2 * net/tipc/socket.c: TIPC socket API
4 * Copyright (c) 2001-2007, 2012-2015, 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>
39 #include "name_table.h"
42 #include "name_distr.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
72 * @conn_timeout: the time we can wait for an unresponded setup request
73 * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue
74 * @link_cong: non-zero if owner must sleep because of link congestion
75 * @sent_unacked: # messages sent by socket, and not yet acked by peer
76 * @rcv_unacked: # messages read by user, but not yet acked back to peer
77 * @remote: 'connected' peer for dgram/rdm
78 * @node: hash table node
79 * @rcu: rcu struct for tipc_sock
90 struct list_head sock_list
;
91 struct list_head publications
;
94 unsigned long probing_intv
;
100 struct sockaddr_tipc remote
;
101 struct rhash_head node
;
105 static int tipc_backlog_rcv(struct sock
*sk
, struct sk_buff
*skb
);
106 static void tipc_data_ready(struct sock
*sk
);
107 static void tipc_write_space(struct sock
*sk
);
108 static int tipc_release(struct socket
*sock
);
109 static int tipc_accept(struct socket
*sock
, struct socket
*new_sock
, int flags
);
110 static int tipc_wait_for_sndmsg(struct socket
*sock
, long *timeo_p
);
111 static void tipc_sk_timeout(unsigned long data
);
112 static int tipc_sk_publish(struct tipc_sock
*tsk
, uint scope
,
113 struct tipc_name_seq
const *seq
);
114 static int tipc_sk_withdraw(struct tipc_sock
*tsk
, uint scope
,
115 struct tipc_name_seq
const *seq
);
116 static struct tipc_sock
*tipc_sk_lookup(struct net
*net
, u32 portid
);
117 static int tipc_sk_insert(struct tipc_sock
*tsk
);
118 static void tipc_sk_remove(struct tipc_sock
*tsk
);
119 static int __tipc_send_stream(struct socket
*sock
, struct msghdr
*m
,
121 static int __tipc_sendmsg(struct socket
*sock
, struct msghdr
*m
, size_t dsz
);
123 static const struct proto_ops packet_ops
;
124 static const struct proto_ops stream_ops
;
125 static const struct proto_ops msg_ops
;
126 static struct proto tipc_proto
;
128 static const struct nla_policy tipc_nl_sock_policy
[TIPC_NLA_SOCK_MAX
+ 1] = {
129 [TIPC_NLA_SOCK_UNSPEC
] = { .type
= NLA_UNSPEC
},
130 [TIPC_NLA_SOCK_ADDR
] = { .type
= NLA_U32
},
131 [TIPC_NLA_SOCK_REF
] = { .type
= NLA_U32
},
132 [TIPC_NLA_SOCK_CON
] = { .type
= NLA_NESTED
},
133 [TIPC_NLA_SOCK_HAS_PUBL
] = { .type
= NLA_FLAG
}
136 static const struct rhashtable_params tsk_rht_params
;
139 * Revised TIPC socket locking policy:
141 * Most socket operations take the standard socket lock when they start
142 * and hold it until they finish (or until they need to sleep). Acquiring
143 * this lock grants the owner exclusive access to the fields of the socket
144 * data structures, with the exception of the backlog queue. A few socket
145 * operations can be done without taking the socket lock because they only
146 * read socket information that never changes during the life of the socket.
148 * Socket operations may acquire the lock for the associated TIPC port if they
149 * need to perform an operation on the port. If any routine needs to acquire
150 * both the socket lock and the port lock it must take the socket lock first
151 * to avoid the risk of deadlock.
153 * The dispatcher handling incoming messages cannot grab the socket lock in
154 * the standard fashion, since invoked it runs at the BH level and cannot block.
155 * Instead, it checks to see if the socket lock is currently owned by someone,
156 * and either handles the message itself or adds it to the socket's backlog
157 * queue; in the latter case the queued message is processed once the process
158 * owning the socket lock releases it.
160 * NOTE: Releasing the socket lock while an operation is sleeping overcomes
161 * the problem of a blocked socket operation preventing any other operations
162 * from occurring. However, applications must be careful if they have
163 * multiple threads trying to send (or receive) on the same socket, as these
164 * operations might interfere with each other. For example, doing a connect
165 * and a receive at the same time might allow the receive to consume the
166 * ACK message meant for the connect. While additional work could be done
167 * to try and overcome this, it doesn't seem to be worthwhile at the present.
169 * NOTE: Releasing the socket lock while an operation is sleeping also ensures
170 * that another operation that must be performed in a non-blocking manner is
171 * not delayed for very long because the lock has already been taken.
173 * NOTE: This code assumes that certain fields of a port/socket pair are
174 * constant over its lifetime; such fields can be examined without taking
175 * the socket lock and/or port lock, and do not need to be re-read even
176 * after resuming processing after waiting. These fields include:
178 * - pointer to socket sk structure (aka tipc_sock structure)
179 * - pointer to port structure
183 static u32
tsk_own_node(struct tipc_sock
*tsk
)
185 return msg_prevnode(&tsk
->phdr
);
188 static u32
tsk_peer_node(struct tipc_sock
*tsk
)
190 return msg_destnode(&tsk
->phdr
);
193 static u32
tsk_peer_port(struct tipc_sock
*tsk
)
195 return msg_destport(&tsk
->phdr
);
198 static bool tsk_unreliable(struct tipc_sock
*tsk
)
200 return msg_src_droppable(&tsk
->phdr
) != 0;
203 static void tsk_set_unreliable(struct tipc_sock
*tsk
, bool unreliable
)
205 msg_set_src_droppable(&tsk
->phdr
, unreliable
? 1 : 0);
208 static bool tsk_unreturnable(struct tipc_sock
*tsk
)
210 return msg_dest_droppable(&tsk
->phdr
) != 0;
213 static void tsk_set_unreturnable(struct tipc_sock
*tsk
, bool unreturnable
)
215 msg_set_dest_droppable(&tsk
->phdr
, unreturnable
? 1 : 0);
218 static int tsk_importance(struct tipc_sock
*tsk
)
220 return msg_importance(&tsk
->phdr
);
223 static int tsk_set_importance(struct tipc_sock
*tsk
, int imp
)
225 if (imp
> TIPC_CRITICAL_IMPORTANCE
)
227 msg_set_importance(&tsk
->phdr
, (u32
)imp
);
231 static struct tipc_sock
*tipc_sk(const struct sock
*sk
)
233 return container_of(sk
, struct tipc_sock
, sk
);
236 static int tsk_conn_cong(struct tipc_sock
*tsk
)
238 return tsk
->sent_unacked
>= TIPC_FLOWCTRL_WIN
;
242 * tsk_advance_rx_queue - discard first buffer in socket receive queue
244 * Caller must hold socket lock
246 static void tsk_advance_rx_queue(struct sock
*sk
)
248 kfree_skb(__skb_dequeue(&sk
->sk_receive_queue
));
251 /* tipc_sk_respond() : send response message back to sender
253 static void tipc_sk_respond(struct sock
*sk
, struct sk_buff
*skb
, int err
)
257 u32 onode
= tipc_own_addr(sock_net(sk
));
259 if (!tipc_msg_reverse(onode
, &skb
, err
))
262 dnode
= msg_destnode(buf_msg(skb
));
263 selector
= msg_origport(buf_msg(skb
));
264 tipc_node_xmit_skb(sock_net(sk
), skb
, dnode
, selector
);
268 * tsk_rej_rx_queue - reject all buffers in socket receive queue
270 * Caller must hold socket lock
272 static void tsk_rej_rx_queue(struct sock
*sk
)
276 while ((skb
= __skb_dequeue(&sk
->sk_receive_queue
)))
277 tipc_sk_respond(sk
, skb
, TIPC_ERR_NO_PORT
);
280 /* tsk_peer_msg - verify if message was sent by connected port's peer
282 * Handles cases where the node's network address has changed from
283 * the default of <0.0.0> to its configured setting.
285 static bool tsk_peer_msg(struct tipc_sock
*tsk
, struct tipc_msg
*msg
)
287 struct tipc_net
*tn
= net_generic(sock_net(&tsk
->sk
), tipc_net_id
);
288 u32 peer_port
= tsk_peer_port(tsk
);
292 if (unlikely(!tsk
->connected
))
295 if (unlikely(msg_origport(msg
) != peer_port
))
298 orig_node
= msg_orignode(msg
);
299 peer_node
= tsk_peer_node(tsk
);
301 if (likely(orig_node
== peer_node
))
304 if (!orig_node
&& (peer_node
== tn
->own_addr
))
307 if (!peer_node
&& (orig_node
== tn
->own_addr
))
314 * tipc_sk_create - create a TIPC socket
315 * @net: network namespace (must be default network)
316 * @sock: pre-allocated socket structure
317 * @protocol: protocol indicator (must be 0)
318 * @kern: caused by kernel or by userspace?
320 * This routine creates additional data structures used by the TIPC socket,
321 * initializes them, and links them together.
323 * Returns 0 on success, errno otherwise
325 static int tipc_sk_create(struct net
*net
, struct socket
*sock
,
326 int protocol
, int kern
)
329 const struct proto_ops
*ops
;
332 struct tipc_sock
*tsk
;
333 struct tipc_msg
*msg
;
335 /* Validate arguments */
336 if (unlikely(protocol
!= 0))
337 return -EPROTONOSUPPORT
;
339 switch (sock
->type
) {
342 state
= SS_UNCONNECTED
;
346 state
= SS_UNCONNECTED
;
357 /* Allocate socket's protocol area */
358 sk
= sk_alloc(net
, AF_TIPC
, GFP_KERNEL
, &tipc_proto
, kern
);
363 tsk
->max_pkt
= MAX_PKT_DEFAULT
;
364 INIT_LIST_HEAD(&tsk
->publications
);
366 tn
= net_generic(sock_net(sk
), tipc_net_id
);
367 tipc_msg_init(tn
->own_addr
, msg
, TIPC_LOW_IMPORTANCE
, TIPC_NAMED_MSG
,
370 /* Finish initializing socket data structures */
373 sock_init_data(sock
, sk
);
374 if (tipc_sk_insert(tsk
)) {
375 pr_warn("Socket create failed; port numbrer exhausted\n");
378 msg_set_origport(msg
, tsk
->portid
);
379 setup_timer(&sk
->sk_timer
, tipc_sk_timeout
, (unsigned long)tsk
);
380 sk
->sk_backlog_rcv
= tipc_backlog_rcv
;
381 sk
->sk_rcvbuf
= sysctl_tipc_rmem
[1];
382 sk
->sk_data_ready
= tipc_data_ready
;
383 sk
->sk_write_space
= tipc_write_space
;
384 tsk
->conn_timeout
= CONN_TIMEOUT_DEFAULT
;
385 tsk
->sent_unacked
= 0;
386 atomic_set(&tsk
->dupl_rcvcnt
, 0);
388 if (sock
->state
== SS_READY
) {
389 tsk_set_unreturnable(tsk
, true);
390 if (sock
->type
== SOCK_DGRAM
)
391 tsk_set_unreliable(tsk
, true);
396 static void tipc_sk_callback(struct rcu_head
*head
)
398 struct tipc_sock
*tsk
= container_of(head
, struct tipc_sock
, rcu
);
404 * tipc_release - destroy a TIPC socket
405 * @sock: socket to destroy
407 * This routine cleans up any messages that are still queued on the socket.
