sk
= sock
->sk
;
560 struct sockaddr_tipc
*addr
= (struct sockaddr_tipc
*)uaddr
;
561 struct tipc_sock
*tsk
= tipc_sk(sk
);
565 if (unlikely(!uaddr_len
)) {
566 res
= tipc_sk_withdraw(tsk
, 0, NULL
);
570 if (uaddr_len
< sizeof(struct sockaddr_tipc
)) {
574 if (addr
->family
!= AF_TIPC
) {
579 if (addr
->addrtype
== TIPC_ADDR_NAME
)
580 addr
->addr
.nameseq
.upper
= addr
->addr
.nameseq
.lower
;
581 else if (addr
->addrtype
!= TIPC_ADDR_NAMESEQ
) {
586 if ((addr
->addr
.nameseq
.type
< TIPC_RESERVED_TYPES
) &&
587 (addr
->addr
.nameseq
.type
!= TIPC_TOP_SRV
) &&
588 (addr
->addr
.nameseq
.type
!= TIPC_CFG_SRV
)) {
593 res
= (addr
->scope
> 0) ?
594 tipc_sk_publish(tsk
, addr
->scope
, &addr
->addr
.nameseq
) :
595 tipc_sk_withdraw(tsk
, -addr
->scope
, &addr
->addr
.nameseq
);
602 * tipc_getname - get port ID of socket or peer socket
603 * @sock: socket structure
604 * @uaddr: area for returned socket address
605 * @uaddr_len: area for returned length of socket address
606 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
608 * Returns 0 on success, errno otherwise
610 * NOTE: This routine doesn't need to take the socket lock since it only
611 * accesses socket information that is unchanging (or which changes in
612 * a completely predictable manner).
614 static int tipc_getname(struct socket
*sock
, struct sockaddr
*uaddr
,
615 int *uaddr_len
, int peer
)
617 struct sockaddr_tipc
*addr
= (struct sockaddr_tipc
*)uaddr
;
618 struct tipc_sock
*tsk
= tipc_sk(sock
->sk
);
620 memset(addr
, 0, sizeof(*addr
));
622 if ((sock
->state
!= SS_CONNECTED
) &&
623 ((peer
!= 2) || (sock
->state
!= SS_DISCONNECTING
)))
625 addr
->addr
.id
.ref
= tsk_peer_port(tsk
);
626 addr
->addr
.id
.node
= tsk_peer_node(tsk
);
628 addr
->addr
.id
.ref
= tsk
->portid
;
629 addr
->addr
.id
.node
= tipc_own_addr
;
632 *uaddr_len
= sizeof(*addr
);
633 addr
->addrtype
= TIPC_ADDR_ID
;
634 addr
->family
= AF_TIPC
;
636 addr
->addr
.name
.domain
= 0;
642 * tipc_poll - read and possibly block on pollmask
643 * @file: file structure associated with the socket
644 * @sock: socket for which to calculate the poll bits
647 * Returns pollmask value
650 * It appears that the usual socket locking mechanisms are not useful here
651 * since the pollmask info is potentially out-of-date the moment this routine
652 * exits. TCP and other protocols seem to rely on higher level poll routines
653 * to handle any preventable race conditions, so TIPC will do the same ...
655 * TIPC sets the returned events as follows:
657 * socket state flags set
658 * ------------ ---------
659 * unconnected no read flags
660 * POLLOUT if port is not congested
662 * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue
665 * connected POLLIN/POLLRDNORM if data in rx queue
666 * POLLOUT if port is not congested
668 * disconnecting POLLIN/POLLRDNORM/POLLHUP
671 * listening POLLIN if SYN in rx queue
674 * ready POLLIN/POLLRDNORM if data in rx queue
675 * [connectionless] POLLOUT (since port cannot be congested)
677 * IMPORTANT: The fact that a read or write operation is indicated does NOT
678 * imply that the operation will succeed, merely that it should be performed
679 * and will not block.
681 static unsigned int tipc_poll(struct file
*file
, struct socket
*sock
,
684 struct sock
*sk
= sock
->sk
;
685 struct tipc_sock
*tsk
= tipc_sk(sk
);
688 sock_poll_wait(file
, sk_sleep(sk
), wait
);
690 switch ((int)sock
->state
) {
697 if (!tsk
->link_cong
&& !tsk_conn_cong(tsk
))
702 if (!skb_queue_empty(&sk
->sk_receive_queue
))
703 mask
|= (POLLIN
| POLLRDNORM
);
705 case SS_DISCONNECTING
:
706 mask
= (POLLIN
| POLLRDNORM
| POLLHUP
);
714 * tipc_sendmcast - send multicast message
715 * @sock: socket structure
716 * @seq: destination address
717 * @msg: message to send
718 * @dsz: total length of message data
719 * @timeo: timeout to wait for wakeup
721 * Called from function tipc_sendmsg(), which has done all sanity checks
722 * Returns the number of bytes sent on success, or errno
724 static int tipc_sendmcast(struct socket
*sock
, struct tipc_name_seq
*seq
,
725 struct msghdr
*msg
, size_t dsz
, long timeo
)
727 struct sock
*sk
= sock
->sk
;
728 struct tipc_msg
*mhdr
= &tipc_sk(sk
)->phdr
;
729 struct sk_buff_head head
;
733 msg_set_type(mhdr
, TIPC_MCAST_MSG
);
734 msg_set_lookup_scope(mhdr
, TIPC_CLUSTER_SCOPE
);
735 msg_set_destport(mhdr
, 0);
736 msg_set_destnode(mhdr
, 0);
737 msg_set_nametype(mhdr
, seq
->type
);
738 msg_set_namelower(mhdr
, seq
->lower
);
739 msg_set_nameupper(mhdr
, seq
->upper
);
740 msg_set_hdr_sz(mhdr
, MCAST_H_SIZE
);
743 mtu
= tipc_bclink_get_mtu();
744 __skb_queue_head_init(&head
);
745 rc
= tipc_msg_build(mhdr
, msg
, 0, dsz
, mtu
, &head
);
746 if (unlikely(rc
< 0))
750 rc
= tipc_bclink_xmit(&head
);
751 if (likely(rc
>= 0)) {
757 if (rc
!= -ELINKCONG
)
759 tipc_sk(sk
)->link_cong
= 1;
760 rc
= tipc_wait_for_sndmsg(sock
, &timeo
);
762 __skb_queue_purge(&head
);
767 /* tipc_sk_mcast_rcv - Deliver multicast message to all destination sockets
769 void tipc_sk_mcast_rcv(struct sk_buff
*buf
)
771 struct tipc_msg
*msg
= buf_msg(buf
);
772 struct tipc_port_list dports
= {0, NULL
, };
773 struct tipc_port_list
*item
;
775 uint i
, last
, dst
= 0;
776 u32 scope
= TIPC_CLUSTER_SCOPE
;
778 if (in_own_node(msg_orignode(msg
)))
779 scope
= TIPC_NODE_SCOPE
;
781 /* Create destination port list: */
782 tipc_nametbl_mc_translate(msg_nametype(msg
),
793 for (item
= &dports
; item
; item
= item
->next
) {
794 for (i
= 0; i
< PLSIZE
&& ++dst
<= last
; i
++) {
795 b
= (dst
!= last
) ? skb_clone(buf
, GFP_ATOMIC
) : buf
;
797 pr_warn("Failed do clone mcast rcv buffer\n");
800 msg_set_destport(msg
, item
->ports
[i
]);
804 tipc_port_list_free(&dports
);
808 * tipc_sk_proto_rcv - receive a connection mng protocol message
809 * @tsk: receiving socket
810 * @dnode: node to send response message to, if any
811 * @buf: buffer containing protocol message
812 * Returns 0 (TIPC_OK) if message was consumed, 1 (TIPC_FWD_MSG) if
813 * (CONN_PROBE_REPLY) message should be forwarded.
815 static int tipc_sk_proto_rcv(struct tipc_sock
*tsk
, u32
*dnode
,
818 struct tipc_msg
*msg
= buf_msg(buf
);
821 /* Ignore if connection cannot be validated: */
822 if (!tsk_peer_msg(tsk
, msg
))
825 tsk
->probing_state
= TIPC_CONN_OK
;
827 if (msg_type(msg
) == CONN_ACK
) {
828 conn_cong
= tsk_conn_cong(tsk
);
829 tsk
->sent_unacked
-= msg_msgcnt(msg
);
831 tsk
->sk
.sk_write_space(&tsk
->sk
);
832 } else if (msg_type(msg
) == CONN_PROBE
) {
833 if (!tipc_msg_reverse(buf
, dnode
, TIPC_OK
))
835 msg_set_type(msg
, CONN_PROBE_REPLY
);
838 /* Do nothing if msg_type() == CONN_PROBE_REPLY */
844 static int tipc_wait_for_sndmsg(struct socket
*sock
, long *timeo_p
)
846 struct sock
*sk
= sock
->sk
;
847 struct tipc_sock
*tsk
= tipc_sk(sk
);
852 int err
= sock_error(sk
);
855 if (sock
->state
== SS_DISCONNECTING
)
859 if (signal_pending(current
))
860 return sock_intr_errno(*timeo_p
);
862 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
863 done
= sk_wait_event(sk
, timeo_p
, !tsk
->link_cong
);
864 finish_wait(sk_sleep(sk
), &wait
);
870 * tipc_sendmsg - send message in connectionless manner
871 * @iocb: if NULL, indicates that socket lock is already held
872 * @sock: socket structure
873 * @m: message to send
874 * @dsz: amount of user data to be sent
876 * Message must have an destination specified explicitly.
