Commit | Line | Data |
---|---|---|
b97bf3fd PL |
1 | /* |
2 | * net/tipc/socket.c: TIPC socket API | |
c4307285 | 3 | * |
5eee6a6d | 4 | * Copyright (c) 2001-2007, Ericsson AB |
4132faca | 5 | * Copyright (c) 2004-2008, 2010-2011, Wind River Systems |
b97bf3fd PL |
6 | * All rights reserved. |
7 | * | |
9ea1fd3c | 8 | * Redistribution and use in source and binary forms, with or without |
b97bf3fd PL |
9 | * modification, are permitted provided that the following conditions are met: |
10 | * | |
9ea1fd3c PL |
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. | |
b97bf3fd | 19 | * |
9ea1fd3c PL |
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. | |
23 | * | |
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 | |
b97bf3fd PL |
34 | * POSSIBILITY OF SUCH DAMAGE. |
35 | */ | |
36 | ||
b97bf3fd PL |
37 | #include <net/sock.h> |
38 | ||
39 | #include <linux/tipc.h> | |
ea714ccd | 40 | #include <linux/tipc_config.h> |
b97bf3fd PL |
41 | |
42 | #include "core.h" | |
d265fef6 | 43 | #include "port.h" |
b97bf3fd PL |
44 | |
45 | #define SS_LISTENING -1 /* socket is listening */ | |
46 | #define SS_READY -2 /* socket is connectionless */ | |
47 | ||
3654ea02 AS |
48 | #define OVERLOAD_LIMIT_BASE 5000 |
49 | #define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */ | |
b97bf3fd PL |
50 | |
51 | struct tipc_sock { | |
52 | struct sock sk; | |
53 | struct tipc_port *p; | |
2da59918 | 54 | struct tipc_portid peer_name; |
564e83b5 | 55 | long conn_timeout; |
b97bf3fd PL |
56 | }; |
57 | ||
0c3141e9 AS |
58 | #define tipc_sk(sk) ((struct tipc_sock *)(sk)) |
59 | #define tipc_sk_port(sk) ((struct tipc_port *)(tipc_sk(sk)->p)) | |
b97bf3fd | 60 | |
0c3141e9 | 61 | static int backlog_rcv(struct sock *sk, struct sk_buff *skb); |
b97bf3fd PL |
62 | static u32 dispatch(struct tipc_port *tport, struct sk_buff *buf); |
63 | static void wakeupdispatch(struct tipc_port *tport); | |
64 | ||
bca65eae FW |
65 | static const struct proto_ops packet_ops; |
66 | static const struct proto_ops stream_ops; | |
67 | static const struct proto_ops msg_ops; | |
b97bf3fd PL |
68 | |
69 | static struct proto tipc_proto; | |
70 | ||
e3ec9c7d | 71 | static int sockets_enabled; |
b97bf3fd PL |
72 | |
73 | static atomic_t tipc_queue_size = ATOMIC_INIT(0); | |
74 | ||
c4307285 | 75 | /* |
0c3141e9 AS |
76 | * Revised TIPC socket locking policy: |
77 | * | |
78 | * Most socket operations take the standard socket lock when they start | |
79 | * and hold it until they finish (or until they need to sleep). Acquiring | |
80 | * this lock grants the owner exclusive access to the fields of the socket | |
81 | * data structures, with the exception of the backlog queue. A few socket | |
82 | * operations can be done without taking the socket lock because they only | |
83 | * read socket information that never changes during the life of the socket. | |
84 | * | |
85 | * Socket operations may acquire the lock for the associated TIPC port if they | |
86 | * need to perform an operation on the port. If any routine needs to acquire | |
87 | * both the socket lock and the port lock it must take the socket lock first | |
88 | * to avoid the risk of deadlock. | |
89 | * | |
90 | * The dispatcher handling incoming messages cannot grab the socket lock in | |
91 | * the standard fashion, since invoked it runs at the BH level and cannot block. | |
92 | * Instead, it checks to see if the socket lock is currently owned by someone, | |
93 | * and either handles the message itself or adds it to the socket's backlog | |
94 | * queue; in the latter case the queued message is processed once the process | |
95 | * owning the socket lock releases it. | |
96 | * | |
97 | * NOTE: Releasing the socket lock while an operation is sleeping overcomes | |
98 | * the problem of a blocked socket operation preventing any other operations | |
99 | * from occurring. However, applications must be careful if they have | |
100 | * multiple threads trying to send (or receive) on the same socket, as these | |
101 | * operations might interfere with each other. For example, doing a connect | |
102 | * and a receive at the same time might allow the receive to consume the | |
103 | * ACK message meant for the connect. While additional work could be done | |
104 | * to try and overcome this, it doesn't seem to be worthwhile at the present. | |
105 | * | |
106 | * NOTE: Releasing the socket lock while an operation is sleeping also ensures | |
107 | * that another operation that must be performed in a non-blocking manner is | |
108 | * not delayed for very long because the lock has already been taken. | |
109 | * | |
110 | * NOTE: This code assumes that certain fields of a port/socket pair are | |
111 | * constant over its lifetime; such fields can be examined without taking | |
112 | * the socket lock and/or port lock, and do not need to be re-read even | |
113 | * after resuming processing after waiting. These fields include: | |
114 | * - socket type | |
115 | * - pointer to socket sk structure (aka tipc_sock structure) | |
116 | * - pointer to port structure | |
117 | * - port reference | |
118 | */ | |
119 | ||
120 | /** | |
121 | * advance_rx_queue - discard first buffer in socket receive queue | |
122 | * | |
123 | * Caller must hold socket lock | |
b97bf3fd | 124 | */ |
0c3141e9 AS |
125 | |
126 | static void advance_rx_queue(struct sock *sk) | |
b97bf3fd | 127 | { |
0c3141e9 AS |
128 | buf_discard(__skb_dequeue(&sk->sk_receive_queue)); |
129 | atomic_dec(&tipc_queue_size); | |
b97bf3fd PL |
130 | } |
131 | ||
0c3141e9 AS |
132 | /** |
133 | * discard_rx_queue - discard all buffers in socket receive queue | |
134 | * | |
135 | * Caller must hold socket lock | |
b97bf3fd | 136 | */ |
0c3141e9 AS |
137 | |
138 | static void discard_rx_queue(struct sock *sk) | |
b97bf3fd | 139 | { |
0c3141e9 AS |
140 | struct sk_buff *buf; |
141 | ||
142 | while ((buf = __skb_dequeue(&sk->sk_receive_queue))) { | |
143 | atomic_dec(&tipc_queue_size); | |
144 | buf_discard(buf); | |
145 | } | |
b97bf3fd PL |
146 | } |
147 | ||
b97bf3fd | 148 | /** |
0c3141e9 AS |
149 | * reject_rx_queue - reject all buffers in socket receive queue |
150 | * | |
151 | * Caller must hold socket lock | |
b97bf3fd PL |
152 | */ |
153 | ||
0c3141e9 | 154 | static void reject_rx_queue(struct sock *sk) |
b97bf3fd | 155 | { |
0c3141e9 AS |
156 | struct sk_buff *buf; |
157 | ||
158 | while ((buf = __skb_dequeue(&sk->sk_receive_queue))) { | |
159 | tipc_reject_msg(buf, TIPC_ERR_NO_PORT); | |
160 | atomic_dec(&tipc_queue_size); | |
161 | } | |
b97bf3fd PL |
162 | } |
163 | ||
164 | /** | |
165 | * tipc_create - create a TIPC socket | |
0c3141e9 | 166 | * @net: network namespace (must be default network) |
b97bf3fd PL |
167 | * @sock: pre-allocated socket structure |
168 | * @protocol: protocol indicator (must be 0) | |
3f378b68 | 169 | * @kern: caused by kernel or by userspace? |
c4307285 | 170 | * |
0c3141e9 AS |
171 | * This routine creates additional data structures used by the TIPC socket, |
172 | * initializes them, and links them together. | |
b97bf3fd PL |
173 | * |
174 | * Returns 0 on success, errno otherwise | |
175 | */ | |
0c3141e9 | 176 | |
3f378b68 EP |
177 | static int tipc_create(struct net *net, struct socket *sock, int protocol, |
178 | int kern) | |
b97bf3fd | 179 | { |
0c3141e9 AS |
180 | const struct proto_ops *ops; |
181 | socket_state state; | |
b97bf3fd | 182 | struct sock *sk; |
7ef43eba | 183 | struct tipc_port *tp_ptr; |
0c3141e9 AS |
184 | |
185 | /* Validate arguments */ | |
b97bf3fd | 186 | |
09ad9bc7 | 187 | if (!net_eq(net, &init_net)) |
1b8d7ae4 EB |
188 | return -EAFNOSUPPORT; |
189 | ||
b97bf3fd PL |
190 | if (unlikely(protocol != 0)) |
191 | return -EPROTONOSUPPORT; | |
192 | ||
b97bf3fd PL |
193 | switch (sock->type) { |
194 | case SOCK_STREAM: | |
0c3141e9 AS |
195 | ops = &stream_ops; |
196 | state = SS_UNCONNECTED; | |
b97bf3fd PL |
197 | break; |
198 | case SOCK_SEQPACKET: | |
0c3141e9 AS |
199 | ops = &packet_ops; |
200 | state = SS_UNCONNECTED; | |
b97bf3fd PL |
201 | break; |
202 | case SOCK_DGRAM: | |
b97bf3fd | 203 | case SOCK_RDM: |
0c3141e9 AS |
204 | ops = &msg_ops; |
205 | state = SS_READY; | |
b97bf3fd | 206 | break; |
49978651 | 207 | default: |
49978651 | 208 | return -EPROTOTYPE; |
b97bf3fd PL |
209 | } |
210 | ||
0c3141e9 AS |
211 | /* Allocate socket's protocol area */ |
212 | ||
6257ff21 | 213 | sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto); |
0c3141e9 | 214 | if (sk == NULL) |
b97bf3fd | 215 | return -ENOMEM; |
b97bf3fd | 216 | |
0c3141e9 | 217 | /* Allocate TIPC port for socket to use */ |
b97bf3fd | 218 | |
0ea52241 AS |
219 | tp_ptr = tipc_createport_raw(sk, &dispatch, &wakeupdispatch, |
220 | TIPC_LOW_IMPORTANCE); | |
221 | if (unlikely(!tp_ptr)) { | |