408 * For DGRAM and RDM socket types, all queued messages are rejected.
409 * For SEQPACKET and STREAM socket types, the first message is rejected
410 * and any others are discarded. (If the first message on a STREAM socket
411 * is partially-read, it is discarded and the next one is rejected instead.)
413 * NOTE: Rejected messages are not necessarily returned to the sender! They
414 * are returned or discarded according to the "destination droppable" setting
415 * specified for the message by the sender.
417 * Returns 0 on success, errno otherwise
419 static int tipc_release(struct socket
*sock
)
421 struct sock
*sk
= sock
->sk
;
423 struct tipc_sock
*tsk
;
428 * Exit if socket isn't fully initialized (occurs when a failed accept()
429 * releases a pre-allocated child socket that was never used)
439 * Reject all unreceived messages, except on an active connection
440 * (which disconnects locally & sends a 'FIN+' to peer)
442 dnode
= tsk_peer_node(tsk
);
443 while (sock
->state
!= SS_DISCONNECTING
) {
444 skb
= __skb_dequeue(&sk
->sk_receive_queue
);
447 if (TIPC_SKB_CB(skb
)->handle
!= NULL
)
450 if ((sock
->state
== SS_CONNECTING
) ||
451 (sock
->state
== SS_CONNECTED
)) {
452 sock
->state
= SS_DISCONNECTING
;
454 tipc_node_remove_conn(net
, dnode
, tsk
->portid
);
456 tipc_sk_respond(sk
, skb
, TIPC_ERR_NO_PORT
);
460 tipc_sk_withdraw(tsk
, 0, NULL
);
461 sk_stop_timer(sk
, &sk
->sk_timer
);
463 if (tsk
->connected
) {
464 skb
= tipc_msg_create(TIPC_CRITICAL_IMPORTANCE
,
465 TIPC_CONN_MSG
, SHORT_H_SIZE
, 0, dnode
,
466 tsk_own_node(tsk
), tsk_peer_port(tsk
),
467 tsk
->portid
, TIPC_ERR_NO_PORT
);
469 tipc_node_xmit_skb(net
, skb
, dnode
, tsk
->portid
);
470 tipc_node_remove_conn(net
, dnode
, tsk
->portid
);
473 /* Discard any remaining (connection-based) messages in receive queue */
474 __skb_queue_purge(&sk
->sk_receive_queue
);
476 /* Reject any messages that accumulated in backlog queue */
477 sock
->state
= SS_DISCONNECTING
;
480 call_rcu(&tsk
->rcu
, tipc_sk_callback
);
487 * tipc_bind - associate or disassocate TIPC name(s) with a socket
488 * @sock: socket structure
489 * @uaddr: socket address describing name(s) and desired operation
490 * @uaddr_len: size of socket address data structure
492 * Name and name sequence binding is indicated using a positive scope value;
493 * a negative scope value unbinds the specified name. Specifying no name
494 * (i.e. a socket address length of 0) unbinds all names from the socket.
496 * Returns 0 on success, errno otherwise
498 * NOTE: This routine doesn't need to take the socket lock since it doesn't
499 * access any non-constant socket information.
501 static int tipc_bind(struct socket
*sock
, struct sockaddr
*uaddr
,
504 struct sock
*sk
= sock
->sk
;
505 struct sockaddr_tipc
*addr
= (struct sockaddr_tipc
*)uaddr
;
506 struct tipc_sock
*tsk
= tipc_sk(sk
);
510 if (unlikely(!uaddr_len
)) {
511 res
= tipc_sk_withdraw(tsk
, 0, NULL
);
515 if (uaddr_len
< sizeof(struct sockaddr_tipc
)) {
519 if (addr
->family
!= AF_TIPC
) {
524 if (addr
->addrtype
== TIPC_ADDR_NAME
)
525 addr
->addr
.nameseq
.upper
= addr
->addr
.nameseq
.lower
;
526 else if (addr
->addrtype
!= TIPC_ADDR_NAMESEQ
) {
531 if ((addr
->addr
.nameseq
.type
< TIPC_RESERVED_TYPES
) &&
532 (addr
->addr
.nameseq
.type
!= TIPC_TOP_SRV
) &&
533 (addr
->addr
.nameseq
.type
!= TIPC_CFG_SRV
)) {
538 res
= (addr
->scope
> 0) ?
539 tipc_sk_publish(tsk
, addr
->scope
, &addr
->addr
.nameseq
) :
540 tipc_sk_withdraw(tsk
, -addr
->scope
, &addr
->addr
.nameseq
);
547 * tipc_getname - get port ID of socket or peer socket
548 * @sock: socket structure
549 * @uaddr: area for returned socket address
550 * @uaddr_len: area for returned length of socket address
551 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
553 * Returns 0 on success, errno otherwise
555 * NOTE: This routine doesn't need to take the socket lock since it only
556 * accesses socket information that is unchanging (or which changes in
557 * a completely predictable manner).
559 static int tipc_getname(struct socket
*sock
, struct sockaddr
*uaddr
,
560 int *uaddr_len
, int peer
)
562 struct sockaddr_tipc
*addr
= (struct sockaddr_tipc
*)uaddr
;
563 struct tipc_sock
*tsk
= tipc_sk(sock
->sk
);
564 struct tipc_net
*tn
= net_generic(sock_net(sock
->sk
), tipc_net_id
);
566 memset(addr
, 0, sizeof(*addr
));
568 if ((sock
->state
!= SS_CONNECTED
) &&
569 ((peer
!= 2) || (sock
->state
!= SS_DISCONNECTING
)))
571 addr
->addr
.id
.ref
= tsk_peer_port(tsk
);
572 addr
->addr
.id
.node
= tsk_peer_node(tsk
);
574 addr
->addr
.id
.ref
= tsk
->portid
;
575 addr
->addr
.id
.node
= tn
->own_addr
;
578 *uaddr_len
= sizeof(*addr
);
579 addr
->addrtype
= TIPC_ADDR_ID
;
580 addr
->family
= AF_TIPC
;
582 addr
->addr
.name
.domain
= 0;
588 * tipc_poll - read and possibly block on pollmask
589 * @file: file structure associated with the socket
590 * @sock: socket for which to calculate the poll bits
593 * Returns pollmask value
596 * It appears that the usual socket locking mechanisms are not useful here
597 * since the pollmask info is potentially out-of-date the moment this routine
598 * exits. TCP and other protocols seem to rely on higher level poll routines
599 * to handle any preventable race conditions, so TIPC will do the same ...
601 * TIPC sets the returned events as follows:
603 * socket state flags set
604 * ------------ ---------
605 * unconnected no read flags
606 * POLLOUT if port is not congested
608 * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue
611 * connected POLLIN/POLLRDNORM if data in rx queue
612 * POLLOUT if port is not congested
614 * disconnecting POLLIN/POLLRDNORM/POLLHUP
617 * listening POLLIN if SYN in rx queue
620 * ready POLLIN/POLLRDNORM if data in rx queue
621 * [connectionless] POLLOUT (since port cannot be congested)
623 * IMPORTANT: The fact that a read or write operation is indicated does NOT
624 * imply that the operation will succeed, merely that it should be performed
625 * and will not block.
627 static unsigned int tipc_poll(struct file
*file
, struct socket
*sock
,
630 struct sock
*sk
= sock
->sk
;
631 struct tipc_sock
*tsk
= tipc_sk(sk
);
634 sock_poll_wait(file
, sk_sleep(sk
), wait
);
636 switch ((int)sock
->state
) {
643 if (!tsk
->link_cong
&& !tsk_conn_cong(tsk
))
648 if (!skb_queue_empty(&sk
->sk_receive_queue
))
649 mask
|= (POLLIN
| POLLRDNORM
);
651 case SS_DISCONNECTING
:
652 mask
= (POLLIN
| POLLRDNORM
| POLLHUP
);
660 * tipc_sendmcast - send multicast message
661 * @sock: socket structure
662 * @seq: destination address
663 * @msg: message to send
664 * @dsz: total length of message data
665 * @timeo: timeout to wait for wakeup
667 * Called from function tipc_sendmsg(), which has done all sanity checks
668 * Returns the number of bytes sent on success, or errno
670 static int tipc_sendmcast(struct socket
*sock
, struct tipc_name_seq
*seq
,
671 struct msghdr
*msg
, size_t dsz
, long timeo
)
673 struct sock
*sk
= sock
->sk
;
674 struct tipc_sock
*tsk
= tipc_sk(sk
);
675 struct net
*net
= sock_net(sk
);
676 struct tipc_msg
*mhdr
= &tsk
->phdr
;
677 struct sk_buff_head
*pktchain
= &sk
->sk_write_queue
;
678 struct iov_iter save
= msg
->msg_iter
;
682 msg_set_type(mhdr
, TIPC_MCAST_MSG
);
683 msg_set_lookup_scope(mhdr
, TIPC_CLUSTER_SCOPE
);
684 msg_set_destport(mhdr
, 0);
685 msg_set_destnode(mhdr
, 0);
686 msg_set_nametype(mhdr
, seq
->type
);
687 msg_set_namelower(mhdr
, seq
->lower
);
688 msg_set_nameupper(mhdr
, seq
->upper
);
689 msg_set_hdr_sz(mhdr
, MCAST_H_SIZE
);
692 mtu
= tipc_bcast_get_mtu(net
);
693 rc
= tipc_msg_build(mhdr
, msg
, 0, dsz
, mtu
, pktchain
);
694 if (unlikely(rc
< 0))
698 rc
= tipc_bcast_xmit(net
, pktchain
);
702 if (rc
== -ELINKCONG
) {
704 rc
= tipc_wait_for_sndmsg(sock
, &timeo
);
708 __skb_queue_purge(pktchain
);
709 if (rc
== -EMSGSIZE
) {
710 msg
->msg_iter
= save
;
719 * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
720 * @arrvq: queue with arriving messages, to be cloned after destination lookup
721 * @inputq: queue with cloned messages, delivered to socket after dest lookup
723 * Multi-threaded: parallel calls with reference to same queues may occur
725 void tipc_sk_mcast_rcv(struct net
*net
, struct sk_buff_head
*arrvq
,
726 struct sk_buff_head
*inputq
)
728 struct tipc_msg
*msg
;
729 struct tipc_plist dports
;
731 u32 scope
= TIPC_CLUSTER_SCOPE
;
732 struct sk_buff_head tmpq
;
734 struct sk_buff
*skb
, *_skb
;
736 __skb_queue_head_init(&tmpq
);
737 tipc_plist_init(&dports
);
739 skb
= tipc_skb_peek(arrvq
, &inputq
->lock
);
740 for (; skb
; skb
= tipc_skb_peek(arrvq
, &inputq
->lock
)) {
742 hsz
= skb_headroom(skb
) + msg_hdr_sz(msg
);
744 if (in_own_node(net
, msg_orignode(msg
)))
745 scope
= TIPC_NODE_SCOPE
;
747 /* Create destination port list and message clones: */
748 tipc_nametbl_mc_translate(net
,
749 msg_nametype(msg
), msg_namelower(msg
),
750 msg_nameupper(msg
), scope
, &dports
);
751 portid
= tipc_plist_pop(&dports
);
752 for (; portid
; portid
= tipc_plist_pop(&dports
)) {
753 _skb
= __pskb_copy(skb
, hsz
, GFP_ATOMIC
);
755 msg_set_destport(buf_msg(_skb
), portid
);
756 __skb_queue_tail(&tmpq
, _skb
);
759 pr_warn("Failed to clone mcast rcv buffer\n");
761 /* Append to inputq if not already done by other thread */
762 spin_lock_bh(&inputq
->lock
);
763 if (skb_peek(arrvq
) == skb
) {
764 skb_queue_splice_tail_init(&tmpq
, inputq
);
765 kfree_skb(__skb_dequeue(arrvq
));
767 spin_unlock_bh(&inputq
->lock
);
768 __skb_queue_purge(&tmpq
);
771 tipc_sk_rcv(net
, inputq
);
775 * tipc_sk_proto_rcv - receive a connection mng protocol message
776 * @tsk: receiving socket
777 * @skb: pointer to message buffer.
