]> git.ipfire.org Git - thirdparty/linux.git/blame - net/tipc/socket.c
Merge tag 'io_uring-5.7-2020-05-22' of git://git.kernel.dk/linux-block
[thirdparty/linux.git] / net / tipc / socket.c
CommitLineData
b97bf3fd 1/*
02c00c2a 2 * net/tipc/socket.c: TIPC socket API
c4307285 3 *
75da2163 4 * Copyright (c) 2001-2007, 2012-2017, Ericsson AB
c5fa7b3c 5 * Copyright (c) 2004-2008, 2010-2013, 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
07f6c4bc 37#include <linux/rhashtable.h>
174cd4b1
IM
38#include <linux/sched/signal.h>
39
b97bf3fd 40#include "core.h"
e2dafe87 41#include "name_table.h"
78acb1f9 42#include "node.h"
e2dafe87 43#include "link.h"
c637c103 44#include "name_distr.h"
2e84c60b 45#include "socket.h"
a6bf70f7 46#include "bcast.h"
49cc66ea 47#include "netlink.h"
75da2163 48#include "group.h"
b4b9771b 49#include "trace.h"
2cf8aa19 50
67879274 51#define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */
0d5fcebf 52#define CONN_PROBING_INTV msecs_to_jiffies(3600000) /* [ms] => 1 h */
07f6c4bc 53#define TIPC_FWD_MSG 1
07f6c4bc
YX
54#define TIPC_MAX_PORT 0xffffffff
55#define TIPC_MIN_PORT 1
e9f8b101 56#define TIPC_ACK_RATE 4 /* ACK at 1/4 of of rcv window size */
301bae56 57
0c288c86
PB
58enum {
59 TIPC_LISTEN = TCP_LISTEN,
8ea642ee 60 TIPC_ESTABLISHED = TCP_ESTABLISHED,
438adcaf 61 TIPC_OPEN = TCP_CLOSE,
9fd4b070 62 TIPC_DISCONNECTING = TCP_CLOSE_WAIT,
99a20889 63 TIPC_CONNECTING = TCP_SYN_SENT,
0c288c86
PB
64};
65
31c82a2d
JM
66struct sockaddr_pair {
67 struct sockaddr_tipc sock;
68 struct sockaddr_tipc member;
69};
70
301bae56
JPM
71/**
72 * struct tipc_sock - TIPC socket structure
73 * @sk: socket - interacts with 'port' and with user via the socket API
301bae56
JPM
74 * @conn_type: TIPC type used when connection was established
75 * @conn_instance: TIPC instance used when connection was established
76 * @published: non-zero if port has one or more associated names
77 * @max_pkt: maximum packet size "hint" used when building messages sent by port
c0bceb97 78 * @maxnagle: maximum size of msg which can be subject to nagle
07f6c4bc 79 * @portid: unique port identity in TIPC socket hash table
301bae56 80 * @phdr: preformatted message header used when sending messages
365ad353 81 * #cong_links: list of congested links
301bae56 82 * @publications: list of publications for port
365ad353 83 * @blocking_link: address of the congested link we are currently sleeping on
301bae56 84 * @pub_count: total # of publications port has made during its lifetime
301bae56
JPM
85 * @conn_timeout: the time we can wait for an unresponded setup request
86 * @dupl_rcvcnt: number of bytes counted twice, in both backlog and rcv queue
365ad353 87 * @cong_link_cnt: number of congested links
75da2163 88 * @snt_unacked: # messages sent by socket, and not yet acked by peer
301bae56 89 * @rcv_unacked: # messages read by user, but not yet acked back to peer
aeda16b6 90 * @peer: 'connected' peer for dgram/rdm
07f6c4bc 91 * @node: hash table node
01fd12bb 92 * @mc_method: cookie for use between socket and broadcast layer
07f6c4bc 93 * @rcu: rcu struct for tipc_sock
301bae56
JPM
94 */
95struct tipc_sock {
96 struct sock sk;
301bae56
JPM
97 u32 conn_type;
98 u32 conn_instance;
99 int published;
100 u32 max_pkt;
c0bceb97 101 u32 maxnagle;
07f6c4bc 102 u32 portid;
301bae56 103 struct tipc_msg phdr;
365ad353 104 struct list_head cong_links;
301bae56
JPM
105 struct list_head publications;
106 u32 pub_count;
301bae56 107 atomic_t dupl_rcvcnt;
67879274 108 u16 conn_timeout;
8ea642ee 109 bool probe_unacked;
365ad353 110 u16 cong_link_cnt;
10724cc7
JPM
111 u16 snt_unacked;
112 u16 snd_win;
60020e18 113 u16 peer_caps;
10724cc7
JPM
114 u16 rcv_unacked;
115 u16 rcv_win;
aeda16b6 116 struct sockaddr_tipc peer;
07f6c4bc 117 struct rhash_head node;
01fd12bb 118 struct tipc_mc_method mc_method;
07f6c4bc 119 struct rcu_head rcu;
75da2163 120 struct tipc_group *group;
c0bceb97
JM
121 u32 oneway;
122 u16 snd_backlog;
123 bool expect_ack;
124 bool nodelay;
60c25306 125 bool group_is_open;
301bae56 126};
b97bf3fd 127
64ac5f59 128static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb);
676d2369 129static void tipc_data_ready(struct sock *sk);
f288bef4 130static void tipc_write_space(struct sock *sk);
f4195d1e 131static void tipc_sock_destruct(struct sock *sk);
247f0f3c 132static int tipc_release(struct socket *sock);
cdfbabfb
DH
133static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags,
134 bool kern);
31b102bb 135static void tipc_sk_timeout(struct timer_list *t);
301bae56 136static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
0fc87aae 137 struct tipc_name_seq const *seq);
301bae56 138static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
0fc87aae 139 struct tipc_name_seq const *seq);
75da2163 140static int tipc_sk_leave(struct tipc_sock *tsk);
e05b31f4 141static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid);
07f6c4bc
YX
142static int tipc_sk_insert(struct tipc_sock *tsk);
143static void tipc_sk_remove(struct tipc_sock *tsk);
365ad353 144static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz);
39a0295f 145static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dsz);
c0bceb97 146static void tipc_sk_push_backlog(struct tipc_sock *tsk);
b97bf3fd 147
bca65eae
FW
148static const struct proto_ops packet_ops;
149static const struct proto_ops stream_ops;
150static const struct proto_ops msg_ops;
b97bf3fd 151static struct proto tipc_proto;
6cca7289
HX
152static const struct rhashtable_params tsk_rht_params;
153
c5898636
JPM
154static u32 tsk_own_node(struct tipc_sock *tsk)
155{
156 return msg_prevnode(&tsk->phdr);
157}
158
301bae56 159static u32 tsk_peer_node(struct tipc_sock *tsk)
2e84c60b 160{
301bae56 161 return msg_destnode(&tsk->phdr);
2e84c60b
JPM
162}
163
301bae56 164static u32 tsk_peer_port(struct tipc_sock *tsk)
2e84c60b 165{
301bae56 166 return msg_destport(&tsk->phdr);
2e84c60b
JPM
167}
168
301bae56 169static bool tsk_unreliable(struct tipc_sock *tsk)
2e84c60b 170{
301bae56 171 return msg_src_droppable(&tsk->phdr) != 0;
2e84c60b
JPM
172}
173
301bae56 174static void tsk_set_unreliable(struct tipc_sock *tsk, bool unreliable)
2e84c60b 175{
301bae56 176 msg_set_src_droppable(&tsk->phdr, unreliable ? 1 : 0);
2e84c60b
JPM
177}
178
301bae56 179static bool tsk_unreturnable(struct tipc_sock *tsk)
2e84c60b 180{
301bae56 181 return msg_dest_droppable(&tsk->phdr) != 0;
2e84c60b
JPM
182}
183
301bae56 184static void tsk_set_unreturnable(struct tipc_sock *tsk, bool unreturnable)
2e84c60b 185{
301bae56 186 msg_set_dest_droppable(&tsk->phdr, unreturnable ? 1 : 0);
2e84c60b
JPM
187}
188
301bae56 189static int tsk_importance(struct tipc_sock *tsk)
2e84c60b 190{
301bae56 191 return msg_importance(&tsk->phdr);
2e84c60b
JPM
192}
193
301bae56 194static int tsk_set_importance(struct tipc_sock *tsk, int imp)
2e84c60b
JPM
195{
196 if (imp > TIPC_CRITICAL_IMPORTANCE)
197 return -EINVAL;
301bae56 198 msg_set_importance(&tsk->phdr, (u32)imp);
2e84c60b
JPM
199 return 0;
200}
8826cde6 201
301bae56
JPM
202static struct tipc_sock *tipc_sk(const struct sock *sk)
203{
204 return container_of(sk, struct tipc_sock, sk);
205}
206
10724cc7 207static bool tsk_conn_cong(struct tipc_sock *tsk)
301bae56 208{
6998cc6e 209 return tsk->snt_unacked > tsk->snd_win;
10724cc7
JPM
210}
211
b7d42635
JM
212static u16 tsk_blocks(int len)
213{
214 return ((len / FLOWCTL_BLK_SZ) + 1);
215}
216
10724cc7
JPM
217/* tsk_blocks(): translate a buffer size in bytes to number of
218 * advertisable blocks, taking into account the ratio truesize(len)/len
219 * We can trust that this ratio is always < 4 for len >= FLOWCTL_BLK_SZ
220 */
221static u16 tsk_adv_blocks(int len)
222{
223 return len / FLOWCTL_BLK_SZ / 4;
224}
225
226/* tsk_inc(): increment counter for sent or received data
227 * - If block based flow control is not supported by peer we
228 * fall back to message based ditto, incrementing the counter
229 */
230static u16 tsk_inc(struct tipc_sock *tsk, int msglen)
231{
232 if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
233 return ((msglen / FLOWCTL_BLK_SZ) + 1);
234 return 1;
301bae56
JPM
235}
236
c0bceb97
JM
237/* tsk_set_nagle - enable/disable nagle property by manipulating maxnagle
238 */
239static void tsk_set_nagle(struct tipc_sock *tsk)
240{
241 struct sock *sk = &tsk->sk;
242
243 tsk->maxnagle = 0;
244 if (sk->sk_type != SOCK_STREAM)
245 return;
246 if (tsk->nodelay)
247 return;
248 if (!(tsk->peer_caps & TIPC_NAGLE))
249 return;
250 /* Limit node local buffer size to avoid receive queue overflow */
251 if (tsk->max_pkt == MAX_MSG_SIZE)
252 tsk->maxnagle = 1500;
253 else
254 tsk->maxnagle = tsk->max_pkt;
255}
256
0c3141e9 257/**
2e84c60b 258 * tsk_advance_rx_queue - discard first buffer in socket receive queue
0c3141e9
AS
259 *
260 * Caller must hold socket lock
b97bf3fd 261 */
2e84c60b 262static void tsk_advance_rx_queue(struct sock *sk)
b97bf3fd 263{
01e661eb 264 trace_tipc_sk_advance_rx(sk, NULL, TIPC_DUMP_SK_RCVQ, " ");
5f6d9123 265 kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
b97bf3fd
PL
266}
267
bcd3ffd4
JPM
268/* tipc_sk_respond() : send response message back to sender
269 */
270static void tipc_sk_respond(struct sock *sk, struct sk_buff *skb, int err)
271{
272 u32 selector;
273 u32 dnode;
274 u32 onode = tipc_own_addr(sock_net(sk));
275
276 if (!tipc_msg_reverse(onode, &skb, err))
277 return;
278
01e661eb 279 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE, "@sk_respond!");
bcd3ffd4
JPM
280 dnode = msg_destnode(buf_msg(skb));
281 selector = msg_origport(buf_msg(skb));
282 tipc_node_xmit_skb(sock_net(sk), skb, dnode, selector);
283}
284
b97bf3fd 285/**
2e84c60b 286 * tsk_rej_rx_queue - reject all buffers in socket receive queue
0c3141e9
AS
287 *
288 * Caller must hold socket lock
b97bf3fd 289 */
49afb806 290static void tsk_rej_rx_queue(struct sock *sk, int error)
b97bf3fd 291{
a6ca1094 292 struct sk_buff *skb;
0c3141e9 293
bcd3ffd4 294 while ((skb = __skb_dequeue(&sk->sk_receive_queue)))
49afb806 295 tipc_sk_respond(sk, skb, error);
b97bf3fd
PL
296}
297
d6fb7e9c
PB
298static bool tipc_sk_connected(struct sock *sk)
299{
f40acbaf 300 return sk->sk_state == TIPC_ESTABLISHED;
d6fb7e9c
PB
301}
302
c752023a
PB
303/* tipc_sk_type_connectionless - check if the socket is datagram socket
304 * @sk: socket
305 *
306 * Returns true if connection less, false otherwise
307 */
308static bool tipc_sk_type_connectionless(struct sock *sk)
309{
310 return sk->sk_type == SOCK_RDM || sk->sk_type == SOCK_DGRAM;
311}
312
2e84c60b 313/* tsk_peer_msg - verify if message was sent by connected port's peer
0fc87aae
JPM
314 *
315 * Handles cases where the node's network address has changed from
316 * the default of <0.0.0> to its configured setting.
317 */
2e84c60b 318static bool tsk_peer_msg(struct tipc_sock *tsk, struct tipc_msg *msg)
0fc87aae 319{
d6fb7e9c 320 struct sock *sk = &tsk->sk;
23fd3eac 321 u32 self = tipc_own_addr(sock_net(sk));
301bae56 322 u32 peer_port = tsk_peer_port(tsk);
23fd3eac 323 u32 orig_node, peer_node;
0fc87aae 324
d6fb7e9c 325 if (unlikely(!tipc_sk_connected(sk)))
0fc87aae
JPM
326 return false;
327
328 if (unlikely(msg_origport(msg) != peer_port))
329 return false;
330
331 orig_node = msg_orignode(msg);
301bae56 332 peer_node = tsk_peer_node(tsk);
0fc87aae
JPM
333
334 if (likely(orig_node == peer_node))
335 return true;
336
23fd3eac 337 if (!orig_node && peer_node == self)
0fc87aae
JPM
338 return true;
339
23fd3eac 340 if (!peer_node && orig_node == self)
0fc87aae
JPM
341 return true;
342
343 return false;
344}
345
0c288c86
PB
346/* tipc_set_sk_state - set the sk_state of the socket
347 * @sk: socket
348 *
349 * Caller must hold socket lock
350 *
351 * Returns 0 on success, errno otherwise
352 */
353static int tipc_set_sk_state(struct sock *sk, int state)
354{
438adcaf 355 int oldsk_state = sk->sk_state;
0c288c86
PB
356 int res = -EINVAL;
357
358 switch (state) {
438adcaf
PB
359 case TIPC_OPEN:
360 res = 0;
361 break;
0c288c86 362 case TIPC_LISTEN:
99a20889 363 case TIPC_CONNECTING:
438adcaf 364 if (oldsk_state == TIPC_OPEN)
0c288c86
PB
365 res = 0;
366 break;
8ea642ee 367 case TIPC_ESTABLISHED:
99a20889 368 if (oldsk_state == TIPC_CONNECTING ||
438adcaf 369 oldsk_state == TIPC_OPEN)
8ea642ee
PB
370 res = 0;
371 break;
9fd4b070 372 case TIPC_DISCONNECTING:
99a20889 373 if (oldsk_state == TIPC_CONNECTING ||
9fd4b070
PB
374 oldsk_state == TIPC_ESTABLISHED)
375 res = 0;
376 break;
0c288c86
PB
377 }
378
379 if (!res)
380 sk->sk_state = state;
381
382 return res;
383}
384
8c44e1af
JPM
385static int tipc_sk_sock_err(struct socket *sock, long *timeout)
386{
387 struct sock *sk = sock->sk;
388 int err = sock_error(sk);
389 int typ = sock->type;
390
391 if (err)
392 return err;
393 if (typ == SOCK_STREAM || typ == SOCK_SEQPACKET) {
394 if (sk->sk_state == TIPC_DISCONNECTING)
395 return -EPIPE;
396 else if (!tipc_sk_connected(sk))
397 return -ENOTCONN;
398 }
399 if (!*timeout)
400 return -EAGAIN;
401 if (signal_pending(current))
402 return sock_intr_errno(*timeout);
403
404 return 0;
405}
406
844cf763
JPM
407#define tipc_wait_for_cond(sock_, timeo_, condition_) \
408({ \
bfd07f3d 409 DEFINE_WAIT_FUNC(wait_, woken_wake_function); \
844cf763
JPM
410 struct sock *sk_; \
411 int rc_; \
412 \
413 while ((rc_ = !(condition_))) { \
bfd07f3d
TN
414 /* coupled with smp_wmb() in tipc_sk_proto_rcv() */ \
415 smp_rmb(); \
844cf763
JPM
416 sk_ = (sock_)->sk; \
417 rc_ = tipc_sk_sock_err((sock_), timeo_); \
418 if (rc_) \
419 break; \
223b7329 420 add_wait_queue(sk_sleep(sk_), &wait_); \
844cf763
JPM
421 release_sock(sk_); \
422 *(timeo_) = wait_woken(&wait_, TASK_INTERRUPTIBLE, *(timeo_)); \
423 sched_annotate_sleep(); \
424 lock_sock(sk_); \
425 remove_wait_queue(sk_sleep(sk_), &wait_); \
426 } \
427 rc_; \
8c44e1af
JPM
428})
429
b97bf3fd 430/**
c5fa7b3c 431 * tipc_sk_create - create a TIPC socket
0c3141e9 432 * @net: network namespace (must be default network)
b97bf3fd
PL
433 * @sock: pre-allocated socket structure
434 * @protocol: protocol indicator (must be 0)
3f378b68 435 * @kern: caused by kernel or by userspace?
c4307285 436 *
0c3141e9
AS
437 * This routine creates additional data structures used by the TIPC socket,
438 * initializes them, and links them together.
b97bf3fd
PL
439 *
440 * Returns 0 on success, errno otherwise
441 */
58ed9442
JPM
442static int tipc_sk_create(struct net *net, struct socket *sock,
443 int protocol, int kern)
b97bf3fd 444{
0c3141e9 445 const struct proto_ops *ops;
b97bf3fd 446 struct sock *sk;
58ed9442 447 struct tipc_sock *tsk;
5b8fa7ce 448 struct tipc_msg *msg;
0c3141e9
AS
449
450 /* Validate arguments */
b97bf3fd
PL
451 if (unlikely(protocol != 0))
452 return -EPROTONOSUPPORT;
453
b97bf3fd
PL
454 switch (sock->type) {
455 case SOCK_STREAM:
0c3141e9 456 ops = &stream_ops;
b97bf3fd
PL
457 break;
458 case SOCK_SEQPACKET:
0c3141e9 459 ops = &packet_ops;
b97bf3fd
PL
460 break;
461 case SOCK_DGRAM:
b97bf3fd 462 case SOCK_RDM:
0c3141e9 463 ops = &msg_ops;
b97bf3fd 464 break;
49978651 465 default:
49978651 466 return -EPROTOTYPE;
b97bf3fd
PL
467 }
468
0c3141e9 469 /* Allocate socket's protocol area */
11aa9c28 470 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto, kern);
0c3141e9 471 if (sk == NULL)
b97bf3fd 472 return -ENOMEM;
b97bf3fd 473
58ed9442 474 tsk = tipc_sk(sk);
301bae56 475 tsk->max_pkt = MAX_PKT_DEFAULT;
c0bceb97 476 tsk->maxnagle = 0;
301bae56 477 INIT_LIST_HEAD(&tsk->publications);
365ad353 478 INIT_LIST_HEAD(&tsk->cong_links);
301bae56 479 msg = &tsk->phdr;
b97bf3fd 480
0c3141e9 481 /* Finish initializing socket data structures */
0c3141e9 482 sock->ops = ops;
0c3141e9 483 sock_init_data(sock, sk);
438adcaf 484 tipc_set_sk_state(sk, TIPC_OPEN);
07f6c4bc 485 if (tipc_sk_insert(tsk)) {
c19ca6cb 486 pr_warn("Socket create failed; port number exhausted\n");
07f6c4bc
YX
487 return -EINVAL;
488 }
40f9f439
HX
489
490 /* Ensure tsk is visible before we read own_addr. */
491 smp_mb();
492
23fd3eac
JM
493 tipc_msg_init(tipc_own_addr(net), msg, TIPC_LOW_IMPORTANCE,
494 TIPC_NAMED_MSG, NAMED_H_SIZE, 0);
40f9f439 495
07f6c4bc 496 msg_set_origport(msg, tsk->portid);
31b102bb 497 timer_setup(&sk->sk_timer, tipc_sk_timeout, 0);
6f00089c 498 sk->sk_shutdown = 0;
64ac5f59 499 sk->sk_backlog_rcv = tipc_sk_backlog_rcv;
cc79dd1b 500 sk->sk_rcvbuf = sysctl_tipc_rmem[1];
f288bef4
YX
501 sk->sk_data_ready = tipc_data_ready;
502 sk->sk_write_space = tipc_write_space;
f4195d1e 503 sk->sk_destruct = tipc_sock_destruct;
4f4482dc 504 tsk->conn_timeout = CONN_TIMEOUT_DEFAULT;
1b22bcad 505 tsk->group_is_open = true;
4f4482dc 506 atomic_set(&tsk->dupl_rcvcnt, 0);
7ef43eba 507
10724cc7
JPM
508 /* Start out with safe limits until we receive an advertised window */
509 tsk->snd_win = tsk_adv_blocks(RCVBUF_MIN);
510 tsk->rcv_win = tsk->snd_win;
511
c752023a 512 if (tipc_sk_type_connectionless(sk)) {
301bae56 513 tsk_set_unreturnable(tsk, true);
0c3141e9 514 if (sock->type == SOCK_DGRAM)
301bae56 515 tsk_set_unreliable(tsk, true);
0c3141e9 516 }
2948a1fc 517 __skb_queue_head_init(&tsk->mc_method.deferredq);
01e661eb 518 trace_tipc_sk_create(sk, NULL, TIPC_DUMP_NONE, " ");
b97bf3fd
PL
519 return 0;
520}
521
07f6c4bc
YX
522static void tipc_sk_callback(struct rcu_head *head)
523{
524 struct tipc_sock *tsk = container_of(head, struct tipc_sock, rcu);
525
526 sock_put(&tsk->sk);
527}
528
6f00089c
PB
529/* Caller should hold socket lock for the socket. */
530static void __tipc_shutdown(struct socket *sock, int error)
531{
532 struct sock *sk = sock->sk;
533 struct tipc_sock *tsk = tipc_sk(sk);
534 struct net *net = sock_net(sk);
12db3c80 535 long timeout = msecs_to_jiffies(CONN_TIMEOUT_DEFAULT);
6f00089c
PB
536 u32 dnode = tsk_peer_node(tsk);
537 struct sk_buff *skb;
538
365ad353
JPM
539 /* Avoid that hi-prio shutdown msgs bypass msgs in link wakeup queue */
540 tipc_wait_for_cond(sock, &timeout, (!tsk->cong_link_cnt &&
541 !tsk_conn_cong(tsk)));
542
d34910e1
TN
543 /* Push out delayed messages if in Nagle mode */
544 tipc_sk_push_backlog(tsk);
545 /* Remove pending SYN */
546 __skb_queue_purge(&sk->sk_write_queue);
67879274 547
49afb806
TL
548 /* Remove partially received buffer if any */
549 skb = skb_peek(&sk->sk_receive_queue);
550 if (skb && TIPC_SKB_CB(skb)->bytes_read) {
551 __skb_unlink(skb, &sk->sk_receive_queue);
552 kfree_skb(skb);
6f00089c 553 }
693c5649 554
49afb806
TL
555 /* Reject all unreceived messages if connectionless */
556 if (tipc_sk_type_connectionless(sk)) {
557 tsk_rej_rx_queue(sk, error);
693c5649 558 return;
49afb806 559 }
693c5649 560
49afb806
TL
561 switch (sk->sk_state) {
562 case TIPC_CONNECTING:
563 case TIPC_ESTABLISHED:
564 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
565 tipc_node_remove_conn(net, dnode, tsk->portid);
566 /* Send a FIN+/- to its peer */
567 skb = __skb_dequeue(&sk->sk_receive_queue);
568 if (skb) {
569 __skb_queue_purge(&sk->sk_receive_queue);
570 tipc_sk_respond(sk, skb, error);
571 break;
572 }
6f00089c
PB
573 skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE,
574 TIPC_CONN_MSG, SHORT_H_SIZE, 0, dnode,
575 tsk_own_node(tsk), tsk_peer_port(tsk),
576 tsk->portid, error);
577 if (skb)
578 tipc_node_xmit_skb(net, skb, dnode, tsk->portid);
49afb806
TL
579 break;
580 case TIPC_LISTEN:
581 /* Reject all SYN messages */
582 tsk_rej_rx_queue(sk, error);
583 break;
584 default:
585 __skb_queue_purge(&sk->sk_receive_queue);
586 break;
6f00089c
PB
587 }
588}
589
b97bf3fd 590/**
247f0f3c 591 * tipc_release - destroy a TIPC socket
b97bf3fd
PL
592 * @sock: socket to destroy
593 *
594 * This routine cleans up any messages that are still queued on the socket.