0c3141e9 AS |
222 | sk_free(sk); |
223 | return -ENOMEM; | |
224 | } | |
b97bf3fd | 225 | |
0c3141e9 | 226 | /* Finish initializing socket data structures */ |
b97bf3fd | 227 | |
0c3141e9 AS |
228 | sock->ops = ops; |
229 | sock->state = state; | |
b97bf3fd | 230 | |
0c3141e9 | 231 | sock_init_data(sock, sk); |
0c3141e9 | 232 | sk->sk_backlog_rcv = backlog_rcv; |
0ea52241 | 233 | tipc_sk(sk)->p = tp_ptr; |
564e83b5 | 234 | tipc_sk(sk)->conn_timeout = msecs_to_jiffies(CONN_TIMEOUT_DEFAULT); |
b97bf3fd | 235 | |
7ef43eba AS |
236 | spin_unlock_bh(tp_ptr->lock); |
237 | ||
0c3141e9 | 238 | if (sock->state == SS_READY) { |
0ea52241 | 239 | tipc_set_portunreturnable(tp_ptr->ref, 1); |
0c3141e9 | 240 | if (sock->type == SOCK_DGRAM) |
0ea52241 | 241 | tipc_set_portunreliable(tp_ptr->ref, 1); |
0c3141e9 | 242 | } |
b97bf3fd PL |
243 | |
244 | return 0; | |
245 | } | |
246 | ||
247 | /** | |
248 | * release - destroy a TIPC socket | |
249 | * @sock: socket to destroy | |
250 | * | |
251 | * This routine cleans up any messages that are still queued on the socket. | |
252 | * For DGRAM and RDM socket types, all queued messages are rejected. | |
253 | * For SEQPACKET and STREAM socket types, the first message is rejected | |
254 | * and any others are discarded. (If the first message on a STREAM socket | |
255 | * is partially-read, it is discarded and the next one is rejected instead.) | |
c4307285 | 256 | * |
b97bf3fd PL |
257 | * NOTE: Rejected messages are not necessarily returned to the sender! They |
258 | * are returned or discarded according to the "destination droppable" setting | |
259 | * specified for the message by the sender. | |
260 | * | |
261 | * Returns 0 on success, errno otherwise | |
262 | */ | |
263 | ||
264 | static int release(struct socket *sock) | |
265 | { | |
b97bf3fd | 266 | struct sock *sk = sock->sk; |
0c3141e9 | 267 | struct tipc_port *tport; |
b97bf3fd | 268 | struct sk_buff *buf; |
0c3141e9 | 269 | int res; |
b97bf3fd | 270 | |
0c3141e9 AS |
271 | /* |
272 | * Exit if socket isn't fully initialized (occurs when a failed accept() | |
273 | * releases a pre-allocated child socket that was never used) | |
274 | */ | |
275 | ||
276 | if (sk == NULL) | |
b97bf3fd | 277 | return 0; |
c4307285 | 278 | |
0c3141e9 AS |
279 | tport = tipc_sk_port(sk); |
280 | lock_sock(sk); | |
281 | ||
282 | /* | |
283 | * Reject all unreceived messages, except on an active connection | |
284 | * (which disconnects locally & sends a 'FIN+' to peer) | |
285 | */ | |
b97bf3fd PL |
286 | |
287 | while (sock->state != SS_DISCONNECTING) { | |
0c3141e9 AS |
288 | buf = __skb_dequeue(&sk->sk_receive_queue); |
289 | if (buf == NULL) | |
b97bf3fd | 290 | break; |
0c3141e9 | 291 | atomic_dec(&tipc_queue_size); |
b97bf3fd PL |
292 | if (TIPC_SKB_CB(buf)->handle != msg_data(buf_msg(buf))) |
293 | buf_discard(buf); | |
0c3141e9 AS |
294 | else { |
295 | if ((sock->state == SS_CONNECTING) || | |
296 | (sock->state == SS_CONNECTED)) { | |
297 | sock->state = SS_DISCONNECTING; | |
298 | tipc_disconnect(tport->ref); | |
299 | } | |
b97bf3fd | 300 | tipc_reject_msg(buf, TIPC_ERR_NO_PORT); |
0c3141e9 | 301 | } |
b97bf3fd PL |
302 | } |
303 | ||
0c3141e9 AS |
304 | /* |
305 | * Delete TIPC port; this ensures no more messages are queued | |
306 | * (also disconnects an active connection & sends a 'FIN-' to peer) | |
307 | */ | |
b97bf3fd | 308 | |
0c3141e9 | 309 | res = tipc_deleteport(tport->ref); |
b97bf3fd | 310 | |
0c3141e9 | 311 | /* Discard any remaining (connection-based) messages in receive queue */ |
b97bf3fd | 312 | |
0c3141e9 | 313 | discard_rx_queue(sk); |
b97bf3fd | 314 | |
0c3141e9 AS |
315 | /* Reject any messages that accumulated in backlog queue */ |
316 | ||
317 | sock->state = SS_DISCONNECTING; | |
318 | release_sock(sk); | |
b97bf3fd PL |
319 | |
320 | sock_put(sk); | |
0c3141e9 | 321 | sock->sk = NULL; |
b97bf3fd | 322 | |
b97bf3fd PL |
323 | return res; |
324 | } | |
325 | ||
326 | /** | |
327 | * bind - associate or disassocate TIPC name(s) with a socket | |
328 | * @sock: socket structure | |
329 | * @uaddr: socket address describing name(s) and desired operation | |
330 | * @uaddr_len: size of socket address data structure | |
c4307285 | 331 | * |
b97bf3fd PL |
332 | * Name and name sequence binding is indicated using a positive scope value; |
333 | * a negative scope value unbinds the specified name. Specifying no name | |
334 | * (i.e. a socket address length of 0) unbinds all names from the socket. | |
c4307285 | 335 | * |
b97bf3fd | 336 | * Returns 0 on success, errno otherwise |
0c3141e9 AS |
337 | * |
338 | * NOTE: This routine doesn't need to take the socket lock since it doesn't | |
339 | * access any non-constant socket information. | |
b97bf3fd PL |
340 | */ |
341 | ||
342 | static int bind(struct socket *sock, struct sockaddr *uaddr, int uaddr_len) | |
343 | { | |
b97bf3fd | 344 | struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr; |
0c3141e9 | 345 | u32 portref = tipc_sk_port(sock->sk)->ref; |
b97bf3fd | 346 | |
0c3141e9 AS |
347 | if (unlikely(!uaddr_len)) |
348 | return tipc_withdraw(portref, 0, NULL); | |
c4307285 | 349 | |
0c3141e9 AS |
350 | if (uaddr_len < sizeof(struct sockaddr_tipc)) |
351 | return -EINVAL; | |
352 | if (addr->family != AF_TIPC) | |
353 | return -EAFNOSUPPORT; | |
b97bf3fd | 354 | |
b97bf3fd PL |
355 | if (addr->addrtype == TIPC_ADDR_NAME) |
356 | addr->addr.nameseq.upper = addr->addr.nameseq.lower; | |
0c3141e9 AS |
357 | else if (addr->addrtype != TIPC_ADDR_NAMESEQ) |
358 | return -EAFNOSUPPORT; | |
c4307285 | 359 | |
0c3141e9 AS |
360 | return (addr->scope > 0) ? |
361 | tipc_publish(portref, addr->scope, &addr->addr.nameseq) : | |
362 | tipc_withdraw(portref, -addr->scope, &addr->addr.nameseq); | |
b97bf3fd PL |
363 | } |
364 | ||
c4307285 | 365 | /** |
b97bf3fd PL |
366 | * get_name - get port ID of socket or peer socket |
367 | * @sock: socket structure | |
368 | * @uaddr: area for returned socket address | |
369 | * @uaddr_len: area for returned length of socket address | |
2da59918 | 370 | * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID |
c4307285 | 371 | * |
b97bf3fd | 372 | * Returns 0 on success, errno otherwise |
0c3141e9 | 373 | * |
2da59918 AS |
374 | * NOTE: This routine doesn't need to take the socket lock since it only |
375 | * accesses socket information that is unchanging (or which changes in | |
0e65967e | 376 | * a completely predictable manner). |
b97bf3fd PL |
377 | */ |
378 | ||
c4307285 | 379 | static int get_name(struct socket *sock, struct sockaddr *uaddr, |
b97bf3fd PL |
380 | int *uaddr_len, int peer) |
381 | { | |
b97bf3fd | 382 | struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr; |
2da59918 | 383 | struct tipc_sock *tsock = tipc_sk(sock->sk); |
b97bf3fd | 384 | |
88f8a5e3 | 385 | memset(addr, 0, sizeof(*addr)); |
0c3141e9 | 386 | if (peer) { |
2da59918 AS |
387 | if ((sock->state != SS_CONNECTED) && |
388 | ((peer != 2) || (sock->state != SS_DISCONNECTING))) | |
389 | return -ENOTCONN; | |
390 | addr->addr.id.ref = tsock->peer_name.ref; | |
391 | addr->addr.id.node = tsock->peer_name.node; | |
0c3141e9 | 392 | } else { |
b924dcf0 AS |
393 | addr->addr.id.ref = tsock->p->ref; |
394 | addr->addr.id.node = tipc_own_addr; | |
0c3141e9 | 395 | } |
b97bf3fd PL |
396 | |
397 | *uaddr_len = sizeof(*addr); | |
398 | addr->addrtype = TIPC_ADDR_ID; | |
399 | addr->family = AF_TIPC; | |
400 | addr->scope = 0; | |
b97bf3fd PL |
401 | addr->addr.name.domain = 0; |
402 | ||
0c3141e9 | 403 | return 0; |
b97bf3fd PL |
404 | } |
405 | ||
406 | /** | |
407 | * poll - read and possibly block on pollmask | |
408 | * @file: file structure associated with the socket | |
409 | * @sock: socket for which to calculate the poll bits | |
410 | * @wait: ??? | |
411 | * | |
9b674e82 AS |
412 | * Returns pollmask value |
413 | * | |
414 | * COMMENTARY: | |
415 | * It appears that the usual socket locking mechanisms are not useful here | |
416 | * since the pollmask info is potentially out-of-date the moment this routine | |
417 | * exits. TCP and other protocols seem to rely on higher level poll routines | |
418 | * to handle any preventable race conditions, so TIPC will do the same ... | |
419 | * | |
420 | * TIPC sets the returned events as follows: | |
f662c070 AS |
421 | * |
422 | * socket state flags set | |
423 | * ------------ --------- | |
424 | * unconnected no read flags | |
425 | * no write flags | |
426 | * | |
427 | * connecting POLLIN/POLLRDNORM if ACK/NACK in rx queue | |
428 | * no write flags | |
429 | * | |
430 | * connected POLLIN/POLLRDNORM if data in rx queue | |
431 | * POLLOUT if port is not congested | |
432 | * | |
433 | * disconnecting POLLIN/POLLRDNORM/POLLHUP | |
434 | * no write flags | |