779 static void tipc_sk_proto_rcv(struct tipc_sock
*tsk
, struct sk_buff
*skb
)
781 struct sock
*sk
= &tsk
->sk
;
782 struct tipc_msg
*hdr
= buf_msg(skb
);
783 int mtyp
= msg_type(hdr
);
786 /* Ignore if connection cannot be validated: */
787 if (!tsk_peer_msg(tsk
, hdr
))
790 tsk
->probing_state
= TIPC_CONN_OK
;
792 if (mtyp
== CONN_PROBE
) {
793 msg_set_type(hdr
, CONN_PROBE_REPLY
);
794 tipc_sk_respond(sk
, skb
, TIPC_OK
);
796 } else if (mtyp
== CONN_ACK
) {
797 conn_cong
= tsk_conn_cong(tsk
);
798 tsk
->sent_unacked
-= msg_msgcnt(hdr
);
800 sk
->sk_write_space(sk
);
801 } else if (mtyp
!= CONN_PROBE_REPLY
) {
802 pr_warn("Received unknown CONN_PROTO msg\n");
808 static int tipc_wait_for_sndmsg(struct socket
*sock
, long *timeo_p
)
810 struct sock
*sk
= sock
->sk
;
811 struct tipc_sock
*tsk
= tipc_sk(sk
);
816 int err
= sock_error(sk
);
819 if (sock
->state
== SS_DISCONNECTING
)
823 if (signal_pending(current
))
824 return sock_intr_errno(*timeo_p
);
826 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
827 done
= sk_wait_event(sk
, timeo_p
, !tsk
->link_cong
);
828 finish_wait(sk_sleep(sk
), &wait
);
834 * tipc_sendmsg - send message in connectionless manner
835 * @sock: socket structure
836 * @m: message to send
837 * @dsz: amount of user data to be sent
839 * Message must have an destination specified explicitly.
840 * Used for SOCK_RDM and SOCK_DGRAM messages,
841 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
842 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
844 * Returns the number of bytes sent on success, or errno otherwise
846 static int tipc_sendmsg(struct socket
*sock
,
847 struct msghdr
*m
, size_t dsz
)
849 struct sock
*sk
= sock
->sk
;
853 ret
= __tipc_sendmsg(sock
, m
, dsz
);
859 static int __tipc_sendmsg(struct socket
*sock
, struct msghdr
*m
, size_t dsz
)
861 DECLARE_SOCKADDR(struct sockaddr_tipc
*, dest
, m
->msg_name
);
862 struct sock
*sk
= sock
->sk
;
863 struct tipc_sock
*tsk
= tipc_sk(sk
);
864 struct net
*net
= sock_net(sk
);
865 struct tipc_msg
*mhdr
= &tsk
->phdr
;
867 struct sk_buff_head
*pktchain
= &sk
->sk_write_queue
;
869 struct tipc_name_seq
*seq
;
870 struct iov_iter save
;
875 if (dsz
> TIPC_MAX_USER_MSG_SIZE
)
877 if (unlikely(!dest
)) {
878 if (tsk
->connected
&& sock
->state
== SS_READY
)
881 return -EDESTADDRREQ
;
882 } else if (unlikely(m
->msg_namelen
< sizeof(*dest
)) ||
883 dest
->family
!= AF_TIPC
) {
886 if (unlikely(sock
->state
!= SS_READY
)) {
887 if (sock
->state
== SS_LISTENING
)
889 if (sock
->state
!= SS_UNCONNECTED
)
893 if (dest
->addrtype
== TIPC_ADDR_NAME
) {
894 tsk
->conn_type
= dest
->addr
.name
.name
.type
;
895 tsk
->conn_instance
= dest
->addr
.name
.name
.instance
;
898 seq
= &dest
->addr
.nameseq
;
899 timeo
= sock_sndtimeo(sk
, m
->msg_flags
& MSG_DONTWAIT
);
901 if (dest
->addrtype
== TIPC_ADDR_MCAST
) {
902 return tipc_sendmcast(sock
, seq
, m
, dsz
, timeo
);
903 } else if (dest
->addrtype
== TIPC_ADDR_NAME
) {
904 u32 type
= dest
->addr
.name
.name
.type
;
905 u32 inst
= dest
->addr
.name
.name
.instance
;
906 u32 domain
= dest
->addr
.name
.domain
;
909 msg_set_type(mhdr
, TIPC_NAMED_MSG
);
910 msg_set_hdr_sz(mhdr
, NAMED_H_SIZE
);
911 msg_set_nametype(mhdr
, type
);
912 msg_set_nameinst(mhdr
, inst
);
913 msg_set_lookup_scope(mhdr
, tipc_addr_scope(domain
));
914 dport
= tipc_nametbl_translate(net
, type
, inst
, &dnode
);
915 msg_set_destnode(mhdr
, dnode
);
916 msg_set_destport(mhdr
, dport
);
917 if (unlikely(!dport
&& !dnode
))
918 return -EHOSTUNREACH
;
919 } else if (dest
->addrtype
== TIPC_ADDR_ID
) {
920 dnode
= dest
->addr
.id
.node
;
921 msg_set_type(mhdr
, TIPC_DIRECT_MSG
);
922 msg_set_lookup_scope(mhdr
, 0);
923 msg_set_destnode(mhdr
, dnode
);
924 msg_set_destport(mhdr
, dest
->addr
.id
.ref
);
925 msg_set_hdr_sz(mhdr
, BASIC_H_SIZE
);
930 mtu
= tipc_node_get_mtu(net
, dnode
, tsk
->portid
);
931 rc
= tipc_msg_build(mhdr
, m
, 0, dsz
, mtu
, pktchain
);
936 skb
= skb_peek(pktchain
);
937 TIPC_SKB_CB(skb
)->wakeup_pending
= tsk
->link_cong
;
938 rc
= tipc_node_xmit(net
, pktchain
, dnode
, tsk
->portid
);
940 if (sock
->state
!= SS_READY
)
941 sock
->state
= SS_CONNECTING
;
944 if (rc
== -ELINKCONG
) {
946 rc
= tipc_wait_for_sndmsg(sock
, &timeo
);
950 __skb_queue_purge(pktchain
);
951 if (rc
== -EMSGSIZE
) {
961 static int tipc_wait_for_sndpkt(struct socket
*sock
, long *timeo_p
)
963 struct sock
*sk
= sock
->sk
;
964 struct tipc_sock
*tsk
= tipc_sk(sk
);
969 int err
= sock_error(sk
);
972 if (sock
->state
== SS_DISCONNECTING
)
974 else if (sock
->state
!= SS_CONNECTED
)
978 if (signal_pending(current
))
979 return sock_intr_errno(*timeo_p
);
981 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
982 done
= sk_wait_event(sk
, timeo_p
,
984 !tsk_conn_cong(tsk
)) ||
986 finish_wait(sk_sleep(sk
), &wait
);
992 * tipc_send_stream - send stream-oriented data
993 * @sock: socket structure
995 * @dsz: total length of data to be transmitted
997 * Used for SOCK_STREAM data.
999 * Returns the number of bytes sent on success (or partial success),
1000 * or errno if no data sent
1002 static int tipc_send_stream(struct socket
*sock
, struct msghdr
*m
, size_t dsz
)
1004 struct sock
*sk
= sock
->sk
;
1008 ret
= __tipc_send_stream(sock
, m
, dsz
);
1014 static int __tipc_send_stream(struct socket
*sock
, struct msghdr
*m
, size_t dsz
)
1016 struct sock
*sk
= sock
->sk
;
1017 struct net
*net
= sock_net(sk
);
1018 struct tipc_sock
*tsk
= tipc_sk(sk
);
1019 struct tipc_msg
*mhdr
= &tsk
->phdr
;
1020 struct sk_buff_head
*pktchain
= &sk
->sk_write_queue
;
1021 DECLARE_SOCKADDR(struct sockaddr_tipc
*, dest
, m
->msg_name
);
1022 u32 portid
= tsk
->portid
;
1026 uint mtu
, send
, sent
= 0;
1027 struct iov_iter save
;
1029 /* Handle implied connection establishment */
1030 if (unlikely(dest
)) {
1031 rc
= __tipc_sendmsg(sock
, m
, dsz
);
1032 if (dsz
&& (dsz
== rc
))
1033 tsk
->sent_unacked
= 1;
1036 if (dsz
> (uint
)INT_MAX
)
1039 if (unlikely(sock
->state
!= SS_CONNECTED
)) {
1040 if (sock
->state
== SS_DISCONNECTING
)
1046 timeo
= sock_sndtimeo(sk
, m
->msg_flags
& MSG_DONTWAIT
);
1047 dnode
= tsk_peer_node(tsk
);
1052 send
= min_t(uint
, dsz
- sent
, TIPC_MAX_USER_MSG_SIZE
);
1053 rc
= tipc_msg_build(mhdr
, m
, sent
, send
, mtu
, pktchain
);
1054 if (unlikely(rc
< 0))
1057 if (likely(!tsk_conn_cong(tsk
))) {
1058 rc
= tipc_node_xmit(net
, pktchain
, dnode
, portid
);
1060 tsk
->sent_unacked
++;
1066 if (rc
== -EMSGSIZE
) {
1067 __skb_queue_purge(pktchain
);
1068 tsk
->max_pkt
= tipc_node_get_mtu(net
, dnode
,
1073 if (rc
!= -ELINKCONG
)
1078 rc
= tipc_wait_for_sndpkt(sock
, &timeo
);
1081 __skb_queue_purge(pktchain
);
1082 return sent
? sent
: rc
;
1086 * tipc_send_packet - send a connection-oriented message
1087 * @sock: socket structure
1088 * @m: message to send
1089 * @dsz: length of data to be transmitted
1091 * Used for SOCK_SEQPACKET messages.
