projects / linux-2.6-block.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
Linux 6.10-rc3
[linux-2.6-block.git] / net / tipc / socket.c
CommitLineData
b97bf3fd 1/*
02c00c2a 2 * net/tipc/socket.c: TIPC socket API
c4307285 3 *
60c102ee 4 * Copyright (c) 2001-2007, 2012-2019, Ericsson AB
c5fa7b3c 5 * Copyright (c) 2004-2008, 2010-2013, Wind River Systems
998d3907 6 * Copyright (c) 2020-2021, Red Hat Inc
b97bf3fd
PL		 7 * All rights reserved.
8 *
9ea1fd3c 9 * Redistribution and use in source and binary forms, with or without
b97bf3fd
PL		 10 * modification, are permitted provided that the following conditions are met:
11 *
9ea1fd3c
PL		 12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the names of the copyright holders nor the names of its
18 * contributors may be used to endorse or promote products derived from
19 * this software without specific prior written permission.
b97bf3fd 20 *
9ea1fd3c
PL		 21 * Alternatively, this software may be distributed under the terms of the
22 * GNU General Public License ("GPL") version 2 as published by the Free
23 * Software Foundation.
24 *
25 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
26 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28 * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
29 * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
30 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
31 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
32 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
33 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
34 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
b97bf3fd
PL		 35 * POSSIBILITY OF SUCH DAMAGE.
36 */
37
07f6c4bc 38#include <linux/rhashtable.h>
174cd4b1 39#include <linux/sched/signal.h>
40e0b090 40#include <trace/events/sock.h>
174cd4b1 41
b97bf3fd 42#include "core.h"
e2dafe87 43#include "name_table.h"
78acb1f9 44#include "node.h"
e2dafe87 45#include "link.h"
c637c103 46#include "name_distr.h"
2e84c60b 47#include "socket.h"
a6bf70f7 48#include "bcast.h"
49cc66ea 49#include "netlink.h"
75da2163 50#include "group.h"
b4b9771b 51#include "trace.h"
2cf8aa19 52
0a3e060f
TL		 53#define NAGLE_START_INIT 4
54#define NAGLE_START_MAX 1024
67879274 55#define CONN_TIMEOUT_DEFAULT 8000 /* default connect timeout = 8s */
0d5fcebf 56#define CONN_PROBING_INTV msecs_to_jiffies(3600000) /* [ms] => 1 h */
07f6c4bc
YX		 57#define TIPC_MAX_PORT 0xffffffff
58#define TIPC_MIN_PORT 1
60462191 59#define TIPC_ACK_RATE 4 /* ACK at 1/4 of rcv window size */
301bae56 60
0c288c86
PB		 61enum {
62 TIPC_LISTEN = TCP_LISTEN,
8ea642ee 63 TIPC_ESTABLISHED = TCP_ESTABLISHED,
438adcaf 64 TIPC_OPEN = TCP_CLOSE,
9fd4b070 65 TIPC_DISCONNECTING = TCP_CLOSE_WAIT,
99a20889 66 TIPC_CONNECTING = TCP_SYN_SENT,
0c288c86
PB		 67};
68
31c82a2d
JM		 69struct sockaddr_pair {
70 struct sockaddr_tipc sock;
71 struct sockaddr_tipc member;
72};
73
301bae56
JPM		 74/**
75 * struct tipc_sock - TIPC socket structure
76 * @sk: socket - interacts with 'port' and with user via the socket API
301bae56 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
f172f4b8 81 * @cong_links: list of congested links
301bae56
JPM		 82 * @publications: list of publications for port
83 * @pub_count: total # of publications port has made during its lifetime
301bae56 84 * @conn_timeout: the time we can wait for an unresponded setup request
f172f4b8 85 * @probe_unacked: probe has not received ack yet
301bae56 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
f172f4b8
RD		 89 * @snd_win: send window size
90 * @peer_caps: peer capabilities mask
301bae56 91 * @rcv_unacked: # messages read by user, but not yet acked back to peer
f172f4b8 92 * @rcv_win: receive window size
aeda16b6 93 * @peer: 'connected' peer for dgram/rdm
07f6c4bc 94 * @node: hash table node
01fd12bb 95 * @mc_method: cookie for use between socket and broadcast layer
07f6c4bc 96 * @rcu: rcu struct for tipc_sock
f172f4b8
RD		 97 * @group: TIPC communications group
98 * @oneway: message count in one direction (FIXME)
99 * @nagle_start: current nagle value
100 * @snd_backlog: send backlog count
101 * @msg_acc: messages accepted; used in managing backlog and nagle
102 * @pkt_cnt: TIPC socket packet count
103 * @expect_ack: whether this TIPC socket is expecting an ack
104 * @nodelay: setsockopt() TIPC_NODELAY setting
105 * @group_is_open: TIPC socket group is fully open (FIXME)
14623e00
JM		 106 * @published: true if port has one or more associated names
107 * @conn_addrtype: address type used when establishing connection
301bae56
JPM		 108 */
109struct tipc_sock {
110 struct sock sk;
301bae56 111 u32 max_pkt;
c0bceb97 112 u32 maxnagle;
07f6c4bc 113 u32 portid;
301bae56 114 struct tipc_msg phdr;
365ad353 115 struct list_head cong_links;
301bae56
JPM		 116 struct list_head publications;
117 u32 pub_count;
301bae56 118 atomic_t dupl_rcvcnt;
67879274 119 u16 conn_timeout;
8ea642ee 120 bool probe_unacked;
365ad353 121 u16 cong_link_cnt;
10724cc7
JPM		 122 u16 snt_unacked;
123 u16 snd_win;
60020e18 124 u16 peer_caps;
10724cc7
JPM		 125 u16 rcv_unacked;
126 u16 rcv_win;
aeda16b6 127 struct sockaddr_tipc peer;
07f6c4bc 128 struct rhash_head node;
01fd12bb 129 struct tipc_mc_method mc_method;
07f6c4bc 130 struct rcu_head rcu;
75da2163 131 struct tipc_group *group;
c0bceb97 132 u32 oneway;
0a3e060f 133 u32 nagle_start;
c0bceb97 134 u16 snd_backlog;
0a3e060f
TL		 135 u16 msg_acc;
136 u16 pkt_cnt;
c0bceb97
JM		 137 bool expect_ack;
138 bool nodelay;
60c25306 139 bool group_is_open;
50a3499a 140 bool published;
14623e00 141 u8 conn_addrtype;
301bae56 142};
b97bf3fd 143
64ac5f59 144static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb);
676d2369 145static void tipc_data_ready(struct sock *sk);
f288bef4 146static void tipc_write_space(struct sock *sk);
f4195d1e 147static void tipc_sock_destruct(struct sock *sk);
247f0f3c 148static int tipc_release(struct socket *sock);
31b102bb 149static void tipc_sk_timeout(struct timer_list *t);
50a3499a 150static int tipc_sk_publish(struct tipc_sock *tsk, struct tipc_uaddr *ua);
2c98da07 151static int tipc_sk_withdraw(struct tipc_sock *tsk, struct tipc_uaddr *ua);
75da2163 152static int tipc_sk_leave(struct tipc_sock *tsk);
e05b31f4 153static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid);
07f6c4bc
YX		 154static int tipc_sk_insert(struct tipc_sock *tsk);
155static void tipc_sk_remove(struct tipc_sock *tsk);
365ad353 156static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz);
39a0295f 157static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dsz);
0a3e060f 158static void tipc_sk_push_backlog(struct tipc_sock *tsk, bool nagle_ack);
f8dd60de 159static int tipc_wait_for_connect(struct socket *sock, long *timeo_p);
b97bf3fd 160
bca65eae
FW		 161static const struct proto_ops packet_ops;
162static const struct proto_ops stream_ops;
163static const struct proto_ops msg_ops;
b97bf3fd 164static struct proto tipc_proto;
6cca7289
HX		 165static const struct rhashtable_params tsk_rht_params;
166
c5898636
JPM		 167static u32 tsk_own_node(struct tipc_sock *tsk)
168{
169 return msg_prevnode(&tsk->phdr);
170}
171
301bae56 172static u32 tsk_peer_node(struct tipc_sock *tsk)
2e84c60b 173{
301bae56 174 return msg_destnode(&tsk->phdr);
2e84c60b
JPM		 175}
176
301bae56 177static u32 tsk_peer_port(struct tipc_sock *tsk)
2e84c60b 178{
301bae56 179 return msg_destport(&tsk->phdr);
2e84c60b
JPM		 180}
181
301bae56 182static bool tsk_unreliable(struct tipc_sock *tsk)
2e84c60b 183{
301bae56 184 return msg_src_droppable(&tsk->phdr) != 0;
2e84c60b
JPM		 185}
186
301bae56 187static void tsk_set_unreliable(struct tipc_sock *tsk, bool unreliable)
2e84c60b 188{
301bae56 189 msg_set_src_droppable(&tsk->phdr, unreliable ? 1 : 0);
2e84c60b
JPM		 190}
191
301bae56 192static bool tsk_unreturnable(struct tipc_sock *tsk)
2e84c60b 193{
301bae56 194 return msg_dest_droppable(&tsk->phdr) != 0;
2e84c60b
JPM		 195}
196
301bae56 197static void tsk_set_unreturnable(struct tipc_sock *tsk, bool unreturnable)