595 * For DGRAM and RDM socket types, all queued messages are rejected.
596 * For SEQPACKET and STREAM socket types, the first message is rejected
597 * and any others are discarded. (If the first message on a STREAM socket
598 * is partially-read, it is discarded and the next one is rejected instead.)
c4307285 599 *
b97bf3fd
PL
600 * NOTE: Rejected messages are not necessarily returned to the sender! They
601 * are returned or discarded according to the "destination droppable" setting
602 * specified for the message by the sender.
603 *
604 * Returns 0 on success, errno otherwise
605 */
247f0f3c 606static int tipc_release(struct socket *sock)
b97bf3fd 607{
b97bf3fd 608 struct sock *sk = sock->sk;
58ed9442 609 struct tipc_sock *tsk;
b97bf3fd 610
0c3141e9
AS
611 /*
612 * Exit if socket isn't fully initialized (occurs when a failed accept()
613 * releases a pre-allocated child socket that was never used)
614 */
0c3141e9 615 if (sk == NULL)
b97bf3fd 616 return 0;
c4307285 617
58ed9442 618 tsk = tipc_sk(sk);
0c3141e9
AS
619 lock_sock(sk);
620
01e661eb 621 trace_tipc_sk_release(sk, NULL, TIPC_DUMP_ALL, " ");
6f00089c
PB
622 __tipc_shutdown(sock, TIPC_ERR_NO_PORT);
623 sk->sk_shutdown = SHUTDOWN_MASK;
75da2163 624 tipc_sk_leave(tsk);
301bae56 625 tipc_sk_withdraw(tsk, 0, NULL);
c55c8eda 626 __skb_queue_purge(&tsk->mc_method.deferredq);
1ea23a21 627 sk_stop_timer(sk, &sk->sk_timer);
07f6c4bc 628 tipc_sk_remove(tsk);
b97bf3fd 629
0a3b8b2b 630 sock_orphan(sk);
0c3141e9 631 /* Reject any messages that accumulated in backlog queue */
0c3141e9 632 release_sock(sk);
a80ae530 633 tipc_dest_list_purge(&tsk->cong_links);
365ad353 634 tsk->cong_link_cnt = 0;
07f6c4bc 635 call_rcu(&tsk->rcu, tipc_sk_callback);
0c3141e9 636 sock->sk = NULL;
b97bf3fd 637
065d7e39 638 return 0;
b97bf3fd
PL
639}
640
641/**
247f0f3c 642 * tipc_bind - associate or disassocate TIPC name(s) with a socket
b97bf3fd
PL
643 * @sock: socket structure
644 * @uaddr: socket address describing name(s) and desired operation
645 * @uaddr_len: size of socket address data structure
c4307285 646 *
b97bf3fd
PL
647 * Name and name sequence binding is indicated using a positive scope value;
648 * a negative scope value unbinds the specified name. Specifying no name
649 * (i.e. a socket address length of 0) unbinds all names from the socket.
c4307285 650 *
b97bf3fd 651 * Returns 0 on success, errno otherwise
0c3141e9
AS
652 *
653 * NOTE: This routine doesn't need to take the socket lock since it doesn't
654 * access any non-constant socket information.
b97bf3fd 655 */
247f0f3c
YX
656static int tipc_bind(struct socket *sock, struct sockaddr *uaddr,
657 int uaddr_len)
b97bf3fd 658{
84602761 659 struct sock *sk = sock->sk;
b97bf3fd 660 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
58ed9442 661 struct tipc_sock *tsk = tipc_sk(sk);
84602761 662 int res = -EINVAL;
b97bf3fd 663
84602761
YX
664 lock_sock(sk);
665 if (unlikely(!uaddr_len)) {
301bae56 666 res = tipc_sk_withdraw(tsk, 0, NULL);
84602761
YX
667 goto exit;
668 }
75da2163
JM
669 if (tsk->group) {
670 res = -EACCES;
671 goto exit;
672 }
84602761
YX
673 if (uaddr_len < sizeof(struct sockaddr_tipc)) {
674 res = -EINVAL;
675 goto exit;
676 }
677 if (addr->family != AF_TIPC) {
678 res = -EAFNOSUPPORT;
679 goto exit;
680 }
b97bf3fd 681
b97bf3fd
PL
682 if (addr->addrtype == TIPC_ADDR_NAME)
683 addr->addr.nameseq.upper = addr->addr.nameseq.lower;
84602761
YX
684 else if (addr->addrtype != TIPC_ADDR_NAMESEQ) {
685 res = -EAFNOSUPPORT;
686 goto exit;
687 }
c4307285 688
13a2e898 689 if ((addr->addr.nameseq.type < TIPC_RESERVED_TYPES) &&
7d0ab17b 690 (addr->addr.nameseq.type != TIPC_TOP_SRV) &&
84602761
YX
691 (addr->addr.nameseq.type != TIPC_CFG_SRV)) {
692 res = -EACCES;
693 goto exit;
694 }
c422f1bd 695
928df188 696 res = (addr->scope >= 0) ?
301bae56
JPM
697 tipc_sk_publish(tsk, addr->scope, &addr->addr.nameseq) :
698 tipc_sk_withdraw(tsk, -addr->scope, &addr->addr.nameseq);
84602761
YX
699exit:
700 release_sock(sk);
701 return res;
b97bf3fd
PL
702}
703
c4307285 704/**
247f0f3c 705 * tipc_getname - get port ID of socket or peer socket
b97bf3fd
PL
706 * @sock: socket structure
707 * @uaddr: area for returned socket address
708 * @uaddr_len: area for returned length of socket address
2da59918 709 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
c4307285 710 *
b97bf3fd 711 * Returns 0 on success, errno otherwise
0c3141e9 712 *
2da59918
AS
713 * NOTE: This routine doesn't need to take the socket lock since it only
714 * accesses socket information that is unchanging (or which changes in
0e65967e 715 * a completely predictable manner).
b97bf3fd 716 */
247f0f3c 717static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
9b2c45d4 718 int peer)
b97bf3fd 719{
b97bf3fd 720 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
9fd4b070
PB
721 struct sock *sk = sock->sk;
722 struct tipc_sock *tsk = tipc_sk(sk);
b97bf3fd 723
88f8a5e3 724 memset(addr, 0, sizeof(*addr));
0c3141e9 725 if (peer) {
f40acbaf 726 if ((!tipc_sk_connected(sk)) &&
9fd4b070 727 ((peer != 2) || (sk->sk_state != TIPC_DISCONNECTING)))
2da59918 728 return -ENOTCONN;
301bae56
JPM
729 addr->addr.id.ref = tsk_peer_port(tsk);
730 addr->addr.id.node = tsk_peer_node(tsk);
0c3141e9 731 } else {
07f6c4bc 732 addr->addr.id.ref = tsk->portid;
23fd3eac 733 addr->addr.id.node = tipc_own_addr(sock_net(sk));
0c3141e9 734 }
b97bf3fd 735
b97bf3fd
PL
736 addr->addrtype = TIPC_ADDR_ID;
737 addr->family = AF_TIPC;
738 addr->scope = 0;
b97bf3fd
PL
739 addr->addr.name.domain = 0;
740
9b2c45d4 741 return sizeof(*addr);
b97bf3fd
PL
742}
743
744/**
a11e1d43 745 * tipc_poll - read and possibly block on pollmask
b97bf3fd
PL
746 * @file: file structure associated with the socket
747 * @sock: socket for which to calculate the poll bits
a11e1d43 748 * @wait: ???
b97bf3fd 749 *
9b674e82
AS
750 * Returns pollmask value
751 *
752 * COMMENTARY:
753 * It appears that the usual socket locking mechanisms are not useful here
754 * since the pollmask info is potentially out-of-date the moment this routine
755 * exits. TCP and other protocols seem to rely on higher level poll routines
756 * to handle any preventable race conditions, so TIPC will do the same ...
757 *
f662c070
AS
758 * IMPORTANT: The fact that a read or write operation is indicated does NOT
759 * imply that the operation will succeed, merely that it should be performed
760 * and will not block.
b97bf3fd 761 */
a11e1d43
LT
762static __poll_t tipc_poll(struct file *file, struct socket *sock,
763 poll_table *wait)
b97bf3fd 764{
9b674e82 765 struct sock *sk = sock->sk;
58ed9442 766 struct tipc_sock *tsk = tipc_sk(sk);
ade994f4 767 __poll_t revents = 0;
9b674e82 768
89ab066d 769 sock_poll_wait(file, sock, wait);
01e661eb 770 trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " ");
a11e1d43 771
6f00089c 772 if (sk->sk_shutdown & RCV_SHUTDOWN)
a9a08845 773 revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
6f00089c 774 if (sk->sk_shutdown == SHUTDOWN_MASK)
a9a08845 775 revents |= EPOLLHUP;
6f00089c 776
f40acbaf
PB
777 switch (sk->sk_state) {
778 case TIPC_ESTABLISHED:
365ad353 779 if (!tsk->cong_link_cnt && !tsk_conn_cong(tsk))
a9a08845 780 revents |= EPOLLOUT;
f79e3365 781 /* fall through */
f40acbaf 782 case TIPC_LISTEN:
ff946833 783 case TIPC_CONNECTING:
3ef7cf57 784 if (!skb_queue_empty_lockless(&sk->sk_receive_queue))
a9a08845 785 revents |= EPOLLIN | EPOLLRDNORM;
f40acbaf
PB
786 break;
787 case TIPC_OPEN:
60c25306 788 if (tsk->group_is_open && !tsk->cong_link_cnt)
a9a08845 789 revents |= EPOLLOUT;
ae236fb2
JM
790 if (!tipc_sk_type_connectionless(sk))
791 break;
3ef7cf57 792 if (skb_queue_empty_lockless(&sk->sk_receive_queue))
ae236fb2 793 break;
a9a08845 794 revents |= EPOLLIN | EPOLLRDNORM;
f40acbaf
PB
795 break;
796 case TIPC_DISCONNECTING:
a9a08845 797 revents = EPOLLIN | EPOLLRDNORM | EPOLLHUP;
f40acbaf 798 break;
f662c070 799 }
ae236fb2 800 return revents;
b97bf3fd
PL
801}
802
0abd8ff2
JPM
803/**
804 * tipc_sendmcast - send multicast message
805 * @sock: socket structure
806 * @seq: destination address
562640f3 807 * @msg: message to send
365ad353
JPM
808 * @dlen: length of data to send
809 * @timeout: timeout to wait for wakeup
0abd8ff2
JPM
810 *
811 * Called from function tipc_sendmsg(), which has done all sanity checks
812 * Returns the number of bytes sent on success, or errno
813 */
814static int tipc_sendmcast(struct socket *sock, struct tipc_name_seq *seq,
365ad353 815 struct msghdr *msg, size_t dlen, long timeout)
0abd8ff2
JPM
816{
817 struct sock *sk = sock->sk;
c5898636 818 struct tipc_sock *tsk = tipc_sk(sk);
365ad353 819 struct tipc_msg *hdr = &tsk->phdr;
f2f9800d 820 struct net *net = sock_net(sk);
365ad353 821 int mtu = tipc_bcast_get_mtu(net);
01fd12bb 822 struct tipc_mc_method *method = &tsk->mc_method;
365ad353 823 struct sk_buff_head pkts;
a853e4c6 824 struct tipc_nlist dsts;
0abd8ff2
JPM
825 int rc;
826
75da2163
JM
827 if (tsk->group)
828 return -EACCES;
829
a853e4c6 830 /* Block or return if any destination link is congested */
365ad353
JPM
831 rc = tipc_wait_for_cond(sock, &timeout, !tsk->cong_link_cnt);
832 if (unlikely(rc))
833 return rc;
f214fc40 834
a853e4c6
JPM
835 /* Lookup destination nodes */
836 tipc_nlist_init(&dsts, tipc_own_addr(net));
837 tipc_nametbl_lookup_dst_nodes(net, seq->type, seq->lower,
e9a03445 838 seq->upper, &dsts);
a853e4c6
JPM
839 if (!dsts.local && !dsts.remote)
840 return -EHOSTUNREACH;
841
842 /* Build message header */
365ad353 843 msg_set_type(hdr, TIPC_MCAST_MSG);
a853e4c6 844 msg_set_hdr_sz(hdr, MCAST_H_SIZE);
365ad353
JPM
845 msg_set_lookup_scope(hdr, TIPC_CLUSTER_SCOPE);
846 msg_set_destport(hdr, 0);
847 msg_set_destnode(hdr, 0);
848 msg_set_nametype(hdr, seq->type);
849 msg_set_namelower(hdr, seq->lower);
850 msg_set_nameupper(hdr, seq->upper);
365ad353 851
a853e4c6 852 /* Build message as chain of buffers */
e654f9f5 853 __skb_queue_head_init(&pkts);
365ad353 854 rc = tipc_msg_build(hdr, msg, 0, dlen, mtu, &pkts);
0abd8ff2 855
a853e4c6 856 /* Send message if build was successful */
01e661eb
TL
857 if (unlikely(rc == dlen)) {
858 trace_tipc_sk_sendmcast(sk, skb_peek(&pkts),
859 TIPC_DUMP_SK_SNDQ, " ");
01fd12bb 860 rc = tipc_mcast_xmit(net, &pkts, method, &dsts,
a853e4c6 861 &tsk->cong_link_cnt);
01e661eb 862 }
a853e4c6
JPM
863
864 tipc_nlist_purge(&dsts);
365ad353
JPM
865
866 return rc ? rc : dlen;
0abd8ff2
JPM
867}
868
27bd9ec0
JM
869/**
870 * tipc_send_group_msg - send a message to a member in the group
871 * @net: network namespace
872 * @m: message to send
873 * @mb: group member
874 * @dnode: destination node
875 * @dport: destination port
876 * @dlen: total length of message data
877 */
878static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
879 struct msghdr *m, struct tipc_member *mb,
880 u32 dnode, u32 dport, int dlen)
881{
b87a5ea3 882 u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group);
2f487712 883 struct tipc_mc_method *method = &tsk->mc_method;
27bd9ec0
JM
884 int blks = tsk_blocks(GROUP_H_SIZE + dlen);
885 struct tipc_msg *hdr = &tsk->phdr;
886 struct sk_buff_head pkts;
887 int mtu, rc;
888
889 /* Complete message header */
890 msg_set_type(hdr, TIPC_GRP_UCAST_MSG);
891 msg_set_hdr_sz(hdr, GROUP_H_SIZE);
892 msg_set_destport(hdr, dport);
893 msg_set_destnode(hdr, dnode);
b87a5ea3 894 msg_set_grp_bc_seqno(hdr, bc_snd_nxt);
27bd9ec0
JM
895
896 /* Build message as chain of buffers */
e654f9f5 897 __skb_queue_head_init(&pkts);
f73b1281 898 mtu = tipc_node_get_mtu(net, dnode, tsk->portid, false);
27bd9ec0
JM
899 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
900 if (unlikely(rc != dlen))
901 return rc;
902
903 /* Send message */
904 rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
905 if (unlikely(rc == -ELINKCONG)) {
906 tipc_dest_push(&tsk->cong_links, dnode, 0);
907 tsk->cong_link_cnt++;
908 }
909
2f487712 910 /* Update send window */
27bd9ec0
JM
911 tipc_group_update_member(mb, blks);
912
2f487712
JM
913 /* A broadcast sent within next EXPIRE period must follow same path */
914 method->rcast = true;
915 method->mandatory = true;
27bd9ec0
JM
916 return dlen;
917}
918
919/**
920 * tipc_send_group_unicast - send message to a member in the group
921 * @sock: socket structure
922 * @m: message to send
923 * @dlen: total length of message data
924 * @timeout: timeout to wait for wakeup
925 *
926 * Called from function tipc_sendmsg(), which has done all sanity checks
927 * Returns the number of bytes sent on success, or errno
928 */
929static int tipc_send_group_unicast(struct socket *sock, struct msghdr *m,
930 int dlen, long timeout)
931{
932 struct sock *sk = sock->sk;
933 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
934 int blks = tsk_blocks(GROUP_H_SIZE + dlen);
935 struct tipc_sock *tsk = tipc_sk(sk);
27bd9ec0
JM
936 struct net *net = sock_net(sk);
937 struct tipc_member *mb = NULL;
938 u32 node, port;
939 int rc;
940
941 node = dest->addr.id.node;
942 port = dest->addr.id.ref;
943 if (!port && !node)
944 return -EHOSTUNREACH;
945
946 /* Block or return if destination link or member is congested */
947 rc = tipc_wait_for_cond(sock, &timeout,
948 !tipc_dest_find(&tsk->cong_links, node, 0) &&
143ece65
CW
949 tsk->group &&
950 !tipc_group_cong(tsk->group, node, port, blks,
951 &mb));
27bd9ec0
JM
952 if (unlikely(rc))
953 return rc;
954
955 if (unlikely(!mb))
956 return -EHOSTUNREACH;
957
958 rc = tipc_send_group_msg(net, tsk, m, mb, node, port, dlen);
959
960 return rc ? rc : dlen;
961}
962
ee106d7f
JM
963/**
964 * tipc_send_group_anycast - send message to any member with given identity
965 * @sock: socket structure
966 * @m: message to send
967 * @dlen: total length of message data
968 * @timeout: timeout to wait for wakeup
969 *
970 * Called from function tipc_sendmsg(), which has done all sanity checks
971 * Returns the number of bytes sent on success, or errno
972 */
973static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m,
974 int dlen, long timeout)
975{
976 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
977 struct sock *sk = sock->sk;
978 struct tipc_sock *tsk = tipc_sk(sk);
979 struct list_head *cong_links = &tsk->cong_links;
980 int blks = tsk_blocks(GROUP_H_SIZE + dlen);
232d07b7 981 struct tipc_msg *hdr = &tsk->phdr;
ee106d7f
JM
982 struct tipc_member *first = NULL;
983 struct tipc_member *mbr = NULL;
984 struct net *net = sock_net(sk);
985 u32 node, port, exclude;
ee106d7f 986 struct list_head dsts;
232d07b7 987 u32 type, inst, scope;
ee106d7f
JM
988 int lookups = 0;
989 int dstcnt, rc;
990 bool cong;
991
992 INIT_LIST_HEAD(&dsts);
993
232d07b7 994 type = msg_nametype(hdr);
ee106d7f 995 inst = dest->addr.name.name.instance;
232d07b7 996 scope = msg_lookup_scope(hdr);
ee106d7f
JM
997
998 while (++lookups < 4) {
143ece65
CW
999 exclude = tipc_group_exclude(tsk->group);
1000
ee106d7f
JM
1001 first = NULL;
1002
1003 /* Look for a non-congested destination member, if any */
1004 while (1) {
232d07b7 1005 if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
ee106d7f
JM
1006 &dstcnt, exclude, false))
1007 return -EHOSTUNREACH;
1008 tipc_dest_pop(&dsts, &node, &port);
143ece65
CW
1009 cong = tipc_group_cong(tsk->group, node, port, blks,
1010 &mbr);
ee106d7f
JM
1011 if (!cong)
1012 break;
1013 if (mbr == first)
1014 break;
1015 if (!first)
1016 first = mbr;
1017 }
1018
1019 /* Start over if destination was not in member list */
1020 if (unlikely(!mbr))
1021 continue;
1022
1023 if (likely(!cong && !tipc_dest_find(cong_links, node, 0)))
1024 break;
1025
1026 /* Block or return if destination link or member is congested */
1027 rc = tipc_wait_for_cond(sock, &timeout,
1028 !tipc_dest_find(cong_links, node, 0) &&
143ece65
CW
1029 tsk->group &&
1030 !tipc_group_cong(tsk->group, node, port,
ee106d7f
JM
1031 blks, &mbr));
1032 if (unlikely(rc))
1033 return rc;
1034
1035 /* Send, unless destination disappeared while waiting */
1036 if (likely(mbr))
1037 break;
1038 }
1039
1040 if (unlikely(lookups >= 4))
1041 return -EHOSTUNREACH;
1042
1043 rc = tipc_send_group_msg(net, tsk, m, mbr, node, port, dlen);
1044
1045 return rc ? rc : dlen;
1046}
1047
75da2163
JM
1048/**