435 | * | |
436 | * listening POLLIN if SYN in rx queue | |
437 | * no write flags | |
438 | * | |
439 | * ready POLLIN/POLLRDNORM if data in rx queue | |
440 | * [connectionless] POLLOUT (since port cannot be congested) | |
441 | * | |
442 | * IMPORTANT: The fact that a read or write operation is indicated does NOT | |
443 | * imply that the operation will succeed, merely that it should be performed | |
444 | * and will not block. | |
b97bf3fd PL |
445 | */ |
446 | ||
c4307285 | 447 | static unsigned int poll(struct file *file, struct socket *sock, |
b97bf3fd PL |
448 | poll_table *wait) |
449 | { | |
9b674e82 | 450 | struct sock *sk = sock->sk; |
f662c070 | 451 | u32 mask = 0; |
9b674e82 | 452 | |
aa395145 | 453 | poll_wait(file, sk_sleep(sk), wait); |
9b674e82 | 454 | |
f662c070 AS |
455 | switch ((int)sock->state) { |
456 | case SS_READY: | |
457 | case SS_CONNECTED: | |
458 | if (!tipc_sk_port(sk)->congested) | |
459 | mask |= POLLOUT; | |
460 | /* fall thru' */ | |
461 | case SS_CONNECTING: | |
462 | case SS_LISTENING: | |
463 | if (!skb_queue_empty(&sk->sk_receive_queue)) | |
464 | mask |= (POLLIN | POLLRDNORM); | |
465 | break; | |
466 | case SS_DISCONNECTING: | |
467 | mask = (POLLIN | POLLRDNORM | POLLHUP); | |
468 | break; | |
469 | } | |
9b674e82 AS |
470 | |
471 | return mask; | |
b97bf3fd PL |
472 | } |
473 | ||
c4307285 | 474 | /** |
b97bf3fd PL |
475 | * dest_name_check - verify user is permitted to send to specified port name |
476 | * @dest: destination address | |
477 | * @m: descriptor for message to be sent | |
c4307285 | 478 | * |
b97bf3fd PL |
479 | * Prevents restricted configuration commands from being issued by |
480 | * unauthorized users. | |
c4307285 | 481 | * |
b97bf3fd PL |
482 | * Returns 0 if permission is granted, otherwise errno |
483 | */ | |
484 | ||
05790c64 | 485 | static int dest_name_check(struct sockaddr_tipc *dest, struct msghdr *m) |
b97bf3fd PL |
486 | { |
487 | struct tipc_cfg_msg_hdr hdr; | |
488 | ||
c4307285 YH |
489 | if (likely(dest->addr.name.name.type >= TIPC_RESERVED_TYPES)) |
490 | return 0; | |
491 | if (likely(dest->addr.name.name.type == TIPC_TOP_SRV)) | |
492 | return 0; | |
c4307285 YH |
493 | if (likely(dest->addr.name.name.type != TIPC_CFG_SRV)) |
494 | return -EACCES; | |
b97bf3fd | 495 | |
c4307285 | 496 | if (copy_from_user(&hdr, m->msg_iov[0].iov_base, sizeof(hdr))) |
b97bf3fd | 497 | return -EFAULT; |
70cb2347 | 498 | if ((ntohs(hdr.tcm_type) & 0xC000) && (!capable(CAP_NET_ADMIN))) |
b97bf3fd | 499 | return -EACCES; |
c4307285 | 500 | |
b97bf3fd PL |
501 | return 0; |
502 | } | |
503 | ||
504 | /** | |
505 | * send_msg - send message in connectionless manner | |
0c3141e9 | 506 | * @iocb: if NULL, indicates that socket lock is already held |
b97bf3fd PL |
507 | * @sock: socket structure |
508 | * @m: message to send | |
e9024f0f | 509 | * @total_len: length of message |
c4307285 | 510 | * |
b97bf3fd | 511 | * Message must have an destination specified explicitly. |
c4307285 | 512 | * Used for SOCK_RDM and SOCK_DGRAM messages, |
b97bf3fd PL |
513 | * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections. |
514 | * (Note: 'SYN+' is prohibited on SOCK_STREAM.) | |
c4307285 | 515 | * |
b97bf3fd PL |
516 | * Returns the number of bytes sent on success, or errno otherwise |
517 | */ | |
518 | ||
519 | static int send_msg(struct kiocb *iocb, struct socket *sock, | |
520 | struct msghdr *m, size_t total_len) | |
521 | { | |
0c3141e9 AS |
522 | struct sock *sk = sock->sk; |
523 | struct tipc_port *tport = tipc_sk_port(sk); | |
c4307285 | 524 | struct sockaddr_tipc *dest = (struct sockaddr_tipc *)m->msg_name; |
b97bf3fd PL |
525 | int needs_conn; |
526 | int res = -EINVAL; | |
527 | ||
528 | if (unlikely(!dest)) | |
529 | return -EDESTADDRREQ; | |
51f9cc1f AS |
530 | if (unlikely((m->msg_namelen < sizeof(*dest)) || |
531 | (dest->family != AF_TIPC))) | |
b97bf3fd PL |
532 | return -EINVAL; |
533 | ||
0c3141e9 AS |
534 | if (iocb) |
535 | lock_sock(sk); | |
536 | ||
b97bf3fd PL |
537 | needs_conn = (sock->state != SS_READY); |
538 | if (unlikely(needs_conn)) { | |
0c3141e9 AS |
539 | if (sock->state == SS_LISTENING) { |
540 | res = -EPIPE; | |
541 | goto exit; | |
542 | } | |
543 | if (sock->state != SS_UNCONNECTED) { | |
544 | res = -EISCONN; | |
545 | goto exit; | |
546 | } | |
547 | if ((tport->published) || | |
548 | ((sock->type == SOCK_STREAM) && (total_len != 0))) { | |
549 | res = -EOPNOTSUPP; | |
550 | goto exit; | |
551 | } | |
3388007b | 552 | if (dest->addrtype == TIPC_ADDR_NAME) { |
0c3141e9 AS |
553 | tport->conn_type = dest->addr.name.name.type; |
554 | tport->conn_instance = dest->addr.name.name.instance; | |
3388007b | 555 | } |
b97bf3fd PL |
556 | |
557 | /* Abort any pending connection attempts (very unlikely) */ | |
558 | ||
0c3141e9 | 559 | reject_rx_queue(sk); |
b97bf3fd PL |
560 | } |
561 | ||
c4307285 YH |
562 | do { |
563 | if (dest->addrtype == TIPC_ADDR_NAME) { | |
2db9983a AS |
564 | res = dest_name_check(dest, m); |
565 | if (res) | |
0c3141e9 AS |
566 | break; |
567 | res = tipc_send2name(tport->ref, | |
c4307285 YH |
568 | &dest->addr.name.name, |
569 | dest->addr.name.domain, | |
570 | m->msg_iovlen, | |
571 | m->msg_iov); | |
0e65967e | 572 | } else if (dest->addrtype == TIPC_ADDR_ID) { |
0c3141e9 | 573 | res = tipc_send2port(tport->ref, |
c4307285 YH |
574 | &dest->addr.id, |
575 | m->msg_iovlen, | |
576 | m->msg_iov); | |
0e65967e | 577 | } else if (dest->addrtype == TIPC_ADDR_MCAST) { |
b97bf3fd PL |
578 | if (needs_conn) { |
579 | res = -EOPNOTSUPP; | |
0c3141e9 | 580 | break; |
b97bf3fd | 581 | } |
2db9983a AS |
582 | res = dest_name_check(dest, m); |
583 | if (res) | |
0c3141e9 AS |
584 | break; |
585 | res = tipc_multicast(tport->ref, | |
c4307285 | 586 | &dest->addr.nameseq, |
c4307285 YH |
587 | m->msg_iovlen, |
588 | m->msg_iov); | |
589 | } | |
590 | if (likely(res != -ELINKCONG)) { | |
a016892c | 591 | if (needs_conn && (res >= 0)) |
0c3141e9 | 592 | sock->state = SS_CONNECTING; |
0c3141e9 | 593 | break; |
c4307285 | 594 | } |
b97bf3fd PL |
595 | if (m->msg_flags & MSG_DONTWAIT) { |
596 | res = -EWOULDBLOCK; | |
0c3141e9 | 597 | break; |
c4307285 | 598 | } |
0c3141e9 | 599 | release_sock(sk); |
aa395145 | 600 | res = wait_event_interruptible(*sk_sleep(sk), |
0c3141e9 AS |
601 | !tport->congested); |
602 | lock_sock(sk); | |
603 | if (res) | |
604 | break; | |
c4307285 | 605 | } while (1); |
0c3141e9 AS |
606 | |
607 | exit: | |
608 | if (iocb) | |
609 | release_sock(sk); | |
610 | return res; | |
b97bf3fd PL |
611 | } |
612 | ||
c4307285 | 613 | /** |
b97bf3fd | 614 | * send_packet - send a connection-oriented message |
0c3141e9 | 615 | * @iocb: if NULL, indicates that socket lock is already held |
b97bf3fd PL |
616 | * @sock: socket structure |
617 | * @m: message to send | |
e9024f0f | 618 | * @total_len: length of message |
c4307285 | 619 | * |
b97bf3fd | 620 | * Used for SOCK_SEQPACKET messages and SOCK_STREAM data. |
c4307285 | 621 | * |
b97bf3fd PL |
622 | * Returns the number of bytes sent on success, or errno otherwise |
623 | */ | |
624 | ||
625 | static int send_packet(struct kiocb *iocb, struct socket *sock, | |
626 | struct msghdr *m, size_t total_len) | |
627 | { | |
0c3141e9 AS |
628 | struct sock *sk = sock->sk; |
629 | struct tipc_port *tport = tipc_sk_port(sk); | |
c4307285 | 630 | struct sockaddr_tipc *dest = (struct sockaddr_tipc *)m->msg_name; |
b97bf3fd PL |
631 | int res; |
632 | ||
633 | /* Handle implied connection establishment */ | |
634 | ||
635 | if (unlikely(dest)) | |
636 | return send_msg(iocb, sock, m, total_len); | |
637 | ||
0c3141e9 AS |
638 | if (iocb) |
639 | lock_sock(sk); | |
b97bf3fd | 640 | |
c4307285 | 641 | do { |
bdd94789 AS |
642 | if (unlikely(sock->state != SS_CONNECTED)) { |
643 | if (sock->state == SS_DISCONNECTING) | |
c4307285 | 644 | res = -EPIPE; |
bdd94789 AS |
645 | else |
646 | res = -ENOTCONN; | |
0c3141e9 | 647 | break; |
bdd94789 AS |
648 | } |
649 | ||
0c3141e9 | 650 | res = tipc_send(tport->ref, m->msg_iovlen, m->msg_iov); |
a016892c | 651 | if (likely(res != -ELINKCONG)) |
0c3141e9 | 652 | break; |
b97bf3fd PL |
653 | if (m->msg_flags & MSG_DONTWAIT) { |
654 | res = -EWOULDBLOCK; | |
0c3141e9 | 655 | break; |
c4307285 | 656 | } |
0c3141e9 | 657 | release_sock(sk); |
aa395145 | 658 | res = wait_event_interruptible(*sk_sleep(sk), |
0c3141e9 AS |
659 | (!tport->congested || !tport->connected)); |
660 | lock_sock(sk); | |
661 | if (res) | |
662 | break; | |
c4307285 | 663 | } while (1); |
0c3141e9 AS |
664 | |
665 | if (iocb) | |
666 | release_sock(sk); | |
667 | return res; | |
b97bf3fd PL |
668 | } |
669 | ||
c4307285 | 670 | /** |
b97bf3fd PL |
671 | * send_stream - send stream-oriented data |
672 | * @iocb: (unused) | |