2e84c60b 198{
301bae56 199 msg_set_dest_droppable(&tsk->phdr, unreturnable ? 1 : 0);
2e84c60b
JPM		 200}
201
301bae56 202static int tsk_importance(struct tipc_sock *tsk)
2e84c60b 203{
301bae56 204 return msg_importance(&tsk->phdr);
2e84c60b
JPM		 205}
206
095ae612 207static struct tipc_sock *tipc_sk(const struct sock *sk)
2e84c60b 208{
095ae612 209 return container_of(sk, struct tipc_sock, sk);
2e84c60b 210}
8826cde6 211
095ae612 212int tsk_set_importance(struct sock *sk, int imp)
301bae56 213{
095ae612
CH		 214 if (imp > TIPC_CRITICAL_IMPORTANCE)
215 return -EINVAL;
216 msg_set_importance(&tipc_sk(sk)->phdr, (u32)imp);
217 return 0;
301bae56
JPM		 218}
219
10724cc7 220static bool tsk_conn_cong(struct tipc_sock *tsk)
301bae56 221{
6998cc6e 222 return tsk->snt_unacked > tsk->snd_win;
10724cc7
JPM		 223}
224
b7d42635
JM		 225static u16 tsk_blocks(int len)
226{
227 return ((len / FLOWCTL_BLK_SZ) + 1);
228}
229
10724cc7
JPM		 230/* tsk_blocks(): translate a buffer size in bytes to number of
231 * advertisable blocks, taking into account the ratio truesize(len)/len
232 * We can trust that this ratio is always < 4 for len >= FLOWCTL_BLK_SZ
233 */
234static u16 tsk_adv_blocks(int len)
235{
236 return len / FLOWCTL_BLK_SZ / 4;
237}
238
239/* tsk_inc(): increment counter for sent or received data
240 * - If block based flow control is not supported by peer we
241 * fall back to message based ditto, incrementing the counter
242 */
243static u16 tsk_inc(struct tipc_sock *tsk, int msglen)
244{
245 if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
246 return ((msglen / FLOWCTL_BLK_SZ) + 1);
247 return 1;
301bae56
JPM		 248}
249
c0bceb97
JM		 250/* tsk_set_nagle - enable/disable nagle property by manipulating maxnagle
251 */
252static void tsk_set_nagle(struct tipc_sock *tsk)
253{
254 struct sock *sk = &tsk->sk;
255
256 tsk->maxnagle = 0;
257 if (sk->sk_type != SOCK_STREAM)
258 return;
259 if (tsk->nodelay)
260 return;
261 if (!(tsk->peer_caps & TIPC_NAGLE))
262 return;
263 /* Limit node local buffer size to avoid receive queue overflow */
264 if (tsk->max_pkt == MAX_MSG_SIZE)
265 tsk->maxnagle = 1500;
266 else
267 tsk->maxnagle = tsk->max_pkt;
268}
269
0c3141e9 270/**
2e84c60b 271 * tsk_advance_rx_queue - discard first buffer in socket receive queue
f172f4b8 272 * @sk: network socket
0c3141e9
AS		 273 *
274 * Caller must hold socket lock
b97bf3fd 275 */
2e84c60b 276static void tsk_advance_rx_queue(struct sock *sk)
b97bf3fd 277{
01e661eb 278 trace_tipc_sk_advance_rx(sk, NULL, TIPC_DUMP_SK_RCVQ, " ");
5f6d9123 279 kfree_skb(__skb_dequeue(&sk->sk_receive_queue));
b97bf3fd
PL		 280}
281
bcd3ffd4
JPM		 282/* tipc_sk_respond() : send response message back to sender
283 */
284static void tipc_sk_respond(struct sock *sk, struct sk_buff *skb, int err)
285{
286 u32 selector;
287 u32 dnode;
288 u32 onode = tipc_own_addr(sock_net(sk));
289
290 if (!tipc_msg_reverse(onode, &skb, err))
291 return;
292
01e661eb 293 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE, "@sk_respond!");
bcd3ffd4
JPM		 294 dnode = msg_destnode(buf_msg(skb));
295 selector = msg_origport(buf_msg(skb));
296 tipc_node_xmit_skb(sock_net(sk), skb, dnode, selector);
297}
298
b97bf3fd 299/**
2e84c60b 300 * tsk_rej_rx_queue - reject all buffers in socket receive queue
f172f4b8
RD		 301 * @sk: network socket
302 * @error: response error code
0c3141e9
AS		 303 *
304 * Caller must hold socket lock
b97bf3fd 305 */
49afb806 306static void tsk_rej_rx_queue(struct sock *sk, int error)
b97bf3fd 307{
a6ca1094 308 struct sk_buff *skb;
0c3141e9 309
bcd3ffd4 310 while ((skb = __skb_dequeue(&sk->sk_receive_queue)))
49afb806 311 tipc_sk_respond(sk, skb, error);
b97bf3fd
PL		 312}
313
d0ac89f6 314static bool tipc_sk_connected(const struct sock *sk)
d6fb7e9c 315{
d0ac89f6 316 return READ_ONCE(sk->sk_state) == TIPC_ESTABLISHED;
d6fb7e9c
PB		 317}
318
c752023a
PB		 319/* tipc_sk_type_connectionless - check if the socket is datagram socket
320 * @sk: socket
321 *
322 * Returns true if connection less, false otherwise
323 */
324static bool tipc_sk_type_connectionless(struct sock *sk)
325{
326 return sk->sk_type == SOCK_RDM || sk->sk_type == SOCK_DGRAM;
327}
328
2e84c60b 329/* tsk_peer_msg - verify if message was sent by connected port's peer
0fc87aae
JPM		 330 *
331 * Handles cases where the node's network address has changed from
332 * the default of <0.0.0> to its configured setting.
333 */
2e84c60b 334static bool tsk_peer_msg(struct tipc_sock *tsk, struct tipc_msg *msg)
0fc87aae 335{
d6fb7e9c 336 struct sock *sk = &tsk->sk;
23fd3eac 337 u32 self = tipc_own_addr(sock_net(sk));
301bae56 338 u32 peer_port = tsk_peer_port(tsk);
23fd3eac 339 u32 orig_node, peer_node;
0fc87aae 340
d6fb7e9c 341 if (unlikely(!tipc_sk_connected(sk)))
0fc87aae
JPM		 342 return false;
343
344 if (unlikely(msg_origport(msg) != peer_port))
345 return false;
346
347 orig_node = msg_orignode(msg);
301bae56 348 peer_node = tsk_peer_node(tsk);
0fc87aae
JPM		 349
350 if (likely(orig_node == peer_node))
351 return true;
352
23fd3eac 353 if (!orig_node && peer_node == self)
0fc87aae
JPM		 354 return true;
355
23fd3eac 356 if (!peer_node && orig_node == self)
0fc87aae
JPM		 357 return true;
358
359 return false;
360}
361
0c288c86
PB		 362/* tipc_set_sk_state - set the sk_state of the socket
363 * @sk: socket
364 *
365 * Caller must hold socket lock
366 *
367 * Returns 0 on success, errno otherwise
368 */
369static int tipc_set_sk_state(struct sock *sk, int state)
370{
438adcaf 371 int oldsk_state = sk->sk_state;
0c288c86
PB		 372 int res = -EINVAL;
373
374 switch (state) {
438adcaf
PB		 375 case TIPC_OPEN:
376 res = 0;
377 break;
0c288c86 378 case TIPC_LISTEN:
99a20889 379 case TIPC_CONNECTING:
438adcaf 380 if (oldsk_state == TIPC_OPEN)
0c288c86
PB		 381 res = 0;
382 break;
8ea642ee 383 case TIPC_ESTABLISHED:
99a20889 384 if (oldsk_state == TIPC_CONNECTING ||
438adcaf 385 oldsk_state == TIPC_OPEN)
8ea642ee
PB		 386 res = 0;
387 break;
9fd4b070 388 case TIPC_DISCONNECTING:
99a20889 389 if (oldsk_state == TIPC_CONNECTING ||
9fd4b070
PB		 390 oldsk_state == TIPC_ESTABLISHED)
391 res = 0;
392 break;
0c288c86
PB		 393 }
394
395 if (!res)
396 sk->sk_state = state;
397
398 return res;
399}
400
8c44e1af
JPM		 401static int tipc_sk_sock_err(struct socket *sock, long *timeout)
402{
403 struct sock *sk = sock->sk;
404 int err = sock_error(sk);
405 int typ = sock->type;
406
407 if (err)
408 return err;
409 if (typ == SOCK_STREAM || typ == SOCK_SEQPACKET) {
410 if (sk->sk_state == TIPC_DISCONNECTING)
411 return -EPIPE;
412 else if (!tipc_sk_connected(sk))
413 return -ENOTCONN;
414 }
415 if (!*timeout)
416 return -EAGAIN;
417 if (signal_pending(current))
418 return sock_intr_errno(*timeout);
419
420 return 0;
421}
422
844cf763
JPM		 423#define tipc_wait_for_cond(sock_, timeo_, condition_) \
424({ \
bfd07f3d 425 DEFINE_WAIT_FUNC(wait_, woken_wake_function); \
844cf763
JPM		 426 struct sock *sk_; \
427 int rc_; \
428 \
429 while ((rc_ = !(condition_))) { \
bfd07f3d
TN		 430 /* coupled with smp_wmb() in tipc_sk_proto_rcv() */ \
431 smp_rmb(); \
844cf763
JPM		 432 sk_ = (sock_)->sk; \
433 rc_ = tipc_sk_sock_err((sock_), timeo_); \
434 if (rc_) \
435 break; \
223b7329 436 add_wait_queue(sk_sleep(sk_), &wait_); \
844cf763
JPM		 437 release_sock(sk_); \
438 *(timeo_) = wait_woken(&wait_, TASK_INTERRUPTIBLE, *(timeo_)); \
439 sched_annotate_sleep(); \
440 lock_sock(sk_); \
441 remove_wait_queue(sk_sleep(sk_), &wait_); \
442 } \
443 rc_; \
8c44e1af
JPM		 444})
445
b97bf3fd 446/**
c5fa7b3c 447 * tipc_sk_create - create a TIPC socket
0c3141e9 448 * @net: network namespace (must be default network)
b97bf3fd
PL		 449 * @sock: pre-allocated socket structure
450 * @protocol: protocol indicator (must be 0)
3f378b68 451 * @kern: caused by kernel or by userspace?
c4307285 452 *
0c3141e9
AS		 453 * This routine creates additional data structures used by the TIPC socket,
454 * initializes them, and links them together.