1049 * tipc_send_group_bcast - send message to all members in communication group
1050 * @sk: socket structure
1051 * @m: message to send
1052 * @dlen: total length of message data
1053 * @timeout: timeout to wait for wakeup
1054 *
1055 * Called from function tipc_sendmsg(), which has done all sanity checks
1056 * Returns the number of bytes sent on success, or errno
1057 */
1058static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
1059 int dlen, long timeout)
1060{
5b8dddb6 1061 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
75da2163
JM
1062 struct sock *sk = sock->sk;
1063 struct net *net = sock_net(sk);
1064 struct tipc_sock *tsk = tipc_sk(sk);
3c6306d4 1065 struct tipc_nlist *dsts;
75da2163 1066 struct tipc_mc_method *method = &tsk->mc_method;
2f487712 1067 bool ack = method->mandatory && method->rcast;
b7d42635 1068 int blks = tsk_blocks(MCAST_H_SIZE + dlen);
75da2163
JM
1069 struct tipc_msg *hdr = &tsk->phdr;
1070 int mtu = tipc_bcast_get_mtu(net);
1071 struct sk_buff_head pkts;
1072 int rc = -EHOSTUNREACH;
1073
b7d42635 1074 /* Block or return if any destination link or member is congested */
143ece65
CW
1075 rc = tipc_wait_for_cond(sock, &timeout,
1076 !tsk->cong_link_cnt && tsk->group &&
1077 !tipc_group_bc_cong(tsk->group, blks));
75da2163
JM
1078 if (unlikely(rc))
1079 return rc;
1080
3c6306d4
CW
1081 dsts = tipc_group_dests(tsk->group);
1082 if (!dsts->local && !dsts->remote)
1083 return -EHOSTUNREACH;
1084
75da2163 1085 /* Complete message header */
5b8dddb6
JM
1086 if (dest) {
1087 msg_set_type(hdr, TIPC_GRP_MCAST_MSG);
1088 msg_set_nameinst(hdr, dest->addr.name.name.instance);
1089 } else {
1090 msg_set_type(hdr, TIPC_GRP_BCAST_MSG);
1091 msg_set_nameinst(hdr, 0);
1092 }
b7d42635 1093 msg_set_hdr_sz(hdr, GROUP_H_SIZE);
75da2163
JM
1094 msg_set_destport(hdr, 0);
1095 msg_set_destnode(hdr, 0);
143ece65 1096 msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(tsk->group));
75da2163 1097
2f487712
JM
1098 /* Avoid getting stuck with repeated forced replicasts */
1099 msg_set_grp_bc_ack_req(hdr, ack);
1100
75da2163 1101 /* Build message as chain of buffers */
e654f9f5 1102 __skb_queue_head_init(&pkts);
75da2163
JM
1103 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1104 if (unlikely(rc != dlen))
1105 return rc;
1106
1107 /* Send message */
2f487712 1108 rc = tipc_mcast_xmit(net, &pkts, method, dsts, &tsk->cong_link_cnt);
75da2163
JM
1109 if (unlikely(rc))
1110 return rc;
1111
b7d42635 1112 /* Update broadcast sequence number and send windows */
2f487712
JM
1113 tipc_group_update_bc_members(tsk->group, blks, ack);
1114
1115 /* Broadcast link is now free to choose method for next broadcast */
1116 method->mandatory = false;
1117 method->expires = jiffies;
1118
75da2163
JM
1119 return dlen;
1120}
1121
5b8dddb6
JM
1122/**
1123 * tipc_send_group_mcast - send message to all members with given identity
1124 * @sock: socket structure
1125 * @m: message to send
1126 * @dlen: total length of message data
1127 * @timeout: timeout to wait for wakeup
1128 *
1129 * Called from function tipc_sendmsg(), which has done all sanity checks
1130 * Returns the number of bytes sent on success, or errno
1131 */
1132static int tipc_send_group_mcast(struct socket *sock, struct msghdr *m,
1133 int dlen, long timeout)
1134{
1135 struct sock *sk = sock->sk;
1136 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
5b8dddb6
JM
1137 struct tipc_sock *tsk = tipc_sk(sk);
1138 struct tipc_group *grp = tsk->group;
232d07b7 1139 struct tipc_msg *hdr = &tsk->phdr;
5b8dddb6 1140 struct net *net = sock_net(sk);
232d07b7 1141 u32 type, inst, scope, exclude;
5b8dddb6 1142 struct list_head dsts;
232d07b7 1143 u32 dstcnt;
5b8dddb6
JM
1144
1145 INIT_LIST_HEAD(&dsts);
1146
232d07b7
JM
1147 type = msg_nametype(hdr);
1148 inst = dest->addr.name.name.instance;
1149 scope = msg_lookup_scope(hdr);
5b8dddb6 1150 exclude = tipc_group_exclude(grp);
232d07b7
JM
1151
1152 if (!tipc_nametbl_lookup(net, type, inst, scope, &dsts,
1153 &dstcnt, exclude, true))
5b8dddb6
JM
1154 return -EHOSTUNREACH;
1155
1156 if (dstcnt == 1) {
1157 tipc_dest_pop(&dsts, &dest->addr.id.node, &dest->addr.id.ref);
1158 return tipc_send_group_unicast(sock, m, dlen, timeout);
1159 }
1160
1161 tipc_dest_list_purge(&dsts);
1162 return tipc_send_group_bcast(sock, m, dlen, timeout);
1163}
1164
cb1b7280
JPM
1165/**
1166 * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
1167 * @arrvq: queue with arriving messages, to be cloned after destination lookup
1168 * @inputq: queue with cloned messages, delivered to socket after dest lookup
1169 *
1170 * Multi-threaded: parallel calls with reference to same queues may occur
078bec82 1171 */
cb1b7280
JPM
1172void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
1173 struct sk_buff_head *inputq)
078bec82 1174{
75da2163 1175 u32 self = tipc_own_addr(net);
232d07b7 1176 u32 type, lower, upper, scope;
cb1b7280 1177 struct sk_buff *skb, *_skb;
b053fcc4 1178 u32 portid, onode;
232d07b7 1179 struct sk_buff_head tmpq;
75da2163 1180 struct list_head dports;
232d07b7
JM
1181 struct tipc_msg *hdr;
1182 int user, mtyp, hlen;
1183 bool exact;
3c724acd 1184
cb1b7280 1185 __skb_queue_head_init(&tmpq);
4d8642d8 1186 INIT_LIST_HEAD(&dports);
078bec82 1187
cb1b7280
JPM
1188 skb = tipc_skb_peek(arrvq, &inputq->lock);
1189 for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) {
232d07b7
JM
1190 hdr = buf_msg(skb);
1191 user = msg_user(hdr);
1192 mtyp = msg_type(hdr);
1193 hlen = skb_headroom(skb) + msg_hdr_sz(hdr);
232d07b7
JM
1194 onode = msg_orignode(hdr);
1195 type = msg_nametype(hdr);
1196
2f487712
JM
1197 if (mtyp == TIPC_GRP_UCAST_MSG || user == GROUP_PROTOCOL) {
1198 spin_lock_bh(&inputq->lock);
1199 if (skb_peek(arrvq) == skb) {
1200 __skb_dequeue(arrvq);
1201 __skb_queue_tail(inputq, skb);
1202 }
c545a945 1203 kfree_skb(skb);
2f487712
JM
1204 spin_unlock_bh(&inputq->lock);
1205 continue;
1206 }
232d07b7
JM
1207
1208 /* Group messages require exact scope match */
1209 if (msg_in_group(hdr)) {
1210 lower = 0;
1211 upper = ~0;
1212 scope = msg_lookup_scope(hdr);
1213 exact = true;
1214 } else {
1215 /* TIPC_NODE_SCOPE means "any scope" in this context */
1216 if (onode == self)
1217 scope = TIPC_NODE_SCOPE;
1218 else
1219 scope = TIPC_CLUSTER_SCOPE;
1220 exact = false;
1221 lower = msg_namelower(hdr);
1222 upper = msg_nameupper(hdr);
75da2163 1223 }
232d07b7
JM
1224
1225 /* Create destination port list: */
1226 tipc_nametbl_mc_lookup(net, type, lower, upper,
1227 scope, exact, &dports);
1228
1229 /* Clone message per destination */
a80ae530 1230 while (tipc_dest_pop(&dports, NULL, &portid)) {
232d07b7 1231 _skb = __pskb_copy(skb, hlen, GFP_ATOMIC);
cb1b7280
JPM
1232 if (_skb) {
1233 msg_set_destport(buf_msg(_skb), portid);
1234 __skb_queue_tail(&tmpq, _skb);
1235 continue;
1236 }
1237 pr_warn("Failed to clone mcast rcv buffer\n");
078bec82 1238 }
cb1b7280
JPM
1239 /* Append to inputq if not already done by other thread */
1240 spin_lock_bh(&inputq->lock);
1241 if (skb_peek(arrvq) == skb) {
1242 skb_queue_splice_tail_init(&tmpq, inputq);
1243 kfree_skb(__skb_dequeue(arrvq));
1244 }
1245 spin_unlock_bh(&inputq->lock);
1246 __skb_queue_purge(&tmpq);
1247 kfree_skb(skb);
078bec82 1248 }
cb1b7280 1249 tipc_sk_rcv(net, inputq);
078bec82
JPM
1250}
1251
c0bceb97
JM
1252/* tipc_sk_push_backlog(): send accumulated buffers in socket write queue
1253 * when socket is in Nagle mode
1254 */
1255static void tipc_sk_push_backlog(struct tipc_sock *tsk)
1256{
1257 struct sk_buff_head *txq = &tsk->sk.sk_write_queue;
1258 struct net *net = sock_net(&tsk->sk);
1259 u32 dnode = tsk_peer_node(tsk);
d34910e1 1260 struct sk_buff *skb = skb_peek(txq);
c0bceb97
JM
1261 int rc;
1262
d34910e1
TN
1263 if (!skb || tsk->cong_link_cnt)
1264 return;
1265
1266 /* Do not send SYN again after congestion */
1267 if (msg_is_syn(buf_msg(skb)))
c0bceb97
JM
1268 return;
1269
1270 tsk->snt_unacked += tsk->snd_backlog;
1271 tsk->snd_backlog = 0;
1272 tsk->expect_ack = true;
1273 rc = tipc_node_xmit(net, txq, dnode, tsk->portid);
1274 if (rc == -ELINKCONG)
1275 tsk->cong_link_cnt = 1;
1276}
1277
ac0074ee 1278/**
64ac5f59 1279 * tipc_sk_conn_proto_rcv - receive a connection mng protocol message
ac0074ee 1280 * @tsk: receiving socket
bcd3ffd4 1281 * @skb: pointer to message buffer.
ac0074ee 1282 */
64ac5f59 1283static void tipc_sk_conn_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb,
e7eb0582 1284 struct sk_buff_head *inputq,
64ac5f59 1285 struct sk_buff_head *xmitq)
ac0074ee 1286{
bcd3ffd4 1287 struct tipc_msg *hdr = buf_msg(skb);
64ac5f59
JM
1288 u32 onode = tsk_own_node(tsk);
1289 struct sock *sk = &tsk->sk;
bcd3ffd4 1290 int mtyp = msg_type(hdr);
c0bceb97 1291 bool was_cong;
bcd3ffd4 1292
ac0074ee 1293 /* Ignore if connection cannot be validated: */
01e661eb
TL
1294 if (!tsk_peer_msg(tsk, hdr)) {
1295 trace_tipc_sk_drop_msg(sk, skb, TIPC_DUMP_NONE, "@proto_rcv!");
ac0074ee 1296 goto exit;
01e661eb 1297 }
ac0074ee 1298
c1be7756
PB
1299 if (unlikely(msg_errcode(hdr))) {
1300 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
1301 tipc_node_remove_conn(sock_net(sk), tsk_peer_node(tsk),
1302 tsk_peer_port(tsk));
1303 sk->sk_state_change(sk);
e7eb0582
PB
1304
1305 /* State change is ignored if socket already awake,
1306 * - convert msg to abort msg and add to inqueue
1307 */
1308 msg_set_user(hdr, TIPC_CRITICAL_IMPORTANCE);
1309 msg_set_type(hdr, TIPC_CONN_MSG);
1310 msg_set_size(hdr, BASIC_H_SIZE);
1311 msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1312 __skb_queue_tail(inputq, skb);
1313 return;
c1be7756
PB
1314 }
1315
8ea642ee 1316 tsk->probe_unacked = false;
ac0074ee 1317
bcd3ffd4
JPM
1318 if (mtyp == CONN_PROBE) {
1319 msg_set_type(hdr, CONN_PROBE_REPLY);
f1d048f2
JPM
1320 if (tipc_msg_reverse(onode, &skb, TIPC_OK))
1321 __skb_queue_tail(xmitq, skb);
bcd3ffd4
JPM
1322 return;
1323 } else if (mtyp == CONN_ACK) {
c0bceb97
JM
1324 was_cong = tsk_conn_cong(tsk);
1325 tsk->expect_ack = false;
1326 tipc_sk_push_backlog(tsk);
10724cc7
JPM
1327 tsk->snt_unacked -= msg_conn_ack(hdr);
1328 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1329 tsk->snd_win = msg_adv_win(hdr);
c0bceb97 1330 if (was_cong && !tsk_conn_cong(tsk))
bcd3ffd4
JPM
1331 sk->sk_write_space(sk);
1332 } else if (mtyp != CONN_PROBE_REPLY) {
1333 pr_warn("Received unknown CONN_PROTO msg\n");
ac0074ee 1334 }
ac0074ee 1335exit:
bcd3ffd4 1336 kfree_skb(skb);
ac0074ee
JPM
1337}
1338
b97bf3fd 1339/**
247f0f3c 1340 * tipc_sendmsg - send message in connectionless manner
b97bf3fd
PL
1341 * @sock: socket structure
1342 * @m: message to send
e2dafe87 1343 * @dsz: amount of user data to be sent
c4307285 1344 *
b97bf3fd 1345 * Message must have an destination specified explicitly.
c4307285 1346 * Used for SOCK_RDM and SOCK_DGRAM messages,
b97bf3fd
PL
1347 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
1348 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
c4307285 1349 *
b97bf3fd
PL
1350 * Returns the number of bytes sent on success, or errno otherwise
1351 */
1b784140 1352static int tipc_sendmsg(struct socket *sock,
e2dafe87 1353 struct msghdr *m, size_t dsz)
39a0295f
YX
1354{
1355 struct sock *sk = sock->sk;
1356 int ret;
1357
1358 lock_sock(sk);
1359 ret = __tipc_sendmsg(sock, m, dsz);
1360 release_sock(sk);
1361
1362 return ret;
1363}
1364
365ad353 1365static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)
b97bf3fd 1366{
0c3141e9 1367 struct sock *sk = sock->sk;
f2f9800d 1368 struct net *net = sock_net(sk);
365ad353
JPM
1369 struct tipc_sock *tsk = tipc_sk(sk);
1370 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
1371 long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1372 struct list_head *clinks = &tsk->cong_links;
1373 bool syn = !tipc_sk_type_connectionless(sk);
75da2163 1374 struct tipc_group *grp = tsk->group;
365ad353 1375 struct tipc_msg *hdr = &tsk->phdr;
f2f8036e 1376 struct tipc_name_seq *seq;
365ad353 1377 struct sk_buff_head pkts;
abc9b4e0
TL
1378 u32 dport = 0, dnode = 0;
1379 u32 type = 0, inst = 0;
365ad353 1380 int mtu, rc;
b97bf3fd 1381
365ad353 1382 if (unlikely(dlen > TIPC_MAX_USER_MSG_SIZE))
c29c3f70 1383 return -EMSGSIZE;
365ad353 1384
27bd9ec0
JM
1385 if (likely(dest)) {
1386 if (unlikely(m->msg_namelen < sizeof(*dest)))
1387 return -EINVAL;
1388 if (unlikely(dest->family != AF_TIPC))
1389 return -EINVAL;
1390 }
1391
1392 if (grp) {
1393 if (!dest)
1394 return tipc_send_group_bcast(sock, m, dlen, timeout);
ee106d7f
JM
1395 if (dest->addrtype == TIPC_ADDR_NAME)
1396 return tipc_send_group_anycast(sock, m, dlen, timeout);
27bd9ec0
JM
1397 if (dest->addrtype == TIPC_ADDR_ID)
1398 return tipc_send_group_unicast(sock, m, dlen, timeout);
5b8dddb6
JM
1399 if (dest->addrtype == TIPC_ADDR_MCAST)
1400 return tipc_send_group_mcast(sock, m, dlen, timeout);
27bd9ec0
JM
1401 return -EINVAL;
1402 }
75da2163 1403
f2f8036e 1404 if (unlikely(!dest)) {
365ad353 1405 dest = &tsk->peer;
0e632089 1406 if (!syn && dest->family != AF_TIPC)
f2f8036e 1407 return -EDESTADDRREQ;
f2f8036e 1408 }
365ad353 1409
365ad353 1410 if (unlikely(syn)) {
0c288c86 1411 if (sk->sk_state == TIPC_LISTEN)
39a0295f 1412 return -EPIPE;
438adcaf 1413 if (sk->sk_state != TIPC_OPEN)
39a0295f
YX
1414 return -EISCONN;
1415 if (tsk->published)
1416 return -EOPNOTSUPP;
3388007b 1417 if (dest->addrtype == TIPC_ADDR_NAME) {
301bae56
JPM
1418 tsk->conn_type = dest->addr.name.name.type;
1419 tsk->conn_instance = dest->addr.name.name.instance;
3388007b 1420 }
25b9221b 1421 msg_set_syn(hdr, 1);
b97bf3fd 1422 }
e2dafe87 1423
365ad353
JPM
1424 seq = &dest->addr.nameseq;
1425 if (dest->addrtype == TIPC_ADDR_MCAST)
1426 return tipc_sendmcast(sock, seq, m, dlen, timeout);
e2dafe87 1427
365ad353
JPM
1428 if (dest->addrtype == TIPC_ADDR_NAME) {
1429 type = dest->addr.name.name.type;
1430 inst = dest->addr.name.name.instance;
928df188 1431 dnode = dest->addr.name.domain;
4ac1c8d0 1432 dport = tipc_nametbl_translate(net, type, inst, &dnode);
39a0295f
YX
1433 if (unlikely(!dport && !dnode))
1434 return -EHOSTUNREACH;
e2dafe87
JPM
1435 } else if (dest->addrtype == TIPC_ADDR_ID) {
1436 dnode = dest->addr.id.node;
335b929b
JM
1437 } else {
1438 return -EINVAL;
e2dafe87
JPM
1439 }
1440
365ad353 1441 /* Block or return if destination link is congested */
a80ae530
JM
1442 rc = tipc_wait_for_cond(sock, &timeout,
1443 !tipc_dest_find(clinks, dnode, 0));
365ad353
JPM
1444 if (unlikely(rc))
1445 return rc;
1446
abc9b4e0
TL
1447 if (dest->addrtype == TIPC_ADDR_NAME) {
1448 msg_set_type(hdr, TIPC_NAMED_MSG);
1449 msg_set_hdr_sz(hdr, NAMED_H_SIZE);
1450 msg_set_nametype(hdr, type);
1451 msg_set_nameinst(hdr, inst);
1452 msg_set_lookup_scope(hdr, tipc_node2scope(dnode));
1453 msg_set_destnode(hdr, dnode);
1454 msg_set_destport(hdr, dport);
1455 } else { /* TIPC_ADDR_ID */
1456 msg_set_type(hdr, TIPC_DIRECT_MSG);
1457 msg_set_lookup_scope(hdr, 0);
1458 msg_set_destnode(hdr, dnode);
1459 msg_set_destport(hdr, dest->addr.id.ref);
1460 msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1461 }
1462
e654f9f5 1463 __skb_queue_head_init(&pkts);
8b1e5b0a 1464 mtu = tipc_node_get_mtu(net, dnode, tsk->portid, true);
365ad353
JPM
1465 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1466 if (unlikely(rc != dlen))
39a0295f 1467 return rc;
2fe97a57
TN
1468 if (unlikely(syn && !tipc_msg_skb_clone(&pkts, &sk->sk_write_queue))) {
1469 __skb_queue_purge(&pkts);
67879274 1470 return -ENOMEM;
2fe97a57 1471 }
e2dafe87 1472
01e661eb 1473 trace_tipc_sk_sendmsg(sk, skb_peek(&pkts), TIPC_DUMP_SK_SNDQ, " ");
365ad353
JPM
1474 rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
1475 if (unlikely(rc == -ELINKCONG)) {
a80ae530 1476 tipc_dest_push(clinks, dnode, 0);
365ad353
JPM
1477 tsk->cong_link_cnt++;
1478 rc = 0;
1479 }
e2dafe87 1480
365ad353
JPM
1481 if (unlikely(syn && !rc))
1482 tipc_set_sk_state(sk, TIPC_CONNECTING);
1483
1484 return rc ? rc : dlen;
b97bf3fd
PL
1485}
1486
c4307285 1487/**
365ad353 1488 * tipc_sendstream - send stream-oriented data
b97bf3fd 1489 * @sock: socket structure
4ccfe5e0
JPM
1490 * @m: data to send
1491 * @dsz: total length of data to be transmitted
c4307285 1492 *
4ccfe5e0 1493 * Used for SOCK_STREAM data.