673 | * @sock: socket structure | |
674 | * @m: data to send | |
675 | * @total_len: total length of data to be sent | |
c4307285 | 676 | * |
b97bf3fd | 677 | * Used for SOCK_STREAM data. |
c4307285 YH |
678 | * |
679 | * Returns the number of bytes sent on success (or partial success), | |
1303e8f1 | 680 | * or errno if no data sent |
b97bf3fd PL |
681 | */ |
682 | ||
b97bf3fd PL |
683 | static int send_stream(struct kiocb *iocb, struct socket *sock, |
684 | struct msghdr *m, size_t total_len) | |
685 | { | |
0c3141e9 AS |
686 | struct sock *sk = sock->sk; |
687 | struct tipc_port *tport = tipc_sk_port(sk); | |
b97bf3fd PL |
688 | struct msghdr my_msg; |
689 | struct iovec my_iov; | |
690 | struct iovec *curr_iov; | |
691 | int curr_iovlen; | |
692 | char __user *curr_start; | |
05646c91 | 693 | u32 hdr_size; |
b97bf3fd PL |
694 | int curr_left; |
695 | int bytes_to_send; | |
1303e8f1 | 696 | int bytes_sent; |
b97bf3fd | 697 | int res; |
c4307285 | 698 | |
0c3141e9 AS |
699 | lock_sock(sk); |
700 | ||
05646c91 | 701 | /* Handle special cases where there is no connection */ |
b97bf3fd | 702 | |
c4307285 | 703 | if (unlikely(sock->state != SS_CONNECTED)) { |
0c3141e9 AS |
704 | if (sock->state == SS_UNCONNECTED) { |
705 | res = send_packet(NULL, sock, m, total_len); | |
706 | goto exit; | |
707 | } else if (sock->state == SS_DISCONNECTING) { | |
708 | res = -EPIPE; | |
709 | goto exit; | |
710 | } else { | |
711 | res = -ENOTCONN; | |
712 | goto exit; | |
713 | } | |
c4307285 | 714 | } |
b97bf3fd | 715 | |
0c3141e9 AS |
716 | if (unlikely(m->msg_name)) { |
717 | res = -EISCONN; | |
718 | goto exit; | |
719 | } | |
eb5959c2 | 720 | |
c4307285 | 721 | /* |
b97bf3fd PL |
722 | * Send each iovec entry using one or more messages |
723 | * | |
c4307285 | 724 | * Note: This algorithm is good for the most likely case |
b97bf3fd PL |
725 | * (i.e. one large iovec entry), but could be improved to pass sets |
726 | * of small iovec entries into send_packet(). | |
727 | */ | |
728 | ||
1303e8f1 AS |
729 | curr_iov = m->msg_iov; |
730 | curr_iovlen = m->msg_iovlen; | |
b97bf3fd PL |
731 | my_msg.msg_iov = &my_iov; |
732 | my_msg.msg_iovlen = 1; | |
eb5959c2 AS |
733 | my_msg.msg_flags = m->msg_flags; |
734 | my_msg.msg_name = NULL; | |
1303e8f1 | 735 | bytes_sent = 0; |
b97bf3fd | 736 | |
05646c91 AS |
737 | hdr_size = msg_hdr_sz(&tport->phdr); |
738 | ||
b97bf3fd PL |
739 | while (curr_iovlen--) { |
740 | curr_start = curr_iov->iov_base; | |
741 | curr_left = curr_iov->iov_len; | |
742 | ||
743 | while (curr_left) { | |
05646c91 AS |
744 | bytes_to_send = tport->max_pkt - hdr_size; |
745 | if (bytes_to_send > TIPC_MAX_USER_MSG_SIZE) | |
746 | bytes_to_send = TIPC_MAX_USER_MSG_SIZE; | |
747 | if (curr_left < bytes_to_send) | |
748 | bytes_to_send = curr_left; | |
b97bf3fd PL |
749 | my_iov.iov_base = curr_start; |
750 | my_iov.iov_len = bytes_to_send; | |
2db9983a AS |
751 | res = send_packet(NULL, sock, &my_msg, 0); |
752 | if (res < 0) { | |
0c3141e9 | 753 | if (bytes_sent) |
05646c91 | 754 | res = bytes_sent; |
0c3141e9 | 755 | goto exit; |
1303e8f1 | 756 | } |
b97bf3fd PL |
757 | curr_left -= bytes_to_send; |
758 | curr_start += bytes_to_send; | |
1303e8f1 | 759 | bytes_sent += bytes_to_send; |
b97bf3fd PL |
760 | } |
761 | ||
762 | curr_iov++; | |
763 | } | |
0c3141e9 AS |
764 | res = bytes_sent; |
765 | exit: | |
766 | release_sock(sk); | |
767 | return res; | |
b97bf3fd PL |
768 | } |
769 | ||
770 | /** | |
771 | * auto_connect - complete connection setup to a remote port | |
772 | * @sock: socket structure | |
b97bf3fd | 773 | * @msg: peer's response message |
c4307285 | 774 | * |
b97bf3fd PL |
775 | * Returns 0 on success, errno otherwise |
776 | */ | |
777 | ||
0c3141e9 | 778 | static int auto_connect(struct socket *sock, struct tipc_msg *msg) |
b97bf3fd | 779 | { |
2da59918 | 780 | struct tipc_sock *tsock = tipc_sk(sock->sk); |
b97bf3fd PL |
781 | |
782 | if (msg_errcode(msg)) { | |
783 | sock->state = SS_DISCONNECTING; | |
784 | return -ECONNREFUSED; | |
785 | } | |
786 | ||
2da59918 AS |
787 | tsock->peer_name.ref = msg_origport(msg); |
788 | tsock->peer_name.node = msg_orignode(msg); | |
789 | tipc_connect2port(tsock->p->ref, &tsock->peer_name); | |
790 | tipc_set_portimportance(tsock->p->ref, msg_importance(msg)); | |
b97bf3fd PL |
791 | sock->state = SS_CONNECTED; |
792 | return 0; | |
793 | } | |
794 | ||
795 | /** | |
796 | * set_orig_addr - capture sender's address for received message | |
797 | * @m: descriptor for message info | |
798 | * @msg: received message header | |
c4307285 | 799 | * |
b97bf3fd PL |
800 | * Note: Address is not captured if not requested by receiver. |
801 | */ | |
802 | ||
05790c64 | 803 | static void set_orig_addr(struct msghdr *m, struct tipc_msg *msg) |
b97bf3fd | 804 | { |
c4307285 | 805 | struct sockaddr_tipc *addr = (struct sockaddr_tipc *)m->msg_name; |
b97bf3fd | 806 | |
c4307285 | 807 | if (addr) { |
b97bf3fd PL |
808 | addr->family = AF_TIPC; |
809 | addr->addrtype = TIPC_ADDR_ID; | |
810 | addr->addr.id.ref = msg_origport(msg); | |
811 | addr->addr.id.node = msg_orignode(msg); | |
0e65967e AS |
812 | addr->addr.name.domain = 0; /* could leave uninitialized */ |
813 | addr->scope = 0; /* could leave uninitialized */ | |
b97bf3fd PL |
814 | m->msg_namelen = sizeof(struct sockaddr_tipc); |
815 | } | |
816 | } | |
817 | ||
818 | /** | |
c4307285 | 819 | * anc_data_recv - optionally capture ancillary data for received message |
b97bf3fd PL |
820 | * @m: descriptor for message info |
821 | * @msg: received message header | |
822 | * @tport: TIPC port associated with message | |
c4307285 | 823 | * |
b97bf3fd | 824 | * Note: Ancillary data is not captured if not requested by receiver. |
c4307285 | 825 | * |
b97bf3fd PL |
826 | * Returns 0 if successful, otherwise errno |
827 | */ | |
828 | ||
05790c64 | 829 | static int anc_data_recv(struct msghdr *m, struct tipc_msg *msg, |
b97bf3fd PL |
830 | struct tipc_port *tport) |
831 | { | |
832 | u32 anc_data[3]; | |
833 | u32 err; | |
834 | u32 dest_type; | |
3546c750 | 835 | int has_name; |
b97bf3fd PL |
836 | int res; |
837 | ||
838 | if (likely(m->msg_controllen == 0)) | |
839 | return 0; | |
840 | ||
841 | /* Optionally capture errored message object(s) */ | |
842 | ||
843 | err = msg ? msg_errcode(msg) : 0; | |
844 | if (unlikely(err)) { | |
845 | anc_data[0] = err; | |
846 | anc_data[1] = msg_data_sz(msg); | |
2db9983a AS |
847 | res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data); |
848 | if (res) | |
b97bf3fd | 849 | return res; |
2db9983a AS |
850 | if (anc_data[1]) { |
851 | res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1], | |
852 | msg_data(msg)); | |
853 | if (res) | |
854 | return res; | |
855 | } | |
b97bf3fd PL |
856 | } |
857 | ||
858 | /* Optionally capture message destination object */ | |
859 | ||
860 | dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG; | |
861 | switch (dest_type) { | |
862 | case TIPC_NAMED_MSG: | |
3546c750 | 863 | has_name = 1; |
b97bf3fd PL |
864 | anc_data[0] = msg_nametype(msg); |
865 | anc_data[1] = msg_namelower(msg); | |
866 | anc_data[2] = msg_namelower(msg); | |
867 | break; | |
868 | case TIPC_MCAST_MSG: | |
3546c750 | 869 | has_name = 1; |
b97bf3fd PL |
870 | anc_data[0] = msg_nametype(msg); |
871 | anc_data[1] = msg_namelower(msg); | |
872 | anc_data[2] = msg_nameupper(msg); | |
873 | break; | |
874 | case TIPC_CONN_MSG: | |
3546c750 | 875 | has_name = (tport->conn_type != 0); |
b97bf3fd PL |
876 | anc_data[0] = tport->conn_type; |
877 | anc_data[1] = tport->conn_instance; | |
878 | anc_data[2] = tport->conn_instance; | |
879 | break; | |
880 | default: | |
3546c750 | 881 | has_name = 0; |
b97bf3fd | 882 | } |
2db9983a AS |
883 | if (has_name) { |
884 | res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data); | |
885 | if (res) | |
886 | return res; | |
887 | } | |
b97bf3fd PL |
888 | |
889 | return 0; | |
890 | } | |
891 | ||
c4307285 | 892 | /** |
b97bf3fd PL |
893 | * recv_msg - receive packet-oriented message |
894 | * @iocb: (unused) | |
895 | * @m: descriptor for message info | |
896 | * @buf_len: total size of user buffer area | |
897 | * @flags: receive flags | |
c4307285 | 898 | * |
b97bf3fd PL |
899 | * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages. |
900 | * If the complete message doesn't fit in user area, truncate it. | |
901 | * | |
902 | * Returns size of returned message data, errno otherwise | |
903 | */ | |
904 | ||
905 | static int recv_msg(struct kiocb *iocb, struct socket *sock, | |
906 | struct msghdr *m, size_t buf_len, int flags) | |
907 | { | |
0c3141e9 AS |
908 | struct sock *sk = sock->sk; |
909 | struct tipc_port *tport = tipc_sk_port(sk); | |
b97bf3fd PL |
910 | struct sk_buff *buf; |
911 | struct tipc_msg *msg; | |