1093 * Returns the number of bytes sent on success, or errno otherwise
1095 static int tipc_send_packet(struct socket
*sock
, struct msghdr
*m
, size_t dsz
)
1097 if (dsz
> TIPC_MAX_USER_MSG_SIZE
)
1100 return tipc_send_stream(sock
, m
, dsz
);
1103 /* tipc_sk_finish_conn - complete the setup of a connection
1105 static void tipc_sk_finish_conn(struct tipc_sock
*tsk
, u32 peer_port
,
1108 struct sock
*sk
= &tsk
->sk
;
1109 struct net
*net
= sock_net(sk
);
1110 struct tipc_msg
*msg
= &tsk
->phdr
;
1112 msg_set_destnode(msg
, peer_node
);
1113 msg_set_destport(msg
, peer_port
);
1114 msg_set_type(msg
, TIPC_CONN_MSG
);
1115 msg_set_lookup_scope(msg
, 0);
1116 msg_set_hdr_sz(msg
, SHORT_H_SIZE
);
1118 tsk
->probing_intv
= CONN_PROBING_INTERVAL
;
1119 tsk
->probing_state
= TIPC_CONN_OK
;
1121 sk_reset_timer(sk
, &sk
->sk_timer
, jiffies
+ tsk
->probing_intv
);
1122 tipc_node_add_conn(net
, peer_node
, tsk
->portid
, peer_port
);
1123 tsk
->max_pkt
= tipc_node_get_mtu(net
, peer_node
, tsk
->portid
);
1127 * set_orig_addr - capture sender's address for received message
1128 * @m: descriptor for message info
1129 * @msg: received message header
1131 * Note: Address is not captured if not requested by receiver.
1133 static void set_orig_addr(struct msghdr
*m
, struct tipc_msg
*msg
)
1135 DECLARE_SOCKADDR(struct sockaddr_tipc
*, addr
, m
->msg_name
);
1138 addr
->family
= AF_TIPC
;
1139 addr
->addrtype
= TIPC_ADDR_ID
;
1140 memset(&addr
->addr
, 0, sizeof(addr
->addr
));
1141 addr
->addr
.id
.ref
= msg_origport(msg
);
1142 addr
->addr
.id
.node
= msg_orignode(msg
);
1143 addr
->addr
.name
.domain
= 0; /* could leave uninitialized */
1144 addr
->scope
= 0; /* could leave uninitialized */
1145 m
->msg_namelen
= sizeof(struct sockaddr_tipc
);
1150 * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
1151 * @m: descriptor for message info
1152 * @msg: received message header
1153 * @tsk: TIPC port associated with message
1155 * Note: Ancillary data is not captured if not requested by receiver.
1157 * Returns 0 if successful, otherwise errno
1159 static int tipc_sk_anc_data_recv(struct msghdr
*m
, struct tipc_msg
*msg
,
1160 struct tipc_sock
*tsk
)
1168 if (likely(m
->msg_controllen
== 0))
1171 /* Optionally capture errored message object(s) */
1172 err
= msg
? msg_errcode(msg
) : 0;
1173 if (unlikely(err
)) {
1175 anc_data
[1] = msg_data_sz(msg
);
1176 res
= put_cmsg(m
, SOL_TIPC
, TIPC_ERRINFO
, 8, anc_data
);
1180 res
= put_cmsg(m
, SOL_TIPC
, TIPC_RETDATA
, anc_data
[1],
1187 /* Optionally capture message destination object */
1188 dest_type
= msg
? msg_type(msg
) : TIPC_DIRECT_MSG
;
1189 switch (dest_type
) {
1190 case TIPC_NAMED_MSG
:
1192 anc_data
[0] = msg_nametype(msg
);
1193 anc_data
[1] = msg_namelower(msg
);
1194 anc_data
[2] = msg_namelower(msg
);
1196 case TIPC_MCAST_MSG
:
1198 anc_data
[0] = msg_nametype(msg
);
1199 anc_data
[1] = msg_namelower(msg
);
1200 anc_data
[2] = msg_nameupper(msg
);
1203 has_name
= (tsk
->conn_type
!= 0);
1204 anc_data
[0] = tsk
->conn_type
;
1205 anc_data
[1] = tsk
->conn_instance
;
1206 anc_data
[2] = tsk
->conn_instance
;
1212 res
= put_cmsg(m
, SOL_TIPC
, TIPC_DESTNAME
, 12, anc_data
);
1220 static void tipc_sk_send_ack(struct tipc_sock
*tsk
, uint ack
)
1222 struct net
*net
= sock_net(&tsk
->sk
);
1223 struct sk_buff
*skb
= NULL
;
1224 struct tipc_msg
*msg
;
1225 u32 peer_port
= tsk_peer_port(tsk
);
1226 u32 dnode
= tsk_peer_node(tsk
);
1228 if (!tsk
->connected
)
1230 skb
= tipc_msg_create(CONN_MANAGER
, CONN_ACK
, INT_H_SIZE
, 0,
1231 dnode
, tsk_own_node(tsk
), peer_port
,
1232 tsk
->portid
, TIPC_OK
);
1236 msg_set_msgcnt(msg
, ack
);
1237 tipc_node_xmit_skb(net
, skb
, dnode
, msg_link_selector(msg
));
1240 static int tipc_wait_for_rcvmsg(struct socket
*sock
, long *timeop
)
1242 struct sock
*sk
= sock
->sk
;
1244 long timeo
= *timeop
;
1248 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
1249 if (timeo
&& skb_queue_empty(&sk
->sk_receive_queue
)) {
1250 if (sock
->state
== SS_DISCONNECTING
) {
1255 timeo
= schedule_timeout(timeo
);
1259 if (!skb_queue_empty(&sk
->sk_receive_queue
))
1264 err
= sock_intr_errno(timeo
);
1265 if (signal_pending(current
))
1268 finish_wait(sk_sleep(sk
), &wait
);
1274 * tipc_recvmsg - receive packet-oriented message
1275 * @m: descriptor for message info
1276 * @buf_len: total size of user buffer area
1277 * @flags: receive flags
1279 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1280 * If the complete message doesn't fit in user area, truncate it.
1282 * Returns size of returned message data, errno otherwise
1284 static int tipc_recvmsg(struct socket
*sock
, struct msghdr
*m
, size_t buf_len
,
1287 struct sock
*sk
= sock
->sk
;
1288 struct tipc_sock
*tsk
= tipc_sk(sk
);
1289 struct sk_buff
*buf
;
1290 struct tipc_msg
*msg
;
1296 /* Catch invalid receive requests */
1297 if (unlikely(!buf_len
))
1302 if (unlikely(sock
->state
== SS_UNCONNECTED
)) {
1307 timeo
= sock_rcvtimeo(sk
, flags
& MSG_DONTWAIT
);
1310 /* Look for a message in receive queue; wait if necessary */
1311 res
= tipc_wait_for_rcvmsg(sock
, &timeo
);
1315 /* Look at first message in receive queue */
1316 buf
= skb_peek(&sk
->sk_receive_queue
);
1318 sz
= msg_data_sz(msg
);
1319 err
= msg_errcode(msg
);
1321 /* Discard an empty non-errored message & try again */
1322 if ((!sz
) && (!err
)) {
1323 tsk_advance_rx_queue(sk
);
1327 /* Capture sender's address (optional) */
1328 set_orig_addr(m
, msg
);
1330 /* Capture ancillary data (optional) */
1331 res
= tipc_sk_anc_data_recv(m
, msg
, tsk
);
1335 /* Capture message data (if valid) & compute return value (always) */
1337 if (unlikely(buf_len
< sz
)) {
1339 m
->msg_flags
|= MSG_TRUNC
;
1341 res
= skb_copy_datagram_msg(buf
, msg_hdr_sz(msg
), m
, sz
);
1346 if ((sock
->state
== SS_READY
) ||
1347 ((err
== TIPC_CONN_SHUTDOWN
) || m
->msg_control
))
1353 /* Consume received message (optional) */
1354 if (likely(!(flags
& MSG_PEEK
))) {
1355 if ((sock
->state
!= SS_READY
) &&
1356 (++tsk
->rcv_unacked
>= TIPC_CONNACK_INTV
)) {
1357 tipc_sk_send_ack(tsk
, tsk
->rcv_unacked
);
1358 tsk
->rcv_unacked
= 0;
1360 tsk_advance_rx_queue(sk
);
1368 * tipc_recv_stream - receive stream-oriented data
1369 * @m: descriptor for message info
1370 * @buf_len: total size of user buffer area
1371 * @flags: receive flags
1373 * Used for SOCK_STREAM messages only. If not enough data is available
1374 * will optionally wait for more; never truncates data.