b97bf3fd 455 *
637b77fd 456 * Return: 0 on success, errno otherwise
b97bf3fd 457 */
58ed9442
JPM		 458static int tipc_sk_create(struct net *net, struct socket *sock,
459 int protocol, int kern)
b97bf3fd 460{
0c3141e9 461 const struct proto_ops *ops;
b97bf3fd 462 struct sock *sk;
58ed9442 463 struct tipc_sock *tsk;
5b8fa7ce 464 struct tipc_msg *msg;
0c3141e9
AS		 465
466 /* Validate arguments */
b97bf3fd
PL		 467 if (unlikely(protocol != 0))
468 return -EPROTONOSUPPORT;
469
b97bf3fd
PL		 470 switch (sock->type) {
471 case SOCK_STREAM:
0c3141e9 472 ops = &stream_ops;
b97bf3fd
PL		 473 break;
474 case SOCK_SEQPACKET:
0c3141e9 475 ops = &packet_ops;
b97bf3fd
PL		 476 break;
477 case SOCK_DGRAM:
b97bf3fd 478 case SOCK_RDM:
0c3141e9 479 ops = &msg_ops;
b97bf3fd 480 break;
49978651 481 default:
49978651 482 return -EPROTOTYPE;
b97bf3fd
PL		 483 }
484
0c3141e9 485 /* Allocate socket's protocol area */
11aa9c28 486 sk = sk_alloc(net, AF_TIPC, GFP_KERNEL, &tipc_proto, kern);
0c3141e9 487 if (sk == NULL)
b97bf3fd 488 return -ENOMEM;
b97bf3fd 489
58ed9442 490 tsk = tipc_sk(sk);
301bae56 491 tsk->max_pkt = MAX_PKT_DEFAULT;
c0bceb97 492 tsk->maxnagle = 0;
0a3e060f 493 tsk->nagle_start = NAGLE_START_INIT;
301bae56 494 INIT_LIST_HEAD(&tsk->publications);
365ad353 495 INIT_LIST_HEAD(&tsk->cong_links);
301bae56 496 msg = &tsk->phdr;
b97bf3fd 497
0c3141e9 498 /* Finish initializing socket data structures */
0c3141e9 499 sock->ops = ops;
0c3141e9 500 sock_init_data(sock, sk);
438adcaf 501 tipc_set_sk_state(sk, TIPC_OPEN);
07f6c4bc 502 if (tipc_sk_insert(tsk)) {
00aff359 503 sk_free(sk);
c19ca6cb 504 pr_warn("Socket create failed; port number exhausted\n");
07f6c4bc
YX		 505 return -EINVAL;
506 }
40f9f439
HX		 507
508 /* Ensure tsk is visible before we read own_addr. */
509 smp_mb();
510
23fd3eac
JM		 511 tipc_msg_init(tipc_own_addr(net), msg, TIPC_LOW_IMPORTANCE,
512 TIPC_NAMED_MSG, NAMED_H_SIZE, 0);
40f9f439 513
07f6c4bc 514 msg_set_origport(msg, tsk->portid);
31b102bb 515 timer_setup(&sk->sk_timer, tipc_sk_timeout, 0);
6f00089c 516 sk->sk_shutdown = 0;
64ac5f59 517 sk->sk_backlog_rcv = tipc_sk_backlog_rcv;
02739545 518 sk->sk_rcvbuf = READ_ONCE(sysctl_tipc_rmem[1]);
f288bef4
YX		 519 sk->sk_data_ready = tipc_data_ready;
520 sk->sk_write_space = tipc_write_space;
f4195d1e 521 sk->sk_destruct = tipc_sock_destruct;
4f4482dc 522 tsk->conn_timeout = CONN_TIMEOUT_DEFAULT;
1b22bcad 523 tsk->group_is_open = true;
4f4482dc 524 atomic_set(&tsk->dupl_rcvcnt, 0);
7ef43eba 525
10724cc7
JPM		 526 /* Start out with safe limits until we receive an advertised window */
527 tsk->snd_win = tsk_adv_blocks(RCVBUF_MIN);
528 tsk->rcv_win = tsk->snd_win;
529
c752023a 530 if (tipc_sk_type_connectionless(sk)) {
301bae56 531 tsk_set_unreturnable(tsk, true);
0c3141e9 532 if (sock->type == SOCK_DGRAM)
301bae56 533 tsk_set_unreliable(tsk, true);
0c3141e9 534 }
2948a1fc 535 __skb_queue_head_init(&tsk->mc_method.deferredq);
01e661eb 536 trace_tipc_sk_create(sk, NULL, TIPC_DUMP_NONE, " ");
b97bf3fd
PL		 537 return 0;
538}
539
07f6c4bc
YX		 540static void tipc_sk_callback(struct rcu_head *head)
541{
542 struct tipc_sock *tsk = container_of(head, struct tipc_sock, rcu);
543
544 sock_put(&tsk->sk);
545}
546
6f00089c
PB		 547/* Caller should hold socket lock for the socket. */
548static void __tipc_shutdown(struct socket *sock, int error)
549{
550 struct sock *sk = sock->sk;
551 struct tipc_sock *tsk = tipc_sk(sk);
552 struct net *net = sock_net(sk);
12db3c80 553 long timeout = msecs_to_jiffies(CONN_TIMEOUT_DEFAULT);
6f00089c
PB		 554 u32 dnode = tsk_peer_node(tsk);
555 struct sk_buff *skb;
556
365ad353
JPM		 557 /* Avoid that hi-prio shutdown msgs bypass msgs in link wakeup queue */
558 tipc_wait_for_cond(sock, &timeout, (!tsk->cong_link_cnt &&
559 !tsk_conn_cong(tsk)));
560
d34910e1 561 /* Push out delayed messages if in Nagle mode */
0a3e060f 562 tipc_sk_push_backlog(tsk, false);
d34910e1
TN		 563 /* Remove pending SYN */
564 __skb_queue_purge(&sk->sk_write_queue);
67879274 565
49afb806
TL		 566 /* Remove partially received buffer if any */
567 skb = skb_peek(&sk->sk_receive_queue);
568 if (skb && TIPC_SKB_CB(skb)->bytes_read) {
569 __skb_unlink(skb, &sk->sk_receive_queue);
570 kfree_skb(skb);
6f00089c 571 }
693c5649 572
49afb806
TL		 573 /* Reject all unreceived messages if connectionless */
574 if (tipc_sk_type_connectionless(sk)) {
575 tsk_rej_rx_queue(sk, error);
693c5649 576 return;
49afb806 577 }
693c5649 578
49afb806
TL		 579 switch (sk->sk_state) {
580 case TIPC_CONNECTING:
581 case TIPC_ESTABLISHED:
582 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
583 tipc_node_remove_conn(net, dnode, tsk->portid);
584 /* Send a FIN+/- to its peer */
585 skb = __skb_dequeue(&sk->sk_receive_queue);
586 if (skb) {
587 __skb_queue_purge(&sk->sk_receive_queue);
588 tipc_sk_respond(sk, skb, error);
589 break;
590 }
6f00089c
PB		 591 skb = tipc_msg_create(TIPC_CRITICAL_IMPORTANCE,
592 TIPC_CONN_MSG, SHORT_H_SIZE, 0, dnode,
593 tsk_own_node(tsk), tsk_peer_port(tsk),
594 tsk->portid, error);
595 if (skb)
596 tipc_node_xmit_skb(net, skb, dnode, tsk->portid);
49afb806
TL		 597 break;
598 case TIPC_LISTEN:
599 /* Reject all SYN messages */
600 tsk_rej_rx_queue(sk, error);
601 break;
602 default:
603 __skb_queue_purge(&sk->sk_receive_queue);
604 break;
6f00089c
PB		 605 }
606}
607
b97bf3fd 608/**
247f0f3c 609 * tipc_release - destroy a TIPC socket
b97bf3fd
PL		 610 * @sock: socket to destroy
611 *
612 * This routine cleans up any messages that are still queued on the socket.
613 * For DGRAM and RDM socket types, all queued messages are rejected.
614 * For SEQPACKET and STREAM socket types, the first message is rejected
615 * and any others are discarded. (If the first message on a STREAM socket
616 * is partially-read, it is discarded and the next one is rejected instead.)
c4307285 617 *
b97bf3fd
PL		 618 * NOTE: Rejected messages are not necessarily returned to the sender! They
619 * are returned or discarded according to the "destination droppable" setting
620 * specified for the message by the sender.
621 *
637b77fd 622 * Return: 0 on success, errno otherwise
b97bf3fd 623 */
247f0f3c 624static int tipc_release(struct socket *sock)
b97bf3fd 625{
b97bf3fd 626 struct sock *sk = sock->sk;
58ed9442 627 struct tipc_sock *tsk;
b97bf3fd 628
0c3141e9
AS		 629 /*
630 * Exit if socket isn't fully initialized (occurs when a failed accept()
631 * releases a pre-allocated child socket that was never used)
632 */
0c3141e9 633 if (sk == NULL)
b97bf3fd 634 return 0;
c4307285 635
58ed9442 636 tsk = tipc_sk(sk);
0c3141e9
AS		 637 lock_sock(sk);
638
01e661eb 639 trace_tipc_sk_release(sk, NULL, TIPC_DUMP_ALL, " ");
6f00089c
PB		 640 __tipc_shutdown(sock, TIPC_ERR_NO_PORT);
641 sk->sk_shutdown = SHUTDOWN_MASK;
75da2163 642 tipc_sk_leave(tsk);
2c98da07 643 tipc_sk_withdraw(tsk, NULL);
c55c8eda 644 __skb_queue_purge(&tsk->mc_method.deferredq);
1ea23a21 645 sk_stop_timer(sk, &sk->sk_timer);
07f6c4bc 646 tipc_sk_remove(tsk);
b97bf3fd 647
0a3b8b2b 648 sock_orphan(sk);
0c3141e9 649 /* Reject any messages that accumulated in backlog queue */
0c3141e9 650 release_sock(sk);
a80ae530 651 tipc_dest_list_purge(&tsk->cong_links);
365ad353 652 tsk->cong_link_cnt = 0;
07f6c4bc 653 call_rcu(&tsk->rcu, tipc_sk_callback);
0c3141e9 654 sock->sk = NULL;
b97bf3fd 655
065d7e39 656 return 0;
b97bf3fd
PL		 657}
658
659/**
60c102ee 660 * __tipc_bind - associate or disassocate TIPC name(s) with a socket
b97bf3fd 661 * @sock: socket structure
60c102ee
JM		 662 * @skaddr: socket address describing name(s) and desired operation
663 * @alen: size of socket address data structure
c4307285 664 *
326af505 665 * Name and name sequence binding are indicated using a positive scope value;
b97bf3fd
PL		 666 * a negative scope value unbinds the specified name. Specifying no name
667 * (i.e. a socket address length of 0) unbinds all names from the socket.
c4307285 668 *
637b77fd 669 * Return: 0 on success, errno otherwise
0c3141e9
AS		 670 *
671 * NOTE: This routine doesn't need to take the socket lock since it doesn't
672 * access any non-constant socket information.
b97bf3fd 673 */
60c102ee 674static int __tipc_bind(struct socket *sock, struct sockaddr *skaddr, int alen)
b97bf3fd 675{
50a3499a 676 struct tipc_uaddr *ua = (struct tipc_uaddr *)skaddr;
60c102ee 677 struct tipc_sock *tsk = tipc_sk(sock->sk);
50a3499a 678 bool unbind = false;
b97bf3fd 679
60c102ee 680 if (unlikely(!alen))
2c98da07 681 return tipc_sk_withdraw(tsk, NULL);
b97bf3fd 682
50a3499a
JM		 683 if (ua->addrtype == TIPC_SERVICE_ADDR) {
684 ua->addrtype = TIPC_SERVICE_RANGE;
685 ua->sr.upper = ua->sr.lower;
686 }
687 if (ua->scope < 0) {
688 unbind = true;
689 ua->scope = -ua->scope;
690 }
691 /* Users may still use deprecated TIPC_ZONE_SCOPE */
692 if (ua->scope != TIPC_NODE_SCOPE)
693 ua->scope = TIPC_CLUSTER_SCOPE;
c4307285 694
60c102ee
JM		 695 if (tsk->group)
696 return -EACCES;
697
50a3499a 698 if (unbind)
2c98da07 699 return tipc_sk_withdraw(tsk, ua);
50a3499a 700 return tipc_sk_publish(tsk, ua);
60c102ee
JM		 701}
702
703int tipc_sk_bind(struct socket *sock, struct sockaddr *skaddr, int alen)
704{
705 int res;
706
707 lock_sock(sock->sk);
708 res = __tipc_bind(sock, skaddr, alen);
709 release_sock(sock->sk);
84602761 710 return res;
b97bf3fd
PL		 711}
712
72671b35
JM		 713static int tipc_bind(struct socket *sock, struct sockaddr *skaddr, int alen)
714{
50a3499a
JM		 715 struct tipc_uaddr *ua = (struct tipc_uaddr *)skaddr;
716 u32 atype = ua->addrtype;
72671b35
JM		 717
718 if (alen) {
50a3499a 719 if (!tipc_uaddr_valid(ua, alen))
72671b35 720 return -EINVAL;
50a3499a 721 if (atype == TIPC_SOCKET_ADDR)
60c102ee 722 return -EAFNOSUPPORT;
50a3499a 723 if (ua->sr.type < TIPC_RESERVED_TYPES) {
72671b35 724 pr_warn_once("Can't bind to reserved service type %u\n",
50a3499a 725 ua->sr.type);
72671b35
JM		 726 return -EACCES;
727 }
728 }
729 return tipc_sk_bind(sock, skaddr, alen);
730}
731
c4307285 732/**
247f0f3c 733 * tipc_getname - get port ID of socket or peer socket
b97bf3fd
PL		 734 * @sock: socket structure
735 * @uaddr: area for returned socket address
2da59918 736 * @peer: 0 = own ID, 1 = current peer ID, 2 = current/former peer ID
c4307285 737 *
637b77fd 738 * Return: 0 on success, errno otherwise
0c3141e9 739 *
2da59918
AS		 740 * NOTE: This routine doesn't need to take the socket lock since it only
741 * accesses socket information that is unchanging (or which changes in
0e65967e 742 * a completely predictable manner).