877 * Used for SOCK_RDM and SOCK_DGRAM messages,
878 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
879 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
881 * Returns the number of bytes sent on success, or errno otherwise
883 static int tipc_sendmsg(struct kiocb
*iocb
, struct socket
*sock
,
884 struct msghdr
*m
, size_t dsz
)
886 DECLARE_SOCKADDR(struct sockaddr_tipc
*, dest
, m
->msg_name
);
887 struct sock
*sk
= sock
->sk
;
888 struct tipc_sock
*tsk
= tipc_sk(sk
);
889 struct tipc_msg
*mhdr
= &tsk
->phdr
;
891 struct sk_buff_head head
;
893 struct tipc_name_seq
*seq
= &dest
->addr
.nameseq
;
899 return -EDESTADDRREQ
;
901 if (unlikely((m
->msg_namelen
< sizeof(*dest
)) ||
902 (dest
->family
!= AF_TIPC
)))
905 if (dsz
> TIPC_MAX_USER_MSG_SIZE
)
911 if (unlikely(sock
->state
!= SS_READY
)) {
912 if (sock
->state
== SS_LISTENING
) {
916 if (sock
->state
!= SS_UNCONNECTED
) {
920 if (tsk
->published
) {
924 if (dest
->addrtype
== TIPC_ADDR_NAME
) {
925 tsk
->conn_type
= dest
->addr
.name
.name
.type
;
926 tsk
->conn_instance
= dest
->addr
.name
.name
.instance
;
930 timeo
= sock_sndtimeo(sk
, m
->msg_flags
& MSG_DONTWAIT
);
932 if (dest
->addrtype
== TIPC_ADDR_MCAST
) {
933 rc
= tipc_sendmcast(sock
, seq
, m
, dsz
, timeo
);
935 } else if (dest
->addrtype
== TIPC_ADDR_NAME
) {
936 u32 type
= dest
->addr
.name
.name
.type
;
937 u32 inst
= dest
->addr
.name
.name
.instance
;
938 u32 domain
= dest
->addr
.name
.domain
;
941 msg_set_type(mhdr
, TIPC_NAMED_MSG
);
942 msg_set_hdr_sz(mhdr
, NAMED_H_SIZE
);
943 msg_set_nametype(mhdr
, type
);
944 msg_set_nameinst(mhdr
, inst
);
945 msg_set_lookup_scope(mhdr
, tipc_addr_scope(domain
));
946 dport
= tipc_nametbl_translate(type
, inst
, &dnode
);
947 msg_set_destnode(mhdr
, dnode
);
948 msg_set_destport(mhdr
, dport
);
949 if (unlikely(!dport
&& !dnode
)) {
953 } else if (dest
->addrtype
== TIPC_ADDR_ID
) {
954 dnode
= dest
->addr
.id
.node
;
955 msg_set_type(mhdr
, TIPC_DIRECT_MSG
);
956 msg_set_lookup_scope(mhdr
, 0);
957 msg_set_destnode(mhdr
, dnode
);
958 msg_set_destport(mhdr
, dest
->addr
.id
.ref
);
959 msg_set_hdr_sz(mhdr
, BASIC_H_SIZE
);
963 mtu
= tipc_node_get_mtu(dnode
, tsk
->portid
);
964 __skb_queue_head_init(&head
);
965 rc
= tipc_msg_build(mhdr
, m
, 0, dsz
, mtu
, &head
);
970 skb
= skb_peek(&head
);
971 TIPC_SKB_CB(skb
)->wakeup_pending
= tsk
->link_cong
;
972 rc
= tipc_link_xmit(&head
, dnode
, tsk
->portid
);
973 if (likely(rc
>= 0)) {
974 if (sock
->state
!= SS_READY
)
975 sock
->state
= SS_CONNECTING
;
981 if (rc
!= -ELINKCONG
)
984 rc
= tipc_wait_for_sndmsg(sock
, &timeo
);
986 __skb_queue_purge(&head
);
995 static int tipc_wait_for_sndpkt(struct socket
*sock
, long *timeo_p
)
997 struct sock
*sk
= sock
->sk
;
998 struct tipc_sock
*tsk
= tipc_sk(sk
);
1003 int err
= sock_error(sk
);
1006 if (sock
->state
== SS_DISCONNECTING
)
1008 else if (sock
->state
!= SS_CONNECTED
)
1012 if (signal_pending(current
))
1013 return sock_intr_errno(*timeo_p
);
1015 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
1016 done
= sk_wait_event(sk
, timeo_p
,
1018 !tsk_conn_cong(tsk
)) ||
1020 finish_wait(sk_sleep(sk
), &wait
);
1026 * tipc_send_stream - send stream-oriented data
1028 * @sock: socket structure
1030 * @dsz: total length of data to be transmitted
1032 * Used for SOCK_STREAM data.
1034 * Returns the number of bytes sent on success (or partial success),
1035 * or errno if no data sent
1037 static int tipc_send_stream(struct kiocb
*iocb
, struct socket
*sock
,
1038 struct msghdr
*m
, size_t dsz
)
1040 struct sock
*sk
= sock
->sk
;
1041 struct tipc_sock
*tsk
= tipc_sk(sk
);
1042 struct tipc_msg
*mhdr
= &tsk
->phdr
;
1043 struct sk_buff_head head
;
1044 DECLARE_SOCKADDR(struct sockaddr_tipc
*, dest
, m
->msg_name
);
1045 u32 portid
= tsk
->portid
;
1049 uint mtu
, send
, sent
= 0;
1051 /* Handle implied connection establishment */
1052 if (unlikely(dest
)) {
1053 rc
= tipc_sendmsg(iocb
, sock
, m
, dsz
);
1054 if (dsz
&& (dsz
== rc
))
1055 tsk
->sent_unacked
= 1;
1058 if (dsz
> (uint
)INT_MAX
)
1064 if (unlikely(sock
->state
!= SS_CONNECTED
)) {
1065 if (sock
->state
== SS_DISCONNECTING
)
1072 timeo
= sock_sndtimeo(sk
, m
->msg_flags
& MSG_DONTWAIT
);
1073 dnode
= tsk_peer_node(tsk
);
1077 send
= min_t(uint
, dsz
- sent
, TIPC_MAX_USER_MSG_SIZE
);
1078 __skb_queue_head_init(&head
);
1079 rc
= tipc_msg_build(mhdr
, m
, sent
, send
, mtu
, &head
);
1080 if (unlikely(rc
< 0))
1083 if (likely(!tsk_conn_cong(tsk
))) {
1084 rc
= tipc_link_xmit(&head
, dnode
, portid
);
1086 tsk
->sent_unacked
++;
1092 if (rc
== -EMSGSIZE
) {
1093 tsk
->max_pkt
= tipc_node_get_mtu(dnode
, portid
);
1096 if (rc
!= -ELINKCONG
)
1100 rc
= tipc_wait_for_sndpkt(sock
, &timeo
);
1102 __skb_queue_purge(&head
);
1107 return sent
? sent
: rc
;
1111 * tipc_send_packet - send a connection-oriented message
1112 * @iocb: if NULL, indicates that socket lock is already held
1113 * @sock: socket structure
1114 * @m: message to send
1115 * @dsz: length of data to be transmitted
1117 * Used for SOCK_SEQPACKET messages.