1376 * Returns size of returned message data, errno otherwise
1378 static int tipc_recv_stream(struct socket
*sock
, struct msghdr
*m
,
1379 size_t buf_len
, int flags
)
1381 struct sock
*sk
= sock
->sk
;
1382 struct tipc_sock
*tsk
= tipc_sk(sk
);
1383 struct sk_buff
*buf
;
1384 struct tipc_msg
*msg
;
1387 int sz_to_copy
, target
, needed
;
1392 /* Catch invalid receive attempts */
1393 if (unlikely(!buf_len
))
1398 if (unlikely(sock
->state
== SS_UNCONNECTED
)) {
1403 target
= sock_rcvlowat(sk
, flags
& MSG_WAITALL
, buf_len
);
1404 timeo
= sock_rcvtimeo(sk
, flags
& MSG_DONTWAIT
);
1407 /* Look for a message in receive queue; wait if necessary */
1408 res
= tipc_wait_for_rcvmsg(sock
, &timeo
);
1412 /* Look at first message in receive queue */
1413 buf
= skb_peek(&sk
->sk_receive_queue
);
1415 sz
= msg_data_sz(msg
);
1416 err
= msg_errcode(msg
);
1418 /* Discard an empty non-errored message & try again */
1419 if ((!sz
) && (!err
)) {
1420 tsk_advance_rx_queue(sk
);
1424 /* Optionally capture sender's address & ancillary data of first msg */
1425 if (sz_copied
== 0) {
1426 set_orig_addr(m
, msg
);
1427 res
= tipc_sk_anc_data_recv(m
, msg
, tsk
);
1432 /* Capture message data (if valid) & compute return value (always) */
1434 u32 offset
= (u32
)(unsigned long)(TIPC_SKB_CB(buf
)->handle
);
1437 needed
= (buf_len
- sz_copied
);
1438 sz_to_copy
= (sz
<= needed
) ? sz
: needed
;
1440 res
= skb_copy_datagram_msg(buf
, msg_hdr_sz(msg
) + offset
,
1445 sz_copied
+= sz_to_copy
;
1447 if (sz_to_copy
< sz
) {
1448 if (!(flags
& MSG_PEEK
))
1449 TIPC_SKB_CB(buf
)->handle
=
1450 (void *)(unsigned long)(offset
+ sz_to_copy
);
1455 goto exit
; /* can't add error msg to valid data */
1457 if ((err
== TIPC_CONN_SHUTDOWN
) || m
->msg_control
)
1463 /* Consume received message (optional) */
1464 if (likely(!(flags
& MSG_PEEK
))) {
1465 if (unlikely(++tsk
->rcv_unacked
>= TIPC_CONNACK_INTV
)) {
1466 tipc_sk_send_ack(tsk
, tsk
->rcv_unacked
);
1467 tsk
->rcv_unacked
= 0;
1469 tsk_advance_rx_queue(sk
);
1472 /* Loop around if more data is required */
1473 if ((sz_copied
< buf_len
) && /* didn't get all requested data */
1474 (!skb_queue_empty(&sk
->sk_receive_queue
) ||
1475 (sz_copied
< target
)) && /* and more is ready or required */
1476 (!(flags
& MSG_PEEK
)) && /* and aren't just peeking at data */
1477 (!err
)) /* and haven't reached a FIN */
1482 return sz_copied
? sz_copied
: res
;
1486 * tipc_write_space - wake up thread if port congestion is released
1489 static void tipc_write_space(struct sock
*sk
)
1491 struct socket_wq
*wq
;
1494 wq
= rcu_dereference(sk
->sk_wq
);
1495 if (wq_has_sleeper(wq
))
1496 wake_up_interruptible_sync_poll(&wq
->wait
, POLLOUT
|
1497 POLLWRNORM
| POLLWRBAND
);
1502 * tipc_data_ready - wake up threads to indicate messages have been received
1504 * @len: the length of messages
1506 static void tipc_data_ready(struct sock
*sk
)
1508 struct socket_wq
*wq
;
1511 wq
= rcu_dereference(sk
->sk_wq
);
1512 if (wq_has_sleeper(wq
))
1513 wake_up_interruptible_sync_poll(&wq
->wait
, POLLIN
|
1514 POLLRDNORM
| POLLRDBAND
);
1519 * filter_connect - Handle all incoming messages for a connection-based socket
1521 * @skb: pointer to message buffer. Set to NULL if buffer is consumed
1523 * Returns true if everything ok, false otherwise
1525 static bool filter_connect(struct tipc_sock
*tsk
, struct sk_buff
*skb
)
1527 struct sock
*sk
= &tsk
->sk
;
1528 struct net
*net
= sock_net(sk
);
1529 struct socket
*sock
= sk
->sk_socket
;
1530 struct tipc_msg
*hdr
= buf_msg(skb
);
1532 if (unlikely(msg_mcast(hdr
)))
1535 switch ((int)sock
->state
) {
1538 /* Accept only connection-based messages sent by peer */
1539 if (unlikely(!tsk_peer_msg(tsk
, hdr
)))
1542 if (unlikely(msg_errcode(hdr
))) {
1543 sock
->state
= SS_DISCONNECTING
;
1545 /* Let timer expire on it's own */
1546 tipc_node_remove_conn(net
, tsk_peer_node(tsk
),
1553 /* Accept only ACK or NACK message */
1554 if (unlikely(!msg_connected(hdr
)))
1557 if (unlikely(msg_errcode(hdr
))) {
1558 sock
->state
= SS_DISCONNECTING
;
1559 sk
->sk_err
= ECONNREFUSED
;
1563 if (unlikely(!msg_isdata(hdr
))) {
1564 sock
->state
= SS_DISCONNECTING
;
1565 sk
->sk_err
= EINVAL
;
1569 tipc_sk_finish_conn(tsk
, msg_origport(hdr
), msg_orignode(hdr
));
1570 msg_set_importance(&tsk
->phdr
, msg_importance(hdr
));
1571 sock
->state
= SS_CONNECTED
;
1573 /* If 'ACK+' message, add to socket receive queue */
1574 if (msg_data_sz(hdr
))
1577 /* If empty 'ACK-' message, wake up sleeping connect() */
1578 if (waitqueue_active(sk_sleep(sk
)))
1579 wake_up_interruptible(sk_sleep(sk
));
1581 /* 'ACK-' message is neither accepted nor rejected: */
1582 msg_set_dest_droppable(hdr
, 1);
1586 case SS_UNCONNECTED
:
1588 /* Accept only SYN message */
1589 if (!msg_connected(hdr
) && !(msg_errcode(hdr
)))
1592 case SS_DISCONNECTING
:
1595 pr_err("Unknown socket state %u\n", sock
->state
);
1601 * rcvbuf_limit - get proper overload limit of socket receive queue
1605 * For all connection oriented messages, irrespective of importance,
1606 * the default overload value (i.e. 67MB) is set as limit.
1608 * For all connectionless messages, by default new queue limits are
1611 * TIPC_LOW_IMPORTANCE (4 MB)
1612 * TIPC_MEDIUM_IMPORTANCE (8 MB)
1613 * TIPC_HIGH_IMPORTANCE (16 MB)
1614 * TIPC_CRITICAL_IMPORTANCE (32 MB)
1616 * Returns overload limit according to corresponding message importance
1618 static unsigned int rcvbuf_limit(struct sock
*sk
, struct sk_buff
*buf
)
1620 struct tipc_msg
*msg
= buf_msg(buf
);
1622 if (msg_connected(msg
))
1623 return sysctl_tipc_rmem
[2];
1625 return sk
->sk_rcvbuf
>> TIPC_CRITICAL_IMPORTANCE
<<
1626 msg_importance(msg
);
1630 * filter_rcv - validate incoming message
1632 * @skb: pointer to message.
1634 * Enqueues message on receive queue if acceptable; optionally handles
1635 * disconnect indication for a connected socket.
1637 * Called with socket lock already taken
1639 * Returns true if message was added to socket receive queue, otherwise false
1641 static bool filter_rcv(struct sock
*sk
, struct sk_buff
*skb
)
1643 struct socket
*sock
= sk
->sk_socket
;
1644 struct tipc_sock
*tsk
= tipc_sk(sk
);
1645 struct tipc_msg
*hdr
= buf_msg(skb
);
1646 unsigned int limit
= rcvbuf_limit(sk
, skb
);
1648 int usr
= msg_user(hdr
);
1650 if (unlikely(msg_user(hdr
) == CONN_MANAGER
)) {
1651 tipc_sk_proto_rcv(tsk
, skb
);
1655 if (unlikely(usr
== SOCK_WAKEUP
)) {
1658 sk
->sk_write_space(sk
);
1662 /* Drop if illegal message type */
1663 if (unlikely(msg_type(hdr
) > TIPC_DIRECT_MSG
)) {
1668 /* Reject if wrong message type for current socket state */
1669 if (unlikely(sock
->state
== SS_READY
)) {
1670 if (msg_connected(hdr
)) {
1671 err
= TIPC_ERR_NO_PORT
;
1674 } else if (unlikely(!filter_connect(tsk
, skb
))) {
1675 err
= TIPC_ERR_NO_PORT
;
1679 /* Reject message if there isn't room to queue it */
1680 if (unlikely(sk_rmem_alloc_get(sk
) + skb
->truesize
>= limit
)) {
1681 err
= TIPC_ERR_OVERLOAD
;
1685 /* Enqueue message */
1686 TIPC_SKB_CB(skb
)->handle
= NULL
;
1687 __skb_queue_tail(&sk
->sk_receive_queue
, skb
);
1688 skb_set_owner_r(skb
, sk
);
1690 sk
->sk_data_ready(sk
);
1694 tipc_sk_respond(sk
, skb
, err
);
1699 * tipc_backlog_rcv - handle incoming message from backlog queue
1703 * Caller must hold socket lock
1707 static int tipc_backlog_rcv(struct sock
*sk
, struct sk_buff
*skb
)
1709 unsigned int truesize
= skb
->truesize
;
1711 if (likely(filter_rcv(sk
, skb
)))
1712 atomic_add(truesize
, &tipc_sk(sk
)->dupl_rcvcnt
);
1717 * tipc_sk_enqueue - extract all buffers with destination 'dport' from
1718 * inputq and try adding them to socket or backlog queue
1719 * @inputq: list of incoming buffers with potentially different destinations
1720 * @sk: socket where the buffers should be enqueued
1721 * @dport: port number for the socket
1723 * Caller must hold socket lock
1725 static void tipc_sk_enqueue(struct sk_buff_head
*inputq
, struct sock
*sk
,
1730 struct sk_buff
*skb
;
1731 unsigned long time_limit
= jiffies
+ 2;
1733 while (skb_queue_len(inputq
)) {
1734 if (unlikely(time_after_eq(jiffies
, time_limit
)))
1737 skb
= tipc_skb_dequeue(inputq
, dport
);
1741 /* Add message directly to receive queue if possible */
1742 if (!sock_owned_by_user(sk
)) {
1743 filter_rcv(sk
, skb
);
1747 /* Try backlog, compensating for double-counted bytes */
1748 dcnt
= &tipc_sk(sk
)->dupl_rcvcnt
;
1749 if (sk
->sk_backlog
.len
)
1750 atomic_set(dcnt
, 0);
1751 lim
= rcvbuf_limit(sk
, skb
) + atomic_read(dcnt
);
1752 if (likely(!sk_add_backlog(sk
, skb
, lim
)))
1755 /* Overload => reject message back to sender */
1756 tipc_sk_respond(sk
, skb
, TIPC_ERR_OVERLOAD
);
1762 * tipc_sk_rcv - handle a chain of incoming buffers
1763 * @inputq: buffer list containing the buffers
1764 * Consumes all buffers in list until inputq is empty
1765 * Note: may be called in multiple threads referring to the same queue
1767 void tipc_sk_rcv(struct net
*net
, struct sk_buff_head
*inputq
)
1769 u32 dnode
, dport
= 0;
1771 struct tipc_sock
*tsk
;
1773 struct sk_buff
*skb
;
1775 while (skb_queue_len(inputq
)) {
1776 dport
= tipc_skb_peek_port(inputq
, dport
);
1777 tsk
= tipc_sk_lookup(net
, dport
);
1781 if (likely(spin_trylock_bh(&sk
->sk_lock
.slock
))) {
1782 tipc_sk_enqueue(inputq
, sk
, dport
);
1783 spin_unlock_bh(&sk
->sk_lock
.slock
);
1789 /* No destination socket => dequeue skb if still there */