b97bf3fd 743 */
247f0f3c 744static int tipc_getname(struct socket *sock, struct sockaddr *uaddr,
9b2c45d4 745 int peer)
b97bf3fd 746{
b97bf3fd 747 struct sockaddr_tipc *addr = (struct sockaddr_tipc *)uaddr;
9fd4b070
PB		 748 struct sock *sk = sock->sk;
749 struct tipc_sock *tsk = tipc_sk(sk);
b97bf3fd 750
88f8a5e3 751 memset(addr, 0, sizeof(*addr));
0c3141e9 752 if (peer) {
f40acbaf 753 if ((!tipc_sk_connected(sk)) &&
9fd4b070 754 ((peer != 2) || (sk->sk_state != TIPC_DISCONNECTING)))
2da59918 755 return -ENOTCONN;
301bae56
JPM		 756 addr->addr.id.ref = tsk_peer_port(tsk);
757 addr->addr.id.node = tsk_peer_node(tsk);
0c3141e9 758 } else {
07f6c4bc 759 addr->addr.id.ref = tsk->portid;
23fd3eac 760 addr->addr.id.node = tipc_own_addr(sock_net(sk));
0c3141e9 761 }
b97bf3fd 762
b6f88d9c 763 addr->addrtype = TIPC_SOCKET_ADDR;
b97bf3fd
PL		 764 addr->family = AF_TIPC;
765 addr->scope = 0;
b97bf3fd
PL		 766 addr->addr.name.domain = 0;
767
9b2c45d4 768 return sizeof(*addr);
b97bf3fd
PL		 769}
770
771/**
a11e1d43 772 * tipc_poll - read and possibly block on pollmask
b97bf3fd
PL		 773 * @file: file structure associated with the socket
774 * @sock: socket for which to calculate the poll bits
a11e1d43 775 * @wait: ???
b97bf3fd 776 *
637b77fd 777 * Return: pollmask value
9b674e82
AS		 778 *
779 * COMMENTARY:
780 * It appears that the usual socket locking mechanisms are not useful here
781 * since the pollmask info is potentially out-of-date the moment this routine
782 * exits. TCP and other protocols seem to rely on higher level poll routines
783 * to handle any preventable race conditions, so TIPC will do the same ...
784 *
f662c070
AS		 785 * IMPORTANT: The fact that a read or write operation is indicated does NOT
786 * imply that the operation will succeed, merely that it should be performed
787 * and will not block.
b97bf3fd 788 */
a11e1d43
LT		 789static __poll_t tipc_poll(struct file *file, struct socket *sock,
790 poll_table *wait)
b97bf3fd 791{
9b674e82 792 struct sock *sk = sock->sk;
58ed9442 793 struct tipc_sock *tsk = tipc_sk(sk);
ade994f4 794 __poll_t revents = 0;
9b674e82 795
89ab066d 796 sock_poll_wait(file, sock, wait);
01e661eb 797 trace_tipc_sk_poll(sk, NULL, TIPC_DUMP_ALL, " ");
a11e1d43 798
6f00089c 799 if (sk->sk_shutdown & RCV_SHUTDOWN)
a9a08845 800 revents |= EPOLLRDHUP | EPOLLIN | EPOLLRDNORM;
6f00089c 801 if (sk->sk_shutdown == SHUTDOWN_MASK)
a9a08845 802 revents |= EPOLLHUP;
6f00089c 803
f40acbaf
PB		 804 switch (sk->sk_state) {
805 case TIPC_ESTABLISHED:
365ad353 806 if (!tsk->cong_link_cnt && !tsk_conn_cong(tsk))
a9a08845 807 revents |= EPOLLOUT;
7f8901b7 808 fallthrough;
f40acbaf 809 case TIPC_LISTEN:
ff946833 810 case TIPC_CONNECTING:
3ef7cf57 811 if (!skb_queue_empty_lockless(&sk->sk_receive_queue))
a9a08845 812 revents |= EPOLLIN | EPOLLRDNORM;
f40acbaf
PB		 813 break;
814 case TIPC_OPEN:
60c25306 815 if (tsk->group_is_open && !tsk->cong_link_cnt)
a9a08845 816 revents |= EPOLLOUT;
ae236fb2
JM		 817 if (!tipc_sk_type_connectionless(sk))
818 break;
3ef7cf57 819 if (skb_queue_empty_lockless(&sk->sk_receive_queue))
ae236fb2 820 break;
a9a08845 821 revents |= EPOLLIN | EPOLLRDNORM;
f40acbaf
PB		 822 break;
823 case TIPC_DISCONNECTING:
a9a08845 824 revents = EPOLLIN | EPOLLRDNORM | EPOLLHUP;
f40acbaf 825 break;
f662c070 826 }
ae236fb2 827 return revents;
b97bf3fd
PL		 828}
829
0abd8ff2
JPM		 830/**
831 * tipc_sendmcast - send multicast message
832 * @sock: socket structure
833f8670 833 * @ua: destination address struct
562640f3 834 * @msg: message to send
365ad353
JPM		 835 * @dlen: length of data to send
836 * @timeout: timeout to wait for wakeup
0abd8ff2
JPM		 837 *
838 * Called from function tipc_sendmsg(), which has done all sanity checks
637b77fd 839 * Return: the number of bytes sent on success, or errno
0abd8ff2 840 */
833f8670 841static int tipc_sendmcast(struct socket *sock, struct tipc_uaddr *ua,
365ad353 842 struct msghdr *msg, size_t dlen, long timeout)
0abd8ff2
JPM		 843{
844 struct sock *sk = sock->sk;
c5898636 845 struct tipc_sock *tsk = tipc_sk(sk);
365ad353 846 struct tipc_msg *hdr = &tsk->phdr;
f2f9800d 847 struct net *net = sock_net(sk);
365ad353
JPM		 848 int mtu = tipc_bcast_get_mtu(net);
849 struct sk_buff_head pkts;
a853e4c6 850 struct tipc_nlist dsts;
0abd8ff2
JPM		 851 int rc;
852
75da2163
JM		 853 if (tsk->group)
854 return -EACCES;
855
a853e4c6 856 /* Block or return if any destination link is congested */
365ad353
JPM		 857 rc = tipc_wait_for_cond(sock, &timeout, !tsk->cong_link_cnt);
858 if (unlikely(rc))
859 return rc;
f214fc40 860
a853e4c6
JPM		 861 /* Lookup destination nodes */
862 tipc_nlist_init(&dsts, tipc_own_addr(net));
833f8670 863 tipc_nametbl_lookup_mcast_nodes(net, ua, &dsts);
a853e4c6
JPM		 864 if (!dsts.local && !dsts.remote)
865 return -EHOSTUNREACH;
866
867 /* Build message header */
365ad353 868 msg_set_type(hdr, TIPC_MCAST_MSG);
a853e4c6 869 msg_set_hdr_sz(hdr, MCAST_H_SIZE);
365ad353
JPM		 870 msg_set_lookup_scope(hdr, TIPC_CLUSTER_SCOPE);
871 msg_set_destport(hdr, 0);
872 msg_set_destnode(hdr, 0);
833f8670
JM		 873 msg_set_nametype(hdr, ua->sr.type);
874 msg_set_namelower(hdr, ua->sr.lower);
875 msg_set_nameupper(hdr, ua->sr.upper);
365ad353 876
a853e4c6 877 /* Build message as chain of buffers */
e654f9f5 878 __skb_queue_head_init(&pkts);
365ad353 879 rc = tipc_msg_build(hdr, msg, 0, dlen, mtu, &pkts);
0abd8ff2 880
a853e4c6 881 /* Send message if build was successful */
01e661eb
TL		 882 if (unlikely(rc == dlen)) {
883 trace_tipc_sk_sendmcast(sk, skb_peek(&pkts),
884 TIPC_DUMP_SK_SNDQ, " ");
833f8670 885 rc = tipc_mcast_xmit(net, &pkts, &tsk->mc_method, &dsts,
a853e4c6 886 &tsk->cong_link_cnt);
01e661eb 887 }
a853e4c6
JPM		 888
889 tipc_nlist_purge(&dsts);
365ad353
JPM		 890
891 return rc ? rc : dlen;
0abd8ff2
JPM		 892}
893
27bd9ec0
JM		 894/**
895 * tipc_send_group_msg - send a message to a member in the group
896 * @net: network namespace
f172f4b8 897 * @tsk: tipc socket
27bd9ec0
JM		 898 * @m: message to send
899 * @mb: group member
900 * @dnode: destination node
901 * @dport: destination port
902 * @dlen: total length of message data
903 */
904static int tipc_send_group_msg(struct net *net, struct tipc_sock *tsk,
905 struct msghdr *m, struct tipc_member *mb,
906 u32 dnode, u32 dport, int dlen)
907{
b87a5ea3 908 u16 bc_snd_nxt = tipc_group_bc_snd_nxt(tsk->group);
2f487712 909 struct tipc_mc_method *method = &tsk->mc_method;
27bd9ec0
JM		 910 int blks = tsk_blocks(GROUP_H_SIZE + dlen);
911 struct tipc_msg *hdr = &tsk->phdr;
912 struct sk_buff_head pkts;
913 int mtu, rc;
914
915 /* Complete message header */
916 msg_set_type(hdr, TIPC_GRP_UCAST_MSG);
917 msg_set_hdr_sz(hdr, GROUP_H_SIZE);
918 msg_set_destport(hdr, dport);
919 msg_set_destnode(hdr, dnode);
b87a5ea3 920 msg_set_grp_bc_seqno(hdr, bc_snd_nxt);
27bd9ec0
JM		 921
922 /* Build message as chain of buffers */
e654f9f5 923 __skb_queue_head_init(&pkts);
f73b1281 924 mtu = tipc_node_get_mtu(net, dnode, tsk->portid, false);
27bd9ec0
JM		 925 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
926 if (unlikely(rc != dlen))
927 return rc;
928
929 /* Send message */
930 rc = tipc_node_xmit(net, &pkts, dnode, tsk->portid);
931 if (unlikely(rc == -ELINKCONG)) {
932 tipc_dest_push(&tsk->cong_links, dnode, 0);
933 tsk->cong_link_cnt++;
934 }
935
2f487712 936 /* Update send window */
27bd9ec0
JM		 937 tipc_group_update_member(mb, blks);
938
2f487712
JM		 939 /* A broadcast sent within next EXPIRE period must follow same path */
940 method->rcast = true;
941 method->mandatory = true;
27bd9ec0
JM		 942 return dlen;
943}
944
945/**
946 * tipc_send_group_unicast - send message to a member in the group
947 * @sock: socket structure
948 * @m: message to send
949 * @dlen: total length of message data
950 * @timeout: timeout to wait for wakeup
951 *
952 * Called from function tipc_sendmsg(), which has done all sanity checks
637b77fd 953 * Return: the number of bytes sent on success, or errno