1119 * Returns the number of bytes sent on success, or errno otherwise
1121 static int tipc_send_packet(struct kiocb
*iocb
, struct socket
*sock
,
1122 struct msghdr
*m
, size_t dsz
)
1124 if (dsz
> TIPC_MAX_USER_MSG_SIZE
)
1127 return tipc_send_stream(iocb
, sock
, m
, dsz
);
1130 /* tipc_sk_finish_conn - complete the setup of a connection
1132 static void tipc_sk_finish_conn(struct tipc_sock
*tsk
, u32 peer_port
,
1135 struct tipc_msg
*msg
= &tsk
->phdr
;
1137 msg_set_destnode(msg
, peer_node
);
1138 msg_set_destport(msg
, peer_port
);
1139 msg_set_type(msg
, TIPC_CONN_MSG
);
1140 msg_set_lookup_scope(msg
, 0);
1141 msg_set_hdr_sz(msg
, SHORT_H_SIZE
);
1143 tsk
->probing_intv
= CONN_PROBING_INTERVAL
;
1144 tsk
->probing_state
= TIPC_CONN_OK
;
1146 if (!mod_timer(&tsk
->timer
, jiffies
+ tsk
->probing_intv
))
1147 sock_hold(&tsk
->sk
);
1148 tipc_node_add_conn(peer_node
, tsk
->portid
, peer_port
);
1149 tsk
->max_pkt
= tipc_node_get_mtu(peer_node
, tsk
->portid
);
1153 * set_orig_addr - capture sender's address for received message
1154 * @m: descriptor for message info
1155 * @msg: received message header
1157 * Note: Address is not captured if not requested by receiver.
1159 static void set_orig_addr(struct msghdr
*m
, struct tipc_msg
*msg
)
1161 DECLARE_SOCKADDR(struct sockaddr_tipc
*, addr
, m
->msg_name
);
1164 addr
->family
= AF_TIPC
;
1165 addr
->addrtype
= TIPC_ADDR_ID
;
1166 memset(&addr
->addr
, 0, sizeof(addr
->addr
));
1167 addr
->addr
.id
.ref
= msg_origport(msg
);
1168 addr
->addr
.id
.node
= msg_orignode(msg
);
1169 addr
->addr
.name
.domain
= 0; /* could leave uninitialized */
1170 addr
->scope
= 0; /* could leave uninitialized */
1171 m
->msg_namelen
= sizeof(struct sockaddr_tipc
);
1176 * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
1177 * @m: descriptor for message info
1178 * @msg: received message header
1179 * @tsk: TIPC port associated with message
1181 * Note: Ancillary data is not captured if not requested by receiver.
1183 * Returns 0 if successful, otherwise errno
1185 static int tipc_sk_anc_data_recv(struct msghdr
*m
, struct tipc_msg
*msg
,
1186 struct tipc_sock
*tsk
)
1194 if (likely(m
->msg_controllen
== 0))
1197 /* Optionally capture errored message object(s) */
1198 err
= msg
? msg_errcode(msg
) : 0;
1199 if (unlikely(err
)) {
1201 anc_data
[1] = msg_data_sz(msg
);
1202 res
= put_cmsg(m
, SOL_TIPC
, TIPC_ERRINFO
, 8, anc_data
);
1206 res
= put_cmsg(m
, SOL_TIPC
, TIPC_RETDATA
, anc_data
[1],
1213 /* Optionally capture message destination object */
1214 dest_type
= msg
? msg_type(msg
) : TIPC_DIRECT_MSG
;
1215 switch (dest_type
) {
1216 case TIPC_NAMED_MSG
:
1218 anc_data
[0] = msg_nametype(msg
);
1219 anc_data
[1] = msg_namelower(msg
);
1220 anc_data
[2] = msg_namelower(msg
);
1222 case TIPC_MCAST_MSG
:
1224 anc_data
[0] = msg_nametype(msg
);
1225 anc_data
[1] = msg_namelower(msg
);
1226 anc_data
[2] = msg_nameupper(msg
);
1229 has_name
= (tsk
->conn_type
!= 0);
1230 anc_data
[0] = tsk
->conn_type
;
1231 anc_data
[1] = tsk
->conn_instance
;
1232 anc_data
[2] = tsk
->conn_instance
;
1238 res
= put_cmsg(m
, SOL_TIPC
, TIPC_DESTNAME
, 12, anc_data
);
1246 static void tipc_sk_send_ack(struct tipc_sock
*tsk
, uint ack
)
1248 struct sk_buff
*skb
= NULL
;
1249 struct tipc_msg
*msg
;
1250 u32 peer_port
= tsk_peer_port(tsk
);
1251 u32 dnode
= tsk_peer_node(tsk
);
1253 if (!tsk
->connected
)
1255 skb
= tipc_msg_create(CONN_MANAGER
, CONN_ACK
, INT_H_SIZE
, 0, dnode
,
1256 tipc_own_addr
, peer_port
, tsk
->portid
, TIPC_OK
);
1260 msg_set_msgcnt(msg
, ack
);
1261 tipc_link_xmit_skb(skb
, dnode
, msg_link_selector(msg
));
1264 static int tipc_wait_for_rcvmsg(struct socket
*sock
, long *timeop
)
1266 struct sock
*sk
= sock
->sk
;
1268 long timeo
= *timeop
;
1272 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
1273 if (timeo
&& skb_queue_empty(&sk
->sk_receive_queue
)) {
1274 if (sock
->state
== SS_DISCONNECTING
) {
1279 timeo
= schedule_timeout(timeo
);
1283 if (!skb_queue_empty(&sk
->sk_receive_queue
))
1285 err
= sock_intr_errno(timeo
);
1286 if (signal_pending(current
))
1292 finish_wait(sk_sleep(sk
), &wait
);
1298 * tipc_recvmsg - receive packet-oriented message
1300 * @m: descriptor for message info
1301 * @buf_len: total size of user buffer area
1302 * @flags: receive flags
1304 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1305 * If the complete message doesn't fit in user area, truncate it.
1307 * Returns size of returned message data, errno otherwise
1309 static int tipc_recvmsg(struct kiocb
*iocb
, struct socket
*sock
,
1310 struct msghdr
*m
, size_t buf_len
, int flags
)
1312 struct sock
*sk
= sock
->sk
;
1313 struct tipc_sock
*tsk
= tipc_sk(sk
);
1314 struct sk_buff
*buf
;
1315 struct tipc_msg
*msg
;
1321 /* Catch invalid receive requests */
1322 if (unlikely(!buf_len
))
1327 if (unlikely(sock
->state
== SS_UNCONNECTED
)) {
1332 timeo
= sock_rcvtimeo(sk
, flags
& MSG_DONTWAIT
);
1335 /* Look for a message in receive queue; wait if necessary */
1336 res
= tipc_wait_for_rcvmsg(sock
, &timeo
);
1340 /* Look at first message in receive queue */
1341 buf
= skb_peek(&sk
->sk_receive_queue
);
1343 sz
= msg_data_sz(msg
);
1344 err
= msg_errcode(msg
);
1346 /* Discard an empty non-errored message & try again */
1347 if ((!sz
) && (!err
)) {
1348 tsk_advance_rx_queue(sk
);
1352 /* Capture sender's address (optional) */
1353 set_orig_addr(m
, msg
);
1355 /* Capture ancillary data (optional) */
1356 res
= tipc_sk_anc_data_recv(m
, msg
, tsk
);
1360 /* Capture message data (if valid) & compute return value (always) */
1362 if (unlikely(buf_len
< sz
)) {
1364 m
->msg_flags
|= MSG_TRUNC
;
1366 res
= skb_copy_datagram_msg(buf
, msg_hdr_sz(msg
), m
, sz
);
1371 if ((sock
->state
== SS_READY
) ||
1372 ((err
== TIPC_CONN_SHUTDOWN
) || m
->msg_control
))
1378 /* Consume received message (optional) */
1379 if (likely(!(flags
& MSG_PEEK
))) {
1380 if ((sock
->state
!= SS_READY
) &&
1381 (++tsk
->rcv_unacked
>= TIPC_CONNACK_INTV
)) {
1382 tipc_sk_send_ack(tsk
, tsk
->rcv_unacked
);
1383 tsk
->rcv_unacked
= 0;
1385 tsk_advance_rx_queue(sk
);
1393 * tipc_recv_stream - receive stream-oriented data
1395 * @m: descriptor for message info
1396 * @buf_len: total size of user buffer area
1397 * @flags: receive flags
1399 * Used for SOCK_STREAM messages only. If not enough data is available
1400 * will optionally wait for more; never truncates data.