c4307285 1494 *
4ccfe5e0
JPM
1495 * Returns the number of bytes sent on success (or partial success),
1496 * or errno if no data sent
b97bf3fd 1497 */
365ad353 1498static int tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz)
39a0295f
YX
1499{
1500 struct sock *sk = sock->sk;
1501 int ret;
1502
1503 lock_sock(sk);
365ad353 1504 ret = __tipc_sendstream(sock, m, dsz);
39a0295f
YX
1505 release_sock(sk);
1506
1507 return ret;
1508}
1509
365ad353 1510static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dlen)
b97bf3fd 1511{
0c3141e9 1512 struct sock *sk = sock->sk;
342dfc30 1513 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
365ad353 1514 long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
c0bceb97 1515 struct sk_buff_head *txq = &sk->sk_write_queue;
365ad353
JPM
1516 struct tipc_sock *tsk = tipc_sk(sk);
1517 struct tipc_msg *hdr = &tsk->phdr;
1518 struct net *net = sock_net(sk);
365ad353 1519 u32 dnode = tsk_peer_node(tsk);
c0bceb97
JM
1520 int maxnagle = tsk->maxnagle;
1521 int maxpkt = tsk->max_pkt;
365ad353 1522 int send, sent = 0;
c0bceb97 1523 int blocks, rc = 0;
7cf87fa2 1524
365ad353
JPM
1525 if (unlikely(dlen > INT_MAX))
1526 return -EMSGSIZE;
4ccfe5e0 1527
365ad353
JPM
1528 /* Handle implicit connection setup */
1529 if (unlikely(dest)) {
1530 rc = __tipc_sendmsg(sock, m, dlen);
92ef12b3
PB
1531 if (dlen && dlen == rc) {
1532 tsk->peer_caps = tipc_node_get_capabilities(net, dnode);
365ad353 1533 tsk->snt_unacked = tsk_inc(tsk, dlen + msg_hdr_sz(hdr));
92ef12b3 1534 }
39a0295f 1535 return rc;
365ad353 1536 }
f214fc40 1537
c4307285 1538 do {
365ad353
JPM
1539 rc = tipc_wait_for_cond(sock, &timeout,
1540 (!tsk->cong_link_cnt &&
8c44e1af
JPM
1541 !tsk_conn_cong(tsk) &&
1542 tipc_sk_connected(sk)));
365ad353
JPM
1543 if (unlikely(rc))
1544 break;
365ad353 1545 send = min_t(size_t, dlen - sent, TIPC_MAX_USER_MSG_SIZE);
c0bceb97
JM
1546 blocks = tsk->snd_backlog;
1547 if (tsk->oneway++ >= 4 && send <= maxnagle) {
1548 rc = tipc_msg_append(hdr, m, send, maxnagle, txq);
1549 if (unlikely(rc < 0))
1550 break;
1551 blocks += rc;
1552 if (blocks <= 64 && tsk->expect_ack) {
1553 tsk->snd_backlog = blocks;
1554 sent += send;
1555 break;
1556 }
1557 tsk->expect_ack = true;
1558 } else {
1559 rc = tipc_msg_build(hdr, m, sent, send, maxpkt, txq);
1560 if (unlikely(rc != send))
1561 break;
1562 blocks += tsk_inc(tsk, send + MIN_H_SIZE);
1563 }
1564 trace_tipc_sk_sendstream(sk, skb_peek(txq),
01e661eb 1565 TIPC_DUMP_SK_SNDQ, " ");
c0bceb97 1566 rc = tipc_node_xmit(net, txq, dnode, tsk->portid);
365ad353
JPM
1567 if (unlikely(rc == -ELINKCONG)) {
1568 tsk->cong_link_cnt = 1;
1569 rc = 0;
1570 }
1571 if (likely(!rc)) {
c0bceb97
JM
1572 tsk->snt_unacked += blocks;
1573 tsk->snd_backlog = 0;
365ad353
JPM
1574 sent += send;
1575 }
1576 } while (sent < dlen && !rc);
39a0295f 1577
3364d61c 1578 return sent ? sent : rc;
b97bf3fd
PL
1579}
1580
c4307285 1581/**
4ccfe5e0 1582 * tipc_send_packet - send a connection-oriented message
b97bf3fd 1583 * @sock: socket structure
4ccfe5e0
JPM
1584 * @m: message to send
1585 * @dsz: length of data to be transmitted
c4307285 1586 *
4ccfe5e0 1587 * Used for SOCK_SEQPACKET messages.
c4307285 1588 *
4ccfe5e0 1589 * Returns the number of bytes sent on success, or errno otherwise
b97bf3fd 1590 */
1b784140 1591static int tipc_send_packet(struct socket *sock, struct msghdr *m, size_t dsz)
b97bf3fd 1592{
4ccfe5e0
JPM
1593 if (dsz > TIPC_MAX_USER_MSG_SIZE)
1594 return -EMSGSIZE;
b97bf3fd 1595
365ad353 1596 return tipc_sendstream(sock, m, dsz);
b97bf3fd
PL
1597}
1598
dadebc00 1599/* tipc_sk_finish_conn - complete the setup of a connection
b97bf3fd 1600 */
301bae56 1601static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port,
dadebc00 1602 u32 peer_node)
b97bf3fd 1603{
3721e9c7
YX
1604 struct sock *sk = &tsk->sk;
1605 struct net *net = sock_net(sk);
301bae56 1606 struct tipc_msg *msg = &tsk->phdr;
b97bf3fd 1607
25b9221b 1608 msg_set_syn(msg, 0);
dadebc00
JPM
1609 msg_set_destnode(msg, peer_node);
1610 msg_set_destport(msg, peer_port);
1611 msg_set_type(msg, TIPC_CONN_MSG);
1612 msg_set_lookup_scope(msg, 0);
1613 msg_set_hdr_sz(msg, SHORT_H_SIZE);
584d24b3 1614
0d5fcebf 1615 sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
8ea642ee 1616 tipc_set_sk_state(sk, TIPC_ESTABLISHED);
f2f9800d 1617 tipc_node_add_conn(net, peer_node, tsk->portid, peer_port);
f73b1281 1618 tsk->max_pkt = tipc_node_get_mtu(net, peer_node, tsk->portid, true);
60020e18 1619 tsk->peer_caps = tipc_node_get_capabilities(net, peer_node);
c0bceb97 1620 tsk_set_nagle(tsk);
67879274 1621 __skb_queue_purge(&sk->sk_write_queue);
10724cc7
JPM
1622 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1623 return;
1624
1625 /* Fall back to message based flow control */
1626 tsk->rcv_win = FLOWCTL_MSG_WIN;
1627 tsk->snd_win = FLOWCTL_MSG_WIN;
b97bf3fd
PL
1628}
1629
1630/**
31c82a2d 1631 * tipc_sk_set_orig_addr - capture sender's address for received message
b97bf3fd 1632 * @m: descriptor for message info
31c82a2d 1633 * @hdr: received message header
c4307285 1634 *
b97bf3fd
PL
1635 * Note: Address is not captured if not requested by receiver.
1636 */
31c82a2d 1637static void tipc_sk_set_orig_addr(struct msghdr *m, struct sk_buff *skb)
b97bf3fd 1638{
31c82a2d
JM
1639 DECLARE_SOCKADDR(struct sockaddr_pair *, srcaddr, m->msg_name);
1640 struct tipc_msg *hdr = buf_msg(skb);
1641
1642 if (!srcaddr)
1643 return;
1644
1645 srcaddr->sock.family = AF_TIPC;
1646 srcaddr->sock.addrtype = TIPC_ADDR_ID;
09c8b971 1647 srcaddr->sock.scope = 0;
31c82a2d
JM
1648 srcaddr->sock.addr.id.ref = msg_origport(hdr);
1649 srcaddr->sock.addr.id.node = msg_orignode(hdr);
1650 srcaddr->sock.addr.name.domain = 0;
31c82a2d
JM
1651 m->msg_namelen = sizeof(struct sockaddr_tipc);
1652
1653 if (!msg_in_group(hdr))
1654 return;
1655
1656 /* Group message users may also want to know sending member's id */
1657 srcaddr->member.family = AF_TIPC;
1658 srcaddr->member.addrtype = TIPC_ADDR_NAME;
09c8b971 1659 srcaddr->member.scope = 0;
31c82a2d
JM
1660 srcaddr->member.addr.name.name.type = msg_nametype(hdr);
1661 srcaddr->member.addr.name.name.instance = TIPC_SKB_CB(skb)->orig_member;
1662 srcaddr->member.addr.name.domain = 0;
1663 m->msg_namelen = sizeof(*srcaddr);
b97bf3fd
PL
1664}
1665
1666/**
301bae56 1667 * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
b97bf3fd 1668 * @m: descriptor for message info
1c1274a5 1669 * @skb: received message buffer
301bae56 1670 * @tsk: TIPC port associated with message
c4307285 1671 *
b97bf3fd 1672 * Note: Ancillary data is not captured if not requested by receiver.
c4307285 1673 *
b97bf3fd
PL
1674 * Returns 0 if successful, otherwise errno
1675 */
1c1274a5 1676static int tipc_sk_anc_data_recv(struct msghdr *m, struct sk_buff *skb,
301bae56 1677 struct tipc_sock *tsk)
b97bf3fd 1678{
1c1274a5 1679 struct tipc_msg *msg;
b97bf3fd
PL
1680 u32 anc_data[3];
1681 u32 err;
1682 u32 dest_type;
3546c750 1683 int has_name;
b97bf3fd
PL
1684 int res;
1685
1686 if (likely(m->msg_controllen == 0))
1687 return 0;
1c1274a5 1688 msg = buf_msg(skb);
b97bf3fd
PL
1689
1690 /* Optionally capture errored message object(s) */
b97bf3fd
PL
1691 err = msg ? msg_errcode(msg) : 0;
1692 if (unlikely(err)) {
1693 anc_data[0] = err;
1694 anc_data[1] = msg_data_sz(msg);
2db9983a
AS
1695 res = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, anc_data);
1696 if (res)
b97bf3fd 1697 return res;
2db9983a 1698 if (anc_data[1]) {
1c1274a5
JM
1699 if (skb_linearize(skb))
1700 return -ENOMEM;
1701 msg = buf_msg(skb);
2db9983a
AS
1702 res = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, anc_data[1],
1703 msg_data(msg));
1704 if (res)
1705 return res;
1706 }
b97bf3fd
PL
1707 }
1708
1709 /* Optionally capture message destination object */
b97bf3fd
PL
1710 dest_type = msg ? msg_type(msg) : TIPC_DIRECT_MSG;
1711 switch (dest_type) {
1712 case TIPC_NAMED_MSG:
3546c750 1713 has_name = 1;
b97bf3fd
PL
1714 anc_data[0] = msg_nametype(msg);
1715 anc_data[1] = msg_namelower(msg);
1716 anc_data[2] = msg_namelower(msg);
1717 break;
1718 case TIPC_MCAST_MSG:
3546c750 1719 has_name = 1;
b97bf3fd
PL
1720 anc_data[0] = msg_nametype(msg);
1721 anc_data[1] = msg_namelower(msg);
1722 anc_data[2] = msg_nameupper(msg);
1723 break;
1724 case TIPC_CONN_MSG:
301bae56
JPM
1725 has_name = (tsk->conn_type != 0);
1726 anc_data[0] = tsk->conn_type;
1727 anc_data[1] = tsk->conn_instance;
1728 anc_data[2] = tsk->conn_instance;
b97bf3fd
PL
1729 break;
1730 default:
3546c750 1731 has_name = 0;
b97bf3fd 1732 }
2db9983a
AS
1733 if (has_name) {
1734 res = put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, anc_data);
1735 if (res)
1736 return res;
1737 }
b97bf3fd
PL
1738
1739 return 0;
1740}
1741
c7268589 1742static struct sk_buff *tipc_sk_build_ack(struct tipc_sock *tsk)
739f5e4e 1743{
d6fb7e9c 1744 struct sock *sk = &tsk->sk;
a6ca1094 1745 struct sk_buff *skb = NULL;
739f5e4e 1746 struct tipc_msg *msg;
301bae56
JPM
1747 u32 peer_port = tsk_peer_port(tsk);
1748 u32 dnode = tsk_peer_node(tsk);
739f5e4e 1749
d6fb7e9c 1750 if (!tipc_sk_connected(sk))
c7268589 1751 return NULL;
c5898636
JPM
1752 skb = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0,
1753 dnode, tsk_own_node(tsk), peer_port,
1754 tsk->portid, TIPC_OK);
a6ca1094 1755 if (!skb)
c7268589 1756 return NULL;
a6ca1094 1757 msg = buf_msg(skb);
10724cc7
JPM
1758 msg_set_conn_ack(msg, tsk->rcv_unacked);
1759 tsk->rcv_unacked = 0;
1760
1761 /* Adjust to and advertize the correct window limit */
1762 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) {
1763 tsk->rcv_win = tsk_adv_blocks(tsk->sk.sk_rcvbuf);
1764 msg_set_adv_win(msg, tsk->rcv_win);
1765 }
c7268589
TL
1766 return skb;
1767}
1768
1769static void tipc_sk_send_ack(struct tipc_sock *tsk)
1770{
1771 struct sk_buff *skb;
1772
1773 skb = tipc_sk_build_ack(tsk);
1774 if (!skb)
1775 return;
1776
1777 tipc_node_xmit_skb(sock_net(&tsk->sk), skb, tsk_peer_node(tsk),
1778 msg_link_selector(buf_msg(skb)));
739f5e4e
JPM
1779}
1780
85d3fc94 1781static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop)
9bbb4ecc
YX
1782{
1783 struct sock *sk = sock->sk;
48766a58 1784 DEFINE_WAIT_FUNC(wait, woken_wake_function);
85d3fc94 1785 long timeo = *timeop;
4e0df495
PB
1786 int err = sock_error(sk);
1787
1788 if (err)
1789 return err;
9bbb4ecc
YX
1790
1791 for (;;) {
fe8e4649 1792 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
6f00089c 1793 if (sk->sk_shutdown & RCV_SHUTDOWN) {
9bbb4ecc
YX
1794 err = -ENOTCONN;
1795 break;
1796 }
48766a58 1797 add_wait_queue(sk_sleep(sk), &wait);
9bbb4ecc 1798 release_sock(sk);
48766a58
TN
1799 timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
1800 sched_annotate_sleep();
9bbb4ecc 1801 lock_sock(sk);
48766a58 1802 remove_wait_queue(sk_sleep(sk), &wait);
9bbb4ecc
YX
1803 }
1804 err = 0;
1805 if (!skb_queue_empty(&sk->sk_receive_queue))
1806 break;
9bbb4ecc
YX
1807 err = -EAGAIN;
1808 if (!timeo)
1809 break;
143fe22f
EH
1810 err = sock_intr_errno(timeo);
1811 if (signal_pending(current))
1812 break;
4e0df495
PB
1813
1814 err = sock_error(sk);
1815 if (err)
1816 break;
9bbb4ecc 1817 }
85d3fc94 1818 *timeop = timeo;
9bbb4ecc
YX
1819 return err;
1820}
1821
c4307285 1822/**
247f0f3c 1823 * tipc_recvmsg - receive packet-oriented message
b97bf3fd 1824 * @m: descriptor for message info
e9f8b101 1825 * @buflen: length of user buffer area
b97bf3fd 1826 * @flags: receive flags
c4307285 1827 *
b97bf3fd
PL
1828 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1829 * If the complete message doesn't fit in user area, truncate it.
1830 *
1831 * Returns size of returned message data, errno otherwise
1832 */
e9f8b101
JPM
1833static int tipc_recvmsg(struct socket *sock, struct msghdr *m,
1834 size_t buflen, int flags)
b97bf3fd 1835{
0c3141e9 1836 struct sock *sk = sock->sk;
e9f8b101 1837 bool connected = !tipc_sk_type_connectionless(sk);
ae236fb2 1838 struct tipc_sock *tsk = tipc_sk(sk);
e9f8b101 1839 int rc, err, hlen, dlen, copy;
b7d42635 1840 struct sk_buff_head xmitq;
ae236fb2
JM
1841 struct tipc_msg *hdr;
1842 struct sk_buff *skb;
1843 bool grp_evt;
e9f8b101 1844 long timeout;
b97bf3fd 1845
0c3141e9 1846 /* Catch invalid receive requests */
e9f8b101 1847 if (unlikely(!buflen))
b97bf3fd
PL
1848 return -EINVAL;
1849
0c3141e9 1850 lock_sock(sk);
e9f8b101
JPM
1851 if (unlikely(connected && sk->sk_state == TIPC_OPEN)) {
1852 rc = -ENOTCONN;
b97bf3fd
PL
1853 goto exit;
1854 }
e9f8b101 1855 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
b97bf3fd 1856
b7d42635 1857 /* Step rcv queue to first msg with data or error; wait if necessary */
e9f8b101 1858 do {
e9f8b101
JPM
1859 rc = tipc_wait_for_rcvmsg(sock, &timeout);
1860 if (unlikely(rc))
1861 goto exit;
1862 skb = skb_peek(&sk->sk_receive_queue);
1863 hdr = buf_msg(skb);
1864 dlen = msg_data_sz(hdr);
1865 hlen = msg_hdr_sz(hdr);
1866 err = msg_errcode(hdr);
ae236fb2 1867 grp_evt = msg_is_grp_evt(hdr);
e9f8b101
JPM
1868 if (likely(dlen || err))
1869 break;
2e84c60b 1870 tsk_advance_rx_queue(sk);
e9f8b101 1871 } while (1);
b97bf3fd 1872
e9f8b101 1873 /* Collect msg meta data, including error code and rejected data */
31c82a2d 1874 tipc_sk_set_orig_addr(m, skb);
1c1274a5 1875 rc = tipc_sk_anc_data_recv(m, skb, tsk);
e9f8b101 1876 if (unlikely(rc))
b97bf3fd 1877 goto exit;
1c1274a5 1878 hdr = buf_msg(skb);
b97bf3fd 1879
e9f8b101
JPM
1880 /* Capture data if non-error msg, otherwise just set return value */
1881 if (likely(!err)) {
1882 copy = min_t(int, dlen, buflen);
1883 if (unlikely(copy != dlen))
b97bf3fd 1884 m->msg_flags |= MSG_TRUNC;
e9f8b101 1885 rc = skb_copy_datagram_msg(skb, hlen, m, copy);
b97bf3fd 1886 } else {
e9f8b101
JPM
1887 copy = 0;
1888 rc = 0;
1889 if (err != TIPC_CONN_SHUTDOWN && connected && !m->msg_control)
1890 rc = -ECONNRESET;
b97bf3fd 1891 }
e9f8b101
JPM
1892 if (unlikely(rc))
1893 goto exit;
b97bf3fd 1894
ae236fb2
JM
1895 /* Mark message as group event if applicable */
1896 if (unlikely(grp_evt)) {
1897 if (msg_grp_evt(hdr) == TIPC_WITHDRAWN)
1898 m->msg_flags |= MSG_EOR;
1899 m->msg_flags |= MSG_OOB;
1900 copy = 0;
1901 }
1902
e9f8b101 1903 /* Caption of data or error code/rejected data was successful */
10724cc7
JPM
1904 if (unlikely(flags & MSG_PEEK))
1905 goto exit;
1906
b7d42635
JM
1907 /* Send group flow control advertisement when applicable */
1908 if (tsk->group && msg_in_group(hdr) && !grp_evt) {
e654f9f5 1909 __skb_queue_head_init(&xmitq);
b7d42635
JM
1910 tipc_group_update_rcv_win(tsk->group, tsk_blocks(hlen + dlen),
1911 msg_orignode(hdr), msg_origport(hdr),
1912 &xmitq);
1913 tipc_node_distr_xmit(sock_net(sk), &xmitq);
1914 }
1915
10724cc7 1916 tsk_advance_rx_queue(sk);
ae236fb2 1917
e9f8b101
JPM
1918 if (likely(!connected))
1919 goto exit;
1920
b7d42635 1921 /* Send connection flow control advertisement when applicable */
e9f8b101
JPM
1922 tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
1923 if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
1924 tipc_sk_send_ack(tsk);
b97bf3fd 1925exit:
0c3141e9 1926 release_sock(sk);
e9f8b101 1927 return rc ? rc : copy;
b97bf3fd
PL
1928}
1929
c4307285 1930/**
ec8a09fb 1931 * tipc_recvstream - receive stream-oriented data
b97bf3fd 1932 * @m: descriptor for message info
ec8a09fb 1933 * @buflen: total size of user buffer area
b97bf3fd 1934 * @flags: receive flags
c4307285
YH
1935 *
1936 * Used for SOCK_STREAM messages only. If not enough data is available
b97bf3fd
PL
1937 * will optionally wait for more; never truncates data.