27bd9ec0
JM		 954 */
955static int tipc_send_group_unicast(struct socket *sock, struct msghdr *m,
956 int dlen, long timeout)
957{
958 struct sock *sk = sock->sk;
006ed14e 959 struct tipc_uaddr *ua = (struct tipc_uaddr *)m->msg_name;
27bd9ec0
JM		 960 int blks = tsk_blocks(GROUP_H_SIZE + dlen);
961 struct tipc_sock *tsk = tipc_sk(sk);
27bd9ec0
JM		 962 struct net *net = sock_net(sk);
963 struct tipc_member *mb = NULL;
964 u32 node, port;
965 int rc;
966
006ed14e
JM		 967 node = ua->sk.node;
968 port = ua->sk.ref;
27bd9ec0
JM		 969 if (!port && !node)
970 return -EHOSTUNREACH;
971
972 /* Block or return if destination link or member is congested */
973 rc = tipc_wait_for_cond(sock, &timeout,
974 !tipc_dest_find(&tsk->cong_links, node, 0) &&
143ece65
CW		 975 tsk->group &&
976 !tipc_group_cong(tsk->group, node, port, blks,
977 &mb));
27bd9ec0
JM		 978 if (unlikely(rc))
979 return rc;
980
981 if (unlikely(!mb))
982 return -EHOSTUNREACH;
983
984 rc = tipc_send_group_msg(net, tsk, m, mb, node, port, dlen);
985
986 return rc ? rc : dlen;
987}
988
ee106d7f
JM		 989/**
990 * tipc_send_group_anycast - send message to any member with given identity
991 * @sock: socket structure
992 * @m: message to send
993 * @dlen: total length of message data
994 * @timeout: timeout to wait for wakeup
995 *
996 * Called from function tipc_sendmsg(), which has done all sanity checks
637b77fd 997 * Return: the number of bytes sent on success, or errno
ee106d7f
JM		 998 */
999static int tipc_send_group_anycast(struct socket *sock, struct msghdr *m,
1000 int dlen, long timeout)
1001{
006ed14e 1002 struct tipc_uaddr *ua = (struct tipc_uaddr *)m->msg_name;
ee106d7f
JM		 1003 struct sock *sk = sock->sk;
1004 struct tipc_sock *tsk = tipc_sk(sk);
1005 struct list_head *cong_links = &tsk->cong_links;
1006 int blks = tsk_blocks(GROUP_H_SIZE + dlen);
232d07b7 1007 struct tipc_msg *hdr = &tsk->phdr;
ee106d7f
JM		 1008 struct tipc_member *first = NULL;
1009 struct tipc_member *mbr = NULL;
1010 struct net *net = sock_net(sk);
1011 u32 node, port, exclude;
ee106d7f
JM		 1012 struct list_head dsts;
1013 int lookups = 0;
1014 int dstcnt, rc;
1015 bool cong;
1016
1017 INIT_LIST_HEAD(&dsts);
006ed14e
JM		 1018 ua->sa.type = msg_nametype(hdr);
1019 ua->scope = msg_lookup_scope(hdr);
ee106d7f
JM		 1020
1021 while (++lookups < 4) {
143ece65
CW		 1022 exclude = tipc_group_exclude(tsk->group);
1023
ee106d7f
JM		 1024 first = NULL;
1025
1026 /* Look for a non-congested destination member, if any */
1027 while (1) {
006ed14e
JM		 1028 if (!tipc_nametbl_lookup_group(net, ua, &dsts, &dstcnt,
1029 exclude, false))
ee106d7f
JM		 1030 return -EHOSTUNREACH;
1031 tipc_dest_pop(&dsts, &node, &port);
143ece65
CW		 1032 cong = tipc_group_cong(tsk->group, node, port, blks,
1033 &mbr);
ee106d7f
JM		 1034 if (!cong)
1035 break;
1036 if (mbr == first)
1037 break;
1038 if (!first)
1039 first = mbr;
1040 }
1041
1042 /* Start over if destination was not in member list */
1043 if (unlikely(!mbr))
1044 continue;
1045
1046 if (likely(!cong && !tipc_dest_find(cong_links, node, 0)))
1047 break;
1048
1049 /* Block or return if destination link or member is congested */
1050 rc = tipc_wait_for_cond(sock, &timeout,
1051 !tipc_dest_find(cong_links, node, 0) &&
143ece65
CW		 1052 tsk->group &&
1053 !tipc_group_cong(tsk->group, node, port,
ee106d7f
JM		 1054 blks, &mbr));
1055 if (unlikely(rc))
1056 return rc;
1057
1058 /* Send, unless destination disappeared while waiting */
1059 if (likely(mbr))
1060 break;
1061 }
1062
1063 if (unlikely(lookups >= 4))
1064 return -EHOSTUNREACH;
1065
1066 rc = tipc_send_group_msg(net, tsk, m, mbr, node, port, dlen);
1067
1068 return rc ? rc : dlen;
1069}
1070
75da2163
JM		 1071/**
1072 * tipc_send_group_bcast - send message to all members in communication group
d8141208 1073 * @sock: socket structure
75da2163
JM		 1074 * @m: message to send
1075 * @dlen: total length of message data
1076 * @timeout: timeout to wait for wakeup
1077 *
1078 * Called from function tipc_sendmsg(), which has done all sanity checks
637b77fd 1079 * Return: the number of bytes sent on success, or errno
75da2163
JM		 1080 */
1081static int tipc_send_group_bcast(struct socket *sock, struct msghdr *m,
1082 int dlen, long timeout)
1083{
006ed14e 1084 struct tipc_uaddr *ua = (struct tipc_uaddr *)m->msg_name;
75da2163
JM		 1085 struct sock *sk = sock->sk;
1086 struct net *net = sock_net(sk);
1087 struct tipc_sock *tsk = tipc_sk(sk);
3c6306d4 1088 struct tipc_nlist *dsts;
75da2163 1089 struct tipc_mc_method *method = &tsk->mc_method;
2f487712 1090 bool ack = method->mandatory && method->rcast;
b7d42635 1091 int blks = tsk_blocks(MCAST_H_SIZE + dlen);
75da2163
JM		 1092 struct tipc_msg *hdr = &tsk->phdr;
1093 int mtu = tipc_bcast_get_mtu(net);
1094 struct sk_buff_head pkts;
1095 int rc = -EHOSTUNREACH;
1096
b7d42635 1097 /* Block or return if any destination link or member is congested */
143ece65
CW		 1098 rc = tipc_wait_for_cond(sock, &timeout,
1099 !tsk->cong_link_cnt && tsk->group &&
1100 !tipc_group_bc_cong(tsk->group, blks));
75da2163
JM		 1101 if (unlikely(rc))
1102 return rc;
1103
3c6306d4
CW		 1104 dsts = tipc_group_dests(tsk->group);
1105 if (!dsts->local && !dsts->remote)
1106 return -EHOSTUNREACH;
1107
75da2163 1108 /* Complete message header */
006ed14e 1109 if (ua) {
5b8dddb6 1110 msg_set_type(hdr, TIPC_GRP_MCAST_MSG);
006ed14e 1111 msg_set_nameinst(hdr, ua->sa.instance);
5b8dddb6
JM		 1112 } else {
1113 msg_set_type(hdr, TIPC_GRP_BCAST_MSG);
1114 msg_set_nameinst(hdr, 0);
1115 }
b7d42635 1116 msg_set_hdr_sz(hdr, GROUP_H_SIZE);
75da2163
JM		 1117 msg_set_destport(hdr, 0);
1118 msg_set_destnode(hdr, 0);
143ece65 1119 msg_set_grp_bc_seqno(hdr, tipc_group_bc_snd_nxt(tsk->group));
75da2163 1120
2f487712
JM		 1121 /* Avoid getting stuck with repeated forced replicasts */
1122 msg_set_grp_bc_ack_req(hdr, ack);
1123
75da2163 1124 /* Build message as chain of buffers */
e654f9f5 1125 __skb_queue_head_init(&pkts);
75da2163
JM		 1126 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1127 if (unlikely(rc != dlen))
1128 return rc;
1129
1130 /* Send message */
2f487712 1131 rc = tipc_mcast_xmit(net, &pkts, method, dsts, &tsk->cong_link_cnt);
75da2163
JM		 1132 if (unlikely(rc))
1133 return rc;
1134
b7d42635 1135 /* Update broadcast sequence number and send windows */
2f487712
JM		 1136 tipc_group_update_bc_members(tsk->group, blks, ack);
1137
1138 /* Broadcast link is now free to choose method for next broadcast */
1139 method->mandatory = false;
1140 method->expires = jiffies;
1141
75da2163
JM		 1142 return dlen;
1143}
1144
5b8dddb6
JM		 1145/**
1146 * tipc_send_group_mcast - send message to all members with given identity
1147 * @sock: socket structure
1148 * @m: message to send
1149 * @dlen: total length of message data
1150 * @timeout: timeout to wait for wakeup
1151 *
1152 * Called from function tipc_sendmsg(), which has done all sanity checks
637b77fd 1153 * Return: the number of bytes sent on success, or errno
5b8dddb6
JM		 1154 */
1155static int tipc_send_group_mcast(struct socket *sock, struct msghdr *m,
1156 int dlen, long timeout)
1157{
006ed14e 1158 struct tipc_uaddr *ua = (struct tipc_uaddr *)m->msg_name;
5b8dddb6 1159 struct sock *sk = sock->sk;
5b8dddb6
JM		 1160 struct tipc_sock *tsk = tipc_sk(sk);
1161 struct tipc_group *grp = tsk->group;
232d07b7 1162 struct tipc_msg *hdr = &tsk->phdr;
5b8dddb6 1163 struct net *net = sock_net(sk);
5b8dddb6 1164 struct list_head dsts;
006ed14e 1165 u32 dstcnt, exclude;
5b8dddb6
JM		 1166
1167 INIT_LIST_HEAD(&dsts);
006ed14e
JM		 1168 ua->sa.type = msg_nametype(hdr);
1169 ua->scope = msg_lookup_scope(hdr);
5b8dddb6 1170 exclude = tipc_group_exclude(grp);
232d07b7 1171
006ed14e 1172 if (!tipc_nametbl_lookup_group(net, ua, &dsts, &dstcnt, exclude, true))
5b8dddb6
JM		 1173 return -EHOSTUNREACH;
1174
1175 if (dstcnt == 1) {
006ed14e 1176 tipc_dest_pop(&dsts, &ua->sk.node, &ua->sk.ref);