1402 * Returns size of returned message data, errno otherwise
1404 static int tipc_recv_stream(struct kiocb
*iocb
, struct socket
*sock
,
1405 struct msghdr
*m
, size_t buf_len
, int flags
)
1407 struct sock
*sk
= sock
->sk
;
1408 struct tipc_sock
*tsk
= tipc_sk(sk
);
1409 struct sk_buff
*buf
;
1410 struct tipc_msg
*msg
;
1413 int sz_to_copy
, target
, needed
;
1418 /* Catch invalid receive attempts */
1419 if (unlikely(!buf_len
))
1424 if (unlikely(sock
->state
== SS_UNCONNECTED
)) {
1429 target
= sock_rcvlowat(sk
, flags
& MSG_WAITALL
, buf_len
);
1430 timeo
= sock_rcvtimeo(sk
, flags
& MSG_DONTWAIT
);
1433 /* Look for a message in receive queue; wait if necessary */
1434 res
= tipc_wait_for_rcvmsg(sock
, &timeo
);
1438 /* Look at first message in receive queue */
1439 buf
= skb_peek(&sk
->sk_receive_queue
);
1441 sz
= msg_data_sz(msg
);
1442 err
= msg_errcode(msg
);
1444 /* Discard an empty non-errored message & try again */
1445 if ((!sz
) && (!err
)) {
1446 tsk_advance_rx_queue(sk
);
1450 /* Optionally capture sender's address & ancillary data of first msg */
1451 if (sz_copied
== 0) {
1452 set_orig_addr(m
, msg
);
1453 res
= tipc_sk_anc_data_recv(m
, msg
, tsk
);
1458 /* Capture message data (if valid) & compute return value (always) */
1460 u32 offset
= (u32
)(unsigned long)(TIPC_SKB_CB(buf
)->handle
);
1463 needed
= (buf_len
- sz_copied
);
1464 sz_to_copy
= (sz
<= needed
) ? sz
: needed
;
1466 res
= skb_copy_datagram_msg(buf
, msg_hdr_sz(msg
) + offset
,
1471 sz_copied
+= sz_to_copy
;
1473 if (sz_to_copy
< sz
) {
1474 if (!(flags
& MSG_PEEK
))
1475 TIPC_SKB_CB(buf
)->handle
=
1476 (void *)(unsigned long)(offset
+ sz_to_copy
);
1481 goto exit
; /* can't add error msg to valid data */
1483 if ((err
== TIPC_CONN_SHUTDOWN
) || m
->msg_control
)
1489 /* Consume received message (optional) */
1490 if (likely(!(flags
& MSG_PEEK
))) {
1491 if (unlikely(++tsk
->rcv_unacked
>= TIPC_CONNACK_INTV
)) {
1492 tipc_sk_send_ack(tsk
, tsk
->rcv_unacked
);
1493 tsk
->rcv_unacked
= 0;
1495 tsk_advance_rx_queue(sk
);
1498 /* Loop around if more data is required */
1499 if ((sz_copied
< buf_len
) && /* didn't get all requested data */
1500 (!skb_queue_empty(&sk
->sk_receive_queue
) ||
1501 (sz_copied
< target
)) && /* and more is ready or required */
1502 (!(flags
& MSG_PEEK
)) && /* and aren't just peeking at data */
1503 (!err
)) /* and haven't reached a FIN */
1508 return sz_copied
? sz_copied
: res
;
1512 * tipc_write_space - wake up thread if port congestion is released
1515 static void tipc_write_space(struct sock
*sk
)
1517 struct socket_wq
*wq
;
1520 wq
= rcu_dereference(sk
->sk_wq
);
1521 if (wq_has_sleeper(wq
))
1522 wake_up_interruptible_sync_poll(&wq
->wait
, POLLOUT
|
1523 POLLWRNORM
| POLLWRBAND
);
1528 * tipc_data_ready - wake up threads to indicate messages have been received
1530 * @len: the length of messages
1532 static void tipc_data_ready(struct sock
*sk
)
1534 struct socket_wq
*wq
;
1537 wq
= rcu_dereference(sk
->sk_wq
);
1538 if (wq_has_sleeper(wq
))
1539 wake_up_interruptible_sync_poll(&wq
->wait
, POLLIN
|
1540 POLLRDNORM
| POLLRDBAND
);
1545 * filter_connect - Handle all incoming messages for a connection-based socket
1549 * Returns 0 (TIPC_OK) if everything ok, -TIPC_ERR_NO_PORT otherwise
1551 static int filter_connect(struct tipc_sock
*tsk
, struct sk_buff
**buf
)
1553 struct sock
*sk
= &tsk
->sk
;
1554 struct socket
*sock
= sk
->sk_socket
;
1555 struct tipc_msg
*msg
= buf_msg(*buf
);
1556 int retval
= -TIPC_ERR_NO_PORT
;
1561 switch ((int)sock
->state
) {
1563 /* Accept only connection-based messages sent by peer */
1564 if (tsk_peer_msg(tsk
, msg
)) {
1565 if (unlikely(msg_errcode(msg
))) {
1566 sock
->state
= SS_DISCONNECTING
;
1568 /* let timer expire on it's own */
1569 tipc_node_remove_conn(tsk_peer_node(tsk
),
1576 /* Accept only ACK or NACK message */
1578 if (unlikely(!msg_connected(msg
)))
1581 if (unlikely(msg_errcode(msg
))) {
1582 sock
->state
= SS_DISCONNECTING
;
1583 sk
->sk_err
= ECONNREFUSED
;
1588 if (unlikely(msg_importance(msg
) > TIPC_CRITICAL_IMPORTANCE
)) {
1589 sock
->state
= SS_DISCONNECTING
;
1590 sk
->sk_err
= EINVAL
;
1595 tipc_sk_finish_conn(tsk
, msg_origport(msg
), msg_orignode(msg
));
1596 msg_set_importance(&tsk
->phdr
, msg_importance(msg
));
1597 sock
->state
= SS_CONNECTED
;
1599 /* If an incoming message is an 'ACK-', it should be
1600 * discarded here because it doesn't contain useful
1601 * data. In addition, we should try to wake up
1602 * connect() routine if sleeping.
1604 if (msg_data_sz(msg
) == 0) {
1607 if (waitqueue_active(sk_sleep(sk
)))
1608 wake_up_interruptible(sk_sleep(sk
));
1613 case SS_UNCONNECTED
:
1614 /* Accept only SYN message */
1615 if (!msg_connected(msg
) && !(msg_errcode(msg
)))
1618 case SS_DISCONNECTING
:
1621 pr_err("Unknown socket state %u\n", sock
->state
);
1627 * rcvbuf_limit - get proper overload limit of socket receive queue
1631 * For all connection oriented messages, irrespective of importance,
1632 * the default overload value (i.e. 67MB) is set as limit.
1634 * For all connectionless messages, by default new queue limits are
1637 * TIPC_LOW_IMPORTANCE (4 MB)
1638 * TIPC_MEDIUM_IMPORTANCE (8 MB)
1639 * TIPC_HIGH_IMPORTANCE (16 MB)
1640 * TIPC_CRITICAL_IMPORTANCE (32 MB)
1642 * Returns overload limit according to corresponding message importance
1644 static unsigned int rcvbuf_limit(struct sock
*sk
, struct sk_buff
*buf
)
1646 struct tipc_msg
*msg
= buf_msg(buf
);
1648 if (msg_connected(msg
))
1649 return sysctl_tipc_rmem
[2];
1651 return sk
->sk_rcvbuf
>> TIPC_CRITICAL_IMPORTANCE
<<
1652 msg_importance(msg
);
1656 * filter_rcv - validate incoming message
1660 * Enqueues message on receive queue if acceptable; optionally handles
1661 * disconnect indication for a connected socket.
1663 * Called with socket lock already taken; port lock may also be taken.
1665 * Returns 0 (TIPC_OK) if message was consumed, -TIPC error code if message
1666 * to be rejected, 1 (TIPC_FWD_MSG) if (CONN_MANAGER) message to be forwarded
1668 static int filter_rcv(struct sock
*sk
, struct sk_buff
*buf
)
1670 struct socket
*sock
= sk
->sk_socket
;
1671 struct tipc_sock
*tsk
= tipc_sk(sk
);
1672 struct tipc_msg
*msg
= buf_msg(buf
);
1673 unsigned int limit
= rcvbuf_limit(sk
, buf
);
1677 if (unlikely(msg_user(msg
) == CONN_MANAGER
))
1678 return tipc_sk_proto_rcv(tsk
, &onode
, buf
);
1680 if (unlikely(msg_user(msg
) == SOCK_WAKEUP
)) {
1683 sk
->sk_write_space(sk
);
1687 /* Reject message if it is wrong sort of message for socket */
1688 if (msg_type(msg
) > TIPC_DIRECT_MSG
)
1689 return -TIPC_ERR_NO_PORT
;
1691 if (sock
->state
== SS_READY
) {
1692 if (msg_connected(msg
))
1693 return -TIPC_ERR_NO_PORT
;
1695 rc
= filter_connect(tsk
, &buf
);
1696 if (rc
!= TIPC_OK
|| buf
== NULL
)
1700 /* Reject message if there isn't room to queue it */
1701 if (sk_rmem_alloc_get(sk
) + buf
->truesize
>= limit
)
1702 return -TIPC_ERR_OVERLOAD
;
1704 /* Enqueue message */
1705 TIPC_SKB_CB(buf
)->handle
= NULL
;
1706 __skb_queue_tail(&sk
->sk_receive_queue
, buf
);
1707 skb_set_owner_r(buf
, sk
);
1709 sk
->sk_data_ready(sk
);
1714 * tipc_backlog_rcv - handle incoming message from backlog queue
1718 * Caller must hold socket lock, but not port lock.