b97bf3fd PL |
912 | unsigned int sz; |
913 | u32 err; | |
914 | int res; | |
915 | ||
0c3141e9 | 916 | /* Catch invalid receive requests */ |
b97bf3fd PL |
917 | |
918 | if (m->msg_iovlen != 1) | |
0c3141e9 | 919 | return -EOPNOTSUPP; /* Don't do multiple iovec entries yet */ |
b97bf3fd PL |
920 | |
921 | if (unlikely(!buf_len)) | |
922 | return -EINVAL; | |
923 | ||
0c3141e9 | 924 | lock_sock(sk); |
b97bf3fd | 925 | |
0c3141e9 AS |
926 | if (unlikely(sock->state == SS_UNCONNECTED)) { |
927 | res = -ENOTCONN; | |
b97bf3fd PL |
928 | goto exit; |
929 | } | |
930 | ||
0c3141e9 | 931 | restart: |
b97bf3fd | 932 | |
0c3141e9 | 933 | /* Look for a message in receive queue; wait if necessary */ |
b97bf3fd | 934 | |
0c3141e9 AS |
935 | while (skb_queue_empty(&sk->sk_receive_queue)) { |
936 | if (sock->state == SS_DISCONNECTING) { | |
937 | res = -ENOTCONN; | |
938 | goto exit; | |
939 | } | |
940 | if (flags & MSG_DONTWAIT) { | |
941 | res = -EWOULDBLOCK; | |
942 | goto exit; | |
943 | } | |
944 | release_sock(sk); | |
aa395145 | 945 | res = wait_event_interruptible(*sk_sleep(sk), |
0c3141e9 AS |
946 | (!skb_queue_empty(&sk->sk_receive_queue) || |
947 | (sock->state == SS_DISCONNECTING))); | |
948 | lock_sock(sk); | |
949 | if (res) | |
950 | goto exit; | |
b97bf3fd PL |
951 | } |
952 | ||
0c3141e9 | 953 | /* Look at first message in receive queue */ |
b97bf3fd | 954 | |
0c3141e9 | 955 | buf = skb_peek(&sk->sk_receive_queue); |
b97bf3fd PL |
956 | msg = buf_msg(buf); |
957 | sz = msg_data_sz(msg); | |
958 | err = msg_errcode(msg); | |
959 | ||
960 | /* Complete connection setup for an implied connect */ | |
961 | ||
962 | if (unlikely(sock->state == SS_CONNECTING)) { | |
0c3141e9 AS |
963 | res = auto_connect(sock, msg); |
964 | if (res) | |
b97bf3fd PL |
965 | goto exit; |
966 | } | |
967 | ||
968 | /* Discard an empty non-errored message & try again */ | |
969 | ||
970 | if ((!sz) && (!err)) { | |
0c3141e9 | 971 | advance_rx_queue(sk); |
b97bf3fd PL |
972 | goto restart; |
973 | } | |
974 | ||
975 | /* Capture sender's address (optional) */ | |
976 | ||
977 | set_orig_addr(m, msg); | |
978 | ||
979 | /* Capture ancillary data (optional) */ | |
980 | ||
0c3141e9 AS |
981 | res = anc_data_recv(m, msg, tport); |
982 | if (res) | |
b97bf3fd PL |
983 | goto exit; |
984 | ||
985 | /* Capture message data (if valid) & compute return value (always) */ | |
c4307285 | 986 | |
b97bf3fd PL |
987 | if (!err) { |
988 | if (unlikely(buf_len < sz)) { | |
989 | sz = buf_len; | |
990 | m->msg_flags |= MSG_TRUNC; | |
991 | } | |
992 | if (unlikely(copy_to_user(m->msg_iov->iov_base, msg_data(msg), | |
993 | sz))) { | |
994 | res = -EFAULT; | |
995 | goto exit; | |
996 | } | |
997 | res = sz; | |
998 | } else { | |
999 | if ((sock->state == SS_READY) || | |
1000 | ((err == TIPC_CONN_SHUTDOWN) || m->msg_control)) | |
1001 | res = 0; | |
1002 | else | |
1003 | res = -ECONNRESET; | |
1004 | } | |
1005 | ||
1006 | /* Consume received message (optional) */ | |
1007 | ||
1008 | if (likely(!(flags & MSG_PEEK))) { | |
99009806 | 1009 | if ((sock->state != SS_READY) && |
0c3141e9 AS |
1010 | (++tport->conn_unacked >= TIPC_FLOW_CONTROL_WIN)) |
1011 | tipc_acknowledge(tport->ref, tport->conn_unacked); | |
1012 | advance_rx_queue(sk); | |
c4307285 | 1013 | } |
b97bf3fd | 1014 | exit: |
0c3141e9 | 1015 | release_sock(sk); |
b97bf3fd PL |
1016 | return res; |
1017 | } | |
1018 | ||
c4307285 | 1019 | /** |
b97bf3fd PL |
1020 | * recv_stream - receive stream-oriented data |
1021 | * @iocb: (unused) | |
1022 | * @m: descriptor for message info | |
1023 | * @buf_len: total size of user buffer area | |
1024 | * @flags: receive flags | |
c4307285 YH |
1025 | * |
1026 | * Used for SOCK_STREAM messages only. If not enough data is available | |
b97bf3fd PL |
1027 | * will optionally wait for more; never truncates data. |
1028 | * | |
1029 | * Returns size of returned message data, errno otherwise | |
1030 | */ | |
1031 | ||
1032 | static int recv_stream(struct kiocb *iocb, struct socket *sock, | |
1033 | struct msghdr *m, size_t buf_len, int flags) | |
1034 | { | |
0c3141e9 AS |
1035 | struct sock *sk = sock->sk; |
1036 | struct tipc_port *tport = tipc_sk_port(sk); | |
b97bf3fd PL |
1037 | struct sk_buff *buf; |
1038 | struct tipc_msg *msg; | |
b97bf3fd | 1039 | unsigned int sz; |
3720d40b | 1040 | int sz_to_copy, target, needed; |
b97bf3fd | 1041 | int sz_copied = 0; |
28c4dadd | 1042 | char __user *crs = m->msg_iov->iov_base; |
b97bf3fd PL |
1043 | unsigned char *buf_crs; |
1044 | u32 err; | |
0c3141e9 | 1045 | int res = 0; |
b97bf3fd | 1046 | |
0c3141e9 | 1047 | /* Catch invalid receive attempts */ |
b97bf3fd PL |
1048 | |
1049 | if (m->msg_iovlen != 1) | |
0c3141e9 | 1050 | return -EOPNOTSUPP; /* Don't do multiple iovec entries yet */ |
b97bf3fd PL |
1051 | |
1052 | if (unlikely(!buf_len)) | |
1053 | return -EINVAL; | |
1054 | ||
0c3141e9 | 1055 | lock_sock(sk); |
b97bf3fd | 1056 | |
0c3141e9 AS |
1057 | if (unlikely((sock->state == SS_UNCONNECTED) || |
1058 | (sock->state == SS_CONNECTING))) { | |
1059 | res = -ENOTCONN; | |
b97bf3fd PL |
1060 | goto exit; |
1061 | } | |
1062 | ||
3720d40b FW |
1063 | target = sock_rcvlowat(sk, flags & MSG_WAITALL, buf_len); |
1064 | ||
0c3141e9 | 1065 | restart: |
b97bf3fd | 1066 | |
0c3141e9 | 1067 | /* Look for a message in receive queue; wait if necessary */ |
b97bf3fd | 1068 | |
0c3141e9 AS |
1069 | while (skb_queue_empty(&sk->sk_receive_queue)) { |
1070 | if (sock->state == SS_DISCONNECTING) { | |
1071 | res = -ENOTCONN; | |
1072 | goto exit; | |
1073 | } | |
1074 | if (flags & MSG_DONTWAIT) { | |
1075 | res = -EWOULDBLOCK; | |
1076 | goto exit; | |
1077 | } | |
1078 | release_sock(sk); | |
aa395145 | 1079 | res = wait_event_interruptible(*sk_sleep(sk), |
0c3141e9 AS |
1080 | (!skb_queue_empty(&sk->sk_receive_queue) || |
1081 | (sock->state == SS_DISCONNECTING))); | |
1082 | lock_sock(sk); | |
1083 | if (res) | |
1084 | goto exit; | |
b97bf3fd PL |
1085 | } |
1086 | ||
0c3141e9 | 1087 | /* Look at first message in receive queue */ |
b97bf3fd | 1088 | |
0c3141e9 | 1089 | buf = skb_peek(&sk->sk_receive_queue); |
b97bf3fd PL |
1090 | msg = buf_msg(buf); |
1091 | sz = msg_data_sz(msg); | |
1092 | err = msg_errcode(msg); | |
1093 | ||
1094 | /* Discard an empty non-errored message & try again */ | |
1095 | ||
1096 | if ((!sz) && (!err)) { | |
0c3141e9 | 1097 | advance_rx_queue(sk); |
b97bf3fd PL |
1098 | goto restart; |
1099 | } | |
1100 | ||
1101 | /* Optionally capture sender's address & ancillary data of first msg */ | |
1102 | ||
1103 | if (sz_copied == 0) { | |
1104 | set_orig_addr(m, msg); | |
0c3141e9 AS |
1105 | res = anc_data_recv(m, msg, tport); |
1106 | if (res) | |
b97bf3fd PL |
1107 | goto exit; |
1108 | } | |
1109 | ||
1110 | /* Capture message data (if valid) & compute return value (always) */ | |
c4307285 | 1111 | |
b97bf3fd PL |
1112 | if (!err) { |
1113 | buf_crs = (unsigned char *)(TIPC_SKB_CB(buf)->handle); | |
7a8036c2 | 1114 | sz = (unsigned char *)msg + msg_size(msg) - buf_crs; |
b97bf3fd PL |
1115 | |
1116 | needed = (buf_len - sz_copied); | |
1117 | sz_to_copy = (sz <= needed) ? sz : needed; | |
1118 | if (unlikely(copy_to_user(crs, buf_crs, sz_to_copy))) { | |
1119 | res = -EFAULT; | |
1120 | goto exit; | |
1121 | } | |
1122 | sz_copied += sz_to_copy; | |
1123 | ||
1124 | if (sz_to_copy < sz) { | |
1125 | if (!(flags & MSG_PEEK)) | |
1126 | TIPC_SKB_CB(buf)->handle = buf_crs + sz_to_copy; | |
1127 | goto exit; | |
1128 | } | |
1129 | ||
1130 | crs += sz_to_copy; | |
1131 | } else { | |
1132 | if (sz_copied != 0) | |
1133 | goto exit; /* can't add error msg to valid data */ | |
1134 | ||
1135 | if ((err == TIPC_CONN_SHUTDOWN) || m->msg_control) | |
1136 | res = 0; | |
1137 | else | |
1138 | res = -ECONNRESET; | |
1139 | } | |
1140 | ||
1141 | /* Consume received message (optional) */ | |
1142 | ||
1143 | if (likely(!(flags & MSG_PEEK))) { | |
0c3141e9 AS |
1144 | if (unlikely(++tport->conn_unacked >= TIPC_FLOW_CONTROL_WIN)) |
1145 | tipc_acknowledge(tport->ref, tport->conn_unacked); | |
1146 | advance_rx_queue(sk); | |
c4307285 | 1147 | } |
b97bf3fd PL |
1148 | |
1149 | /* Loop around if more data is required */ | |
1150 | ||
f64f9e71 JP |
1151 | if ((sz_copied < buf_len) && /* didn't get all requested data */ |
1152 | (!skb_queue_empty(&sk->sk_receive_queue) || | |
3720d40b | 1153 | (sz_copied < target)) && /* and more is ready or required */ |
f64f9e71 JP |
1154 | (!(flags & MSG_PEEK)) && /* and aren't just peeking at data */ |
1155 | (!err)) /* and haven't reached a FIN */ | |
b97bf3fd PL |
1156 | goto restart; |
1157 | ||
1158 | exit: | |
0c3141e9 | 1159 | release_sock(sk); |
a3b0a5a9 | 1160 | return sz_copied ? sz_copied : res; |
b97bf3fd PL |
1161 | } |