5b8dddb6
JM		 1177 return tipc_send_group_unicast(sock, m, dlen, timeout);
1178 }
1179
1180 tipc_dest_list_purge(&dsts);
1181 return tipc_send_group_bcast(sock, m, dlen, timeout);
1182}
1183
cb1b7280
JPM		 1184/**
1185 * tipc_sk_mcast_rcv - Deliver multicast messages to all destination sockets
f172f4b8 1186 * @net: the associated network namespace
cb1b7280
JPM		 1187 * @arrvq: queue with arriving messages, to be cloned after destination lookup
1188 * @inputq: queue with cloned messages, delivered to socket after dest lookup
1189 *
1190 * Multi-threaded: parallel calls with reference to same queues may occur
078bec82 1191 */
cb1b7280
JPM		 1192void tipc_sk_mcast_rcv(struct net *net, struct sk_buff_head *arrvq,
1193 struct sk_buff_head *inputq)
078bec82 1194{
75da2163 1195 u32 self = tipc_own_addr(net);
cb1b7280 1196 struct sk_buff *skb, *_skb;
b053fcc4 1197 u32 portid, onode;
232d07b7 1198 struct sk_buff_head tmpq;
75da2163 1199 struct list_head dports;
232d07b7 1200 struct tipc_msg *hdr;
45ceea2d 1201 struct tipc_uaddr ua;
232d07b7 1202 int user, mtyp, hlen;
3c724acd 1203
cb1b7280 1204 __skb_queue_head_init(&tmpq);
4d8642d8 1205 INIT_LIST_HEAD(&dports);
45ceea2d 1206 ua.addrtype = TIPC_SERVICE_RANGE;
078bec82 1207
5ef21325 1208 /* tipc_skb_peek() increments the head skb's reference counter */
cb1b7280
JPM		 1209 skb = tipc_skb_peek(arrvq, &inputq->lock);
1210 for (; skb; skb = tipc_skb_peek(arrvq, &inputq->lock)) {
232d07b7
JM		 1211 hdr = buf_msg(skb);
1212 user = msg_user(hdr);
1213 mtyp = msg_type(hdr);
1214 hlen = skb_headroom(skb) + msg_hdr_sz(hdr);
232d07b7 1215 onode = msg_orignode(hdr);
45ceea2d 1216 ua.sr.type = msg_nametype(hdr);
5ef21325
JM		 1217 ua.sr.lower = msg_namelower(hdr);
1218 ua.sr.upper = msg_nameupper(hdr);
1219 if (onode == self)
1220 ua.scope = TIPC_ANY_SCOPE;
1221 else
1222 ua.scope = TIPC_CLUSTER_SCOPE;
232d07b7 1223
2f487712
JM		 1224 if (mtyp == TIPC_GRP_UCAST_MSG || user == GROUP_PROTOCOL) {
1225 spin_lock_bh(&inputq->lock);
1226 if (skb_peek(arrvq) == skb) {
1227 __skb_dequeue(arrvq);
1228 __skb_queue_tail(inputq, skb);
1229 }
c545a945 1230 kfree_skb(skb);
2f487712
JM		 1231 spin_unlock_bh(&inputq->lock);
1232 continue;
1233 }
232d07b7
JM		 1234
1235 /* Group messages require exact scope match */
1236 if (msg_in_group(hdr)) {
45ceea2d
JM		 1237 ua.sr.lower = 0;
1238 ua.sr.upper = ~0;
1239 ua.scope = msg_lookup_scope(hdr);
75da2163 1240 }
232d07b7
JM		 1241
1242 /* Create destination port list: */
5ef21325 1243 tipc_nametbl_lookup_mcast_sockets(net, &ua, &dports);
232d07b7
JM		 1244
1245 /* Clone message per destination */
a80ae530 1246 while (tipc_dest_pop(&dports, NULL, &portid)) {
232d07b7 1247 _skb = __pskb_copy(skb, hlen, GFP_ATOMIC);
cb1b7280
JPM		 1248 if (_skb) {
1249 msg_set_destport(buf_msg(_skb), portid);
1250 __skb_queue_tail(&tmpq, _skb);
1251 continue;
1252 }
1253 pr_warn("Failed to clone mcast rcv buffer\n");
078bec82 1254 }
5ef21325 1255 /* Append clones to inputq only if skb is still head of arrvq */
cb1b7280
JPM		 1256 spin_lock_bh(&inputq->lock);
1257 if (skb_peek(arrvq) == skb) {
1258 skb_queue_splice_tail_init(&tmpq, inputq);
5ef21325 1259 /* Decrement the skb's refcnt */
75016891 1260 kfree_skb(__skb_dequeue(arrvq));
cb1b7280
JPM		 1261 }
1262 spin_unlock_bh(&inputq->lock);
1263 __skb_queue_purge(&tmpq);
1264 kfree_skb(skb);
078bec82 1265 }
cb1b7280 1266 tipc_sk_rcv(net, inputq);
078bec82
JPM		 1267}
1268
c0bceb97
JM		 1269/* tipc_sk_push_backlog(): send accumulated buffers in socket write queue
1270 * when socket is in Nagle mode
1271 */
0a3e060f 1272static void tipc_sk_push_backlog(struct tipc_sock *tsk, bool nagle_ack)
c0bceb97
JM		 1273{
1274 struct sk_buff_head *txq = &tsk->sk.sk_write_queue;
0a3e060f 1275 struct sk_buff *skb = skb_peek_tail(txq);
c0bceb97
JM		 1276 struct net *net = sock_net(&tsk->sk);
1277 u32 dnode = tsk_peer_node(tsk);
1278 int rc;
1279
0a3e060f
TL		 1280 if (nagle_ack) {
1281 tsk->pkt_cnt += skb_queue_len(txq);
1282 if (!tsk->pkt_cnt || tsk->msg_acc / tsk->pkt_cnt < 2) {
1283 tsk->oneway = 0;
1284 if (tsk->nagle_start < NAGLE_START_MAX)
1285 tsk->nagle_start *= 2;
1286 tsk->expect_ack = false;
1287 pr_debug("tsk %10u: bad nagle %u -> %u, next start %u!\n",
1288 tsk->portid, tsk->msg_acc, tsk->pkt_cnt,
1289 tsk->nagle_start);
1290 } else {
1291 tsk->nagle_start = NAGLE_START_INIT;
1292 if (skb) {
1293 msg_set_ack_required(buf_msg(skb));
1294 tsk->expect_ack = true;
1295 } else {
1296 tsk->expect_ack = false;
1297 }
1298 }
1299 tsk->msg_acc = 0;
1300 tsk->pkt_cnt = 0;
1301 }
1302
d34910e1
TN		 1303 if (!skb || tsk->cong_link_cnt)
1304 return;
1305
1306 /* Do not send SYN again after congestion */
1307 if (msg_is_syn(buf_msg(skb)))
c0bceb97
JM		 1308 return;
1309
0a3e060f
TL		 1310 if (tsk->msg_acc)
1311 tsk->pkt_cnt += skb_queue_len(txq);
c0bceb97
JM		 1312 tsk->snt_unacked += tsk->snd_backlog;
1313 tsk->snd_backlog = 0;
c0bceb97
JM		 1314 rc = tipc_node_xmit(net, txq, dnode, tsk->portid);
1315 if (rc == -ELINKCONG)
1316 tsk->cong_link_cnt = 1;
1317}
1318
ac0074ee 1319/**
64ac5f59 1320 * tipc_sk_conn_proto_rcv - receive a connection mng protocol message
ac0074ee 1321 * @tsk: receiving socket
bcd3ffd4 1322 * @skb: pointer to message buffer.
f172f4b8
RD		 1323 * @inputq: buffer list containing the buffers
1324 * @xmitq: output message area
ac0074ee 1325 */
64ac5f59 1326static void tipc_sk_conn_proto_rcv(struct tipc_sock *tsk, struct sk_buff *skb,
e7eb0582 1327 struct sk_buff_head *inputq,
64ac5f59 1328 struct sk_buff_head *xmitq)
ac0074ee 1329{
bcd3ffd4 1330 struct tipc_msg *hdr = buf_msg(skb);
64ac5f59
JM		 1331 u32 onode = tsk_own_node(tsk);
1332 struct sock *sk = &tsk->sk;
bcd3ffd4 1333 int mtyp = msg_type(hdr);
c0bceb97 1334 bool was_cong;
bcd3ffd4 1335
ac0074ee 1336 /* Ignore if connection cannot be validated: */
01e661eb
TL		 1337 if (!tsk_peer_msg(tsk, hdr)) {
1338 trace_tipc_sk_drop_msg(sk, skb, TIPC_DUMP_NONE, "@proto_rcv!");
ac0074ee 1339 goto exit;
01e661eb 1340 }
ac0074ee 1341
c1be7756
PB		 1342 if (unlikely(msg_errcode(hdr))) {
1343 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
1344 tipc_node_remove_conn(sock_net(sk), tsk_peer_node(tsk),
1345 tsk_peer_port(tsk));
1346 sk->sk_state_change(sk);
e7eb0582
PB		 1347
1348 /* State change is ignored if socket already awake,
1349 * - convert msg to abort msg and add to inqueue
1350 */
1351 msg_set_user(hdr, TIPC_CRITICAL_IMPORTANCE);
1352 msg_set_type(hdr, TIPC_CONN_MSG);
1353 msg_set_size(hdr, BASIC_H_SIZE);
1354 msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1355 __skb_queue_tail(inputq, skb);
1356 return;
c1be7756
PB		 1357 }
1358
8ea642ee 1359 tsk->probe_unacked = false;
ac0074ee 1360
bcd3ffd4
JPM		 1361 if (mtyp == CONN_PROBE) {
1362 msg_set_type(hdr, CONN_PROBE_REPLY);
f1d048f2
JPM		 1363 if (tipc_msg_reverse(onode, &skb, TIPC_OK))
1364 __skb_queue_tail(xmitq, skb);
bcd3ffd4
JPM		 1365 return;
1366 } else if (mtyp == CONN_ACK) {
c0bceb97 1367 was_cong = tsk_conn_cong(tsk);
0a3e060f 1368 tipc_sk_push_backlog(tsk, msg_nagle_ack(hdr));
10724cc7
JPM		 1369 tsk->snt_unacked -= msg_conn_ack(hdr);
1370 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1371 tsk->snd_win = msg_adv_win(hdr);
c0bceb97 1372 if (was_cong && !tsk_conn_cong(tsk))
bcd3ffd4
JPM		 1373 sk->sk_write_space(sk);
1374 } else if (mtyp != CONN_PROBE_REPLY) {
1375 pr_warn("Received unknown CONN_PROTO msg\n");
ac0074ee 1376 }
ac0074ee 1377exit:
bcd3ffd4 1378 kfree_skb(skb);
ac0074ee
JPM		 1379}
1380
b97bf3fd 1381/**
247f0f3c 1382 * tipc_sendmsg - send message in connectionless manner
b97bf3fd
PL		 1383 * @sock: socket structure
1384 * @m: message to send
e2dafe87 1385 * @dsz: amount of user data to be sent
c4307285 1386 *
b97bf3fd 1387 * Message must have an destination specified explicitly.
c4307285 1388 * Used for SOCK_RDM and SOCK_DGRAM messages,
b97bf3fd
PL		 1389 * and for 'SYN' messages on SOCK_SEQPACKET and SOCK_STREAM connections.