1722 static int tipc_backlog_rcv(struct sock
*sk
, struct sk_buff
*skb
)
1726 struct tipc_sock
*tsk
= tipc_sk(sk
);
1727 uint truesize
= skb
->truesize
;
1729 rc
= filter_rcv(sk
, skb
);
1732 if (atomic_read(&tsk
->dupl_rcvcnt
) < TIPC_CONN_OVERLOAD_LIMIT
)
1733 atomic_add(truesize
, &tsk
->dupl_rcvcnt
);
1737 if ((rc
< 0) && !tipc_msg_reverse(skb
, &onode
, -rc
))
1740 tipc_link_xmit_skb(skb
, onode
, 0);
1746 * tipc_sk_rcv - handle incoming message
1747 * @skb: buffer containing arriving message
1749 * Returns 0 if success, or errno: -EHOSTUNREACH
1751 int tipc_sk_rcv(struct sk_buff
*skb
)
1753 struct tipc_sock
*tsk
;
1755 u32 dport
= msg_destport(buf_msg(skb
));
1760 /* Validate destination and message */
1761 tsk
= tipc_sk_lookup(dport
);
1762 if (unlikely(!tsk
)) {
1763 rc
= tipc_msg_eval(skb
, &dnode
);
1769 spin_lock_bh(&sk
->sk_lock
.slock
);
1771 if (!sock_owned_by_user(sk
)) {
1772 rc
= filter_rcv(sk
, skb
);
1774 if (sk
->sk_backlog
.len
== 0)
1775 atomic_set(&tsk
->dupl_rcvcnt
, 0);
1776 limit
= rcvbuf_limit(sk
, skb
) + atomic_read(&tsk
->dupl_rcvcnt
);
1777 if (sk_add_backlog(sk
, skb
, limit
))
1778 rc
= -TIPC_ERR_OVERLOAD
;
1780 spin_unlock_bh(&sk
->sk_lock
.slock
);
1785 if ((rc
< 0) && !tipc_msg_reverse(skb
, &dnode
, -rc
))
1786 return -EHOSTUNREACH
;
1788 tipc_link_xmit_skb(skb
, dnode
, 0);
1789 return (rc
< 0) ? -EHOSTUNREACH
: 0;
1792 static int tipc_wait_for_connect(struct socket
*sock
, long *timeo_p
)
1794 struct sock
*sk
= sock
->sk
;
1799 int err
= sock_error(sk
);
1804 if (signal_pending(current
))
1805 return sock_intr_errno(*timeo_p
);
1807 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
1808 done
= sk_wait_event(sk
, timeo_p
, sock
->state
!= SS_CONNECTING
);
1809 finish_wait(sk_sleep(sk
), &wait
);
1815 * tipc_connect - establish a connection to another TIPC port
1816 * @sock: socket structure
1817 * @dest: socket address for destination port
1818 * @destlen: size of socket address data structure
1819 * @flags: file-related flags associated with socket
1821 * Returns 0 on success, errno otherwise
1823 static int tipc_connect(struct socket
*sock
, struct sockaddr
*dest
,
1824 int destlen
, int flags
)
1826 struct sock
*sk
= sock
->sk
;
1827 struct sockaddr_tipc
*dst
= (struct sockaddr_tipc
*)dest
;
1828 struct msghdr m
= {NULL
,};
1829 long timeout
= (flags
& O_NONBLOCK
) ? 0 : tipc_sk(sk
)->conn_timeout
;
1830 socket_state previous
;
1835 /* For now, TIPC does not allow use of connect() with DGRAM/RDM types */
1836 if (sock
->state
== SS_READY
) {
1842 * Reject connection attempt using multicast address
1844 * Note: send_msg() validates the rest of the address fields,
1845 * so there's no need to do it here
1847 if (dst
->addrtype
== TIPC_ADDR_MCAST
) {
1852 previous
= sock
->state
;
1853 switch (sock
->state
) {
1854 case SS_UNCONNECTED
:
1855 /* Send a 'SYN-' to destination */
1857 m
.msg_namelen
= destlen
;
1859 /* If connect is in non-blocking case, set MSG_DONTWAIT to
1860 * indicate send_msg() is never blocked.
1863 m
.msg_flags
= MSG_DONTWAIT
;
1865 res
= tipc_sendmsg(NULL
, sock
, &m
, 0);
1866 if ((res
< 0) && (res
!= -EWOULDBLOCK
))
1869 /* Just entered SS_CONNECTING state; the only
1870 * difference is that return value in non-blocking
1871 * case is EINPROGRESS, rather than EALREADY.
1875 if (previous
== SS_CONNECTING
)
1879 timeout
= msecs_to_jiffies(timeout
);
1880 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
1881 res
= tipc_wait_for_connect(sock
, &timeout
);
1896 * tipc_listen - allow socket to listen for incoming connections
1897 * @sock: socket structure
1900 * Returns 0 on success, errno otherwise
1902 static int tipc_listen(struct socket
*sock
, int len
)
1904 struct sock
*sk
= sock
->sk
;
1909 if (sock
->state
!= SS_UNCONNECTED
)
1912 sock
->state
= SS_LISTENING
;
1920 static int tipc_wait_for_accept(struct socket
*sock
, long timeo
)
1922 struct sock
*sk
= sock
->sk
;
1926 /* True wake-one mechanism for incoming connections: only
1927 * one process gets woken up, not the 'whole herd'.
1928 * Since we do not 'race & poll' for established sockets
1929 * anymore, the common case will execute the loop only once.
1932 prepare_to_wait_exclusive(sk_sleep(sk
), &wait
,
1933 TASK_INTERRUPTIBLE
);
1934 if (timeo
&& skb_queue_empty(&sk
->sk_receive_queue
)) {
1936 timeo
= schedule_timeout(timeo
);
1940 if (!skb_queue_empty(&sk
->sk_receive_queue
))
1943 if (sock
->state
!= SS_LISTENING
)
1945 err
= sock_intr_errno(timeo
);
1946 if (signal_pending(current
))
1952 finish_wait(sk_sleep(sk
), &wait
);
1957 * tipc_accept - wait for connection request
1958 * @sock: listening socket
1959 * @newsock: new socket that is to be connected
1960 * @flags: file-related flags associated with socket
1962 * Returns 0 on success, errno otherwise
1964 static int tipc_accept(struct socket
*sock
, struct socket
*new_sock
, int flags
)
1966 struct sock
*new_sk
, *sk
= sock
->sk
;
1967 struct sk_buff
*buf
;
1968 struct tipc_sock
*new_tsock
;
1969 struct tipc_msg
*msg
;
1975 if (sock
->state
!= SS_LISTENING
) {
1979 timeo
= sock_rcvtimeo(sk
, flags
& O_NONBLOCK
);
1980 res
= tipc_wait_for_accept(sock
, timeo
);
1984 buf
= skb_peek(&sk
->sk_receive_queue
);
1986 res
= tipc_sk_create(sock_net(sock
->sk
), new_sock
, 0, 1);
1990 new_sk
= new_sock
->sk
;
1991 new_tsock
= tipc_sk(new_sk
);
1994 /* we lock on new_sk; but lockdep sees the lock on sk */
1995 lock_sock_nested(new_sk
, SINGLE_DEPTH_NESTING
);
1998 * Reject any stray messages received by new socket
1999 * before the socket lock was taken (very, very unlikely)
2001 tsk_rej_rx_queue(new_sk
);
2003 /* Connect new socket to it's peer */
2004 tipc_sk_finish_conn(new_tsock
, msg_origport(msg
), msg_orignode(msg
));
2005 new_sock
->state
= SS_CONNECTED
;
2007 tsk_set_importance(new_tsock
, msg_importance(msg
));
2008 if (msg_named(msg
)) {
2009 new_tsock
->conn_type
= msg_nametype(msg
);
2010 new_tsock
->conn_instance
= msg_nameinst(msg
);
2014 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2015 * Respond to 'SYN+' by queuing it on new socket.
2017 if (!msg_data_sz(msg
)) {
2018 struct msghdr m
= {NULL
,};
2020 tsk_advance_rx_queue(sk
);
2021 tipc_send_packet(NULL
, new_sock
, &m
, 0);
2023 __skb_dequeue(&sk
->sk_receive_queue
);
2024 __skb_queue_head(&new_sk
->sk_receive_queue
, buf
);
2025 skb_set_owner_r(buf
, new_sk
);
2027 release_sock(new_sk
);
2034 * tipc_shutdown - shutdown socket connection
2035 * @sock: socket structure
2036 * @how: direction to close (must be SHUT_RDWR)
2038 * Terminates connection (if necessary), then purges socket's receive queue.