1790 skb
= tipc_skb_dequeue(inputq
, dport
);
1794 /* Try secondary lookup if unresolved named message */
1795 err
= TIPC_ERR_NO_PORT
;
1796 if (tipc_msg_lookup_dest(net
, skb
, &err
))
1799 /* Prepare for message rejection */
1800 if (!tipc_msg_reverse(tipc_own_addr(net
), &skb
, err
))
1803 dnode
= msg_destnode(buf_msg(skb
));
1804 tipc_node_xmit_skb(net
, skb
, dnode
, dport
);
1808 static int tipc_wait_for_connect(struct socket
*sock
, long *timeo_p
)
1810 struct sock
*sk
= sock
->sk
;
1815 int err
= sock_error(sk
);
1820 if (signal_pending(current
))
1821 return sock_intr_errno(*timeo_p
);
1823 prepare_to_wait(sk_sleep(sk
), &wait
, TASK_INTERRUPTIBLE
);
1824 done
= sk_wait_event(sk
, timeo_p
, sock
->state
!= SS_CONNECTING
);
1825 finish_wait(sk_sleep(sk
), &wait
);
1831 * tipc_connect - establish a connection to another TIPC port
1832 * @sock: socket structure
1833 * @dest: socket address for destination port
1834 * @destlen: size of socket address data structure
1835 * @flags: file-related flags associated with socket
1837 * Returns 0 on success, errno otherwise
1839 static int tipc_connect(struct socket
*sock
, struct sockaddr
*dest
,
1840 int destlen
, int flags
)
1842 struct sock
*sk
= sock
->sk
;
1843 struct tipc_sock
*tsk
= tipc_sk(sk
);
1844 struct sockaddr_tipc
*dst
= (struct sockaddr_tipc
*)dest
;
1845 struct msghdr m
= {NULL
,};
1846 long timeout
= (flags
& O_NONBLOCK
) ? 0 : tsk
->conn_timeout
;
1847 socket_state previous
;
1852 /* DGRAM/RDM connect(), just save the destaddr */
1853 if (sock
->state
== SS_READY
) {
1854 if (dst
->family
== AF_UNSPEC
) {
1855 memset(&tsk
->remote
, 0, sizeof(struct sockaddr_tipc
));
1857 } else if (destlen
!= sizeof(struct sockaddr_tipc
)) {
1860 memcpy(&tsk
->remote
, dest
, destlen
);
1867 * Reject connection attempt using multicast address
1869 * Note: send_msg() validates the rest of the address fields,
1870 * so there's no need to do it here
1872 if (dst
->addrtype
== TIPC_ADDR_MCAST
) {
1877 previous
= sock
->state
;
1878 switch (sock
->state
) {
1879 case SS_UNCONNECTED
:
1880 /* Send a 'SYN-' to destination */
1882 m
.msg_namelen
= destlen
;
1884 /* If connect is in non-blocking case, set MSG_DONTWAIT to
1885 * indicate send_msg() is never blocked.
1888 m
.msg_flags
= MSG_DONTWAIT
;
1890 res
= __tipc_sendmsg(sock
, &m
, 0);
1891 if ((res
< 0) && (res
!= -EWOULDBLOCK
))
1894 /* Just entered SS_CONNECTING state; the only
1895 * difference is that return value in non-blocking
1896 * case is EINPROGRESS, rather than EALREADY.
1900 if (previous
== SS_CONNECTING
)
1904 timeout
= msecs_to_jiffies(timeout
);
1905 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
1906 res
= tipc_wait_for_connect(sock
, &timeout
);
1921 * tipc_listen - allow socket to listen for incoming connections
1922 * @sock: socket structure
1925 * Returns 0 on success, errno otherwise
1927 static int tipc_listen(struct socket
*sock
, int len
)
1929 struct sock
*sk
= sock
->sk
;
1934 if (sock
->state
!= SS_UNCONNECTED
)
1937 sock
->state
= SS_LISTENING
;
1945 static int tipc_wait_for_accept(struct socket
*sock
, long timeo
)
1947 struct sock
*sk
= sock
->sk
;
1951 /* True wake-one mechanism for incoming connections: only
1952 * one process gets woken up, not the 'whole herd'.
1953 * Since we do not 'race & poll' for established sockets
1954 * anymore, the common case will execute the loop only once.
1957 prepare_to_wait_exclusive(sk_sleep(sk
), &wait
,
1958 TASK_INTERRUPTIBLE
);
1959 if (timeo
&& skb_queue_empty(&sk
->sk_receive_queue
)) {
1961 timeo
= schedule_timeout(timeo
);
1965 if (!skb_queue_empty(&sk
->sk_receive_queue
))
1968 if (sock
->state
!= SS_LISTENING
)
1973 err
= sock_intr_errno(timeo
);
1974 if (signal_pending(current
))
1977 finish_wait(sk_sleep(sk
), &wait
);
1982 * tipc_accept - wait for connection request
1983 * @sock: listening socket
1984 * @newsock: new socket that is to be connected
1985 * @flags: file-related flags associated with socket
1987 * Returns 0 on success, errno otherwise
1989 static int tipc_accept(struct socket
*sock
, struct socket
*new_sock
, int flags
)
1991 struct sock
*new_sk
, *sk
= sock
->sk
;
1992 struct sk_buff
*buf
;
1993 struct tipc_sock
*new_tsock
;
1994 struct tipc_msg
*msg
;
2000 if (sock
->state
!= SS_LISTENING
) {
2004 timeo
= sock_rcvtimeo(sk
, flags
& O_NONBLOCK
);
2005 res
= tipc_wait_for_accept(sock
, timeo
);
2009 buf
= skb_peek(&sk
->sk_receive_queue
);
2011 res
= tipc_sk_create(sock_net(sock
->sk
), new_sock
, 0, 1);
2014 security_sk_clone(sock
->sk
, new_sock
->sk
);
2016 new_sk
= new_sock
->sk
;
2017 new_tsock
= tipc_sk(new_sk
);
2020 /* we lock on new_sk; but lockdep sees the lock on sk */
2021 lock_sock_nested(new_sk
, SINGLE_DEPTH_NESTING
);
2024 * Reject any stray messages received by new socket
2025 * before the socket lock was taken (very, very unlikely)
2027 tsk_rej_rx_queue(new_sk
);
2029 /* Connect new socket to it's peer */
2030 tipc_sk_finish_conn(new_tsock
, msg_origport(msg
), msg_orignode(msg
));
2031 new_sock
->state
= SS_CONNECTED
;
2033 tsk_set_importance(new_tsock
, msg_importance(msg
));
2034 if (msg_named(msg
)) {
2035 new_tsock
->conn_type
= msg_nametype(msg
);
2036 new_tsock
->conn_instance
= msg_nameinst(msg
);
2040 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2041 * Respond to 'SYN+' by queuing it on new socket.
2043 if (!msg_data_sz(msg
)) {
2044 struct msghdr m
= {NULL
,};
2046 tsk_advance_rx_queue(sk
);
2047 __tipc_send_stream(new_sock
, &m
, 0);
2049 __skb_dequeue(&sk
->sk_receive_queue
);
2050 __skb_queue_head(&new_sk
->sk_receive_queue
, buf
);
2051 skb_set_owner_r(buf
, new_sk
);
2053 release_sock(new_sk
);
2060 * tipc_shutdown - shutdown socket connection
2061 * @sock: socket structure
2062 * @how: direction to close (must be SHUT_RDWR)
2064 * Terminates connection (if necessary), then purges socket's receive queue.
2066 * Returns 0 on success, errno otherwise
2068 static int tipc_shutdown(struct socket
*sock
, int how
)
2070 struct sock
*sk
= sock
->sk
;
2071 struct net
*net
= sock_net(sk
);
2072 struct tipc_sock
*tsk
= tipc_sk(sk
);
2073 struct sk_buff
*skb
;
2074 u32 dnode
= tsk_peer_node(tsk
);
2075 u32 dport
= tsk_peer_port(tsk
);
2076 u32 onode
= tipc_own_addr(net
);
2077 u32 oport
= tsk
->portid
;
2080 if (how
!= SHUT_RDWR
)
2085 switch (sock
->state
) {
2090 dnode
= tsk_peer_node(tsk
);
2092 /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */
2093 skb
= __skb_dequeue(&sk
->sk_receive_queue
);
2095 if (TIPC_SKB_CB(skb
)->handle
!= NULL
) {
2099 tipc_sk_respond(sk
, skb
, TIPC_CONN_SHUTDOWN
);
2101 skb
= tipc_msg_create(TIPC_CRITICAL_IMPORTANCE
,
2102 TIPC_CONN_MSG
, SHORT_H_SIZE
,
2103 0, dnode
, onode
, dport
, oport
,
2104 TIPC_CONN_SHUTDOWN
);
2105 tipc_node_xmit_skb(net
, skb
, dnode
, tsk
->portid
);
2108 sock
->state
= SS_DISCONNECTING
;
2109 tipc_node_remove_conn(net
, dnode
, tsk
->portid
);
2112 case SS_DISCONNECTING
:
2114 /* Discard any unreceived messages */
2115 __skb_queue_purge(&sk
->sk_receive_queue
);
2117 /* Wake up anyone sleeping in poll */
2118 sk
->sk_state_change(sk
);
2130 static void tipc_sk_timeout(unsigned long data
)
2132 struct tipc_sock
*tsk
= (struct tipc_sock
*)data
;
2133 struct sock
*sk
= &tsk
->sk
;
2134 struct sk_buff
*skb
= NULL
;
2135 u32 peer_port
, peer_node
;
2136 u32 own_node
= tsk_own_node(tsk
);
2139 if (!tsk
->connected
) {
2143 peer_port
= tsk_peer_port(tsk
);
2144 peer_node
= tsk_peer_node(tsk
);
2146 if (tsk
->probing_state
== TIPC_CONN_PROBING
) {
2147 if (!sock_owned_by_user(sk
)) {
2148 sk
->sk_socket
->state
= SS_DISCONNECTING
;
2150 tipc_node_remove_conn(sock_net(sk
), tsk_peer_node(tsk
),
2151 tsk_peer_port(tsk
));
2152 sk
->sk_state_change(sk
);
2154 /* Try again later */
2155 sk_reset_timer(sk
, &sk
->sk_timer
, (HZ
/ 20));
2159 skb
= tipc_msg_create(CONN_MANAGER
, CONN_PROBE
,
2160 INT_H_SIZE
, 0, peer_node
, own_node
,
2161 peer_port
, tsk
->portid
, TIPC_OK
);
2162 tsk
->probing_state
= TIPC_CONN_PROBING
;
2163 sk_reset_timer(sk
, &sk
->sk_timer
, jiffies
+ tsk
->probing_intv
);
2167 tipc_node_xmit_skb(sock_net(sk
), skb
, peer_node
, tsk
->portid
);
2172 static int tipc_sk_publish(struct tipc_sock
*tsk
, uint scope
,
2173 struct tipc_name_seq
const *seq
)
2175 struct net
*net
= sock_net(&tsk
->sk
);
2176 struct publication
*publ
;
2181 key
= tsk
->portid
+ tsk
->pub_count
+ 1;
2182 if (key
== tsk
->portid
)
2185 publ
= tipc_nametbl_publish(net
, seq
->type
, seq
->lower
, seq
->upper
,
2186 scope
, tsk
->portid
, key
);
2187 if (unlikely(!publ
))
2190 list_add(&publ
->pport_list
, &tsk
->publications
);
2196 static int tipc_sk_withdraw(struct tipc_sock
*tsk
, uint scope
,
2197 struct tipc_name_seq
const *seq
)
2199 struct net
*net
= sock_net(&tsk
->sk
);
2200 struct publication
*publ
;
2201 struct publication
*safe
;
2204 list_for_each_entry_safe(publ
, safe
, &tsk
->publications
, pport_list
) {
2206 if (publ
->scope
!= scope
)
2208 if (publ
->type
!= seq
->type
)
2210 if (publ
->lower
!= seq
->lower
)
2212 if (publ
->upper
!= seq
->upper
)
2214 tipc_nametbl_withdraw(net
, publ
->type
, publ
->lower
,
2215 publ
->ref
, publ
->key
);
2219 tipc_nametbl_withdraw(net
, publ
->type
, publ
->lower
,
2220 publ
->ref
, publ
->key
);
2223 if (list_empty(&tsk
->publications
))
2228 /* tipc_sk_reinit: set non-zero address in all existing sockets
2229 * when we go from standalone to network mode.