1390 * (Note: 'SYN+' is prohibited on SOCK_STREAM.)
c4307285 1391 *
637b77fd 1392 * Return: the number of bytes sent on success, or errno otherwise
b97bf3fd 1393 */
1b784140 1394static int tipc_sendmsg(struct socket *sock,
e2dafe87 1395 struct msghdr *m, size_t dsz)
39a0295f
YX		 1396{
1397 struct sock *sk = sock->sk;
1398 int ret;
1399
1400 lock_sock(sk);
1401 ret = __tipc_sendmsg(sock, m, dsz);
1402 release_sock(sk);
1403
1404 return ret;
1405}
1406
365ad353 1407static int __tipc_sendmsg(struct socket *sock, struct msghdr *m, size_t dlen)
b97bf3fd 1408{
0c3141e9 1409 struct sock *sk = sock->sk;
f2f9800d 1410 struct net *net = sock_net(sk);
365ad353 1411 struct tipc_sock *tsk = tipc_sk(sk);
908148bc 1412 struct tipc_uaddr *ua = (struct tipc_uaddr *)m->msg_name;
365ad353
JPM		 1413 long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
1414 struct list_head *clinks = &tsk->cong_links;
1415 bool syn = !tipc_sk_type_connectionless(sk);
75da2163 1416 struct tipc_group *grp = tsk->group;
365ad353 1417 struct tipc_msg *hdr = &tsk->phdr;
908148bc 1418 struct tipc_socket_addr skaddr;
365ad353 1419 struct sk_buff_head pkts;
908148bc 1420 int atype, mtu, rc;
b97bf3fd 1421
365ad353 1422 if (unlikely(dlen > TIPC_MAX_USER_MSG_SIZE))
c29c3f70 1423 return -EMSGSIZE;
365ad353 1424
908148bc
JM		 1425 if (ua) {
1426 if (!tipc_uaddr_valid(ua, m->msg_namelen))
27bd9ec0 1427 return -EINVAL;
743902c5 1428 atype = ua->addrtype;
27bd9ec0
JM		 1429 }
1430
908148bc 1431 /* If socket belongs to a communication group follow other paths */
27bd9ec0 1432 if (grp) {
908148bc 1433 if (!ua)
27bd9ec0 1434 return tipc_send_group_bcast(sock, m, dlen, timeout);
908148bc 1435 if (atype == TIPC_SERVICE_ADDR)
ee106d7f 1436 return tipc_send_group_anycast(sock, m, dlen, timeout);
908148bc 1437 if (atype == TIPC_SOCKET_ADDR)
27bd9ec0 1438 return tipc_send_group_unicast(sock, m, dlen, timeout);
908148bc 1439 if (atype == TIPC_SERVICE_RANGE)
5b8dddb6 1440 return tipc_send_group_mcast(sock, m, dlen, timeout);
27bd9ec0
JM		 1441 return -EINVAL;
1442 }
75da2163 1443
908148bc
JM		 1444 if (!ua) {
1445 ua = (struct tipc_uaddr *)&tsk->peer;
1446 if (!syn && ua->family != AF_TIPC)
f2f8036e 1447 return -EDESTADDRREQ;
02fdc14d 1448 atype = ua->addrtype;
f2f8036e 1449 }
365ad353 1450
365ad353 1451 if (unlikely(syn)) {
0c288c86 1452 if (sk->sk_state == TIPC_LISTEN)
39a0295f 1453 return -EPIPE;
438adcaf 1454 if (sk->sk_state != TIPC_OPEN)
39a0295f
YX		 1455 return -EISCONN;
1456 if (tsk->published)
1457 return -EOPNOTSUPP;
14623e00
JM		 1458 if (atype == TIPC_SERVICE_ADDR)
1459 tsk->conn_addrtype = atype;
25b9221b 1460 msg_set_syn(hdr, 1);
b97bf3fd 1461 }
e2dafe87 1462
d6d86830
HZ		 1463 memset(&skaddr, 0, sizeof(skaddr));
1464
908148bc
JM		 1465 /* Determine destination */
1466 if (atype == TIPC_SERVICE_RANGE) {
833f8670 1467 return tipc_sendmcast(sock, ua, m, dlen, timeout);
908148bc
JM		 1468 } else if (atype == TIPC_SERVICE_ADDR) {
1469 skaddr.node = ua->lookup_node;
1470 ua->scope = tipc_node2scope(skaddr.node);
1471 if (!tipc_nametbl_lookup_anycast(net, ua, &skaddr))
39a0295f 1472 return -EHOSTUNREACH;
908148bc
JM		 1473 } else if (atype == TIPC_SOCKET_ADDR) {
1474 skaddr = ua->sk;
335b929b
JM		 1475 } else {
1476 return -EINVAL;
e2dafe87
JPM		 1477 }
1478
365ad353 1479 /* Block or return if destination link is congested */
a80ae530 1480 rc = tipc_wait_for_cond(sock, &timeout,
908148bc 1481 !tipc_dest_find(clinks, skaddr.node, 0));
365ad353
JPM		 1482 if (unlikely(rc))
1483 return rc;
1484
908148bc
JM		 1485 /* Finally build message header */
1486 msg_set_destnode(hdr, skaddr.node);
1487 msg_set_destport(hdr, skaddr.ref);
1488 if (atype == TIPC_SERVICE_ADDR) {
abc9b4e0
TL		 1489 msg_set_type(hdr, TIPC_NAMED_MSG);
1490 msg_set_hdr_sz(hdr, NAMED_H_SIZE);
908148bc
JM		 1491 msg_set_nametype(hdr, ua->sa.type);
1492 msg_set_nameinst(hdr, ua->sa.instance);
1493 msg_set_lookup_scope(hdr, ua->scope);
b6f88d9c 1494 } else { /* TIPC_SOCKET_ADDR */
abc9b4e0
TL		 1495 msg_set_type(hdr, TIPC_DIRECT_MSG);
1496 msg_set_lookup_scope(hdr, 0);
abc9b4e0
TL		 1497 msg_set_hdr_sz(hdr, BASIC_H_SIZE);
1498 }
1499
908148bc 1500 /* Add message body */
e654f9f5 1501 __skb_queue_head_init(&pkts);
908148bc 1502 mtu = tipc_node_get_mtu(net, skaddr.node, tsk->portid, true);
365ad353
JPM		 1503 rc = tipc_msg_build(hdr, m, 0, dlen, mtu, &pkts);
1504 if (unlikely(rc != dlen))
39a0295f 1505 return rc;
2fe97a57
TN		 1506 if (unlikely(syn && !tipc_msg_skb_clone(&pkts, &sk->sk_write_queue))) {
1507 __skb_queue_purge(&pkts);
67879274 1508 return -ENOMEM;
2fe97a57 1509 }
e2dafe87 1510
908148bc 1511 /* Send message */
01e661eb 1512 trace_tipc_sk_sendmsg(sk, skb_peek(&pkts), TIPC_DUMP_SK_SNDQ, " ");
908148bc 1513 rc = tipc_node_xmit(net, &pkts, skaddr.node, tsk->portid);
365ad353 1514 if (unlikely(rc == -ELINKCONG)) {
908148bc 1515 tipc_dest_push(clinks, skaddr.node, 0);
365ad353
JPM		 1516 tsk->cong_link_cnt++;
1517 rc = 0;
1518 }
e2dafe87 1519
f8dd60de 1520 if (unlikely(syn && !rc)) {
365ad353 1521 tipc_set_sk_state(sk, TIPC_CONNECTING);
7387a72c 1522 if (dlen && timeout) {
f8dd60de
XL		 1523 timeout = msecs_to_jiffies(timeout);
1524 tipc_wait_for_connect(sock, &timeout);
1525 }
1526 }
365ad353
JPM		 1527
1528 return rc ? rc : dlen;
b97bf3fd
PL		 1529}
1530
c4307285 1531/**
365ad353 1532 * tipc_sendstream - send stream-oriented data
b97bf3fd 1533 * @sock: socket structure
4ccfe5e0
JPM		 1534 * @m: data to send
1535 * @dsz: total length of data to be transmitted
c4307285 1536 *
4ccfe5e0 1537 * Used for SOCK_STREAM data.