2040 * Returns 0 on success, errno otherwise
2042 static int tipc_shutdown(struct socket
*sock
, int how
)
2044 struct sock
*sk
= sock
->sk
;
2045 struct tipc_sock
*tsk
= tipc_sk(sk
);
2046 struct sk_buff
*skb
;
2050 if (how
!= SHUT_RDWR
)
2055 switch (sock
->state
) {
2060 /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */
2061 skb
= __skb_dequeue(&sk
->sk_receive_queue
);
2063 if (TIPC_SKB_CB(skb
)->handle
!= NULL
) {
2067 if (tipc_msg_reverse(skb
, &dnode
, TIPC_CONN_SHUTDOWN
))
2068 tipc_link_xmit_skb(skb
, dnode
, tsk
->portid
);
2069 tipc_node_remove_conn(dnode
, tsk
->portid
);
2071 dnode
= tsk_peer_node(tsk
);
2072 skb
= tipc_msg_create(TIPC_CRITICAL_IMPORTANCE
,
2073 TIPC_CONN_MSG
, SHORT_H_SIZE
,
2074 0, dnode
, tipc_own_addr
,
2076 tsk
->portid
, TIPC_CONN_SHUTDOWN
);
2077 tipc_link_xmit_skb(skb
, dnode
, tsk
->portid
);
2080 sock
->state
= SS_DISCONNECTING
;
2081 tipc_node_remove_conn(dnode
, tsk
->portid
);
2084 case SS_DISCONNECTING
:
2086 /* Discard any unreceived messages */
2087 __skb_queue_purge(&sk
->sk_receive_queue
);
2089 /* Wake up anyone sleeping in poll */
2090 sk
->sk_state_change(sk
);
2102 static void tipc_sk_timeout(unsigned long data
)
2104 struct tipc_sock
*tsk
= (struct tipc_sock
*)data
;
2105 struct sock
*sk
= &tsk
->sk
;
2106 struct sk_buff
*skb
= NULL
;
2107 u32 peer_port
, peer_node
;
2110 if (!tsk
->connected
) {
2114 peer_port
= tsk_peer_port(tsk
);
2115 peer_node
= tsk_peer_node(tsk
);
2117 if (tsk
->probing_state
== TIPC_CONN_PROBING
) {
2118 /* Previous probe not answered -> self abort */
2119 skb
= tipc_msg_create(TIPC_CRITICAL_IMPORTANCE
, TIPC_CONN_MSG
,
2120 SHORT_H_SIZE
, 0, tipc_own_addr
,
2121 peer_node
, tsk
->portid
, peer_port
,
2124 skb
= tipc_msg_create(CONN_MANAGER
, CONN_PROBE
, INT_H_SIZE
,
2125 0, peer_node
, tipc_own_addr
,
2126 peer_port
, tsk
->portid
, TIPC_OK
);
2127 tsk
->probing_state
= TIPC_CONN_PROBING
;
2128 if (!mod_timer(&tsk
->timer
, jiffies
+ tsk
->probing_intv
))
2133 tipc_link_xmit_skb(skb
, peer_node
, tsk
->portid
);
2138 static int tipc_sk_publish(struct tipc_sock
*tsk
, uint scope
,
2139 struct tipc_name_seq
const *seq
)
2141 struct publication
*publ
;
2146 key
= tsk
->portid
+ tsk
->pub_count
+ 1;
2147 if (key
== tsk
->portid
)
2150 publ
= tipc_nametbl_publish(seq
->type
, seq
->lower
, seq
->upper
,
2151 scope
, tsk
->portid
, key
);
2152 if (unlikely(!publ
))
2155 list_add(&publ
->pport_list
, &tsk
->publications
);
2161 static int tipc_sk_withdraw(struct tipc_sock
*tsk
, uint scope
,
2162 struct tipc_name_seq
const *seq
)
2164 struct publication
*publ
;
2165 struct publication
*safe
;
2168 list_for_each_entry_safe(publ
, safe
, &tsk
->publications
, pport_list
) {
2170 if (publ
->scope
!= scope
)
2172 if (publ
->type
!= seq
->type
)
2174 if (publ
->lower
!= seq
->lower
)
2176 if (publ
->upper
!= seq
->upper
)
2178 tipc_nametbl_withdraw(publ
->type
, publ
->lower
,
2179 publ
->ref
, publ
->key
);
2183 tipc_nametbl_withdraw(publ
->type
, publ
->lower
,
2184 publ
->ref
, publ
->key
);
2187 if (list_empty(&tsk
->publications
))
2192 static int tipc_sk_show(struct tipc_sock
*tsk
, char *buf
,
2193 int len
, int full_id
)
2195 struct publication
*publ
;
2199 ret
= tipc_snprintf(buf
, len
, "<%u.%u.%u:%u>:",
2200 tipc_zone(tipc_own_addr
),
2201 tipc_cluster(tipc_own_addr
),
2202 tipc_node(tipc_own_addr
), tsk
->portid
);
2204 ret
= tipc_snprintf(buf
, len
, "%-10u:", tsk
->portid
);
2206 if (tsk
->connected
) {
2207 u32 dport
= tsk_peer_port(tsk
);
2208 u32 destnode
= tsk_peer_node(tsk
);
2210 ret
+= tipc_snprintf(buf
+ ret
, len
- ret
,
2211 " connected to <%u.%u.%u:%u>",
2212 tipc_zone(destnode
),
2213 tipc_cluster(destnode
),
2214 tipc_node(destnode
), dport
);
2215 if (tsk
->conn_type
!= 0)
2216 ret
+= tipc_snprintf(buf
+ ret
, len
- ret
,
2217 " via {%u,%u}", tsk
->conn_type
,
2218 tsk
->conn_instance
);
2219 } else if (tsk
->published
) {
2220 ret
+= tipc_snprintf(buf
+ ret
, len
- ret
, " bound to");
2221 list_for_each_entry(publ
, &tsk
->publications
, pport_list
) {
2222 if (publ
->lower
== publ
->upper
)
2223 ret
+= tipc_snprintf(buf
+ ret
, len
- ret
,
2224 " {%u,%u}", publ
->type
,
2227 ret
+= tipc_snprintf(buf
+ ret
, len
- ret
,
2228 " {%u,%u,%u}", publ
->type
,
2229 publ
->lower
, publ
->upper
);
2232 ret
+= tipc_snprintf(buf
+ ret
, len
- ret
, "\n");
2236 struct sk_buff
*tipc_sk_socks_show(void)
2238 const struct bucket_table
*tbl
;
2239 struct rhash_head
*pos
;
2240 struct sk_buff
*buf
;
2241 struct tlv_desc
*rep_tlv
;
2244 struct tipc_sock
*tsk
;
2248 buf
= tipc_cfg_reply_alloc(TLV_SPACE(ULTRA_STRING_MAX_LEN
));
2251 rep_tlv
= (struct tlv_desc
*)buf
->data
;
2252 pb
= TLV_DATA(rep_tlv
);
2253 pb_len
= ULTRA_STRING_MAX_LEN
;
2256 tbl
= rht_dereference_rcu((&tipc_sk_rht
)->tbl
, &tipc_sk_rht
);
2257 for (i
= 0; i
< tbl
->size
; i
++) {
2258 rht_for_each_entry_rcu(tsk
, pos
, tbl
, i
, node
) {
2259 spin_lock_bh(&tsk
->sk
.sk_lock
.slock
);
2260 str_len
+= tipc_sk_show(tsk
, pb
+ str_len
,
2261 pb_len
- str_len
, 0);
2262 spin_unlock_bh(&tsk
->sk
.sk_lock
.slock
);
2267 str_len
+= 1; /* for "\0" */
2268 skb_put(buf
, TLV_SPACE(str_len
));
2269 TLV_SET(rep_tlv
, TIPC_TLV_ULTRA_STRING
, NULL
, str_len
);
2274 /* tipc_sk_reinit: set non-zero address in all existing sockets
2275 * when we go from standalone to network mode.