1938 *
1939 * Returns size of returned message data, errno otherwise
1940 */
ec8a09fb
JPM
1941static int tipc_recvstream(struct socket *sock, struct msghdr *m,
1942 size_t buflen, int flags)
b97bf3fd 1943{
0c3141e9 1944 struct sock *sk = sock->sk;
58ed9442 1945 struct tipc_sock *tsk = tipc_sk(sk);
ec8a09fb
JPM
1946 struct sk_buff *skb;
1947 struct tipc_msg *hdr;
1948 struct tipc_skb_cb *skb_cb;
1949 bool peek = flags & MSG_PEEK;
1950 int offset, required, copy, copied = 0;
1951 int hlen, dlen, err, rc;
1952 long timeout;
b97bf3fd 1953
0c3141e9 1954 /* Catch invalid receive attempts */
ec8a09fb 1955 if (unlikely(!buflen))
b97bf3fd
PL
1956 return -EINVAL;
1957
0c3141e9 1958 lock_sock(sk);
b97bf3fd 1959
438adcaf 1960 if (unlikely(sk->sk_state == TIPC_OPEN)) {
ec8a09fb 1961 rc = -ENOTCONN;
9bbb4ecc 1962 goto exit;
b97bf3fd 1963 }
ec8a09fb
JPM
1964 required = sock_rcvlowat(sk, flags & MSG_WAITALL, buflen);
1965 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
b97bf3fd 1966
ec8a09fb
JPM
1967 do {
1968 /* Look at first msg in receive queue; wait if necessary */
1969 rc = tipc_wait_for_rcvmsg(sock, &timeout);
1970 if (unlikely(rc))
1971 break;
1972 skb = skb_peek(&sk->sk_receive_queue);
1973 skb_cb = TIPC_SKB_CB(skb);
1974 hdr = buf_msg(skb);
1975 dlen = msg_data_sz(hdr);
1976 hlen = msg_hdr_sz(hdr);
1977 err = msg_errcode(hdr);
0232fd0a 1978
ec8a09fb
JPM
1979 /* Discard any empty non-errored (SYN-) message */
1980 if (unlikely(!dlen && !err)) {
1981 tsk_advance_rx_queue(sk);
1982 continue;
1983 }
0232fd0a 1984
ec8a09fb
JPM
1985 /* Collect msg meta data, incl. error code and rejected data */
1986 if (!copied) {
31c82a2d 1987 tipc_sk_set_orig_addr(m, skb);
1c1274a5 1988 rc = tipc_sk_anc_data_recv(m, skb, tsk);
ec8a09fb
JPM
1989 if (rc)
1990 break;
1c1274a5 1991 hdr = buf_msg(skb);
ec8a09fb 1992 }
b97bf3fd 1993
ec8a09fb
JPM
1994 /* Copy data if msg ok, otherwise return error/partial data */
1995 if (likely(!err)) {
1996 offset = skb_cb->bytes_read;
1997 copy = min_t(int, dlen - offset, buflen - copied);
1998 rc = skb_copy_datagram_msg(skb, hlen + offset, m, copy);
1999 if (unlikely(rc))
2000 break;
2001 copied += copy;
2002 offset += copy;
2003 if (unlikely(offset < dlen)) {
2004 if (!peek)
2005 skb_cb->bytes_read = offset;
2006 break;
2007 }
2008 } else {
2009 rc = 0;
2010 if ((err != TIPC_CONN_SHUTDOWN) && !m->msg_control)
2011 rc = -ECONNRESET;
2012 if (copied || rc)
2013 break;
b97bf3fd 2014 }
b97bf3fd 2015
ec8a09fb
JPM
2016 if (unlikely(peek))
2017 break;
b97bf3fd 2018
ec8a09fb 2019 tsk_advance_rx_queue(sk);
10724cc7 2020
ec8a09fb
JPM
2021 /* Send connection flow control advertisement when applicable */
2022 tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
c7268589 2023 if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
ec8a09fb 2024 tipc_sk_send_ack(tsk);
b97bf3fd 2025
ec8a09fb
JPM
2026 /* Exit if all requested data or FIN/error received */
2027 if (copied == buflen || err)
2028 break;
b97bf3fd 2029
ec8a09fb 2030 } while (!skb_queue_empty(&sk->sk_receive_queue) || copied < required);
b97bf3fd 2031exit:
0c3141e9 2032 release_sock(sk);
ec8a09fb 2033 return copied ? copied : rc;
b97bf3fd
PL
2034}
2035
f288bef4
YX
2036/**
2037 * tipc_write_space - wake up thread if port congestion is released
2038 * @sk: socket
2039 */
2040static void tipc_write_space(struct sock *sk)
2041{
2042 struct socket_wq *wq;
2043
2044 rcu_read_lock();
2045 wq = rcu_dereference(sk->sk_wq);
1ce0bf50 2046 if (skwq_has_sleeper(wq))
a9a08845
LT
2047 wake_up_interruptible_sync_poll(&wq->wait, EPOLLOUT |
2048 EPOLLWRNORM | EPOLLWRBAND);
f288bef4
YX
2049 rcu_read_unlock();
2050}
2051
2052/**
2053 * tipc_data_ready - wake up threads to indicate messages have been received
2054 * @sk: socket
2055 * @len: the length of messages
2056 */
676d2369 2057static void tipc_data_ready(struct sock *sk)
f288bef4
YX
2058{
2059 struct socket_wq *wq;
2060
2061 rcu_read_lock();
2062 wq = rcu_dereference(sk->sk_wq);
1ce0bf50 2063 if (skwq_has_sleeper(wq))
a9a08845
LT
2064 wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN |
2065 EPOLLRDNORM | EPOLLRDBAND);
f288bef4
YX
2066 rcu_read_unlock();
2067}
2068
f4195d1e
YX
2069static void tipc_sock_destruct(struct sock *sk)
2070{
2071 __skb_queue_purge(&sk->sk_receive_queue);
2072}
2073
64ac5f59
JM
2074static void tipc_sk_proto_rcv(struct sock *sk,
2075 struct sk_buff_head *inputq,
2076 struct sk_buff_head *xmitq)
2077{
2078 struct sk_buff *skb = __skb_dequeue(inputq);
2079 struct tipc_sock *tsk = tipc_sk(sk);
2080 struct tipc_msg *hdr = buf_msg(skb);
75da2163 2081 struct tipc_group *grp = tsk->group;
b7d42635 2082 bool wakeup = false;
64ac5f59
JM
2083
2084 switch (msg_user(hdr)) {
2085 case CONN_MANAGER:
e7eb0582 2086 tipc_sk_conn_proto_rcv(tsk, skb, inputq, xmitq);
64ac5f59
JM
2087 return;
2088 case SOCK_WAKEUP:
a80ae530 2089 tipc_dest_del(&tsk->cong_links, msg_orignode(hdr), 0);
bfd07f3d
TN
2090 /* coupled with smp_rmb() in tipc_wait_for_cond() */
2091 smp_wmb();
64ac5f59 2092 tsk->cong_link_cnt--;
b7d42635 2093 wakeup = true;
c0bceb97 2094 tipc_sk_push_backlog(tsk);
64ac5f59 2095 break;
75da2163 2096 case GROUP_PROTOCOL:
b7d42635 2097 tipc_group_proto_rcv(grp, &wakeup, hdr, inputq, xmitq);
75da2163 2098 break;
64ac5f59 2099 case TOP_SRV:
b7d42635 2100 tipc_group_member_evt(tsk->group, &wakeup, &sk->sk_rcvbuf,
7ad32bcb 2101 hdr, inputq, xmitq);
64ac5f59
JM
2102 break;
2103 default:
2104 break;
2105 }
2106
b7d42635
JM
2107 if (wakeup)
2108 sk->sk_write_space(sk);
2109
64ac5f59
JM
2110 kfree_skb(skb);
2111}
2112
7e6c131e 2113/**
39fdc9c7 2114 * tipc_sk_filter_connect - check incoming message for a connection-based socket
58ed9442 2115 * @tsk: TIPC socket
39fdc9c7 2116 * @skb: pointer to message buffer.
c7268589 2117 * @xmitq: for Nagle ACK if any
39fdc9c7 2118 * Returns true if message should be added to receive queue, false otherwise
7e6c131e 2119 */
c7268589
TL
2120static bool tipc_sk_filter_connect(struct tipc_sock *tsk, struct sk_buff *skb,
2121 struct sk_buff_head *xmitq)
7e6c131e 2122{
58ed9442 2123 struct sock *sk = &tsk->sk;
f2f9800d 2124 struct net *net = sock_net(sk);
cda3696d 2125 struct tipc_msg *hdr = buf_msg(skb);
39fdc9c7
JM
2126 bool con_msg = msg_connected(hdr);
2127 u32 pport = tsk_peer_port(tsk);
2128 u32 pnode = tsk_peer_node(tsk);
2129 u32 oport = msg_origport(hdr);
2130 u32 onode = msg_orignode(hdr);
2131 int err = msg_errcode(hdr);
67879274 2132 unsigned long delay;
7e6c131e 2133
cda3696d
JPM
2134 if (unlikely(msg_mcast(hdr)))
2135 return false;
c0bceb97 2136 tsk->oneway = 0;
7e6c131e 2137
99a20889
PB
2138 switch (sk->sk_state) {
2139 case TIPC_CONNECTING:
39fdc9c7
JM
2140 /* Setup ACK */
2141 if (likely(con_msg)) {
2142 if (err)
2143 break;
2144 tipc_sk_finish_conn(tsk, oport, onode);
2145 msg_set_importance(&tsk->phdr, msg_importance(hdr));
2146 /* ACK+ message with data is added to receive queue */
2147 if (msg_data_sz(hdr))
2148 return true;
2149 /* Empty ACK-, - wake up sleeping connect() and drop */
ff946833 2150 sk->sk_state_change(sk);
39fdc9c7
JM
2151 msg_set_dest_droppable(hdr, 1);
2152 return false;
584d24b3 2153 }
39fdc9c7
JM
2154 /* Ignore connectionless message if not from listening socket */
2155 if (oport != pport || onode != pnode)
2156 return false;
584d24b3 2157
67879274
TN
2158 /* Rejected SYN */
2159 if (err != TIPC_ERR_OVERLOAD)
2160 break;
2161
2162 /* Prepare for new setup attempt if we have a SYN clone */
2163 if (skb_queue_empty(&sk->sk_write_queue))
2164 break;
2165 get_random_bytes(&delay, 2);
2166 delay %= (tsk->conn_timeout / 4);
2167 delay = msecs_to_jiffies(delay + 100);
2168 sk_reset_timer(sk, &sk->sk_timer, jiffies + delay);
2169 return false;
438adcaf 2170 case TIPC_OPEN:
9fd4b070 2171 case TIPC_DISCONNECTING:
39fdc9c7 2172 return false;
438adcaf 2173 case TIPC_LISTEN:
7e6c131e 2174 /* Accept only SYN message */
25b9221b
JM
2175 if (!msg_is_syn(hdr) &&
2176 tipc_node_get_capabilities(net, onode) & TIPC_SYN_BIT)
2177 return false;
39fdc9c7 2178 if (!con_msg && !err)
cda3696d 2179 return true;
39fdc9c7 2180 return false;
f40acbaf 2181 case TIPC_ESTABLISHED:
c0bceb97
JM
2182 if (!skb_queue_empty(&sk->sk_write_queue))
2183 tipc_sk_push_backlog(tsk);
f40acbaf 2184 /* Accept only connection-based messages sent by peer */
c7268589
TL
2185 if (likely(con_msg && !err && pport == oport &&
2186 pnode == onode)) {
2187 if (msg_ack_required(hdr)) {
2188 struct sk_buff *skb;
2189
2190 skb = tipc_sk_build_ack(tsk);
2191 if (skb)
2192 __skb_queue_tail(xmitq, skb);
2193 }
39fdc9c7 2194 return true;
c7268589 2195 }
39fdc9c7 2196 if (!tsk_peer_msg(tsk, hdr))
f40acbaf 2197 return false;
39fdc9c7
JM
2198 if (!err)
2199 return true;
2200 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2201 tipc_node_remove_conn(net, pnode, tsk->portid);
2202 sk->sk_state_change(sk);
f40acbaf 2203 return true;
7e6c131e 2204 default:
438adcaf 2205 pr_err("Unknown sk_state %u\n", sk->sk_state);
7e6c131e 2206 }
39fdc9c7
JM
2207 /* Abort connection setup attempt */
2208 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2209 sk->sk_err = ECONNREFUSED;
2210 sk->sk_state_change(sk);
2211 return true;
7e6c131e
YX
2212}
2213
aba79f33
YX
2214/**
2215 * rcvbuf_limit - get proper overload limit of socket receive queue
2216 * @sk: socket
10724cc7 2217 * @skb: message
aba79f33 2218 *
10724cc7
JPM
2219 * For connection oriented messages, irrespective of importance,
2220 * default queue limit is 2 MB.
aba79f33 2221 *
10724cc7
JPM
2222 * For connectionless messages, queue limits are based on message
2223 * importance as follows:
aba79f33 2224 *
10724cc7
JPM
2225 * TIPC_LOW_IMPORTANCE (2 MB)
2226 * TIPC_MEDIUM_IMPORTANCE (4 MB)
2227 * TIPC_HIGH_IMPORTANCE (8 MB)
2228 * TIPC_CRITICAL_IMPORTANCE (16 MB)
aba79f33
YX
2229 *
2230 * Returns overload limit according to corresponding message importance
2231 */
10724cc7 2232static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *skb)
aba79f33 2233{
10724cc7
JPM
2234 struct tipc_sock *tsk = tipc_sk(sk);
2235 struct tipc_msg *hdr = buf_msg(skb);
2236
b7d42635 2237 if (unlikely(msg_in_group(hdr)))
8265792b 2238 return READ_ONCE(sk->sk_rcvbuf);
b7d42635 2239
10724cc7 2240 if (unlikely(!msg_connected(hdr)))
8265792b 2241 return READ_ONCE(sk->sk_rcvbuf) << msg_importance(hdr);
aba79f33 2242
10724cc7 2243 if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
8265792b 2244 return READ_ONCE(sk->sk_rcvbuf);
0cee6bbe 2245
10724cc7 2246 return FLOWCTL_MSG_LIM;
aba79f33
YX
2247}
2248
c4307285 2249/**
64ac5f59 2250 * tipc_sk_filter_rcv - validate incoming message
0c3141e9 2251 * @sk: socket
cda3696d 2252 * @skb: pointer to message.
c4307285 2253 *
0c3141e9
AS
2254 * Enqueues message on receive queue if acceptable; optionally handles
2255 * disconnect indication for a connected socket.
2256 *
1186adf7 2257 * Called with socket lock already taken
c4307285 2258 *
b97bf3fd 2259 */
64ac5f59
JM
2260static void tipc_sk_filter_rcv(struct sock *sk, struct sk_buff *skb,
2261 struct sk_buff_head *xmitq)
b97bf3fd 2262{
64ac5f59 2263 bool sk_conn = !tipc_sk_type_connectionless(sk);
58ed9442 2264 struct tipc_sock *tsk = tipc_sk(sk);
75da2163 2265 struct tipc_group *grp = tsk->group;
cda3696d 2266 struct tipc_msg *hdr = buf_msg(skb);
64ac5f59
JM
2267 struct net *net = sock_net(sk);
2268 struct sk_buff_head inputq;
77d5ad40 2269 int mtyp = msg_type(hdr);
64ac5f59 2270 int limit, err = TIPC_OK;
ec8a2e56 2271
01e661eb 2272 trace_tipc_sk_filter_rcv(sk, skb, TIPC_DUMP_ALL, " ");
64ac5f59
JM
2273 TIPC_SKB_CB(skb)->bytes_read = 0;
2274 __skb_queue_head_init(&inputq);
2275 __skb_queue_tail(&inputq, skb);
50100a5e 2276
64ac5f59
JM
2277 if (unlikely(!msg_isdata(hdr)))
2278 tipc_sk_proto_rcv(sk, &inputq, xmitq);
0c3141e9 2279
75da2163
JM
2280 if (unlikely(grp))
2281 tipc_group_filter_msg(grp, &inputq, xmitq);
2282
77d5ad40 2283 if (unlikely(!grp) && mtyp == TIPC_MCAST_MSG)
08e046c8 2284 tipc_mcast_filter_msg(net, &tsk->mc_method.deferredq, &inputq);
c55c8eda 2285
64ac5f59
JM
2286 /* Validate and add to receive buffer if there is space */
2287 while ((skb = __skb_dequeue(&inputq))) {
2288 hdr = buf_msg(skb);
2289 limit = rcvbuf_limit(sk, skb);
c7268589 2290 if ((sk_conn && !tipc_sk_filter_connect(tsk, skb, xmitq)) ||
75da2163
JM
2291 (!sk_conn && msg_connected(hdr)) ||
2292 (!grp && msg_in_group(hdr)))
cda3696d 2293 err = TIPC_ERR_NO_PORT;
872619d8 2294 else if (sk_rmem_alloc_get(sk) + skb->truesize >= limit) {
01e661eb
TL
2295 trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL,
2296 "err_overload2!");
872619d8 2297 atomic_inc(&sk->sk_drops);
64ac5f59 2298 err = TIPC_ERR_OVERLOAD;
872619d8 2299 }
b97bf3fd 2300
64ac5f59 2301 if (unlikely(err)) {
01e661eb
TL
2302 if (tipc_msg_reverse(tipc_own_addr(net), &skb, err)) {
2303 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE,
2304 "@filter_rcv!");
2305 __skb_queue_tail(xmitq, skb);
2306 }
64ac5f59
JM
2307 err = TIPC_OK;
2308 continue;
2309 }
2310 __skb_queue_tail(&sk->sk_receive_queue, skb);
2311 skb_set_owner_r(skb, sk);
01e661eb
TL
2312 trace_tipc_sk_overlimit2(sk, skb, TIPC_DUMP_ALL,
2313 "rcvq >90% allocated!");
64ac5f59 2314 sk->sk_data_ready(sk);
cda3696d 2315 }
0c3141e9 2316}
b97bf3fd 2317
0c3141e9 2318/**
64ac5f59 2319 * tipc_sk_backlog_rcv - handle incoming message from backlog queue
0c3141e9 2320 * @sk: socket
a6ca1094 2321 * @skb: message
0c3141e9 2322 *
e3a77561 2323 * Caller must hold socket lock
0c3141e9 2324 */
64ac5f59 2325static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb)
0c3141e9 2326{
64ac5f59 2327 unsigned int before = sk_rmem_alloc_get(sk);
f1d048f2 2328 struct sk_buff_head xmitq;
64ac5f59 2329 unsigned int added;
0c3141e9 2330
f1d048f2
JPM
2331 __skb_queue_head_init(&xmitq);
2332
64ac5f59
JM
2333 tipc_sk_filter_rcv(sk, skb, &xmitq);
2334 added = sk_rmem_alloc_get(sk) - before;
2335 atomic_add(added, &tipc_sk(sk)->dupl_rcvcnt);
f1d048f2 2336
64ac5f59 2337 /* Send pending response/rejected messages, if any */
f70d37b7 2338 tipc_node_distr_xmit(sock_net(sk), &xmitq);
0c3141e9
AS
2339 return 0;
2340}
2341
d570d864 2342/**
c637c103
JPM
2343 * tipc_sk_enqueue - extract all buffers with destination 'dport' from
2344 * inputq and try adding them to socket or backlog queue
2345 * @inputq: list of incoming buffers with potentially different destinations
2346 * @sk: socket where the buffers should be enqueued
2347 * @dport: port number for the socket
d570d864
JPM
2348 *
2349 * Caller must hold socket lock
d570d864 2350 */
cda3696d 2351static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk,
f1d048f2 2352 u32 dport, struct sk_buff_head *xmitq)
d570d864 2353{
f1d048f2
JPM
2354 unsigned long time_limit = jiffies + 2;
2355 struct sk_buff *skb;
d570d864
JPM
2356 unsigned int lim;
2357 atomic_t *dcnt;
f1d048f2 2358 u32 onode;
c637c103
JPM
2359
2360 while (skb_queue_len(inputq)) {
51a00daf 2361 if (unlikely(time_after_eq(jiffies, time_limit)))
cda3696d
JPM
2362 return;
2363
c637c103
JPM
2364 skb = tipc_skb_dequeue(inputq, dport);
2365 if (unlikely(!skb))
cda3696d
JPM
2366 return;
2367
2368 /* Add message directly to receive queue if possible */
c637c103 2369 if (!sock_owned_by_user(sk)) {
64ac5f59 2370 tipc_sk_filter_rcv(sk, skb, xmitq);
cda3696d 2371 continue;
c637c103 2372 }
cda3696d
JPM
2373
2374 /* Try backlog, compensating for double-counted bytes */
c637c103 2375 dcnt = &tipc_sk(sk)->dupl_rcvcnt;
7c8bcfb1 2376 if (!sk->sk_backlog.len)
c637c103
JPM
2377 atomic_set(dcnt, 0);
2378 lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
01e661eb
TL
2379 if (likely(!sk_add_backlog(sk, skb, lim))) {
2380 trace_tipc_sk_overlimit1(sk, skb, TIPC_DUMP_ALL,
2381 "bklg & rcvq >90% allocated!");
c637c103 2382 continue;
01e661eb 2383 }
cda3696d 2384
01e661eb 2385 trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL, "err_overload!");
cda3696d 2386 /* Overload => reject message back to sender */
f1d048f2 2387 onode = tipc_own_addr(sock_net(sk));
872619d8 2388 atomic_inc(&sk->sk_drops);
01e661eb
TL
2389 if (tipc_msg_reverse(onode, &skb, TIPC_ERR_OVERLOAD)) {
2390 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_ALL,
2391 "@sk_enqueue!");
f1d048f2 2392 __skb_queue_tail(xmitq, skb);
01e661eb 2393 }
cda3696d 2394 break;
c637c103 2395 }
d570d864
JPM
2396}
2397
0c3141e9 2398/**
c637c103
JPM
2399 * tipc_sk_rcv - handle a chain of incoming buffers
2400 * @inputq: buffer list containing the buffers
2401 * Consumes all buffers in list until inputq is empty
2402 * Note: may be called in multiple threads referring to the same queue
0c3141e9 2403 */
cda3696d 2404void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq)
0c3141e9 2405{
f1d048f2 2406 struct sk_buff_head xmitq;
c637c103 2407 u32 dnode, dport = 0;
9871b27f 2408 int err;
9816f061 2409 struct tipc_sock *tsk;
9816f061 2410 struct sock *sk;
cda3696d 2411 struct sk_buff *skb;
9816f061 2412
f1d048f2 2413 __skb_queue_head_init(&xmitq);
c637c103 2414 while (skb_queue_len(inputq)) {
c637c103
JPM
2415 dport = tipc_skb_peek_port(inputq, dport);
2416 tsk = tipc_sk_lookup(net, dport);
cda3696d 2417
c637c103
JPM
2418 if (likely(tsk)) {
2419 sk = &tsk->sk;
2420 if (likely(spin_trylock_bh(&sk->sk_lock.slock))) {
f1d048f2 2421 tipc_sk_enqueue(inputq, sk, dport, &xmitq);
c637c103 2422 spin_unlock_bh(&sk->sk_lock.slock);
c637c103 2423 }
f1d048f2 2424 /* Send pending response/rejected messages, if any */
f70d37b7 2425 tipc_node_distr_xmit(sock_net(sk), &xmitq);
c637c103 2426 sock_put(sk);
c637c103 2427 continue;
c637c103 2428 }
cda3696d
JPM
2429 /* No destination socket => dequeue skb if still there */
2430 skb = tipc_skb_dequeue(inputq, dport);
2431 if (!skb)
2432 return;
2433
2434 /* Try secondary lookup if unresolved named message */
2435 err = TIPC_ERR_NO_PORT;
2436 if (tipc_msg_lookup_dest(net, skb, &err))
2437 goto xmit;
2438
2439 /* Prepare for message rejection */
2440 if (!tipc_msg_reverse(tipc_own_addr(net), &skb, err))
c637c103 2441 continue;
01e661eb
TL
2442
2443 trace_tipc_sk_rej_msg(NULL, skb, TIPC_DUMP_NONE, "@sk_rcv!");
e3a77561 2444xmit:
cda3696d 2445 dnode = msg_destnode(buf_msg(skb));
af9b028e 2446 tipc_node_xmit_skb(net, skb, dnode, dport);
c637c103 2447 }
b97bf3fd
PL
2448}
2449
78eb3a53
YX
2450static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
2451{
d9dc8b0f 2452 DEFINE_WAIT_FUNC(wait, woken_wake_function);
78eb3a53 2453 struct sock *sk = sock->sk;
78eb3a53
YX
2454 int done;
2455
2456 do {
2457 int err = sock_error(sk);
2458 if (err)
2459 return err;
2460 if (!*timeo_p)
2461 return -ETIMEDOUT;
2462 if (signal_pending(current))
2463 return sock_intr_errno(*timeo_p);
5391a877
TL
2464 if (sk->sk_state == TIPC_DISCONNECTING)
2465 break;
78eb3a53 2466
d9dc8b0f 2467 add_wait_queue(sk_sleep(sk), &wait);
9546a0b7
TL
2468 done = sk_wait_event(sk, timeo_p, tipc_sk_connected(sk),
2469 &wait);
d9dc8b0f 2470 remove_wait_queue(sk_sleep(sk), &wait);
78eb3a53
YX
2471 } while (!done);
2472 return 0;
2473}
2474
ea239314
EH
2475static bool tipc_sockaddr_is_sane(struct sockaddr_tipc *addr)
2476{
2477 if (addr->family != AF_TIPC)
2478 return false;
2479 if (addr->addrtype == TIPC_SERVICE_RANGE)
2480 return (addr->addr.nameseq.lower <= addr->addr.nameseq.upper);
2481 return (addr->addrtype == TIPC_SERVICE_ADDR ||
2482 addr->addrtype == TIPC_SOCKET_ADDR);
2483}
2484
b97bf3fd 2485/**
247f0f3c 2486 * tipc_connect - establish a connection to another TIPC port
b97bf3fd
PL
2487 * @sock: socket structure
2488 * @dest: socket address for destination port
2489 * @destlen: size of socket address data structure
0c3141e9 2490 * @flags: file-related flags associated with socket
b97bf3fd
PL
2491 *
2492 * Returns 0 on success, errno otherwise
2493 */
247f0f3c
YX
2494static int tipc_connect(struct socket *sock, struct sockaddr *dest,
2495 int destlen, int flags)
b97bf3fd 2496{
0c3141e9 2497 struct sock *sk = sock->sk;
f2f8036e 2498 struct tipc_sock *tsk = tipc_sk(sk);
b89741a0
AS
2499 struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
2500 struct msghdr m = {NULL,};
f2f8036e 2501 long timeout = (flags & O_NONBLOCK) ? 0 : tsk->conn_timeout;
99a20889 2502 int previous;
f2f8036e 2503 int res = 0;
b89741a0 2504
23998835
JM
2505 if (destlen != sizeof(struct sockaddr_tipc))
2506 return -EINVAL;
2507
0c3141e9
AS
2508 lock_sock(sk);
2509
75da2163
JM
2510 if (tsk->group) {
2511 res = -EINVAL;
2512 goto exit;
2513 }
2514
23998835
JM
2515 if (dst->family == AF_UNSPEC) {
2516 memset(&tsk->peer, 0, sizeof(struct sockaddr_tipc));
2517 if (!tipc_sk_type_connectionless(sk))
610600c8 2518 res = -EINVAL;
0c3141e9
AS
2519 goto exit;
2520 }
ea239314 2521 if (!tipc_sockaddr_is_sane(dst)) {
0c3141e9 2522 res = -EINVAL;
23998835 2523 goto exit;
ea239314 2524 }
23998835
JM
2525 /* DGRAM/RDM connect(), just save the destaddr */
2526 if (tipc_sk_type_connectionless(sk)) {
2527 memcpy(&tsk->peer, dest, destlen);
0c3141e9 2528 goto exit;
ea239314
EH
2529 } else if (dst->addrtype == TIPC_SERVICE_RANGE) {
2530 res = -EINVAL;
2531 goto exit;
0c3141e9
AS
2532 }
2533
99a20889 2534 previous = sk->sk_state;
438adcaf
PB
2535
2536 switch (sk->sk_state) {
2537 case TIPC_OPEN:
584d24b3
YX
2538 /* Send a 'SYN-' to destination */
2539 m.msg_name = dest;
2540 m.msg_namelen = destlen;
2541
2542 /* If connect is in non-blocking case, set MSG_DONTWAIT to
2543 * indicate send_msg() is never blocked.