1162 | ||
1163 | /** | |
1819b837 AS |
1164 | * rx_queue_full - determine if receive queue can accept another message |
1165 | * @msg: message to be added to queue | |
b97bf3fd PL |
1166 | * @queue_size: current size of queue |
1167 | * @base: nominal maximum size of queue | |
c4307285 | 1168 | * |
1819b837 | 1169 | * Returns 1 if queue is unable to accept message, 0 otherwise |
b97bf3fd PL |
1170 | */ |
1171 | ||
1819b837 | 1172 | static int rx_queue_full(struct tipc_msg *msg, u32 queue_size, u32 base) |
b97bf3fd PL |
1173 | { |
1174 | u32 threshold; | |
1175 | u32 imp = msg_importance(msg); | |
1176 | ||
1177 | if (imp == TIPC_LOW_IMPORTANCE) | |
1178 | threshold = base; | |
1179 | else if (imp == TIPC_MEDIUM_IMPORTANCE) | |
1180 | threshold = base * 2; | |
1181 | else if (imp == TIPC_HIGH_IMPORTANCE) | |
1182 | threshold = base * 100; | |
1183 | else | |
1184 | return 0; | |
1185 | ||
1186 | if (msg_connected(msg)) | |
1187 | threshold *= 4; | |
1188 | ||
a02cec21 | 1189 | return queue_size >= threshold; |
b97bf3fd PL |
1190 | } |
1191 | ||
c4307285 | 1192 | /** |
0c3141e9 AS |
1193 | * filter_rcv - validate incoming message |
1194 | * @sk: socket | |
b97bf3fd | 1195 | * @buf: message |
c4307285 | 1196 | * |
0c3141e9 AS |
1197 | * Enqueues message on receive queue if acceptable; optionally handles |
1198 | * disconnect indication for a connected socket. | |
1199 | * | |
1200 | * Called with socket lock already taken; port lock may also be taken. | |
c4307285 | 1201 | * |
b97bf3fd PL |
1202 | * Returns TIPC error status code (TIPC_OK if message is not to be rejected) |
1203 | */ | |
1204 | ||
0c3141e9 | 1205 | static u32 filter_rcv(struct sock *sk, struct sk_buff *buf) |
b97bf3fd | 1206 | { |
0c3141e9 | 1207 | struct socket *sock = sk->sk_socket; |
b97bf3fd | 1208 | struct tipc_msg *msg = buf_msg(buf); |
b97bf3fd PL |
1209 | u32 recv_q_len; |
1210 | ||
b97bf3fd PL |
1211 | /* Reject message if it is wrong sort of message for socket */ |
1212 | ||
1213 | /* | |
1214 | * WOULD IT BE BETTER TO JUST DISCARD THESE MESSAGES INSTEAD? | |
1215 | * "NO PORT" ISN'T REALLY THE RIGHT ERROR CODE, AND THERE MAY | |
1216 | * BE SECURITY IMPLICATIONS INHERENT IN REJECTING INVALID TRAFFIC | |
1217 | */ | |
0c3141e9 | 1218 | |
b97bf3fd | 1219 | if (sock->state == SS_READY) { |
b29f1428 | 1220 | if (msg_connected(msg)) |
b97bf3fd | 1221 | return TIPC_ERR_NO_PORT; |
b97bf3fd | 1222 | } else { |
b29f1428 | 1223 | if (msg_mcast(msg)) |
b97bf3fd | 1224 | return TIPC_ERR_NO_PORT; |
b97bf3fd | 1225 | if (sock->state == SS_CONNECTED) { |
b29f1428 | 1226 | if (!msg_connected(msg)) |
b97bf3fd | 1227 | return TIPC_ERR_NO_PORT; |
b29f1428 AS |
1228 | } else if (sock->state == SS_CONNECTING) { |
1229 | if (!msg_connected(msg) && (msg_errcode(msg) == 0)) | |
b97bf3fd | 1230 | return TIPC_ERR_NO_PORT; |
b29f1428 AS |
1231 | } else if (sock->state == SS_LISTENING) { |
1232 | if (msg_connected(msg) || msg_errcode(msg)) | |
b97bf3fd | 1233 | return TIPC_ERR_NO_PORT; |
b29f1428 | 1234 | } else if (sock->state == SS_DISCONNECTING) { |
b97bf3fd | 1235 | return TIPC_ERR_NO_PORT; |
b29f1428 AS |
1236 | } else /* (sock->state == SS_UNCONNECTED) */ { |
1237 | if (msg_connected(msg) || msg_errcode(msg)) | |
b97bf3fd | 1238 | return TIPC_ERR_NO_PORT; |
b97bf3fd PL |
1239 | } |
1240 | } | |
1241 | ||
1242 | /* Reject message if there isn't room to queue it */ | |
1243 | ||
1819b837 AS |
1244 | recv_q_len = (u32)atomic_read(&tipc_queue_size); |
1245 | if (unlikely(recv_q_len >= OVERLOAD_LIMIT_BASE)) { | |
1246 | if (rx_queue_full(msg, recv_q_len, OVERLOAD_LIMIT_BASE)) | |
b97bf3fd | 1247 | return TIPC_ERR_OVERLOAD; |
c4307285 | 1248 | } |
0c3141e9 | 1249 | recv_q_len = skb_queue_len(&sk->sk_receive_queue); |
1819b837 AS |
1250 | if (unlikely(recv_q_len >= (OVERLOAD_LIMIT_BASE / 2))) { |
1251 | if (rx_queue_full(msg, recv_q_len, OVERLOAD_LIMIT_BASE / 2)) | |
b97bf3fd | 1252 | return TIPC_ERR_OVERLOAD; |
c4307285 | 1253 | } |
b97bf3fd | 1254 | |
0c3141e9 AS |
1255 | /* Enqueue message (finally!) */ |
1256 | ||
0c3141e9 AS |
1257 | TIPC_SKB_CB(buf)->handle = msg_data(msg); |
1258 | atomic_inc(&tipc_queue_size); | |
1259 | __skb_queue_tail(&sk->sk_receive_queue, buf); | |
1260 | ||
b97bf3fd PL |
1261 | /* Initiate connection termination for an incoming 'FIN' */ |
1262 | ||
1263 | if (unlikely(msg_errcode(msg) && (sock->state == SS_CONNECTED))) { | |
1264 | sock->state = SS_DISCONNECTING; | |
0c3141e9 | 1265 | tipc_disconnect_port(tipc_sk_port(sk)); |
b97bf3fd PL |
1266 | } |
1267 | ||
aa395145 ED |
1268 | if (waitqueue_active(sk_sleep(sk))) |
1269 | wake_up_interruptible(sk_sleep(sk)); | |
0c3141e9 AS |
1270 | return TIPC_OK; |
1271 | } | |
b97bf3fd | 1272 | |
0c3141e9 AS |
1273 | /** |
1274 | * backlog_rcv - handle incoming message from backlog queue | |
1275 | * @sk: socket | |
1276 | * @buf: message | |
1277 | * | |
1278 | * Caller must hold socket lock, but not port lock. | |
1279 | * | |
1280 | * Returns 0 | |
1281 | */ | |
b97bf3fd | 1282 | |
0c3141e9 AS |
1283 | static int backlog_rcv(struct sock *sk, struct sk_buff *buf) |
1284 | { | |
1285 | u32 res; | |
1286 | ||
1287 | res = filter_rcv(sk, buf); | |
1288 | if (res) | |
1289 | tipc_reject_msg(buf, res); | |
1290 | return 0; | |
1291 | } | |
1292 | ||
1293 | /** | |
1294 | * dispatch - handle incoming message | |
1295 | * @tport: TIPC port that received message | |
1296 | * @buf: message | |
1297 | * | |
1298 | * Called with port lock already taken. | |
1299 | * | |
1300 | * Returns TIPC error status code (TIPC_OK if message is not to be rejected) | |
1301 | */ | |
1302 | ||
1303 | static u32 dispatch(struct tipc_port *tport, struct sk_buff *buf) | |
1304 | { | |
1305 | struct sock *sk = (struct sock *)tport->usr_handle; | |
1306 | u32 res; | |
1307 | ||
1308 | /* | |
1309 | * Process message if socket is unlocked; otherwise add to backlog queue | |
1310 | * | |
1311 | * This code is based on sk_receive_skb(), but must be distinct from it | |
1312 | * since a TIPC-specific filter/reject mechanism is utilized | |
1313 | */ | |
1314 | ||
1315 | bh_lock_sock(sk); | |
1316 | if (!sock_owned_by_user(sk)) { | |
1317 | res = filter_rcv(sk, buf); | |
1318 | } else { | |
a3a858ff | 1319 | if (sk_add_backlog(sk, buf)) |
53eecb1b ZY |
1320 | res = TIPC_ERR_OVERLOAD; |
1321 | else | |
1322 | res = TIPC_OK; | |
0c3141e9 AS |
1323 | } |
1324 | bh_unlock_sock(sk); | |
1325 | ||
1326 | return res; | |
b97bf3fd PL |
1327 | } |
1328 | ||
c4307285 | 1329 | /** |
b97bf3fd PL |
1330 | * wakeupdispatch - wake up port after congestion |
1331 | * @tport: port to wakeup | |
c4307285 | 1332 | * |
0c3141e9 | 1333 | * Called with port lock already taken. |
b97bf3fd PL |
1334 | */ |
1335 | ||
1336 | static void wakeupdispatch(struct tipc_port *tport) | |
1337 | { | |
0c3141e9 | 1338 | struct sock *sk = (struct sock *)tport->usr_handle; |
b97bf3fd | 1339 | |
aa395145 ED |
1340 | if (waitqueue_active(sk_sleep(sk))) |
1341 | wake_up_interruptible(sk_sleep(sk)); | |
b97bf3fd PL |
1342 | } |
1343 | ||
1344 | /** | |
1345 | * connect - establish a connection to another TIPC port | |
1346 | * @sock: socket structure | |
1347 | * @dest: socket address for destination port | |
1348 | * @destlen: size of socket address data structure | |
0c3141e9 | 1349 | * @flags: file-related flags associated with socket |
b97bf3fd PL |
1350 | * |
1351 | * Returns 0 on success, errno otherwise | |
1352 | */ | |
1353 | ||
c4307285 | 1354 | static int connect(struct socket *sock, struct sockaddr *dest, int destlen, |
b97bf3fd PL |
1355 | int flags) |
1356 | { | |
0c3141e9 | 1357 | struct sock *sk = sock->sk; |
b89741a0 AS |
1358 | struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest; |
1359 | struct msghdr m = {NULL,}; | |
1360 | struct sk_buff *buf; | |
1361 | struct tipc_msg *msg; | |
564e83b5 | 1362 | long timeout; |
b89741a0 AS |
1363 | int res; |
1364 | ||
0c3141e9 AS |
1365 | lock_sock(sk); |
1366 | ||
b89741a0 AS |
1367 | /* For now, TIPC does not allow use of connect() with DGRAM/RDM types */ |
1368 | ||
0c3141e9 AS |
1369 | if (sock->state == SS_READY) { |
1370 | res = -EOPNOTSUPP; | |
1371 | goto exit; | |
1372 | } | |
b89741a0 AS |
1373 | |
1374 | /* For now, TIPC does not support the non-blocking form of connect() */ | |
1375 | ||
0c3141e9 | 1376 | if (flags & O_NONBLOCK) { |
35997e31 | 1377 | res = -EOPNOTSUPP; |
0c3141e9 AS |
1378 | goto exit; |
1379 | } | |
b89741a0 AS |
1380 | |
1381 | /* Issue Posix-compliant error code if socket is in the wrong state */ | |
1382 | ||
0c3141e9 AS |
1383 | if (sock->state == SS_LISTENING) { |
1384 | res = -EOPNOTSUPP; | |
1385 | goto exit; | |
1386 | } | |
1387 | if (sock->state == SS_CONNECTING) { | |
1388 | res = -EALREADY; | |
1389 | goto exit; | |
1390 | } | |