2231 void tipc_sk_reinit(struct net
*net
)
2233 struct tipc_net
*tn
= net_generic(net
, tipc_net_id
);
2234 const struct bucket_table
*tbl
;
2235 struct rhash_head
*pos
;
2236 struct tipc_sock
*tsk
;
2237 struct tipc_msg
*msg
;
2241 tbl
= rht_dereference_rcu((&tn
->sk_rht
)->tbl
, &tn
->sk_rht
);
2242 for (i
= 0; i
< tbl
->size
; i
++) {
2243 rht_for_each_entry_rcu(tsk
, pos
, tbl
, i
, node
) {
2244 spin_lock_bh(&tsk
->sk
.sk_lock
.slock
);
2246 msg_set_prevnode(msg
, tn
->own_addr
);
2247 msg_set_orignode(msg
, tn
->own_addr
);
2248 spin_unlock_bh(&tsk
->sk
.sk_lock
.slock
);
2254 static struct tipc_sock
*tipc_sk_lookup(struct net
*net
, u32 portid
)
2256 struct tipc_net
*tn
= net_generic(net
, tipc_net_id
);
2257 struct tipc_sock
*tsk
;
2260 tsk
= rhashtable_lookup_fast(&tn
->sk_rht
, &portid
, tsk_rht_params
);
2262 sock_hold(&tsk
->sk
);
2268 static int tipc_sk_insert(struct tipc_sock
*tsk
)
2270 struct sock
*sk
= &tsk
->sk
;
2271 struct net
*net
= sock_net(sk
);
2272 struct tipc_net
*tn
= net_generic(net
, tipc_net_id
);
2273 u32 remaining
= (TIPC_MAX_PORT
- TIPC_MIN_PORT
) + 1;
2274 u32 portid
= prandom_u32() % remaining
+ TIPC_MIN_PORT
;
2276 while (remaining
--) {
2278 if ((portid
< TIPC_MIN_PORT
) || (portid
> TIPC_MAX_PORT
))
2279 portid
= TIPC_MIN_PORT
;
2280 tsk
->portid
= portid
;
2281 sock_hold(&tsk
->sk
);
2282 if (!rhashtable_lookup_insert_fast(&tn
->sk_rht
, &tsk
->node
,
2291 static void tipc_sk_remove(struct tipc_sock
*tsk
)
2293 struct sock
*sk
= &tsk
->sk
;
2294 struct tipc_net
*tn
= net_generic(sock_net(sk
), tipc_net_id
);
2296 if (!rhashtable_remove_fast(&tn
->sk_rht
, &tsk
->node
, tsk_rht_params
)) {
2297 WARN_ON(atomic_read(&sk
->sk_refcnt
) == 1);
2302 static const struct rhashtable_params tsk_rht_params
= {
2304 .head_offset
= offsetof(struct tipc_sock
, node
),
2305 .key_offset
= offsetof(struct tipc_sock
, portid
),
2306 .key_len
= sizeof(u32
), /* portid */
2307 .max_size
= 1048576,
2309 .automatic_shrinking
= true,
2312 int tipc_sk_rht_init(struct net
*net
)
2314 struct tipc_net
*tn
= net_generic(net
, tipc_net_id
);
2316 return rhashtable_init(&tn
->sk_rht
, &tsk_rht_params
);
2319 void tipc_sk_rht_destroy(struct net
*net
)
2321 struct tipc_net
*tn
= net_generic(net
, tipc_net_id
);
2323 /* Wait for socket readers to complete */
2326 rhashtable_destroy(&tn
->sk_rht
);
2330 * tipc_setsockopt - set socket option
2331 * @sock: socket structure
2332 * @lvl: option level
2333 * @opt: option identifier
2334 * @ov: pointer to new option value
2335 * @ol: length of option value
2337 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
2338 * (to ease compatibility).
2340 * Returns 0 on success, errno otherwise
2342 static int tipc_setsockopt(struct socket
*sock
, int lvl
, int opt
,
2343 char __user
*ov
, unsigned int ol
)
2345 struct sock
*sk
= sock
->sk
;
2346 struct tipc_sock
*tsk
= tipc_sk(sk
);
2350 if ((lvl
== IPPROTO_TCP
) && (sock
->type
== SOCK_STREAM
))
2352 if (lvl
!= SOL_TIPC
)
2353 return -ENOPROTOOPT
;
2354 if (ol
< sizeof(value
))
2356 res
= get_user(value
, (u32 __user
*)ov
);
2363 case TIPC_IMPORTANCE
:
2364 res
= tsk_set_importance(tsk
, value
);
2366 case TIPC_SRC_DROPPABLE
:
2367 if (sock
->type
!= SOCK_STREAM
)
2368 tsk_set_unreliable(tsk
, value
);
2372 case TIPC_DEST_DROPPABLE
:
2373 tsk_set_unreturnable(tsk
, value
);
2375 case TIPC_CONN_TIMEOUT
:
2376 tipc_sk(sk
)->conn_timeout
= value
;
2377 /* no need to set "res", since already 0 at this point */
2389 * tipc_getsockopt - get socket option
2390 * @sock: socket structure
2391 * @lvl: option level
2392 * @opt: option identifier
2393 * @ov: receptacle for option value
2394 * @ol: receptacle for length of option value
2396 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
2397 * (to ease compatibility).
2399 * Returns 0 on success, errno otherwise
2401 static int tipc_getsockopt(struct socket
*sock
, int lvl
, int opt
,
2402 char __user
*ov
, int __user
*ol
)
2404 struct sock
*sk
= sock
->sk
;
2405 struct tipc_sock
*tsk
= tipc_sk(sk
);
2410 if ((lvl
== IPPROTO_TCP
) && (sock
->type
== SOCK_STREAM
))
2411 return put_user(0, ol
);
2412 if (lvl
!= SOL_TIPC
)
2413 return -ENOPROTOOPT
;
2414 res
= get_user(len
, ol
);
2421 case TIPC_IMPORTANCE
:
2422 value
= tsk_importance(tsk
);
2424 case TIPC_SRC_DROPPABLE
:
2425 value
= tsk_unreliable(tsk
);
2427 case TIPC_DEST_DROPPABLE
:
2428 value
= tsk_unreturnable(tsk
);
2430 case TIPC_CONN_TIMEOUT
:
2431 value
= tsk
->conn_timeout
;
2432 /* no need to set "res", since already 0 at this point */
2434 case TIPC_NODE_RECVQ_DEPTH
:
2435 value
= 0; /* was tipc_queue_size, now obsolete */
2437 case TIPC_SOCK_RECVQ_DEPTH
:
2438 value
= skb_queue_len(&sk
->sk_receive_queue
);
2447 return res
; /* "get" failed */
2449 if (len
< sizeof(value
))
2452 if (copy_to_user(ov
, &value
, sizeof(value
)))
2455 return put_user(sizeof(value
), ol
);
2458 static int tipc_ioctl(struct socket
*sock
, unsigned int cmd
, unsigned long arg
)
2460 struct sock
*sk
= sock
->sk
;
2461 struct tipc_sioc_ln_req lnr
;
2462 void __user
*argp
= (void __user
*)arg
;
2465 case SIOCGETLINKNAME
:
2466 if (copy_from_user(&lnr
, argp
, sizeof(lnr
)))
2468 if (!tipc_node_get_linkname(sock_net(sk
),
2469 lnr
.bearer_id
& 0xffff, lnr
.peer
,
2470 lnr
.linkname
, TIPC_MAX_LINK_NAME
)) {
2471 if (copy_to_user(argp
, &lnr
, sizeof(lnr
)))
2475 return -EADDRNOTAVAIL
;
2477 return -ENOIOCTLCMD
;
2481 /* Protocol switches for the various types of TIPC sockets */
2483 static const struct proto_ops msg_ops
= {
2484 .owner
= THIS_MODULE
,
2486 .release
= tipc_release
,
2488 .connect
= tipc_connect
,
2489 .socketpair
= sock_no_socketpair
,
2490 .accept
= sock_no_accept
,
2491 .getname
= tipc_getname
,
2493 .ioctl
= tipc_ioctl
,
2494 .listen
= sock_no_listen
,
2495 .shutdown
= tipc_shutdown
,
2496 .setsockopt
= tipc_setsockopt
,
2497 .getsockopt
= tipc_getsockopt
,
2498 .sendmsg
= tipc_sendmsg
,
2499 .recvmsg
= tipc_recvmsg
,
2500 .mmap
= sock_no_mmap
,
2501 .sendpage
= sock_no_sendpage
2504 static const struct proto_ops packet_ops
= {
2505 .owner
= THIS_MODULE
,
2507 .release
= tipc_release
,
2509 .connect
= tipc_connect
,
2510 .socketpair
= sock_no_socketpair
,
2511 .accept
= tipc_accept
,
2512 .getname
= tipc_getname
,
2514 .ioctl
= tipc_ioctl
,
2515 .listen
= tipc_listen
,
2516 .shutdown
= tipc_shutdown
,
2517 .setsockopt
= tipc_setsockopt
,
2518 .getsockopt
= tipc_getsockopt
,
2519 .sendmsg
= tipc_send_packet
,
2520 .recvmsg
= tipc_recvmsg
,
2521 .mmap
= sock_no_mmap
,
2522 .sendpage
= sock_no_sendpage
2525 static const struct proto_ops stream_ops
= {
2526 .owner
= THIS_MODULE
,
2528 .release
= tipc_release