2544 */
2545 if (!timeout)
2546 m.msg_flags = MSG_DONTWAIT;
2547
39a0295f 2548 res = __tipc_sendmsg(sock, &m, 0);
584d24b3
YX
2549 if ((res < 0) && (res != -EWOULDBLOCK))
2550 goto exit;
2551
99a20889 2552 /* Just entered TIPC_CONNECTING state; the only
584d24b3
YX
2553 * difference is that return value in non-blocking
2554 * case is EINPROGRESS, rather than EALREADY.
2555 */
2556 res = -EINPROGRESS;
f79e3365 2557 /* fall through */
99a20889
PB
2558 case TIPC_CONNECTING:
2559 if (!timeout) {
2560 if (previous == TIPC_CONNECTING)
2561 res = -EALREADY;
78eb3a53 2562 goto exit;
99a20889 2563 }
78eb3a53
YX
2564 timeout = msecs_to_jiffies(timeout);
2565 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
2566 res = tipc_wait_for_connect(sock, &timeout);
f40acbaf
PB
2567 break;
2568 case TIPC_ESTABLISHED:
584d24b3 2569 res = -EISCONN;
f40acbaf
PB
2570 break;
2571 default:
584d24b3 2572 res = -EINVAL;
f40acbaf 2573 }
99a20889 2574
0c3141e9
AS
2575exit:
2576 release_sock(sk);
b89741a0 2577 return res;
b97bf3fd
PL
2578}
2579
c4307285 2580/**
247f0f3c 2581 * tipc_listen - allow socket to listen for incoming connections
b97bf3fd
PL
2582 * @sock: socket structure
2583 * @len: (unused)
c4307285 2584 *
b97bf3fd
PL
2585 * Returns 0 on success, errno otherwise
2586 */
247f0f3c 2587static int tipc_listen(struct socket *sock, int len)
b97bf3fd 2588{
0c3141e9
AS
2589 struct sock *sk = sock->sk;
2590 int res;
2591
2592 lock_sock(sk);
0c288c86 2593 res = tipc_set_sk_state(sk, TIPC_LISTEN);
0c3141e9 2594 release_sock(sk);
0c288c86 2595
0c3141e9 2596 return res;
b97bf3fd
PL
2597}
2598
6398e23c
YX
2599static int tipc_wait_for_accept(struct socket *sock, long timeo)
2600{
2601 struct sock *sk = sock->sk;
2602 DEFINE_WAIT(wait);
2603 int err;
2604
2605 /* True wake-one mechanism for incoming connections: only
2606 * one process gets woken up, not the 'whole herd'.
2607 * Since we do not 'race & poll' for established sockets
2608 * anymore, the common case will execute the loop only once.
2609 */
2610 for (;;) {
2611 prepare_to_wait_exclusive(sk_sleep(sk), &wait,
2612 TASK_INTERRUPTIBLE);
fe8e4649 2613 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
6398e23c
YX
2614 release_sock(sk);
2615 timeo = schedule_timeout(timeo);
2616 lock_sock(sk);
2617 }
2618 err = 0;
2619 if (!skb_queue_empty(&sk->sk_receive_queue))
2620 break;
6398e23c
YX
2621 err = -EAGAIN;
2622 if (!timeo)
2623 break;
143fe22f
EH
2624 err = sock_intr_errno(timeo);
2625 if (signal_pending(current))
2626 break;
6398e23c
YX
2627 }
2628 finish_wait(sk_sleep(sk), &wait);
2629 return err;
2630}
2631
c4307285 2632/**
247f0f3c 2633 * tipc_accept - wait for connection request
b97bf3fd
PL
2634 * @sock: listening socket
2635 * @newsock: new socket that is to be connected
2636 * @flags: file-related flags associated with socket
c4307285 2637 *
b97bf3fd
PL
2638 * Returns 0 on success, errno otherwise
2639 */
cdfbabfb
DH
2640static int tipc_accept(struct socket *sock, struct socket *new_sock, int flags,
2641 bool kern)
b97bf3fd 2642{
0fef8f20 2643 struct sock *new_sk, *sk = sock->sk;
b97bf3fd 2644 struct sk_buff *buf;
301bae56 2645 struct tipc_sock *new_tsock;
0fef8f20 2646 struct tipc_msg *msg;
6398e23c 2647 long timeo;
0c3141e9 2648 int res;
b97bf3fd 2649
0c3141e9 2650 lock_sock(sk);
b97bf3fd 2651
0c288c86 2652 if (sk->sk_state != TIPC_LISTEN) {
0c3141e9 2653 res = -EINVAL;
b97bf3fd
PL
2654 goto exit;
2655 }
6398e23c
YX
2656 timeo = sock_rcvtimeo(sk, flags & O_NONBLOCK);
2657 res = tipc_wait_for_accept(sock, timeo);
2658 if (res)
2659 goto exit;
0c3141e9
AS
2660
2661 buf = skb_peek(&sk->sk_receive_queue);
2662
cdfbabfb 2663 res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, kern);
0fef8f20
PG
2664 if (res)
2665 goto exit;
fdd75ea8 2666 security_sk_clone(sock->sk, new_sock->sk);
b97bf3fd 2667
0fef8f20 2668 new_sk = new_sock->sk;
301bae56 2669 new_tsock = tipc_sk(new_sk);
0fef8f20 2670 msg = buf_msg(buf);
b97bf3fd 2671
0fef8f20
PG
2672 /* we lock on new_sk; but lockdep sees the lock on sk */
2673 lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
2674
2675 /*
2676 * Reject any stray messages received by new socket
2677 * before the socket lock was taken (very, very unlikely)
2678 */
49afb806 2679 tsk_rej_rx_queue(new_sk, TIPC_ERR_NO_PORT);
0fef8f20
PG
2680
2681 /* Connect new socket to it's peer */
301bae56 2682 tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg));
0fef8f20 2683
301bae56 2684 tsk_set_importance(new_tsock, msg_importance(msg));
0fef8f20 2685 if (msg_named(msg)) {
301bae56
JPM
2686 new_tsock->conn_type = msg_nametype(msg);
2687 new_tsock->conn_instance = msg_nameinst(msg);
b97bf3fd 2688 }
0fef8f20
PG
2689
2690 /*
2691 * Respond to 'SYN-' by discarding it & returning 'ACK'-.
2692 * Respond to 'SYN+' by queuing it on new socket.
2693 */
2694 if (!msg_data_sz(msg)) {
2695 struct msghdr m = {NULL,};
2696
2e84c60b 2697 tsk_advance_rx_queue(sk);
365ad353 2698 __tipc_sendstream(new_sock, &m, 0);
0fef8f20
PG
2699 } else {
2700 __skb_dequeue(&sk->sk_receive_queue);
2701 __skb_queue_head(&new_sk->sk_receive_queue, buf);
aba79f33 2702 skb_set_owner_r(buf, new_sk);
0fef8f20
PG
2703 }
2704 release_sock(new_sk);
b97bf3fd 2705exit:
0c3141e9 2706 release_sock(sk);
b97bf3fd
PL
2707 return res;
2708}
2709
2710/**
247f0f3c 2711 * tipc_shutdown - shutdown socket connection
b97bf3fd 2712 * @sock: socket structure
e247a8f5 2713 * @how: direction to close (must be SHUT_RDWR)
b97bf3fd
PL
2714 *
2715 * Terminates connection (if necessary), then purges socket's receive queue.
c4307285 2716 *
b97bf3fd
PL
2717 * Returns 0 on success, errno otherwise
2718 */
247f0f3c 2719static int tipc_shutdown(struct socket *sock, int how)
b97bf3fd 2720{
0c3141e9 2721 struct sock *sk = sock->sk;
b97bf3fd
PL
2722 int res;
2723
e247a8f5
AS
2724 if (how != SHUT_RDWR)
2725 return -EINVAL;
b97bf3fd 2726
0c3141e9 2727 lock_sock(sk);
b97bf3fd 2728
01e661eb 2729 trace_tipc_sk_shutdown(sk, NULL, TIPC_DUMP_ALL, " ");
6f00089c
PB
2730 __tipc_shutdown(sock, TIPC_CONN_SHUTDOWN);
2731 sk->sk_shutdown = SEND_SHUTDOWN;
b97bf3fd 2732
6f00089c 2733 if (sk->sk_state == TIPC_DISCONNECTING) {
75031151 2734 /* Discard any unreceived messages */
57467e56 2735 __skb_queue_purge(&sk->sk_receive_queue);
75031151
YX
2736
2737 /* Wake up anyone sleeping in poll */
2738 sk->sk_state_change(sk);
b97bf3fd 2739 res = 0;
6f00089c 2740 } else {
b97bf3fd
PL
2741 res = -ENOTCONN;
2742 }
2743
0c3141e9 2744 release_sock(sk);
b97bf3fd
PL
2745 return res;
2746}
2747
afe8792f
JM
2748static void tipc_sk_check_probing_state(struct sock *sk,
2749 struct sk_buff_head *list)
2750{
2751 struct tipc_sock *tsk = tipc_sk(sk);
2752 u32 pnode = tsk_peer_node(tsk);
2753 u32 pport = tsk_peer_port(tsk);
2754 u32 self = tsk_own_node(tsk);
2755 u32 oport = tsk->portid;
2756 struct sk_buff *skb;
2757
2758 if (tsk->probe_unacked) {
2759 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2760 sk->sk_err = ECONNABORTED;
2761 tipc_node_remove_conn(sock_net(sk), pnode, pport);
2762 sk->sk_state_change(sk);
2763 return;
2764 }
2765 /* Prepare new probe */
2766 skb = tipc_msg_create(CONN_MANAGER, CONN_PROBE, INT_H_SIZE, 0,
2767 pnode, self, pport, oport, TIPC_OK);
2768 if (skb)
2769 __skb_queue_tail(list, skb);
2770 tsk->probe_unacked = true;
2771 sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
2772}
2773
67879274
TN
2774static void tipc_sk_retry_connect(struct sock *sk, struct sk_buff_head *list)
2775{
2776 struct tipc_sock *tsk = tipc_sk(sk);
2777
2778 /* Try again later if dest link is congested */
2779 if (tsk->cong_link_cnt) {
2780 sk_reset_timer(sk, &sk->sk_timer, msecs_to_jiffies(100));
2781 return;
2782 }
2783 /* Prepare SYN for retransmit */
2784 tipc_msg_skb_clone(&sk->sk_write_queue, list);
2785}
2786
31b102bb 2787static void tipc_sk_timeout(struct timer_list *t)
57289015 2788{
31b102bb
KC
2789 struct sock *sk = from_timer(sk, t, sk_timer);
2790 struct tipc_sock *tsk = tipc_sk(sk);
afe8792f
JM
2791 u32 pnode = tsk_peer_node(tsk);
2792 struct sk_buff_head list;
67879274 2793 int rc = 0;
57289015 2794
e654f9f5 2795 __skb_queue_head_init(&list);
6c9808ce 2796 bh_lock_sock(sk);
0d5fcebf
JM
2797
2798 /* Try again later if socket is busy */
2799 if (sock_owned_by_user(sk)) {
2800 sk_reset_timer(sk, &sk->sk_timer, jiffies + HZ / 20);
afe8792f 2801 bh_unlock_sock(sk);
91a4a3eb 2802 sock_put(sk);
afe8792f 2803 return;
57289015 2804 }
57289015 2805
afe8792f
JM
2806 if (sk->sk_state == TIPC_ESTABLISHED)
2807 tipc_sk_check_probing_state(sk, &list);
67879274
TN
2808 else if (sk->sk_state == TIPC_CONNECTING)
2809 tipc_sk_retry_connect(sk, &list);
afe8792f 2810
57289015 2811 bh_unlock_sock(sk);
afe8792f
JM
2812
2813 if (!skb_queue_empty(&list))
67879274 2814 rc = tipc_node_xmit(sock_net(sk), &list, pnode, tsk->portid);
afe8792f 2815
67879274
TN
2816 /* SYN messages may cause link congestion */
2817 if (rc == -ELINKCONG) {
2818 tipc_dest_push(&tsk->cong_links, pnode, 0);
2819 tsk->cong_link_cnt = 1;
2820 }
07f6c4bc 2821 sock_put(sk);
57289015
JPM
2822}
2823
301bae56 2824static int tipc_sk_publish(struct tipc_sock *tsk, uint scope,
0fc87aae
JPM
2825 struct tipc_name_seq const *seq)
2826{
d6fb7e9c
PB
2827 struct sock *sk = &tsk->sk;
2828 struct net *net = sock_net(sk);
0fc87aae
JPM
2829 struct publication *publ;
2830 u32 key;
2831
928df188
JM
2832 if (scope != TIPC_NODE_SCOPE)
2833 scope = TIPC_CLUSTER_SCOPE;
2834
d6fb7e9c 2835 if (tipc_sk_connected(sk))
0fc87aae 2836 return -EINVAL;
07f6c4bc
YX
2837 key = tsk->portid + tsk->pub_count + 1;
2838 if (key == tsk->portid)
0fc87aae
JPM
2839 return -EADDRINUSE;
2840
f2f9800d 2841 publ = tipc_nametbl_publish(net, seq->type, seq->lower, seq->upper,
07f6c4bc 2842 scope, tsk->portid, key);
0fc87aae
JPM
2843 if (unlikely(!publ))
2844 return -EINVAL;
2845
e50e73e1 2846 list_add(&publ->binding_sock, &tsk->publications);
301bae56
JPM
2847 tsk->pub_count++;
2848 tsk->published = 1;
0fc87aae
JPM
2849 return 0;
2850}
2851
301bae56 2852static int tipc_sk_withdraw(struct tipc_sock *tsk, uint scope,
0fc87aae
JPM
2853 struct tipc_name_seq const *seq)
2854{
f2f9800d 2855 struct net *net = sock_net(&tsk->sk);
0fc87aae
JPM
2856 struct publication *publ;
2857 struct publication *safe;
2858 int rc = -EINVAL;
2859
928df188
JM
2860 if (scope != TIPC_NODE_SCOPE)
2861 scope = TIPC_CLUSTER_SCOPE;
2862
e50e73e1 2863 list_for_each_entry_safe(publ, safe, &tsk->publications, binding_sock) {
0fc87aae
JPM
2864 if (seq) {
2865 if (publ->scope != scope)
2866 continue;
2867 if (publ->type != seq->type)
2868 continue;
2869 if (publ->lower != seq->lower)
2870 continue;
2871 if (publ->upper != seq->upper)
2872 break;
f2f9800d 2873 tipc_nametbl_withdraw(net, publ->type, publ->lower,
37922ea4 2874 publ->upper, publ->key);
0fc87aae
JPM
2875 rc = 0;
2876 break;
2877 }
f2f9800d 2878 tipc_nametbl_withdraw(net, publ->type, publ->lower,
37922ea4 2879 publ->upper, publ->key);
0fc87aae
JPM
2880 rc = 0;
2881 }
301bae56
JPM
2882 if (list_empty(&tsk->publications))
2883 tsk->published = 0;
0fc87aae
JPM
2884 return rc;
2885}
2886
5a9ee0be
JPM
2887/* tipc_sk_reinit: set non-zero address in all existing sockets
2888 * when we go from standalone to network mode.