2277 void tipc_sk_reinit(void)
2279 const struct bucket_table
*tbl
;
2280 struct rhash_head
*pos
;
2281 struct tipc_sock
*tsk
;
2282 struct tipc_msg
*msg
;
2286 tbl
= rht_dereference_rcu((&tipc_sk_rht
)->tbl
, &tipc_sk_rht
);
2287 for (i
= 0; i
< tbl
->size
; i
++) {
2288 rht_for_each_entry_rcu(tsk
, pos
, tbl
, i
, node
) {
2289 spin_lock_bh(&tsk
->sk
.sk_lock
.slock
);
2291 msg_set_prevnode(msg
, tipc_own_addr
);
2292 msg_set_orignode(msg
, tipc_own_addr
);
2293 spin_unlock_bh(&tsk
->sk
.sk_lock
.slock
);
2299 static struct tipc_sock
*tipc_sk_lookup(u32 portid
)
2301 struct tipc_sock
*tsk
;
2304 tsk
= rhashtable_lookup(&tipc_sk_rht
, &portid
);
2306 sock_hold(&tsk
->sk
);
2312 static int tipc_sk_insert(struct tipc_sock
*tsk
)
2314 u32 remaining
= (TIPC_MAX_PORT
- TIPC_MIN_PORT
) + 1;
2315 u32 portid
= prandom_u32() % remaining
+ TIPC_MIN_PORT
;
2317 while (remaining
--) {
2319 if ((portid
< TIPC_MIN_PORT
) || (portid
> TIPC_MAX_PORT
))
2320 portid
= TIPC_MIN_PORT
;
2321 tsk
->portid
= portid
;
2322 sock_hold(&tsk
->sk
);
2323 if (rhashtable_lookup_insert(&tipc_sk_rht
, &tsk
->node
))
2331 static void tipc_sk_remove(struct tipc_sock
*tsk
)
2333 struct sock
*sk
= &tsk
->sk
;
2335 if (rhashtable_remove(&tipc_sk_rht
, &tsk
->node
)) {
2336 WARN_ON(atomic_read(&sk
->sk_refcnt
) == 1);
2341 int tipc_sk_rht_init(void)
2343 struct rhashtable_params rht_params
= {
2345 .head_offset
= offsetof(struct tipc_sock
, node
),
2346 .key_offset
= offsetof(struct tipc_sock
, portid
),
2347 .key_len
= sizeof(u32
), /* portid */
2349 .max_shift
= 20, /* 1M */
2350 .min_shift
= 8, /* 256 */
2351 .grow_decision
= rht_grow_above_75
,
2352 .shrink_decision
= rht_shrink_below_30
,
2355 return rhashtable_init(&tipc_sk_rht
, &rht_params
);
2358 void tipc_sk_rht_destroy(void)
2360 /* Wait for socket readers to complete */
2363 rhashtable_destroy(&tipc_sk_rht
);
2367 * tipc_setsockopt - set socket option
2368 * @sock: socket structure
2369 * @lvl: option level
2370 * @opt: option identifier
2371 * @ov: pointer to new option value
2372 * @ol: length of option value
2374 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
2375 * (to ease compatibility).
2377 * Returns 0 on success, errno otherwise
2379 static int tipc_setsockopt(struct socket
*sock
, int lvl
, int opt
,
2380 char __user
*ov
, unsigned int ol
)
2382 struct sock
*sk
= sock
->sk
;
2383 struct tipc_sock
*tsk
= tipc_sk(sk
);
2387 if ((lvl
== IPPROTO_TCP
) && (sock
->type
== SOCK_STREAM
))
2389 if (lvl
!= SOL_TIPC
)
2390 return -ENOPROTOOPT
;
2391 if (ol
< sizeof(value
))
2393 res
= get_user(value
, (u32 __user
*)ov
);
2400 case TIPC_IMPORTANCE
:
2401 res
= tsk_set_importance(tsk
, value
);
2403 case TIPC_SRC_DROPPABLE
:
2404 if (sock
->type
!= SOCK_STREAM
)
2405 tsk_set_unreliable(tsk
, value
);
2409 case TIPC_DEST_DROPPABLE
:
2410 tsk_set_unreturnable(tsk
, value
);
2412 case TIPC_CONN_TIMEOUT
:
2413 tipc_sk(sk
)->conn_timeout
= value
;
2414 /* no need to set "res", since already 0 at this point */
2426 * tipc_getsockopt - get socket option
2427 * @sock: socket structure
2428 * @lvl: option level
2429 * @opt: option identifier
2430 * @ov: receptacle for option value
2431 * @ol: receptacle for length of option value
2433 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
2434 * (to ease compatibility).
2436 * Returns 0 on success, errno otherwise
2438 static int tipc_getsockopt(struct socket
*sock
, int lvl
, int opt
,
2439 char __user
*ov
, int __user
*ol
)
2441 struct sock
*sk
= sock
->sk
;
2442 struct tipc_sock
*tsk
= tipc_sk(sk
);
2447 if ((lvl
== IPPROTO_TCP
) && (sock
->type
== SOCK_STREAM
))
2448 return put_user(0, ol
);
2449 if (lvl
!= SOL_TIPC
)
2450 return -ENOPROTOOPT
;
2451 res
= get_user(len
, ol
);
2458 case TIPC_IMPORTANCE
:
2459 value
= tsk_importance(tsk
);
2461 case TIPC_SRC_DROPPABLE
:
2462 value
= tsk_unreliable(tsk
);
2464 case TIPC_DEST_DROPPABLE
:
2465 value
= tsk_unreturnable(tsk
);
2467 case TIPC_CONN_TIMEOUT
:
2468 value
= tsk
->conn_timeout
;
2469 /* no need to set "res", since already 0 at this point */
2471 case TIPC_NODE_RECVQ_DEPTH
:
2472 value
= 0; /* was tipc_queue_size, now obsolete */
2474 case TIPC_SOCK_RECVQ_DEPTH
:
2475 value
= skb_queue_len(&sk
->sk_receive_queue
);
2484 return res
; /* "get" failed */
2486 if (len
< sizeof(value
))
2489 if (copy_to_user(ov
, &value
, sizeof(value
)))
2492 return put_user(sizeof(value
), ol
);
2495 static int tipc_ioctl(struct socket
*sk
, unsigned int cmd
, unsigned long arg
)
2497 struct tipc_sioc_ln_req lnr
;
2498 void __user
*argp
= (void __user
*)arg
;
2501 case SIOCGETLINKNAME
:
2502 if (copy_from_user(&lnr
, argp
, sizeof(lnr
)))
2504 if (!tipc_node_get_linkname(lnr
.bearer_id
& 0xffff, lnr
.peer
,
2505 lnr
.linkname
, TIPC_MAX_LINK_NAME
)) {
2506 if (copy_to_user(argp
, &lnr
, sizeof(lnr
)))
2510 return -EADDRNOTAVAIL
;
2512 return -ENOIOCTLCMD
;
2516 /* Protocol switches for the various types of TIPC sockets */
2518 static const struct proto_ops msg_ops
= {
2519 .owner
= THIS_MODULE
,
2521 .release
= tipc_release
,
2523 .connect
= tipc_connect
,
2524 .socketpair
= sock_no_socketpair
,
2525 .accept
= sock_no_accept
,
2526 .getname
= tipc_getname
,
2528 .ioctl
= tipc_ioctl
,
2529 .listen
= sock_no_listen
,
2530 .shutdown
= tipc_shutdown
,
2531 .setsockopt
= tipc_setsockopt
,
2532 .getsockopt
= tipc_getsockopt
,
2533 .sendmsg
= tipc_sendmsg
,
2534 .recvmsg
= tipc_recvmsg
,
2535 .mmap
= sock_no_mmap
,
2536 .sendpage
= sock_no_sendpage
2539 static const struct proto_ops packet_ops
= {
2540 .owner
= THIS_MODULE
,
2542 .release
= tipc_release
,
2544 .connect
= tipc_connect
,
2545 .socketpair
= sock_no_socketpair
,
2546 .accept
= tipc_accept
,
2547 .getname
= tipc_getname
,
2549 .ioctl
= tipc_ioctl
,
2550 .listen
= tipc_listen
,
2551 .shutdown
= tipc_shutdown
,
2552 .setsockopt
= tipc_setsockopt
,
2553 .getsockopt
= tipc_getsockopt
,
2554 .sendmsg
= tipc_send_packet
,
2555 .recvmsg
= tipc_recvmsg
,
2556 .mmap
= sock_no_mmap
,
2557 .sendpage
= sock_no_sendpage
2560 static const struct proto_ops stream_ops
= {
2561 .owner
= THIS_MODULE
,
2563 .release
= tipc_release
,
2565 .connect
= tipc_connect
,
2566 .socketpair
= sock_no_socketpair
,
2567 .accept
= tipc_accept
,
2568 .getname