,
2530 .connect
= tipc_connect
,
2531 .socketpair
= sock_no_socketpair
,
2532 .accept
= tipc_accept
,
2533 .getname
= tipc_getname
,
2535 .ioctl
= tipc_ioctl
,
2536 .listen
= tipc_listen
,
2537 .shutdown
= tipc_shutdown
,
2538 .setsockopt
= tipc_setsockopt
,
2539 .getsockopt
= tipc_getsockopt
,
2540 .sendmsg
= tipc_send_stream
,
2541 .recvmsg
= tipc_recv_stream
,
2542 .mmap
= sock_no_mmap
,
2543 .sendpage
= sock_no_sendpage
2546 static const struct net_proto_family tipc_family_ops
= {
2547 .owner
= THIS_MODULE
,
2549 .create
= tipc_sk_create
2552 static struct proto tipc_proto
= {
2554 .owner
= THIS_MODULE
,
2555 .obj_size
= sizeof(struct tipc_sock
),
2556 .sysctl_rmem
= sysctl_tipc_rmem
2560 * tipc_socket_init - initialize TIPC socket interface
2562 * Returns 0 on success, errno otherwise
2564 int tipc_socket_init(void)
2568 res
= proto_register(&tipc_proto
, 1);
2570 pr_err("Failed to register TIPC protocol type\n");
2574 res
= sock_register(&tipc_family_ops
);
2576 pr_err("Failed to register TIPC socket type\n");
2577 proto_unregister(&tipc_proto
);
2585 * tipc_socket_stop - stop TIPC socket interface
2587 void tipc_socket_stop(void)
2589 sock_unregister(tipc_family_ops
.family
);
2590 proto_unregister(&tipc_proto
);
2593 /* Caller should hold socket lock for the passed tipc socket. */
2594 static int __tipc_nl_add_sk_con(struct sk_buff
*skb
, struct tipc_sock
*tsk
)
2598 struct nlattr
*nest
;
2600 peer_node
= tsk_peer_node(tsk
);
2601 peer_port
= tsk_peer_port(tsk
);
2603 nest
= nla_nest_start(skb
, TIPC_NLA_SOCK_CON
);
2605 if (nla_put_u32(skb
, TIPC_NLA_CON_NODE
, peer_node
))
2607 if (nla_put_u32(skb
, TIPC_NLA_CON_SOCK
, peer_port
))
2610 if (tsk
->conn_type
!= 0) {
2611 if (nla_put_flag(skb
, TIPC_NLA_CON_FLAG
))
2613 if (nla_put_u32(skb
, TIPC_NLA_CON_TYPE
, tsk
->conn_type
))
2615 if (nla_put_u32(skb
, TIPC_NLA_CON_INST
, tsk
->conn_instance
))
2618 nla_nest_end(skb
, nest
);
2623 nla_nest_cancel(skb
, nest
);
2628 /* Caller should hold socket lock for the passed tipc socket. */
2629 static int __tipc_nl_add_sk(struct sk_buff
*skb
, struct netlink_callback
*cb
,
2630 struct tipc_sock
*tsk
)
2634 struct nlattr
*attrs
;
2635 struct net
*net
= sock_net(skb
->sk
);
2636 struct tipc_net
*tn
= net_generic(net
, tipc_net_id
);
2638 hdr
= genlmsg_put(skb
, NETLINK_CB(cb
->skb
).portid
, cb
->nlh
->nlmsg_seq
,
2639 &tipc_genl_family
, NLM_F_MULTI
, TIPC_NL_SOCK_GET
);
2643 attrs
= nla_nest_start(skb
, TIPC_NLA_SOCK
);
2645 goto genlmsg_cancel
;
2646 if (nla_put_u32(skb
, TIPC_NLA_SOCK_REF
, tsk
->portid
))
2647 goto attr_msg_cancel
;
2648 if (nla_put_u32(skb
, TIPC_NLA_SOCK_ADDR
, tn
->own_addr
))
2649 goto attr_msg_cancel
;
2651 if (tsk
->connected
) {
2652 err
= __tipc_nl_add_sk_con(skb
, tsk
);
2654 goto attr_msg_cancel
;
2655 } else if (!list_empty(&tsk
->publications
)) {
2656 if (nla_put_flag(skb
, TIPC_NLA_SOCK_HAS_PUBL
))
2657 goto attr_msg_cancel
;
2659 nla_nest_end(skb
, attrs
);
2660 genlmsg_end(skb
, hdr
);
2665 nla_nest_cancel(skb
, attrs
);
2667 genlmsg_cancel(skb
, hdr
);
2672 int tipc_nl_sk_dump(struct sk_buff
*skb
, struct netlink_callback
*cb
)
2675 struct tipc_sock
*tsk
;
2676 const struct bucket_table
*tbl
;
2677 struct rhash_head
*pos
;
2678 struct net
*net
= sock_net(skb
->sk
);
2679 struct tipc_net
*tn
= net_generic(net
, tipc_net_id
);
2680 u32 tbl_id
= cb
->args
[0];
2681 u32 prev_portid
= cb
->args
[1];
2684 tbl
= rht_dereference_rcu((&tn
->sk_rht
)->tbl
, &tn
->sk_rht
);
2685 for (; tbl_id
< tbl
->size
; tbl_id
++) {
2686 rht_for_each_entry_rcu(tsk
, pos
, tbl
, tbl_id
, node
) {
2687 spin_lock_bh(&tsk
->sk
.sk_lock
.slock
);
2688 if (prev_portid
&& prev_portid
!= tsk
->portid
) {
2689 spin_unlock_bh(&tsk
->sk
.sk_lock
.slock
);
2693 err
= __tipc_nl_add_sk(skb
, cb
, tsk
);
2695 prev_portid
= tsk
->portid
;
2696 spin_unlock_bh(&tsk
->sk
.sk_lock
.slock
);
2700 spin_unlock_bh(&tsk
->sk
.sk_lock
.slock
);
2705 cb
->args
[0] = tbl_id
;
2706 cb
->args
[1] = prev_portid
;
2711 /* Caller should hold socket lock for the passed tipc socket. */
2712 static int __tipc_nl_add_sk_publ(struct sk_buff
*skb
,
2713 struct netlink_callback
*cb
,
2714 struct publication
*publ
)
2717 struct nlattr
*attrs
;
2719 hdr
= genlmsg_put(skb
, NETLINK_CB(cb
->skb
).portid
, cb
->nlh
->nlmsg_seq
,
2720 &tipc_genl_family
, NLM_F_MULTI
, TIPC_NL_PUBL_GET
);
2724 attrs
= nla_nest_start(skb
, TIPC_NLA_PUBL
);
2726 goto genlmsg_cancel
;
2728 if (nla_put_u32(skb
, TIPC_NLA_PUBL_KEY
, publ
->key
))
2729 goto attr_msg_cancel
;
2730 if (nla_put_u32(skb
, TIPC_NLA_PUBL_TYPE
, publ
->type
))
2731 goto attr_msg_cancel
;
2732 if (nla_put_u32(skb
, TIPC_NLA_PUBL_LOWER
, publ
->lower
))
2733 goto attr_msg_cancel
;
2734 if (nla_put_u32(skb
, TIPC_NLA_PUBL_UPPER
, publ
->upper
))
2735 goto attr_msg_cancel
;
2737 nla_nest_end(skb
, attrs
);
2738 genlmsg_end(skb
, hdr
);
2743 nla_nest_cancel(skb
, attrs
);
2745 genlmsg_cancel(skb
, hdr
);
2750 /* Caller should hold socket lock for the passed tipc socket. */
2751 static int __tipc_nl_list_sk_publ(struct sk_buff
*skb
,
2752 struct netlink_callback
*cb
,
2753 struct tipc_sock
*tsk
, u32
*last_publ
)
2756 struct publication
*p
;
2759 list_for_each_entry(p
, &tsk
->publications
, pport_list
) {
2760 if (p
->key
== *last_publ
)
2763 if (p
->key
!= *last_publ
) {
2764 /* We never set seq or call nl_dump_check_consistent()
2765 * this means that setting prev_seq here will cause the
2766 * consistence check to fail in the netlink callback
2767 * handler. Resulting in the last NLMSG_DONE message
2768 * having the NLM_F_DUMP_INTR flag set.
2775 p
= list_first_entry(&tsk
->publications
, struct publication
,
2779 list_for_each_entry_from(p
, &tsk
->publications
, pport_list
) {
2780 err
= __tipc_nl_add_sk_publ(skb
, cb
, p
);
2782 *last_publ
= p
->key
;
2791 int tipc_nl_publ_dump(struct sk_buff
*skb
, struct netlink_callback
*cb
)
2794 u32 tsk_portid
= cb
->args
[0];
2795 u32 last_publ
= cb
->args
[1];
2796 u32 done
= cb
->args
[2];
2797 struct net
*net
= sock_net(skb
->sk
);
2798 struct tipc_sock
*tsk
;
2801 struct nlattr
**attrs
;
2802 struct nlattr
*sock
[TIPC_NLA_SOCK_MAX
+ 1];
2804 err
= tipc_nlmsg_parse(cb
->nlh
, &attrs
);
2808 err
= nla_parse_nested(sock
, TIPC_NLA_SOCK_MAX
,
2809 attrs
[TIPC_NLA_SOCK
],
2810 tipc_nl_sock_policy
);
2814 if (!sock
[TIPC_NLA_SOCK_REF
])
2817 tsk_portid
= nla_get_u32(sock
[TIPC_NLA_SOCK_REF
]);
2823 tsk
= tipc_sk_lookup(net
, tsk_portid
);
2827 lock_sock(&tsk
->sk
);
2828 err
= __tipc_nl_list_sk_publ(skb
, cb
, tsk
, &last_publ
);
2831 release_sock(&tsk
->sk
);
2834 cb
->args
[0] = tsk_portid
;
2835 cb
->args
[1] = last_publ
;