2889 */
e05b31f4 2890void tipc_sk_reinit(struct net *net)
5a9ee0be 2891{
e05b31f4 2892 struct tipc_net *tn = net_generic(net, tipc_net_id);
40f9f439 2893 struct rhashtable_iter iter;
07f6c4bc 2894 struct tipc_sock *tsk;
5a9ee0be 2895 struct tipc_msg *msg;
5a9ee0be 2896
40f9f439
HX
2897 rhashtable_walk_enter(&tn->sk_rht, &iter);
2898
2899 do {
97a6ec4a 2900 rhashtable_walk_start(&iter);
40f9f439
HX
2901
2902 while ((tsk = rhashtable_walk_next(&iter)) && !IS_ERR(tsk)) {
15ef70e2
CW
2903 sock_hold(&tsk->sk);
2904 rhashtable_walk_stop(&iter);
2905 lock_sock(&tsk->sk);
07f6c4bc 2906 msg = &tsk->phdr;
23fd3eac
JM
2907 msg_set_prevnode(msg, tipc_own_addr(net));
2908 msg_set_orignode(msg, tipc_own_addr(net));
15ef70e2
CW
2909 release_sock(&tsk->sk);
2910 rhashtable_walk_start(&iter);
2911 sock_put(&tsk->sk);
07f6c4bc 2912 }
97a6ec4a 2913
40f9f439
HX
2914 rhashtable_walk_stop(&iter);
2915 } while (tsk == ERR_PTR(-EAGAIN));
bd583fe3
CW
2916
2917 rhashtable_walk_exit(&iter);
5a9ee0be
JPM
2918}
2919
e05b31f4 2920static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid)
808d90f9 2921{
e05b31f4 2922 struct tipc_net *tn = net_generic(net, tipc_net_id);
07f6c4bc 2923 struct tipc_sock *tsk;
808d90f9 2924
07f6c4bc 2925 rcu_read_lock();
ab818362 2926 tsk = rhashtable_lookup(&tn->sk_rht, &portid, tsk_rht_params);
07f6c4bc
YX
2927 if (tsk)
2928 sock_hold(&tsk->sk);
2929 rcu_read_unlock();
808d90f9 2930
07f6c4bc 2931 return tsk;
808d90f9
JPM
2932}
2933
07f6c4bc 2934static int tipc_sk_insert(struct tipc_sock *tsk)
808d90f9 2935{
e05b31f4
YX
2936 struct sock *sk = &tsk->sk;
2937 struct net *net = sock_net(sk);
2938 struct tipc_net *tn = net_generic(net, tipc_net_id);
07f6c4bc
YX
2939 u32 remaining = (TIPC_MAX_PORT - TIPC_MIN_PORT) + 1;
2940 u32 portid = prandom_u32() % remaining + TIPC_MIN_PORT;
808d90f9 2941
07f6c4bc
YX
2942 while (remaining--) {
2943 portid++;
2944 if ((portid < TIPC_MIN_PORT) || (portid > TIPC_MAX_PORT))
2945 portid = TIPC_MIN_PORT;
2946 tsk->portid = portid;
2947 sock_hold(&tsk->sk);
6cca7289
HX
2948 if (!rhashtable_lookup_insert_fast(&tn->sk_rht, &tsk->node,
2949 tsk_rht_params))
07f6c4bc
YX
2950 return 0;
2951 sock_put(&tsk->sk);
808d90f9
JPM
2952 }
2953
07f6c4bc 2954 return -1;
808d90f9
JPM
2955}
2956
07f6c4bc 2957static void tipc_sk_remove(struct tipc_sock *tsk)
808d90f9 2958{
07f6c4bc 2959 struct sock *sk = &tsk->sk;
e05b31f4 2960 struct tipc_net *tn = net_generic(sock_net(sk), tipc_net_id);
808d90f9 2961
6cca7289 2962 if (!rhashtable_remove_fast(&tn->sk_rht, &tsk->node, tsk_rht_params)) {
41c6d650 2963 WARN_ON(refcount_read(&sk->sk_refcnt) == 1);
07f6c4bc 2964 __sock_put(sk);
808d90f9 2965 }
808d90f9
JPM
2966}
2967
6cca7289
HX
2968static const struct rhashtable_params tsk_rht_params = {
2969 .nelem_hint = 192,
2970 .head_offset = offsetof(struct tipc_sock, node),
2971 .key_offset = offsetof(struct tipc_sock, portid),
2972 .key_len = sizeof(u32), /* portid */
6cca7289
HX
2973 .max_size = 1048576,
2974 .min_size = 256,
b5e2c150 2975 .automatic_shrinking = true,
6cca7289
HX
2976};
2977
e05b31f4 2978int tipc_sk_rht_init(struct net *net)
808d90f9 2979{
e05b31f4 2980 struct tipc_net *tn = net_generic(net, tipc_net_id);
6cca7289
HX
2981
2982 return rhashtable_init(&tn->sk_rht, &tsk_rht_params);
808d90f9
JPM
2983}
2984
e05b31f4 2985void tipc_sk_rht_destroy(struct net *net)
808d90f9 2986{
e05b31f4
YX
2987 struct tipc_net *tn = net_generic(net, tipc_net_id);
2988
07f6c4bc
YX
2989 /* Wait for socket readers to complete */
2990 synchronize_net();
808d90f9 2991
e05b31f4 2992 rhashtable_destroy(&tn->sk_rht);
808d90f9
JPM
2993}
2994
75da2163
JM
2995static int tipc_sk_join(struct tipc_sock *tsk, struct tipc_group_req *mreq)
2996{
2997 struct net *net = sock_net(&tsk->sk);
75da2163
JM
2998 struct tipc_group *grp = tsk->group;
2999 struct tipc_msg *hdr = &tsk->phdr;
3000 struct tipc_name_seq seq;
3001 int rc;
3002
3003 if (mreq->type < TIPC_RESERVED_TYPES)
3004 return -EACCES;
232d07b7
JM
3005 if (mreq->scope > TIPC_NODE_SCOPE)
3006 return -EINVAL;
75da2163
JM
3007 if (grp)
3008 return -EACCES;
60c25306 3009 grp = tipc_group_create(net, tsk->portid, mreq, &tsk->group_is_open);
75da2163
JM
3010 if (!grp)
3011 return -ENOMEM;
3012 tsk->group = grp;
3013 msg_set_lookup_scope(hdr, mreq->scope);
3014 msg_set_nametype(hdr, mreq->type);
3015 msg_set_dest_droppable(hdr, true);
3016 seq.type = mreq->type;
3017 seq.lower = mreq->instance;
3018 seq.upper = seq.lower;
232d07b7 3019 tipc_nametbl_build_group(net, grp, mreq->type, mreq->scope);
75da2163 3020 rc = tipc_sk_publish(tsk, mreq->scope, &seq);
e233df01 3021 if (rc) {
75da2163 3022 tipc_group_delete(net, grp);
e233df01 3023 tsk->group = NULL;
febafc84 3024 return rc;
e233df01 3025 }
d12d2e12 3026 /* Eliminate any risk that a broadcast overtakes sent JOINs */
399574d4
JM
3027 tsk->mc_method.rcast = true;
3028 tsk->mc_method.mandatory = true;
d12d2e12 3029 tipc_group_join(net, grp, &tsk->sk.sk_rcvbuf);
75da2163
JM
3030 return rc;
3031}
3032
3033static int tipc_sk_leave(struct tipc_sock *tsk)
3034{
3035 struct net *net = sock_net(&tsk->sk);
3036 struct tipc_group *grp = tsk->group;
3037 struct tipc_name_seq seq;
3038 int scope;
3039
3040 if (!grp)
3041 return -EINVAL;
3042 tipc_group_self(grp, &seq, &scope);
3043 tipc_group_delete(net, grp);
3044 tsk->group = NULL;
3045 tipc_sk_withdraw(tsk, scope, &seq);
3046 return 0;
3047}
3048
b97bf3fd 3049/**
247f0f3c 3050 * tipc_setsockopt - set socket option
b97bf3fd
PL
3051 * @sock: socket structure
3052 * @lvl: option level
3053 * @opt: option identifier
3054 * @ov: pointer to new option value
3055 * @ol: length of option value
c4307285
YH
3056 *
3057 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
b97bf3fd 3058 * (to ease compatibility).
c4307285 3059 *
b97bf3fd
PL
3060 * Returns 0 on success, errno otherwise
3061 */
247f0f3c
YX
3062static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
3063 char __user *ov, unsigned int ol)
b97bf3fd 3064{
0c3141e9 3065 struct sock *sk = sock->sk;
58ed9442 3066 struct tipc_sock *tsk = tipc_sk(sk);
75da2163 3067 struct tipc_group_req mreq;
01fd12bb 3068 u32 value = 0;
a08ef476 3069 int res = 0;
b97bf3fd 3070
c4307285
YH
3071 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3072 return 0;
b97bf3fd
PL
3073 if (lvl != SOL_TIPC)
3074 return -ENOPROTOOPT;
01fd12bb
JPM
3075
3076 switch (opt) {
3077 case TIPC_IMPORTANCE:
3078 case TIPC_SRC_DROPPABLE:
3079 case TIPC_DEST_DROPPABLE:
3080 case TIPC_CONN_TIMEOUT:
c0bceb97 3081 case TIPC_NODELAY:
01fd12bb
JPM
3082 if (ol < sizeof(value))
3083 return -EINVAL;
75da2163
JM
3084 if (get_user(value, (u32 __user *)ov))
3085 return -EFAULT;
3086 break;
3087 case TIPC_GROUP_JOIN:
3088 if (ol < sizeof(mreq))
3089 return -EINVAL;
3090 if (copy_from_user(&mreq, ov, sizeof(mreq)))
3091 return -EFAULT;
01fd12bb
JPM
3092 break;
3093 default:
3094 if (ov || ol)
3095 return -EINVAL;
3096 }
b97bf3fd 3097
0c3141e9 3098 lock_sock(sk);
c4307285 3099
b97bf3fd
PL
3100 switch (opt) {
3101 case TIPC_IMPORTANCE:
301bae56 3102 res = tsk_set_importance(tsk, value);
b97bf3fd
PL
3103 break;
3104 case TIPC_SRC_DROPPABLE:
3105 if (sock->type != SOCK_STREAM)
301bae56 3106 tsk_set_unreliable(tsk, value);
c4307285 3107 else
b97bf3fd
PL
3108 res = -ENOPROTOOPT;
3109 break;
3110 case TIPC_DEST_DROPPABLE:
301bae56 3111 tsk_set_unreturnable(tsk, value);
b97bf3fd
PL
3112 break;
3113 case TIPC_CONN_TIMEOUT:
a0f40f02 3114 tipc_sk(sk)->conn_timeout = value;
b97bf3fd 3115 break;
01fd12bb
JPM
3116 case TIPC_MCAST_BROADCAST:
3117 tsk->mc_method.rcast = false;
3118 tsk->mc_method.mandatory = true;
3119 break;
3120 case TIPC_MCAST_REPLICAST:
3121 tsk->mc_method.rcast = true;
3122 tsk->mc_method.mandatory = true;
3123 break;
75da2163
JM
3124 case TIPC_GROUP_JOIN:
3125 res = tipc_sk_join(tsk, &mreq);
3126 break;
3127 case TIPC_GROUP_LEAVE:
3128 res = tipc_sk_leave(tsk);
3129 break;
c0bceb97
JM
3130 case TIPC_NODELAY:
3131 tsk->nodelay = !!value;
3132 tsk_set_nagle(tsk);
3133 break;
b97bf3fd
PL
3134 default:
3135 res = -EINVAL;
3136 }
3137
0c3141e9
AS
3138 release_sock(sk);
3139
b97bf3fd
PL
3140 return res;
3141}
3142
3143/**
247f0f3c 3144 * tipc_getsockopt - get socket option
b97bf3fd
PL
3145 * @sock: socket structure
3146 * @lvl: option level
3147 * @opt: option identifier
3148 * @ov: receptacle for option value
3149 * @ol: receptacle for length of option value
c4307285
YH
3150 *
3151 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
b97bf3fd 3152 * (to ease compatibility).
c4307285 3153 *
b97bf3fd
PL
3154 * Returns 0 on success, errno otherwise
3155 */
247f0f3c
YX
3156static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
3157 char __user *ov, int __user *ol)
b97bf3fd 3158{
0c3141e9 3159 struct sock *sk = sock->sk;
58ed9442 3160 struct tipc_sock *tsk = tipc_sk(sk);
75da2163
JM
3161 struct tipc_name_seq seq;
3162 int len, scope;
b97bf3fd 3163 u32 value;
c4307285 3164 int res;
b97bf3fd 3165
c4307285
YH
3166 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3167 return put_user(0, ol);
b97bf3fd
PL
3168 if (lvl != SOL_TIPC)
3169 return -ENOPROTOOPT;
2db9983a
AS
3170 res = get_user(len, ol);
3171 if (res)
c4307285 3172 return res;
b97bf3fd 3173
0c3141e9 3174 lock_sock(sk);
b97bf3fd
PL
3175
3176 switch (opt) {
3177 case TIPC_IMPORTANCE:
301bae56 3178 value = tsk_importance(tsk);
b97bf3fd
PL
3179 break;
3180 case TIPC_SRC_DROPPABLE:
301bae56 3181 value = tsk_unreliable(tsk);
b97bf3fd
PL
3182 break;
3183 case TIPC_DEST_DROPPABLE:
301bae56 3184 value = tsk_unreturnable(tsk);
b97bf3fd
PL
3185 break;
3186 case TIPC_CONN_TIMEOUT:
301bae56 3187 value = tsk->conn_timeout;
0c3141e9 3188 /* no need to set "res", since already 0 at this point */
b97bf3fd 3189 break;
0e65967e 3190 case TIPC_NODE_RECVQ_DEPTH:
9da3d475 3191 value = 0; /* was tipc_queue_size, now obsolete */
6650613d 3192 break;
0e65967e 3193 case TIPC_SOCK_RECVQ_DEPTH:
6650613d 3194 value = skb_queue_len(&sk->sk_receive_queue);
3195 break;
42e5425a
TN
3196 case TIPC_SOCK_RECVQ_USED:
3197 value = sk_rmem_alloc_get(sk);
3198 break;
75da2163
JM
3199 case TIPC_GROUP_JOIN:
3200 seq.type = 0;
3201 if (tsk->group)
3202 tipc_group_self(tsk->group, &seq, &scope);
3203 value = seq.type;
3204 break;
b97bf3fd
PL
3205 default:
3206 res = -EINVAL;
3207 }
3208
0c3141e9
AS
3209 release_sock(sk);
3210
25860c3b
PG
3211 if (res)
3212 return res; /* "get" failed */
b97bf3fd 3213
25860c3b
PG
3214 if (len < sizeof(value))
3215 return -EINVAL;
3216
3217 if (copy_to_user(ov, &value, sizeof(value)))
3218 return -EFAULT;
3219
3220 return put_user(sizeof(value), ol);
b97bf3fd
PL
3221}
3222
f2f9800d 3223static int tipc_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
78acb1f9 3224{
3e5cf362
JM
3225 struct net *net = sock_net(sock->sk);
3226 struct tipc_sioc_nodeid_req nr = {0};
78acb1f9
EH
3227 struct tipc_sioc_ln_req lnr;
3228 void __user *argp = (void __user *)arg;
3229
3230 switch (cmd) {
3231 case SIOCGETLINKNAME:
3232 if (copy_from_user(&lnr, argp, sizeof(lnr)))
3233 return -EFAULT;
3e5cf362 3234 if (!tipc_node_get_linkname(net,
f2f9800d 3235 lnr.bearer_id & 0xffff, lnr.peer,
78acb1f9
EH
3236 lnr.linkname, TIPC_MAX_LINK_NAME)) {
3237 if (copy_to_user(argp, &lnr, sizeof(lnr)))
3238 return -EFAULT;
3239 return 0;
3240 }
3241 return -EADDRNOTAVAIL;
3e5cf362
JM
3242 case SIOCGETNODEID:
3243 if (copy_from_user(&nr, argp, sizeof(nr)))
3244 return -EFAULT;
3245 if (!tipc_node_get_id(net, nr.peer, nr.node_id))
3246 return -EADDRNOTAVAIL;
3247 if (copy_to_user(argp, &nr, sizeof(nr)))
3248 return -EFAULT;
3249 return 0;
78acb1f9
EH
3250 default:
3251 return -ENOIOCTLCMD;
3252 }
3253}
3254
70b03759
EH
3255static int tipc_socketpair(struct socket *sock1, struct socket *sock2)
3256{
3257 struct tipc_sock *tsk2 = tipc_sk(sock2->sk);
3258 struct tipc_sock *tsk1 = tipc_sk(sock1->sk);
66bc1e8d
EH
3259 u32 onode = tipc_own_addr(sock_net(sock1->sk));
3260
3261 tsk1->peer.family = AF_TIPC;
3262 tsk1->peer.addrtype = TIPC_ADDR_ID;
3263 tsk1->peer.scope = TIPC_NODE_SCOPE;
3264 tsk1->peer.addr.id.ref = tsk2->portid;
3265 tsk1->peer.addr.id.node = onode;
3266 tsk2->peer.family = AF_TIPC;
3267 tsk2->peer.addrtype = TIPC_ADDR_ID;
3268 tsk2->peer.scope = TIPC_NODE_SCOPE;
3269 tsk2->peer.addr.id.ref = tsk1->portid;
3270 tsk2->peer.addr.id.node = onode;
3271
3272 tipc_sk_finish_conn(tsk1, tsk2->portid, onode);
3273 tipc_sk_finish_conn(tsk2, tsk1->portid, onode);
70b03759
EH
3274 return 0;
3275}
3276
ae86b9e3
BH
3277/* Protocol switches for the various types of TIPC sockets */
3278
bca65eae 3279static const struct proto_ops msg_ops = {
0e65967e 3280 .owner = THIS_MODULE,
b97bf3fd 3281 .family = AF_TIPC,
247f0f3c
YX
3282 .release = tipc_release,
3283 .bind = tipc_bind,
3284 .connect = tipc_connect,
66bc1e8d 3285 .socketpair = tipc_socketpair,
245f3d34 3286 .accept = sock_no_accept,
247f0f3c 3287 .getname = tipc_getname,
a11e1d43 3288 .poll = tipc_poll,
78acb1f9 3289 .ioctl = tipc_ioctl,
245f3d34 3290 .listen = sock_no_listen,
247f0f3c
YX
3291 .shutdown = tipc_shutdown,
3292 .setsockopt = tipc_setsockopt,
3293 .getsockopt = tipc_getsockopt,
3294 .sendmsg = tipc_sendmsg,
3295 .recvmsg = tipc_recvmsg,
8238745a
YH
3296 .mmap = sock_no_mmap,
3297 .sendpage = sock_no_sendpage
b97bf3fd
PL
3298};
3299
bca65eae 3300static const struct proto_ops packet_ops = {
0e65967e 3301 .owner = THIS_MODULE,
b97bf3fd 3302 .family = AF_TIPC,
247f0f3c
YX
3303 .release = tipc_release,
3304 .bind = tipc_bind,
3305 .connect = tipc_connect,
70b03759 3306 .socketpair = tipc_socketpair,
247f0f3c
YX
3307 .accept = tipc_accept,
3308 .getname = tipc_getname,
a11e1d43 3309 .poll = tipc_poll,
78acb1f9 3310 .ioctl = tipc_ioctl,
247f0f3c
YX
3311 .listen = tipc_listen,
3312 .shutdown = tipc_shutdown,
3313 .setsockopt = tipc_setsockopt,
3314 .getsockopt = tipc_getsockopt,
3315 .sendmsg = tipc_send_packet,
3316 .recvmsg = tipc_recvmsg,
8238745a
YH
3317 .mmap = sock_no_mmap,
3318 .sendpage = sock_no_sendpage
b97bf3fd
PL
3319};
3320
bca65eae 3321static const struct proto_ops stream_ops = {
0e65967e 3322 .owner = THIS_MODULE,
b97bf3fd 3323 .family = AF_TIPC,
247f0f3c
YX
3324 .release = tipc_release,
3325 .bind = tipc_bind,
3326 .connect = tipc_connect,
70b03759 3327 .socketpair = tipc_socketpair,
247f0f3c
YX
3328 .accept = tipc_accept,
3329 .getname = tipc_getname,
a11e1d43 3330 .poll = tipc_poll,
78acb1f9 3331 .ioctl = tipc_ioctl,
247f0f3c
YX
3332 .listen = tipc_listen,
3333 .shutdown = tipc_shutdown,
3334 .setsockopt = tipc_setsockopt,
3335 .getsockopt = tipc_getsockopt,
365ad353 3336 .sendmsg = tipc_sendstream,
ec8a09fb 3337 .recvmsg = tipc_recvstream,
8238745a
YH
3338 .mmap = sock_no_mmap,
3339 .sendpage = sock_no_sendpage
b97bf3fd
PL
3340};
3341
bca65eae 3342static const struct net_proto_family tipc_family_ops = {
0e65967e 3343 .owner = THIS_MODULE,
b97bf3fd 3344 .family = AF_TIPC,
c5fa7b3c 3345 .create = tipc_sk_create
b97bf3fd
PL
3346};
3347
3348static struct proto tipc_proto = {
3349 .name = "TIPC",
3350 .owner = THIS_MODULE,
cc79dd1b
YX
3351 .obj_size = sizeof(struct tipc_sock),
3352 .sysctl_rmem = sysctl_tipc_rmem
b97bf3fd
PL
3353};
3354
3355/**
4323add6 3356 * tipc_socket_init - initialize TIPC socket interface
c4307285 3357 *
b97bf3fd
PL
3358 * Returns 0 on success, errno otherwise
3359 */
4323add6 3360int tipc_socket_init(void)
b97bf3fd
PL
3361{
3362 int res;
3363
c4307285 3364 res = proto_register(&tipc_proto, 1);
b97bf3fd 3365 if (res) {
2cf8aa19 3366 pr_err("Failed to register TIPC protocol type\n");
b97bf3fd
PL
3367 goto out;
3368 }
3369
3370 res = sock_register(&tipc_family_ops);
3371 if (res) {
2cf8aa19 3372 pr_err("Failed to register TIPC socket type\n");
b97bf3fd
PL
3373 proto_unregister(&tipc_proto);
3374 goto out;
3375 }
b97bf3fd
PL
3376 out:
3377 return res;
3378}
3379
3380/**
4323add6 3381 * tipc_socket_stop - stop TIPC socket interface
b97bf3fd 3382 */
4323add6 3383void tipc_socket_stop(void)
b97bf3fd 3384{
b97bf3fd
PL
3385 sock_unregister(tipc_family_ops.family);
3386 proto_unregister(&tipc_proto);
3387}
34b78a12
RA
3388
3389/* Caller should hold socket lock for the passed tipc socket. */
d8182804 3390static int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk)
34b78a12
RA
3391{
3392 u32 peer_node;
3393 u32 peer_port;
3394 struct nlattr *nest;
3395
3396 peer_node = tsk_peer_node(tsk);
3397 peer_port = tsk_peer_port(tsk);
3398
ae0be8de 3399 nest = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_CON);
517ccc2a
KL
3400 if (!nest)
3401 return -EMSGSIZE;
34b78a12
RA
3402
3403 if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node))
3404 goto msg_full;
3405 if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port))
3406 goto msg_full;
3407
3408 if (tsk->conn_type != 0) {
3409 if (nla_put_flag(skb, TIPC_NLA_CON_FLAG))
3410 goto msg_full;