1391 | if (sock->state != SS_UNCONNECTED) { | |
1392 | res = -EISCONN; | |
1393 | goto exit; | |
1394 | } | |
b89741a0 AS |
1395 | |
1396 | /* | |
1397 | * Reject connection attempt using multicast address | |
1398 | * | |
1399 | * Note: send_msg() validates the rest of the address fields, | |
1400 | * so there's no need to do it here | |
1401 | */ | |
1402 | ||
0c3141e9 AS |
1403 | if (dst->addrtype == TIPC_ADDR_MCAST) { |
1404 | res = -EINVAL; | |
1405 | goto exit; | |
1406 | } | |
1407 | ||
1408 | /* Reject any messages already in receive queue (very unlikely) */ | |
1409 | ||
1410 | reject_rx_queue(sk); | |
b89741a0 AS |
1411 | |
1412 | /* Send a 'SYN-' to destination */ | |
1413 | ||
1414 | m.msg_name = dest; | |
1415 | m.msg_namelen = destlen; | |
1416 | res = send_msg(NULL, sock, &m, 0); | |
a016892c | 1417 | if (res < 0) |
0c3141e9 | 1418 | goto exit; |
b89741a0 | 1419 | |
0c3141e9 | 1420 | /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */ |
b89741a0 | 1421 | |
564e83b5 | 1422 | timeout = tipc_sk(sk)->conn_timeout; |
0c3141e9 | 1423 | release_sock(sk); |
aa395145 | 1424 | res = wait_event_interruptible_timeout(*sk_sleep(sk), |
0c3141e9 AS |
1425 | (!skb_queue_empty(&sk->sk_receive_queue) || |
1426 | (sock->state != SS_CONNECTING)), | |
564e83b5 | 1427 | timeout ? timeout : MAX_SCHEDULE_TIMEOUT); |
0c3141e9 | 1428 | lock_sock(sk); |
b89741a0 | 1429 | |
b89741a0 | 1430 | if (res > 0) { |
0c3141e9 AS |
1431 | buf = skb_peek(&sk->sk_receive_queue); |
1432 | if (buf != NULL) { | |
1433 | msg = buf_msg(buf); | |
1434 | res = auto_connect(sock, msg); | |
1435 | if (!res) { | |
1436 | if (!msg_data_sz(msg)) | |
1437 | advance_rx_queue(sk); | |
1438 | } | |
1439 | } else { | |
a016892c | 1440 | if (sock->state == SS_CONNECTED) |
0c3141e9 | 1441 | res = -EISCONN; |
a016892c | 1442 | else |
0c3141e9 | 1443 | res = -ECONNREFUSED; |
b89741a0 AS |
1444 | } |
1445 | } else { | |
1446 | if (res == 0) | |
1447 | res = -ETIMEDOUT; | |
1448 | else | |
1449 | ; /* leave "res" unchanged */ | |
1450 | sock->state = SS_DISCONNECTING; | |
1451 | } | |
1452 | ||
0c3141e9 AS |
1453 | exit: |
1454 | release_sock(sk); | |
b89741a0 | 1455 | return res; |
b97bf3fd PL |
1456 | } |
1457 | ||
c4307285 | 1458 | /** |
b97bf3fd PL |
1459 | * listen - allow socket to listen for incoming connections |
1460 | * @sock: socket structure | |
1461 | * @len: (unused) | |
c4307285 | 1462 | * |
b97bf3fd PL |
1463 | * Returns 0 on success, errno otherwise |
1464 | */ | |
1465 | ||
1466 | static int listen(struct socket *sock, int len) | |
1467 | { | |
0c3141e9 AS |
1468 | struct sock *sk = sock->sk; |
1469 | int res; | |
1470 | ||
1471 | lock_sock(sk); | |
b97bf3fd PL |
1472 | |
1473 | if (sock->state == SS_READY) | |
0c3141e9 AS |
1474 | res = -EOPNOTSUPP; |
1475 | else if (sock->state != SS_UNCONNECTED) | |
1476 | res = -EINVAL; | |
1477 | else { | |
1478 | sock->state = SS_LISTENING; | |
1479 | res = 0; | |
1480 | } | |
1481 | ||
1482 | release_sock(sk); | |
1483 | return res; | |
b97bf3fd PL |
1484 | } |
1485 | ||
c4307285 | 1486 | /** |
b97bf3fd PL |
1487 | * accept - wait for connection request |
1488 | * @sock: listening socket | |
1489 | * @newsock: new socket that is to be connected | |
1490 | * @flags: file-related flags associated with socket | |
c4307285 | 1491 | * |
b97bf3fd PL |
1492 | * Returns 0 on success, errno otherwise |
1493 | */ | |
1494 | ||
0c3141e9 | 1495 | static int accept(struct socket *sock, struct socket *new_sock, int flags) |
b97bf3fd | 1496 | { |
0c3141e9 | 1497 | struct sock *sk = sock->sk; |
b97bf3fd | 1498 | struct sk_buff *buf; |
0c3141e9 | 1499 | int res; |
b97bf3fd | 1500 | |
0c3141e9 | 1501 | lock_sock(sk); |
b97bf3fd | 1502 | |
0c3141e9 AS |
1503 | if (sock->state == SS_READY) { |
1504 | res = -EOPNOTSUPP; | |
1505 | goto exit; | |
1506 | } | |
1507 | if (sock->state != SS_LISTENING) { | |
1508 | res = -EINVAL; | |
b97bf3fd PL |
1509 | goto exit; |
1510 | } | |
b97bf3fd | 1511 | |
0c3141e9 AS |
1512 | while (skb_queue_empty(&sk->sk_receive_queue)) { |
1513 | if (flags & O_NONBLOCK) { | |
1514 | res = -EWOULDBLOCK; | |
1515 | goto exit; | |
1516 | } | |
1517 | release_sock(sk); | |
aa395145 | 1518 | res = wait_event_interruptible(*sk_sleep(sk), |
0c3141e9 AS |
1519 | (!skb_queue_empty(&sk->sk_receive_queue))); |
1520 | lock_sock(sk); | |
1521 | if (res) | |
1522 | goto exit; | |
1523 | } | |
1524 | ||
1525 | buf = skb_peek(&sk->sk_receive_queue); | |
1526 | ||
3f378b68 | 1527 | res = tipc_create(sock_net(sock->sk), new_sock, 0, 0); |
b97bf3fd | 1528 | if (!res) { |
0c3141e9 | 1529 | struct sock *new_sk = new_sock->sk; |
2da59918 AS |
1530 | struct tipc_sock *new_tsock = tipc_sk(new_sk); |
1531 | struct tipc_port *new_tport = new_tsock->p; | |
0c3141e9 | 1532 | u32 new_ref = new_tport->ref; |
b97bf3fd | 1533 | struct tipc_msg *msg = buf_msg(buf); |
0c3141e9 AS |
1534 | |
1535 | lock_sock(new_sk); | |
1536 | ||
1537 | /* | |
1538 | * Reject any stray messages received by new socket | |
1539 | * before the socket lock was taken (very, very unlikely) | |
1540 | */ | |
1541 | ||
1542 | reject_rx_queue(new_sk); | |
1543 | ||
1544 | /* Connect new socket to it's peer */ | |
b97bf3fd | 1545 | |
2da59918 AS |
1546 | new_tsock->peer_name.ref = msg_origport(msg); |
1547 | new_tsock->peer_name.node = msg_orignode(msg); | |
1548 | tipc_connect2port(new_ref, &new_tsock->peer_name); | |
0c3141e9 | 1549 | new_sock->state = SS_CONNECTED; |
b97bf3fd PL |
1550 | |
1551 | tipc_set_portimportance(new_ref, msg_importance(msg)); | |
1552 | if (msg_named(msg)) { | |
0c3141e9 AS |
1553 | new_tport->conn_type = msg_nametype(msg); |
1554 | new_tport->conn_instance = msg_nameinst(msg); | |
b97bf3fd PL |
1555 | } |
1556 | ||
0c3141e9 | 1557 | /* |
b97bf3fd PL |
1558 | * Respond to 'SYN-' by discarding it & returning 'ACK'-. |
1559 | * Respond to 'SYN+' by queuing it on new socket. | |
1560 | */ | |
1561 | ||
c4307285 YH |
1562 | if (!msg_data_sz(msg)) { |
1563 | struct msghdr m = {NULL,}; | |
b97bf3fd | 1564 | |
0c3141e9 AS |
1565 | advance_rx_queue(sk); |
1566 | send_packet(NULL, new_sock, &m, 0); | |
c4307285 | 1567 | } else { |
0c3141e9 AS |
1568 | __skb_dequeue(&sk->sk_receive_queue); |
1569 | __skb_queue_head(&new_sk->sk_receive_queue, buf); | |
b97bf3fd | 1570 | } |
0c3141e9 | 1571 | release_sock(new_sk); |
b97bf3fd PL |
1572 | } |
1573 | exit: | |
0c3141e9 | 1574 | release_sock(sk); |
b97bf3fd PL |
1575 | return res; |
1576 | } | |
1577 | ||
1578 | /** | |
1579 | * shutdown - shutdown socket connection | |
1580 | * @sock: socket structure | |
e247a8f5 | 1581 | * @how: direction to close (must be SHUT_RDWR) |
b97bf3fd PL |
1582 | * |
1583 | * Terminates connection (if necessary), then purges socket's receive queue. | |
c4307285 | 1584 | * |
b97bf3fd PL |
1585 | * Returns 0 on success, errno otherwise |
1586 | */ | |
1587 | ||
1588 | static int shutdown(struct socket *sock, int how) | |
1589 | { | |
0c3141e9 AS |
1590 | struct sock *sk = sock->sk; |
1591 | struct tipc_port *tport = tipc_sk_port(sk); | |
b97bf3fd PL |
1592 | struct sk_buff *buf; |
1593 | int res; | |
1594 | ||
e247a8f5 AS |
1595 | if (how != SHUT_RDWR) |
1596 | return -EINVAL; | |
b97bf3fd | 1597 | |
0c3141e9 | 1598 | lock_sock(sk); |
b97bf3fd PL |
1599 | |
1600 | switch (sock->state) { | |
0c3141e9 | 1601 | case SS_CONNECTING: |
b97bf3fd PL |
1602 | case SS_CONNECTED: |
1603 | ||
0c3141e9 | 1604 | /* Disconnect and send a 'FIN+' or 'FIN-' message to peer */ |
b97bf3fd | 1605 | restart: |
0c3141e9 AS |
1606 | buf = __skb_dequeue(&sk->sk_receive_queue); |
1607 | if (buf) { | |
b97bf3fd PL |
1608 | atomic_dec(&tipc_queue_size); |
1609 | if (TIPC_SKB_CB(buf)->handle != msg_data(buf_msg(buf))) { | |
1610 | buf_discard(buf); | |
1611 | goto restart; | |
1612 | } | |
0c3141e9 | 1613 | tipc_disconnect(tport->ref); |
b97bf3fd | 1614 | tipc_reject_msg(buf, TIPC_CONN_SHUTDOWN); |
0c3141e9 AS |
1615 | } else { |
1616 | tipc_shutdown(tport->ref); | |
b97bf3fd | 1617 | } |
0c3141e9 AS |
1618 | |
1619 | sock->state = SS_DISCONNECTING; | |
b97bf3fd PL |
1620 | |
1621 | /* fall through */ | |
1622 | ||
1623 | case SS_DISCONNECTING: | |
1624 | ||
0c3141e9 | 1625 | /* Discard any unreceived messages; wake up sleeping tasks */ |
b97bf3fd | 1626 | |
0c3141e9 | 1627 | discard_rx_queue(sk); |
aa395145 ED |
1628 | if (waitqueue_active(sk_sleep(sk))) |
1629 | wake_up_interruptible(sk_sleep(sk)); | |
b97bf3fd PL |
1630 | res = 0; |
1631 | break; | |
1632 | ||
1633 | default: | |
1634 | res = -ENOTCONN; | |
1635 | } | |
1636 | ||
0c3141e9 | 1637 | release_sock(sk); |
b97bf3fd PL |
1638 | return res; |
1639 | } | |
1640 | ||
1641 | /** | |
1642 | * setsockopt - set socket option | |
1643 | * @sock: socket structure | |
1644 | * @lvl: option level | |
1645 | * @opt: option identifier | |