= tipc_getname
,
2570 .ioctl
= tipc_ioctl
,
2571 .listen
= tipc_listen
,
2572 .shutdown
= tipc_shutdown
,
2573 .setsockopt
= tipc_setsockopt
,
2574 .getsockopt
= tipc_getsockopt
,
2575 .sendmsg
= tipc_send_stream
,
2576 .recvmsg
= tipc_recv_stream
,
2577 .mmap
= sock_no_mmap
,
2578 .sendpage
= sock_no_sendpage
2581 static const struct net_proto_family tipc_family_ops
= {
2582 .owner
= THIS_MODULE
,
2584 .create
= tipc_sk_create
2587 static struct proto tipc_proto
= {
2589 .owner
= THIS_MODULE
,
2590 .obj_size
= sizeof(struct tipc_sock
),
2591 .sysctl_rmem
= sysctl_tipc_rmem
2594 static struct proto tipc_proto_kern
= {
2596 .obj_size
= sizeof(struct tipc_sock
),
2597 .sysctl_rmem
= sysctl_tipc_rmem
2601 * tipc_socket_init - initialize TIPC socket interface
2603 * Returns 0 on success, errno otherwise
2605 int tipc_socket_init(void)
2609 res
= proto_register(&tipc_proto
, 1);
2611 pr_err("Failed to register TIPC protocol type\n");
2615 res
= sock_register(&tipc_family_ops
);
2617 pr_err("Failed to register TIPC socket type\n");
2618 proto_unregister(&tipc_proto
);
2626 * tipc_socket_stop - stop TIPC socket interface
2628 void tipc_socket_stop(void)
2630 sock_unregister(tipc_family_ops
.family
);
2631 proto_unregister(&tipc_proto
);
2634 /* Caller should hold socket lock for the passed tipc socket. */
2635 static int __tipc_nl_add_sk_con(struct sk_buff
*skb
, struct tipc_sock
*tsk
)
2639 struct nlattr
*nest
;
2641 peer_node
= tsk_peer_node(tsk
);
2642 peer_port
= tsk_peer_port(tsk
);
2644 nest
= nla_nest_start(skb
, TIPC_NLA_SOCK_CON
);
2646 if (nla_put_u32(skb
, TIPC_NLA_CON_NODE
, peer_node
))
2648 if (nla_put_u32(skb
, TIPC_NLA_CON_SOCK
, peer_port
))
2651 if (tsk
->conn_type
!= 0) {
2652 if (nla_put_flag(skb
, TIPC_NLA_CON_FLAG
))
2654 if (nla_put_u32(skb
, TIPC_NLA_CON_TYPE
, tsk
->conn_type
))
2656 if (nla_put_u32(skb
, TIPC_NLA_CON_INST
, tsk
->conn_instance
))
2659 nla_nest_end(skb
, nest
);
2664 nla_nest_cancel(skb
, nest
);
2669 /* Caller should hold socket lock for the passed tipc socket. */
2670 static int __tipc_nl_add_sk(struct sk_buff
*skb
, struct netlink_callback
*cb
,
2671 struct tipc_sock
*tsk
)
2675 struct nlattr
*attrs
;
2677 hdr
= genlmsg_put(skb
, NETLINK_CB(cb
->skb
).portid
, cb
->nlh
->nlmsg_seq
,
2678 &tipc_genl_v2_family
, NLM_F_MULTI
, TIPC_NL_SOCK_GET
);
2682 attrs
= nla_nest_start(skb
, TIPC_NLA_SOCK
);
2684 goto genlmsg_cancel
;
2685 if (nla_put_u32(skb
, TIPC_NLA_SOCK_REF
, tsk
->portid
))
2686 goto attr_msg_cancel
;
2687 if (nla_put_u32(skb
, TIPC_NLA_SOCK_ADDR
, tipc_own_addr
))
2688 goto attr_msg_cancel
;
2690 if (tsk
->connected
) {
2691 err
= __tipc_nl_add_sk_con(skb
, tsk
);
2693 goto attr_msg_cancel
;
2694 } else if (!list_empty(&tsk
->publications
)) {
2695 if (nla_put_flag(skb
, TIPC_NLA_SOCK_HAS_PUBL
))
2696 goto attr_msg_cancel
;
2698 nla_nest_end(skb
, attrs
);
2699 genlmsg_end(skb
, hdr
);
2704 nla_nest_cancel(skb
, attrs
);
2706 genlmsg_cancel(skb
, hdr
);
2711 int tipc_nl_sk_dump(struct sk_buff
*skb
, struct netlink_callback
*cb
)
2714 struct tipc_sock
*tsk
;
2715 const struct bucket_table
*tbl
;
2716 struct rhash_head
*pos
;
2717 u32 prev_portid
= cb
->args
[0];
2718 u32 portid
= prev_portid
;
2722 tbl
= rht_dereference_rcu((&tipc_sk_rht
)->tbl
, &tipc_sk_rht
);
2723 for (i
= 0; i
< tbl
->size
; i
++) {
2724 rht_for_each_entry_rcu(tsk
, pos
, tbl
, i
, node
) {
2725 spin_lock_bh(&tsk
->sk
.sk_lock
.slock
);
2726 portid
= tsk
->portid
;
2727 err
= __tipc_nl_add_sk(skb
, cb
, tsk
);
2728 spin_unlock_bh(&tsk
->sk
.sk_lock
.slock
);
2732 prev_portid
= portid
;
2737 cb
->args
[0] = prev_portid
;
2742 /* Caller should hold socket lock for the passed tipc socket. */
2743 static int __tipc_nl_add_sk_publ(struct sk_buff
*skb
,
2744 struct netlink_callback
*cb
,
2745 struct publication
*publ
)
2748 struct nlattr
*attrs
;
2750 hdr
= genlmsg_put(skb
, NETLINK_CB(cb
->skb
).portid
, cb
->nlh
->nlmsg_seq
,
2751 &tipc_genl_v2_family
, NLM_F_MULTI
, TIPC_NL_PUBL_GET
);
2755 attrs
= nla_nest_start(skb
, TIPC_NLA_PUBL
);
2757 goto genlmsg_cancel
;
2759 if (nla_put_u32(skb
, TIPC_NLA_PUBL_KEY
, publ
->key
))
2760 goto attr_msg_cancel
;
2761 if (nla_put_u32(skb
, TIPC_NLA_PUBL_TYPE
, publ
->type
))
2762 goto attr_msg_cancel
;
2763 if (nla_put_u32(skb
, TIPC_NLA_PUBL_LOWER
, publ
->lower
))
2764 goto attr_msg_cancel
;
2765 if (nla_put_u32(skb
, TIPC_NLA_PUBL_UPPER
, publ
->upper
))
2766 goto attr_msg_cancel
;
2768 nla_nest_end(skb
, attrs
);
2769 genlmsg_end(skb
, hdr
);
2774 nla_nest_cancel(skb
, attrs
);
2776 genlmsg_cancel(skb
, hdr
);
2781 /* Caller should hold socket lock for the passed tipc socket. */
2782 static int __tipc_nl_list_sk_publ(struct sk_buff
*skb
,
2783 struct netlink_callback
*cb
,
2784 struct tipc_sock
*tsk
, u32
*last_publ
)
2787 struct publication
*p
;
2790 list_for_each_entry(p
, &tsk
->publications
, pport_list
) {
2791 if (p
->key
== *last_publ
)
2794 if (p
->key
!= *last_publ
) {
2795 /* We never set seq or call nl_dump_check_consistent()
2796 * this means that setting prev_seq here will cause the
2797 * consistence check to fail in the netlink callback
2798 * handler. Resulting in the last NLMSG_DONE message
2799 * having the NLM_F_DUMP_INTR flag set.
2806 p
= list_first_entry(&tsk
->publications
, struct publication
,
2810 list_for_each_entry_from(p
, &tsk
->publications
, pport_list
) {
2811 err
= __tipc_nl_add_sk_publ(skb
, cb
, p
);
2813 *last_publ
= p
->key
;
2822 int tipc_nl_publ_dump(struct sk_buff
*skb
, struct netlink_callback
*cb
)
2825 u32 tsk_portid
= cb
->args
[0];
2826 u32 last_publ
= cb
->args
[1];
2827 u32 done
= cb
->args
[2];
2828 struct tipc_sock
*tsk
;
2831 struct nlattr
**attrs
;
2832 struct nlattr
*sock
[TIPC_NLA_SOCK_MAX
+ 1];
2834 err
= tipc_nlmsg_parse(cb
->nlh
, &attrs
);
2838 err
= nla_parse_nested(sock
, TIPC_NLA_SOCK_MAX
,
2839 attrs
[TIPC_NLA_SOCK
],
2840 tipc_nl_sock_policy
);
2844 if (!sock
[TIPC_NLA_SOCK_REF
])
2847 tsk_portid
= nla_get_u32(sock
[TIPC_NLA_SOCK_REF
]);
2853 tsk
= tipc_sk_lookup(tsk_portid
);
2857 lock_sock(&tsk
->sk
);
2858 err
= __tipc_nl_list_sk_publ(skb
, cb
, tsk
, &last_publ
);
2861 release_sock(&tsk
->sk
);
2864 cb
->args
[0] = tsk_portid
;
2865 cb
->args
[1] = last_publ
;