3411 if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, tsk->conn_type))
3412 goto msg_full;
3413 if (nla_put_u32(skb, TIPC_NLA_CON_INST, tsk->conn_instance))
3414 goto msg_full;
3415 }
3416 nla_nest_end(skb, nest);
3417
3418 return 0;
3419
3420msg_full:
3421 nla_nest_cancel(skb, nest);
3422
3423 return -EMSGSIZE;
3424}
3425
dfde331e
GM
3426static int __tipc_nl_add_sk_info(struct sk_buff *skb, struct tipc_sock
3427 *tsk)
3428{
3429 struct net *net = sock_net(skb->sk);
dfde331e
GM
3430 struct sock *sk = &tsk->sk;
3431
3432 if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->portid) ||
23fd3eac 3433 nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tipc_own_addr(net)))
dfde331e
GM
3434 return -EMSGSIZE;
3435
3436 if (tipc_sk_connected(sk)) {
3437 if (__tipc_nl_add_sk_con(skb, tsk))
3438 return -EMSGSIZE;
3439 } else if (!list_empty(&tsk->publications)) {
3440 if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL))
3441 return -EMSGSIZE;
3442 }
3443 return 0;
3444}
3445
34b78a12 3446/* Caller should hold socket lock for the passed tipc socket. */
d8182804
RA
3447static int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb,
3448 struct tipc_sock *tsk)
34b78a12 3449{
34b78a12 3450 struct nlattr *attrs;
dfde331e 3451 void *hdr;
34b78a12
RA
3452
3453 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
bfb3e5dd 3454 &tipc_genl_family, NLM_F_MULTI, TIPC_NL_SOCK_GET);
34b78a12
RA
3455 if (!hdr)
3456 goto msg_cancel;
3457
ae0be8de 3458 attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK);
34b78a12
RA
3459 if (!attrs)
3460 goto genlmsg_cancel;
dfde331e
GM
3461
3462 if (__tipc_nl_add_sk_info(skb, tsk))
34b78a12
RA
3463 goto attr_msg_cancel;
3464
34b78a12
RA
3465 nla_nest_end(skb, attrs);
3466 genlmsg_end(skb, hdr);
3467
3468 return 0;
3469
3470attr_msg_cancel:
3471 nla_nest_cancel(skb, attrs);
3472genlmsg_cancel:
3473 genlmsg_cancel(skb, hdr);
3474msg_cancel:
3475 return -EMSGSIZE;
3476}
3477
c30b70de
GM
3478int tipc_nl_sk_walk(struct sk_buff *skb, struct netlink_callback *cb,
3479 int (*skb_handler)(struct sk_buff *skb,
3480 struct netlink_callback *cb,
3481 struct tipc_sock *tsk))
34b78a12 3482{
8f5c5fcf 3483 struct rhashtable_iter *iter = (void *)cb->args[4];
dfde331e
GM
3484 struct tipc_sock *tsk;
3485 int err;
34b78a12 3486
9a07efa9
CW
3487 rhashtable_walk_start(iter);
3488 while ((tsk = rhashtable_walk_next(iter)) != NULL) {
3489 if (IS_ERR(tsk)) {
3490 err = PTR_ERR(tsk);
3491 if (err == -EAGAIN) {
3492 err = 0;
d6e164e3
RA
3493 continue;
3494 }
9a07efa9
CW
3495 break;
3496 }
d6e164e3 3497
9a07efa9
CW
3498 sock_hold(&tsk->sk);
3499 rhashtable_walk_stop(iter);
3500 lock_sock(&tsk->sk);
3501 err = skb_handler(skb, cb, tsk);
3502 if (err) {
3503 release_sock(&tsk->sk);
3504 sock_put(&tsk->sk);
3505 goto out;
07f6c4bc 3506 }
9a07efa9
CW
3507 release_sock(&tsk->sk);
3508 rhashtable_walk_start(iter);
3509 sock_put(&tsk->sk);
34b78a12 3510 }
9a07efa9 3511 rhashtable_walk_stop(iter);
d6e164e3 3512out:
34b78a12
RA
3513 return skb->len;
3514}
c30b70de
GM
3515EXPORT_SYMBOL(tipc_nl_sk_walk);
3516
9a07efa9
CW
3517int tipc_dump_start(struct netlink_callback *cb)
3518{
8f5c5fcf
CW
3519 return __tipc_dump_start(cb, sock_net(cb->skb->sk));
3520}
3521EXPORT_SYMBOL(tipc_dump_start);
3522
3523int __tipc_dump_start(struct netlink_callback *cb, struct net *net)
3524{
3525 /* tipc_nl_name_table_dump() uses cb->args[0...3]. */
3526 struct rhashtable_iter *iter = (void *)cb->args[4];
9a07efa9
CW
3527 struct tipc_net *tn = tipc_net(net);
3528
3529 if (!iter) {
3530 iter = kmalloc(sizeof(*iter), GFP_KERNEL);
3531 if (!iter)
3532 return -ENOMEM;
3533
8f5c5fcf 3534 cb->args[4] = (long)iter;
9a07efa9
CW
3535 }
3536
3537 rhashtable_walk_enter(&tn->sk_rht, iter);
3538 return 0;
3539}
9a07efa9
CW
3540
3541int tipc_dump_done(struct netlink_callback *cb)
3542{
8f5c5fcf 3543 struct rhashtable_iter *hti = (void *)cb->args[4];
9a07efa9
CW
3544
3545 rhashtable_walk_exit(hti);
3546 kfree(hti);
3547 return 0;
3548}
3549EXPORT_SYMBOL(tipc_dump_done);
3550
e41f0548
CW
3551int tipc_sk_fill_sock_diag(struct sk_buff *skb, struct netlink_callback *cb,
3552 struct tipc_sock *tsk, u32 sk_filter_state,
c30b70de
GM
3553 u64 (*tipc_diag_gen_cookie)(struct sock *sk))
3554{
3555 struct sock *sk = &tsk->sk;
3556 struct nlattr *attrs;
3557 struct nlattr *stat;
3558
3559 /*filter response w.r.t sk_state*/
3560 if (!(sk_filter_state & (1 << sk->sk_state)))
3561 return 0;
3562
ae0be8de 3563 attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK);
c30b70de
GM
3564 if (!attrs)
3565 goto msg_cancel;
3566
3567 if (__tipc_nl_add_sk_info(skb, tsk))
3568 goto attr_msg_cancel;
3569
3570 if (nla_put_u32(skb, TIPC_NLA_SOCK_TYPE, (u32)sk->sk_type) ||
3571 nla_put_u32(skb, TIPC_NLA_SOCK_TIPC_STATE, (u32)sk->sk_state) ||
3572 nla_put_u32(skb, TIPC_NLA_SOCK_INO, sock_i_ino(sk)) ||
3573 nla_put_u32(skb, TIPC_NLA_SOCK_UID,
e41f0548 3574 from_kuid_munged(sk_user_ns(NETLINK_CB(cb->skb).sk),
4b2e6877 3575 sock_i_uid(sk))) ||
c30b70de
GM
3576 nla_put_u64_64bit(skb, TIPC_NLA_SOCK_COOKIE,
3577 tipc_diag_gen_cookie(sk),
3578 TIPC_NLA_SOCK_PAD))
3579 goto attr_msg_cancel;
3580
ae0be8de 3581 stat = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_STAT);
c30b70de
GM
3582 if (!stat)
3583 goto attr_msg_cancel;
3584
3585 if (nla_put_u32(skb, TIPC_NLA_SOCK_STAT_RCVQ,
3586 skb_queue_len(&sk->sk_receive_queue)) ||
3587 nla_put_u32(skb, TIPC_NLA_SOCK_STAT_SENDQ,
872619d8
GM
3588 skb_queue_len(&sk->sk_write_queue)) ||
3589 nla_put_u32(skb, TIPC_NLA_SOCK_STAT_DROP,
3590 atomic_read(&sk->sk_drops)))
c30b70de
GM
3591 goto stat_msg_cancel;
3592
3593 if (tsk->cong_link_cnt &&
3594 nla_put_flag(skb, TIPC_NLA_SOCK_STAT_LINK_CONG))
3595 goto stat_msg_cancel;
3596
3597 if (tsk_conn_cong(tsk) &&
3598 nla_put_flag(skb, TIPC_NLA_SOCK_STAT_CONN_CONG))
3599 goto stat_msg_cancel;
3600
3601 nla_nest_end(skb, stat);
a1be5a20
GM
3602
3603 if (tsk->group)
3604 if (tipc_group_fill_sock_diag(tsk->group, skb))
3605 goto stat_msg_cancel;
3606
c30b70de
GM
3607 nla_nest_end(skb, attrs);
3608
3609 return 0;
3610
3611stat_msg_cancel:
3612 nla_nest_cancel(skb, stat);
3613attr_msg_cancel:
3614 nla_nest_cancel(skb, attrs);
3615msg_cancel:
3616 return -EMSGSIZE;
3617}
3618EXPORT_SYMBOL(tipc_sk_fill_sock_diag);
1a1a143d 3619
dfde331e
GM
3620int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb)
3621{
c30b70de 3622 return tipc_nl_sk_walk(skb, cb, __tipc_nl_add_sk);
dfde331e
GM
3623}
3624
1a1a143d 3625/* Caller should hold socket lock for the passed tipc socket. */
d8182804
RA
3626static int __tipc_nl_add_sk_publ(struct sk_buff *skb,
3627 struct netlink_callback *cb,
3628 struct publication *publ)
1a1a143d
RA
3629{
3630 void *hdr;
3631 struct nlattr *attrs;
3632
3633 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
bfb3e5dd 3634 &tipc_genl_family, NLM_F_MULTI, TIPC_NL_PUBL_GET);
1a1a143d
RA
3635 if (!hdr)
3636 goto msg_cancel;
3637
ae0be8de 3638 attrs = nla_nest_start_noflag(skb, TIPC_NLA_PUBL);
1a1a143d
RA
3639 if (!attrs)
3640 goto genlmsg_cancel;
3641
3642 if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key))
3643 goto attr_msg_cancel;
3644 if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->type))
3645 goto attr_msg_cancel;
3646 if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->lower))
3647 goto attr_msg_cancel;
3648 if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->upper))
3649 goto attr_msg_cancel;
3650
3651 nla_nest_end(skb, attrs);
3652 genlmsg_end(skb, hdr);
3653
3654 return 0;
3655
3656attr_msg_cancel:
3657 nla_nest_cancel(skb, attrs);
3658genlmsg_cancel:
3659 genlmsg_cancel(skb, hdr);
3660msg_cancel:
3661 return -EMSGSIZE;
3662}
3663
3664/* Caller should hold socket lock for the passed tipc socket. */
d8182804
RA
3665static int __tipc_nl_list_sk_publ(struct sk_buff *skb,
3666 struct netlink_callback *cb,
3667 struct tipc_sock *tsk, u32 *last_publ)
1a1a143d
RA
3668{
3669 int err;
3670 struct publication *p;
3671
3672 if (*last_publ) {
e50e73e1 3673 list_for_each_entry(p, &tsk->publications, binding_sock) {
1a1a143d
RA
3674 if (p->key == *last_publ)
3675 break;
3676 }
3677 if (p->key != *last_publ) {
3678 /* We never set seq or call nl_dump_check_consistent()
3679 * this means that setting prev_seq here will cause the
3680 * consistence check to fail in the netlink callback
3681 * handler. Resulting in the last NLMSG_DONE message
3682 * having the NLM_F_DUMP_INTR flag set.
3683 */
3684 cb->prev_seq = 1;
3685 *last_publ = 0;
3686 return -EPIPE;
3687 }
3688 } else {
3689 p = list_first_entry(&tsk->publications, struct publication,
e50e73e1 3690 binding_sock);
1a1a143d
RA
3691 }
3692
e50e73e1 3693 list_for_each_entry_from(p, &tsk->publications, binding_sock) {
1a1a143d
RA
3694 err = __tipc_nl_add_sk_publ(skb, cb, p);
3695 if (err) {
3696 *last_publ = p->key;
3697 return err;
3698 }
3699 }
3700 *last_publ = 0;
3701
3702 return 0;
3703}
3704
3705int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)
3706{
3707 int err;
07f6c4bc 3708 u32 tsk_portid = cb->args[0];
1a1a143d
RA
3709 u32 last_publ = cb->args[1];
3710 u32 done = cb->args[2];
e05b31f4 3711 struct net *net = sock_net(skb->sk);
1a1a143d
RA
3712 struct tipc_sock *tsk;
3713
07f6c4bc 3714 if (!tsk_portid) {
057af707 3715 struct nlattr **attrs = genl_dumpit_info(cb)->attrs;
1a1a143d
RA
3716 struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1];
3717
45e093ae
RA
3718 if (!attrs[TIPC_NLA_SOCK])
3719 return -EINVAL;
3720
8cb08174
JB
3721 err = nla_parse_nested_deprecated(sock, TIPC_NLA_SOCK_MAX,
3722 attrs[TIPC_NLA_SOCK],
3723 tipc_nl_sock_policy, NULL);
1a1a143d
RA
3724 if (err)
3725 return err;
3726
3727 if (!sock[TIPC_NLA_SOCK_REF])
3728 return -EINVAL;
3729
07f6c4bc 3730 tsk_portid = nla_get_u32(sock[TIPC_NLA_SOCK_REF]);
1a1a143d
RA
3731 }
3732
3733 if (done)
3734 return 0;
3735
e05b31f4 3736 tsk = tipc_sk_lookup(net, tsk_portid);
1a1a143d
RA
3737 if (!tsk)
3738 return -EINVAL;
3739
3740 lock_sock(&tsk->sk);
3741 err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ);
3742 if (!err)
3743 done = 1;
3744 release_sock(&tsk->sk);
07f6c4bc 3745 sock_put(&tsk->sk);
1a1a143d 3746
07f6c4bc 3747 cb->args[0] = tsk_portid;
1a1a143d
RA
3748 cb->args[1] = last_publ;
3749 cb->args[2] = done;
3750
3751 return skb->len;
3752}
b4b9771b 3753
01e661eb
TL
3754/**
3755 * tipc_sk_filtering - check if a socket should be traced
3756 * @sk: the socket to be examined
3757 * @sysctl_tipc_sk_filter[]: the socket tuple for filtering,
3758 * (portid, sock type, name type, name lower, name upper)
3759 *
3760 * Returns true if the socket meets the socket tuple data
3761 * (value 0 = 'any') or when there is no tuple set (all = 0),
3762 * otherwise false
3763 */
3764bool tipc_sk_filtering(struct sock *sk)
3765{
3766 struct tipc_sock *tsk;
3767 struct publication *p;
3768 u32 _port, _sktype, _type, _lower, _upper;
3769 u32 type = 0, lower = 0, upper = 0;
3770
3771 if (!sk)
3772 return true;
3773
3774 tsk = tipc_sk(sk);
3775
3776 _port = sysctl_tipc_sk_filter[0];
3777 _sktype = sysctl_tipc_sk_filter[1];
3778 _type = sysctl_tipc_sk_filter[2];
3779 _lower = sysctl_tipc_sk_filter[3];
3780 _upper = sysctl_tipc_sk_filter[4];
3781
3782 if (!_port && !_sktype && !_type && !_lower && !_upper)
3783 return true;
3784
3785 if (_port)
3786 return (_port == tsk->portid);
3787
3788 if (_sktype && _sktype != sk->sk_type)
3789 return false;
3790
3791 if (tsk->published) {
3792 p = list_first_entry_or_null(&tsk->publications,
3793 struct publication, binding_sock);
3794 if (p) {
3795 type = p->type;
3796 lower = p->lower;
3797 upper = p->upper;
3798 }
3799 }
3800
3801 if (!tipc_sk_type_connectionless(sk)) {
3802 type = tsk->conn_type;
3803 lower = tsk->conn_instance;
3804 upper = tsk->conn_instance;
3805 }
3806
3807 if ((_type && _type != type) || (_lower && _lower != lower) ||
3808 (_upper && _upper != upper))
3809 return false;
3810
3811 return true;
3812}
3813
b4b9771b
TL
3814u32 tipc_sock_get_portid(struct sock *sk)
3815{
3816 return (sk) ? (tipc_sk(sk))->portid : 0;
3817}
3818
01e661eb
TL
3819/**
3820 * tipc_sk_overlimit1 - check if socket rx queue is about to be overloaded,
3821 * both the rcv and backlog queues are considered
3822 * @sk: tipc sk to be checked
3823 * @skb: tipc msg to be checked
3824 *
3825 * Returns true if the socket rx queue allocation is > 90%, otherwise false
3826 */
3827
3828bool tipc_sk_overlimit1(struct sock *sk, struct sk_buff *skb)
3829{
3830 atomic_t *dcnt = &tipc_sk(sk)->dupl_rcvcnt;
3831 unsigned int lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
3832 unsigned int qsize = sk->sk_backlog.len + sk_rmem_alloc_get(sk);
3833
3834 return (qsize > lim * 90 / 100);
3835}
3836
3837/**
3838 * tipc_sk_overlimit2 - check if socket rx queue is about to be overloaded,
3839 * only the rcv queue is considered
3840 * @sk: tipc sk to be checked
3841 * @skb: tipc msg to be checked
3842 *
3843 * Returns true if the socket rx queue allocation is > 90%, otherwise false
3844 */
3845
3846bool tipc_sk_overlimit2(struct sock *sk, struct sk_buff *skb)
3847{
3848 unsigned int lim = rcvbuf_limit(sk, skb);
3849 unsigned int qsize = sk_rmem_alloc_get(sk);
3850
3851 return (qsize > lim * 90 / 100);
3852}
3853
b4b9771b
TL
3854/**
3855 * tipc_sk_dump - dump TIPC socket
3856 * @sk: tipc sk to be dumped
3857 * @dqueues: bitmask to decide if any socket queue to be dumped?
3858 * - TIPC_DUMP_NONE: don't dump socket queues
3859 * - TIPC_DUMP_SK_SNDQ: dump socket send queue
3860 * - TIPC_DUMP_SK_RCVQ: dump socket rcv queue
3861 * - TIPC_DUMP_SK_BKLGQ: dump socket backlog queue
3862 * - TIPC_DUMP_ALL: dump all the socket queues above
3863 * @buf: returned buffer of dump data in format
3864 */
3865int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)
3866{
3867 int i = 0;
3868 size_t sz = (dqueues) ? SK_LMAX : SK_LMIN;
3869 struct tipc_sock *tsk;
3870 struct publication *p;
3871 bool tsk_connected;
3872
3873 if (!sk) {
3874 i += scnprintf(buf, sz, "sk data: (null)\n");
3875 return i;
3876 }
3877
3878 tsk = tipc_sk(sk);
3879 tsk_connected = !tipc_sk_type_connectionless(sk);
3880
3881 i += scnprintf(buf, sz, "sk data: %u", sk->sk_type);
3882 i += scnprintf(buf + i, sz - i, " %d", sk->sk_state);
3883 i += scnprintf(buf + i, sz - i, " %x", tsk_own_node(tsk));
3884 i += scnprintf(buf + i, sz - i, " %u", tsk->portid);
3885 i += scnprintf(buf + i, sz - i, " | %u", tsk_connected);
3886 if (tsk_connected) {
3887 i += scnprintf(buf + i, sz - i, " %x", tsk_peer_node(tsk));
3888 i += scnprintf(buf + i, sz - i, " %u", tsk_peer_port(tsk));
3889 i += scnprintf(buf + i, sz - i, " %u", tsk->conn_type);
3890 i += scnprintf(buf + i, sz - i, " %u", tsk->conn_instance);
3891 }
3892 i += scnprintf(buf + i, sz - i, " | %u", tsk->published);
3893 if (tsk->published) {
3894 p = list_first_entry_or_null(&tsk->publications,
3895 struct publication, binding_sock);
3896 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->type : 0);
3897 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->lower : 0);
3898 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->upper : 0);
3899 }
3900 i += scnprintf(buf + i, sz - i, " | %u", tsk->snd_win);
3901 i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_win);
3902 i += scnprintf(buf + i, sz - i, " %u", tsk->max_pkt);
3903 i += scnprintf(buf + i, sz - i, " %x", tsk->peer_caps);
3904 i += scnprintf(buf + i, sz - i, " %u", tsk->cong_link_cnt);
3905 i += scnprintf(buf + i, sz - i, " %u", tsk->snt_unacked);
3906 i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_unacked);
3907 i += scnprintf(buf + i, sz - i, " %u", atomic_read(&tsk->dupl_rcvcnt));
3908 i += scnprintf(buf + i, sz - i, " %u", sk->sk_shutdown);
3909 i += scnprintf(buf + i, sz - i, " | %d", sk_wmem_alloc_get(sk));
3910 i += scnprintf(buf + i, sz - i, " %d", sk->sk_sndbuf);
3911 i += scnprintf(buf + i, sz - i, " | %d", sk_rmem_alloc_get(sk));
3912 i += scnprintf(buf + i, sz - i, " %d", sk->sk_rcvbuf);
70c26558 3913 i += scnprintf(buf + i, sz - i, " | %d\n", READ_ONCE(sk->sk_backlog.len));
b4b9771b
TL
3914
3915 if (dqueues & TIPC_DUMP_SK_SNDQ) {
3916 i += scnprintf(buf + i, sz - i, "sk_write_queue: ");
3917 i += tipc_list_dump(&sk->sk_write_queue, false, buf + i);
3918 }
3919
3920 if (dqueues & TIPC_DUMP_SK_RCVQ) {
3921 i += scnprintf(buf + i, sz - i, "sk_receive_queue: ");
3922 i += tipc_list_dump(&sk->sk_receive_queue, false, buf + i);
3923 }
3924
3925 if (dqueues & TIPC_DUMP_SK_BKLGQ) {
3926 i += scnprintf(buf + i, sz - i, "sk_backlog:\n head ");
3927 i += tipc_skb_dump(sk->sk_backlog.head, false, buf + i);
3928 if (sk->sk_backlog.tail != sk->sk_backlog.head) {
3929 i += scnprintf(buf + i, sz - i, " tail ");
3930 i += tipc_skb_dump(sk->sk_backlog.tail, false,
3931 buf + i);
3932 }
3933 }
3934
3935 return i;
3936}