1646 | * @ov: pointer to new option value | |
1647 | * @ol: length of option value | |
c4307285 YH |
1648 | * |
1649 | * For stream sockets only, accepts and ignores all IPPROTO_TCP options | |
b97bf3fd | 1650 | * (to ease compatibility). |
c4307285 | 1651 | * |
b97bf3fd PL |
1652 | * Returns 0 on success, errno otherwise |
1653 | */ | |
1654 | ||
c4307285 | 1655 | static int setsockopt(struct socket *sock, |
b7058842 | 1656 | int lvl, int opt, char __user *ov, unsigned int ol) |
b97bf3fd | 1657 | { |
0c3141e9 AS |
1658 | struct sock *sk = sock->sk; |
1659 | struct tipc_port *tport = tipc_sk_port(sk); | |
b97bf3fd PL |
1660 | u32 value; |
1661 | int res; | |
1662 | ||
c4307285 YH |
1663 | if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM)) |
1664 | return 0; | |
b97bf3fd PL |
1665 | if (lvl != SOL_TIPC) |
1666 | return -ENOPROTOOPT; | |
1667 | if (ol < sizeof(value)) | |
1668 | return -EINVAL; | |
2db9983a AS |
1669 | res = get_user(value, (u32 __user *)ov); |
1670 | if (res) | |
b97bf3fd PL |
1671 | return res; |
1672 | ||
0c3141e9 | 1673 | lock_sock(sk); |
c4307285 | 1674 | |
b97bf3fd PL |
1675 | switch (opt) { |
1676 | case TIPC_IMPORTANCE: | |
0c3141e9 | 1677 | res = tipc_set_portimportance(tport->ref, value); |
b97bf3fd PL |
1678 | break; |
1679 | case TIPC_SRC_DROPPABLE: | |
1680 | if (sock->type != SOCK_STREAM) | |
0c3141e9 | 1681 | res = tipc_set_portunreliable(tport->ref, value); |
c4307285 | 1682 | else |
b97bf3fd PL |
1683 | res = -ENOPROTOOPT; |
1684 | break; | |
1685 | case TIPC_DEST_DROPPABLE: | |
0c3141e9 | 1686 | res = tipc_set_portunreturnable(tport->ref, value); |
b97bf3fd PL |
1687 | break; |
1688 | case TIPC_CONN_TIMEOUT: | |
564e83b5 | 1689 | tipc_sk(sk)->conn_timeout = msecs_to_jiffies(value); |
0c3141e9 | 1690 | /* no need to set "res", since already 0 at this point */ |
b97bf3fd PL |
1691 | break; |
1692 | default: | |
1693 | res = -EINVAL; | |
1694 | } | |
1695 | ||
0c3141e9 AS |
1696 | release_sock(sk); |
1697 | ||
b97bf3fd PL |
1698 | return res; |
1699 | } | |
1700 | ||
1701 | /** | |
1702 | * getsockopt - get socket option | |
1703 | * @sock: socket structure | |
1704 | * @lvl: option level | |
1705 | * @opt: option identifier | |
1706 | * @ov: receptacle for option value | |
1707 | * @ol: receptacle for length of option value | |
c4307285 YH |
1708 | * |
1709 | * For stream sockets only, returns 0 length result for all IPPROTO_TCP options | |
b97bf3fd | 1710 | * (to ease compatibility). |
c4307285 | 1711 | * |
b97bf3fd PL |
1712 | * Returns 0 on success, errno otherwise |
1713 | */ | |
1714 | ||
c4307285 | 1715 | static int getsockopt(struct socket *sock, |
28c4dadd | 1716 | int lvl, int opt, char __user *ov, int __user *ol) |
b97bf3fd | 1717 | { |
0c3141e9 AS |
1718 | struct sock *sk = sock->sk; |
1719 | struct tipc_port *tport = tipc_sk_port(sk); | |
c4307285 | 1720 | int len; |
b97bf3fd | 1721 | u32 value; |
c4307285 | 1722 | int res; |
b97bf3fd | 1723 | |
c4307285 YH |
1724 | if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM)) |
1725 | return put_user(0, ol); | |
b97bf3fd PL |
1726 | if (lvl != SOL_TIPC) |
1727 | return -ENOPROTOOPT; | |
2db9983a AS |
1728 | res = get_user(len, ol); |
1729 | if (res) | |
c4307285 | 1730 | return res; |
b97bf3fd | 1731 | |
0c3141e9 | 1732 | lock_sock(sk); |
b97bf3fd PL |
1733 | |
1734 | switch (opt) { | |
1735 | case TIPC_IMPORTANCE: | |
0c3141e9 | 1736 | res = tipc_portimportance(tport->ref, &value); |
b97bf3fd PL |
1737 | break; |
1738 | case TIPC_SRC_DROPPABLE: | |
0c3141e9 | 1739 | res = tipc_portunreliable(tport->ref, &value); |
b97bf3fd PL |
1740 | break; |
1741 | case TIPC_DEST_DROPPABLE: | |
0c3141e9 | 1742 | res = tipc_portunreturnable(tport->ref, &value); |
b97bf3fd PL |
1743 | break; |
1744 | case TIPC_CONN_TIMEOUT: | |
564e83b5 | 1745 | value = jiffies_to_msecs(tipc_sk(sk)->conn_timeout); |
0c3141e9 | 1746 | /* no need to set "res", since already 0 at this point */ |
b97bf3fd | 1747 | break; |
0e65967e | 1748 | case TIPC_NODE_RECVQ_DEPTH: |
6650613d | 1749 | value = (u32)atomic_read(&tipc_queue_size); |
1750 | break; | |
0e65967e | 1751 | case TIPC_SOCK_RECVQ_DEPTH: |
6650613d | 1752 | value = skb_queue_len(&sk->sk_receive_queue); |
1753 | break; | |
b97bf3fd PL |
1754 | default: |
1755 | res = -EINVAL; | |
1756 | } | |
1757 | ||
0c3141e9 AS |
1758 | release_sock(sk); |
1759 | ||
25860c3b PG |
1760 | if (res) |
1761 | return res; /* "get" failed */ | |
b97bf3fd | 1762 | |
25860c3b PG |
1763 | if (len < sizeof(value)) |
1764 | return -EINVAL; | |
1765 | ||
1766 | if (copy_to_user(ov, &value, sizeof(value))) | |
1767 | return -EFAULT; | |
1768 | ||
1769 | return put_user(sizeof(value), ol); | |
b97bf3fd PL |
1770 | } |
1771 | ||
b97bf3fd PL |
1772 | /** |
1773 | * Protocol switches for the various types of TIPC sockets | |
1774 | */ | |
1775 | ||
bca65eae | 1776 | static const struct proto_ops msg_ops = { |
0e65967e | 1777 | .owner = THIS_MODULE, |
b97bf3fd PL |
1778 | .family = AF_TIPC, |
1779 | .release = release, | |
1780 | .bind = bind, | |
1781 | .connect = connect, | |
5eee6a6d | 1782 | .socketpair = sock_no_socketpair, |
b97bf3fd PL |
1783 | .accept = accept, |
1784 | .getname = get_name, | |
1785 | .poll = poll, | |
5eee6a6d | 1786 | .ioctl = sock_no_ioctl, |
b97bf3fd PL |
1787 | .listen = listen, |
1788 | .shutdown = shutdown, | |
1789 | .setsockopt = setsockopt, | |
1790 | .getsockopt = getsockopt, | |
1791 | .sendmsg = send_msg, | |
1792 | .recvmsg = recv_msg, | |
8238745a YH |
1793 | .mmap = sock_no_mmap, |
1794 | .sendpage = sock_no_sendpage | |
b97bf3fd PL |
1795 | }; |
1796 | ||
bca65eae | 1797 | static const struct proto_ops packet_ops = { |
0e65967e | 1798 | .owner = THIS_MODULE, |
b97bf3fd PL |
1799 | .family = AF_TIPC, |
1800 | .release = release, | |
1801 | .bind = bind, | |
1802 | .connect = connect, | |
5eee6a6d | 1803 | .socketpair = sock_no_socketpair, |
b97bf3fd PL |
1804 | .accept = accept, |
1805 | .getname = get_name, | |
1806 | .poll = poll, | |
5eee6a6d | 1807 | .ioctl = sock_no_ioctl, |
b97bf3fd PL |
1808 | .listen = listen, |
1809 | .shutdown = shutdown, | |
1810 | .setsockopt = setsockopt, | |
1811 | .getsockopt = getsockopt, | |
1812 | .sendmsg = send_packet, | |
1813 | .recvmsg = recv_msg, | |
8238745a YH |
1814 | .mmap = sock_no_mmap, |
1815 | .sendpage = sock_no_sendpage | |
b97bf3fd PL |
1816 | }; |
1817 | ||
bca65eae | 1818 | static const struct proto_ops stream_ops = { |
0e65967e | 1819 | .owner = THIS_MODULE, |
b97bf3fd PL |
1820 | .family = AF_TIPC, |
1821 | .release = release, | |
1822 | .bind = bind, | |
1823 | .connect = connect, | |
5eee6a6d | 1824 | .socketpair = sock_no_socketpair, |
b97bf3fd PL |
1825 | .accept = accept, |
1826 | .getname = get_name, | |
1827 | .poll = poll, | |
5eee6a6d | 1828 | .ioctl = sock_no_ioctl, |
b97bf3fd PL |
1829 | .listen = listen, |
1830 | .shutdown = shutdown, | |
1831 | .setsockopt = setsockopt, | |
1832 | .getsockopt = getsockopt, | |
1833 | .sendmsg = send_stream, | |
1834 | .recvmsg = recv_stream, | |
8238745a YH |
1835 | .mmap = sock_no_mmap, |
1836 | .sendpage = sock_no_sendpage | |
b97bf3fd PL |
1837 | }; |
1838 | ||
bca65eae | 1839 | static const struct net_proto_family tipc_family_ops = { |
0e65967e | 1840 | .owner = THIS_MODULE, |
b97bf3fd PL |
1841 | .family = AF_TIPC, |
1842 | .create = tipc_create | |
1843 | }; | |
1844 | ||
1845 | static struct proto tipc_proto = { | |
1846 | .name = "TIPC", | |
1847 | .owner = THIS_MODULE, | |
1848 | .obj_size = sizeof(struct tipc_sock) | |
1849 | }; | |
1850 | ||
1851 | /** | |
4323add6 | 1852 | * tipc_socket_init - initialize TIPC socket interface |
c4307285 | 1853 | * |
b97bf3fd PL |
1854 | * Returns 0 on success, errno otherwise |
1855 | */ | |
4323add6 | 1856 | int tipc_socket_init(void) |
b97bf3fd PL |
1857 | { |
1858 | int res; | |
1859 | ||
c4307285 | 1860 | res = proto_register(&tipc_proto, 1); |
b97bf3fd | 1861 | if (res) { |
d0a14a9d | 1862 | err("Failed to register TIPC protocol type\n"); |
b97bf3fd PL |
1863 | goto out; |
1864 | } | |
1865 | ||
1866 | res = sock_register(&tipc_family_ops); | |
1867 | if (res) { | |
d0a14a9d | 1868 | err("Failed to register TIPC socket type\n"); |
b97bf3fd PL |
1869 | proto_unregister(&tipc_proto); |
1870 | goto out; | |
1871 | } | |
1872 | ||
1873 | sockets_enabled = 1; | |
1874 | out: | |
1875 | return res; | |
1876 | } | |
1877 | ||
1878 | /** | |
4323add6 | 1879 | * tipc_socket_stop - stop TIPC socket interface |
b97bf3fd | 1880 | */ |
0c3141e9 | 1881 | |
4323add6 | 1882 | void tipc_socket_stop(void) |
b97bf3fd PL |
1883 | { |
1884 | if (!sockets_enabled) | |
1885 | return; | |
1886 | ||
1887 | sockets_enabled = 0; | |
1888 | sock_unregister(tipc_family_ops.family); | |
1889 | proto_unregister(&tipc_proto); | |
1890 | } | |
1891 |