c4307285 1538 *
637b77fd 1539 * Return: the number of bytes sent on success (or partial success),
4ccfe5e0 1540 * or errno if no data sent
b97bf3fd 1541 */
365ad353 1542static int tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dsz)
39a0295f
YX		 1543{
1544 struct sock *sk = sock->sk;
1545 int ret;
1546
1547 lock_sock(sk);
365ad353 1548 ret = __tipc_sendstream(sock, m, dsz);
39a0295f
YX		 1549 release_sock(sk);
1550
1551 return ret;
1552}
1553
365ad353 1554static int __tipc_sendstream(struct socket *sock, struct msghdr *m, size_t dlen)
b97bf3fd 1555{
0c3141e9 1556 struct sock *sk = sock->sk;
342dfc30 1557 DECLARE_SOCKADDR(struct sockaddr_tipc *, dest, m->msg_name);
365ad353 1558 long timeout = sock_sndtimeo(sk, m->msg_flags & MSG_DONTWAIT);
c0bceb97 1559 struct sk_buff_head *txq = &sk->sk_write_queue;
365ad353
JPM		 1560 struct tipc_sock *tsk = tipc_sk(sk);
1561 struct tipc_msg *hdr = &tsk->phdr;
1562 struct net *net = sock_net(sk);
0a3e060f 1563 struct sk_buff *skb;
365ad353 1564 u32 dnode = tsk_peer_node(tsk);
c0bceb97
JM		 1565 int maxnagle = tsk->maxnagle;
1566 int maxpkt = tsk->max_pkt;
365ad353 1567 int send, sent = 0;
c0bceb97 1568 int blocks, rc = 0;
7cf87fa2 1569
365ad353
JPM		 1570 if (unlikely(dlen > INT_MAX))
1571 return -EMSGSIZE;
4ccfe5e0 1572
365ad353 1573 /* Handle implicit connection setup */
f8dd60de 1574 if (unlikely(dest && sk->sk_state == TIPC_OPEN)) {
365ad353 1575 rc = __tipc_sendmsg(sock, m, dlen);
92ef12b3
PB		 1576 if (dlen && dlen == rc) {
1577 tsk->peer_caps = tipc_node_get_capabilities(net, dnode);
365ad353 1578 tsk->snt_unacked = tsk_inc(tsk, dlen + msg_hdr_sz(hdr));
92ef12b3 1579 }
39a0295f 1580 return rc;
365ad353 1581 }
f214fc40 1582
c4307285 1583 do {
365ad353
JPM		 1584 rc = tipc_wait_for_cond(sock, &timeout,
1585 (!tsk->cong_link_cnt &&
8c44e1af
JPM		 1586 !tsk_conn_cong(tsk) &&
1587 tipc_sk_connected(sk)));
365ad353
JPM		 1588 if (unlikely(rc))
1589 break;
365ad353 1590 send = min_t(size_t, dlen - sent, TIPC_MAX_USER_MSG_SIZE);
c0bceb97 1591 blocks = tsk->snd_backlog;
c9aa81fa
TL		 1592 if (tsk->oneway++ >= tsk->nagle_start && maxnagle &&
1593 send <= maxnagle) {
c0bceb97
JM		 1594 rc = tipc_msg_append(hdr, m, send, maxnagle, txq);
1595 if (unlikely(rc < 0))
1596 break;
1597 blocks += rc;
0a3e060f 1598 tsk->msg_acc++;
c0bceb97
JM		 1599 if (blocks <= 64 && tsk->expect_ack) {
1600 tsk->snd_backlog = blocks;
1601 sent += send;
1602 break;
0a3e060f
TL		 1603 } else if (blocks > 64) {
1604 tsk->pkt_cnt += skb_queue_len(txq);
1605 } else {
1606 skb = skb_peek_tail(txq);
4c21daae
Y		 1607 if (skb) {
1608 msg_set_ack_required(buf_msg(skb));
1609 tsk->expect_ack = true;
1610 } else {
1611 tsk->expect_ack = false;
1612 }
0a3e060f
TL		 1613 tsk->msg_acc = 0;
1614 tsk->pkt_cnt = 0;
c0bceb97 1615 }
c0bceb97
JM		 1616 } else {
1617 rc = tipc_msg_build(hdr, m, sent, send, maxpkt, txq);
1618 if (unlikely(rc != send))
1619 break;
1620 blocks += tsk_inc(tsk, send + MIN_H_SIZE);
1621 }
1622 trace_tipc_sk_sendstream(sk, skb_peek(txq),
01e661eb 1623 TIPC_DUMP_SK_SNDQ, " ");
c0bceb97 1624 rc = tipc_node_xmit(net, txq, dnode, tsk->portid);
365ad353
JPM		 1625 if (unlikely(rc == -ELINKCONG)) {
1626 tsk->cong_link_cnt = 1;
1627 rc = 0;
1628 }
1629 if (likely(!rc)) {
c0bceb97
JM		 1630 tsk->snt_unacked += blocks;
1631 tsk->snd_backlog = 0;
365ad353
JPM		 1632 sent += send;
1633 }
1634 } while (sent < dlen && !rc);
39a0295f 1635
3364d61c 1636 return sent ? sent : rc;
b97bf3fd
PL		 1637}
1638
c4307285 1639/**
4ccfe5e0 1640 * tipc_send_packet - send a connection-oriented message
b97bf3fd 1641 * @sock: socket structure
4ccfe5e0
JPM		 1642 * @m: message to send
1643 * @dsz: length of data to be transmitted
c4307285 1644 *
4ccfe5e0 1645 * Used for SOCK_SEQPACKET messages.
c4307285 1646 *
637b77fd 1647 * Return: the number of bytes sent on success, or errno otherwise
b97bf3fd 1648 */
1b784140 1649static int tipc_send_packet(struct socket *sock, struct msghdr *m, size_t dsz)
b97bf3fd 1650{
4ccfe5e0
JPM		 1651 if (dsz > TIPC_MAX_USER_MSG_SIZE)
1652 return -EMSGSIZE;
b97bf3fd 1653
365ad353 1654 return tipc_sendstream(sock, m, dsz);
b97bf3fd
PL		 1655}
1656
dadebc00 1657/* tipc_sk_finish_conn - complete the setup of a connection
b97bf3fd 1658 */
301bae56 1659static void tipc_sk_finish_conn(struct tipc_sock *tsk, u32 peer_port,
dadebc00 1660 u32 peer_node)
b97bf3fd 1661{
3721e9c7
YX		 1662 struct sock *sk = &tsk->sk;
1663 struct net *net = sock_net(sk);
301bae56 1664 struct tipc_msg *msg = &tsk->phdr;
b97bf3fd 1665
25b9221b 1666 msg_set_syn(msg, 0);
dadebc00
JPM		 1667 msg_set_destnode(msg, peer_node);
1668 msg_set_destport(msg, peer_port);
1669 msg_set_type(msg, TIPC_CONN_MSG);
1670 msg_set_lookup_scope(msg, 0);
1671 msg_set_hdr_sz(msg, SHORT_H_SIZE);
584d24b3 1672
0d5fcebf 1673 sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
8ea642ee 1674 tipc_set_sk_state(sk, TIPC_ESTABLISHED);
f2f9800d 1675 tipc_node_add_conn(net, peer_node, tsk->portid, peer_port);
f73b1281 1676 tsk->max_pkt = tipc_node_get_mtu(net, peer_node, tsk->portid, true);
60020e18 1677 tsk->peer_caps = tipc_node_get_capabilities(net, peer_node);
c0bceb97 1678 tsk_set_nagle(tsk);
67879274 1679 __skb_queue_purge(&sk->sk_write_queue);
10724cc7
JPM		 1680 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL)
1681 return;
1682
1683 /* Fall back to message based flow control */
1684 tsk->rcv_win = FLOWCTL_MSG_WIN;
1685 tsk->snd_win = FLOWCTL_MSG_WIN;
b97bf3fd
PL		 1686}
1687
1688/**
31c82a2d 1689 * tipc_sk_set_orig_addr - capture sender's address for received message
b97bf3fd 1690 * @m: descriptor for message info
d8141208 1691 * @skb: received message
c4307285 1692 *
b97bf3fd
PL		 1693 * Note: Address is not captured if not requested by receiver.
1694 */
31c82a2d 1695static void tipc_sk_set_orig_addr(struct msghdr *m, struct sk_buff *skb)
b97bf3fd 1696{
31c82a2d
JM		 1697 DECLARE_SOCKADDR(struct sockaddr_pair *, srcaddr, m->msg_name);
1698 struct tipc_msg *hdr = buf_msg(skb);
1699
1700 if (!srcaddr)
1701 return;
1702
1703 srcaddr->sock.family = AF_TIPC;
b6f88d9c 1704 srcaddr->sock.addrtype = TIPC_SOCKET_ADDR;
09c8b971 1705 srcaddr->sock.scope = 0;
31c82a2d
JM		 1706 srcaddr->sock.addr.id.ref = msg_origport(hdr);
1707 srcaddr->sock.addr.id.node = msg_orignode(hdr);
1708 srcaddr->sock.addr.name.domain = 0;
31c82a2d
JM		 1709 m->msg_namelen = sizeof(struct sockaddr_tipc);
1710
1711 if (!msg_in_group(hdr))
1712 return;
1713
1714 /* Group message users may also want to know sending member's id */
1715 srcaddr->member.family = AF_TIPC;
b6f88d9c 1716 srcaddr->member.addrtype = TIPC_SERVICE_ADDR;
09c8b971 1717 srcaddr->member.scope = 0;
31c82a2d
JM		 1718 srcaddr->member.addr.name.name.type = msg_nametype(hdr);
1719 srcaddr->member.addr.name.name.instance = TIPC_SKB_CB(skb)->orig_member;
1720 srcaddr->member.addr.name.domain = 0;
1721 m->msg_namelen = sizeof(*srcaddr);
b97bf3fd
PL		 1722}
1723
1724/**
301bae56 1725 * tipc_sk_anc_data_recv - optionally capture ancillary data for received message
b97bf3fd 1726 * @m: descriptor for message info
1c1274a5 1727 * @skb: received message buffer
301bae56 1728 * @tsk: TIPC port associated with message
c4307285 1729 *
b97bf3fd 1730 * Note: Ancillary data is not captured if not requested by receiver.
c4307285 1731 *
637b77fd 1732 * Return: 0 if successful, otherwise errno
b97bf3fd 1733 */
1c1274a5 1734static int tipc_sk_anc_data_recv(struct msghdr *m, struct sk_buff *skb,
301bae56 1735 struct tipc_sock *tsk)
b97bf3fd 1736{
62633c2f
JM		 1737 struct tipc_msg *hdr;
1738 u32 data[3] = {0,};
1739 bool has_addr;
1740 int dlen, rc;
b97bf3fd
PL		 1741
1742 if (likely(m->msg_controllen == 0))
1743 return 0;
1744
62633c2f
JM		 1745 hdr = buf_msg(skb);
1746 dlen = msg_data_sz(hdr);
1747
1748 /* Capture errored message object, if any */
1749 if (msg_errcode(hdr)) {
1750 if (skb_linearize(skb))
1751 return -ENOMEM;
1752 hdr = buf_msg(skb);
1753 data[0] = msg_errcode(hdr);
1754 data[1] = dlen;
1755 rc = put_cmsg(m, SOL_TIPC, TIPC_ERRINFO, 8, data);
1756 if (rc || !dlen)
1757 return rc;
1758 rc = put_cmsg(m, SOL_TIPC, TIPC_RETDATA, dlen, msg_data(hdr));
1759 if (rc)
1760 return rc;
b97bf3fd
PL		 1761 }
1762
62633c2f
JM		 1763 /* Capture TIPC_SERVICE_ADDR/RANGE destination address, if any */
1764 switch (msg_type(hdr)) {
b97bf3fd 1765 case TIPC_NAMED_MSG:
62633c2f
JM		 1766 has_addr = true;
1767 data[0] = msg_nametype(hdr);
1768 data[1] = msg_namelower(hdr);
1769 data[2] = data[1];
b97bf3fd
PL		 1770 break;
1771 case TIPC_MCAST_MSG:
62633c2f
JM		 1772 has_addr = true;
1773 data[0] = msg_nametype(hdr);
1774 data[1] = msg_namelower(hdr);
1775 data[2] = msg_nameupper(hdr);
b97bf3fd
PL		 1776 break;
1777 case TIPC_CONN_MSG:
62633c2f
JM		 1778 has_addr = !!tsk->conn_addrtype;
1779 data[0] = msg_nametype(&tsk->phdr);
1780 data[1] = msg_nameinst(&tsk->phdr);
1781 data[2] = data[1];
b97bf3fd
PL		 1782 break;
1783 default:
62633c2f 1784 has_addr = false;
2db9983a 1785 }
62633c2f
JM		 1786 if (!has_addr)
1787 return 0;
1788 return put_cmsg(m, SOL_TIPC, TIPC_DESTNAME, 12, data);
b97bf3fd
PL		 1789}
1790
c7268589 1791static struct sk_buff *tipc_sk_build_ack(struct tipc_sock *tsk)
739f5e4e 1792{
d6fb7e9c 1793 struct sock *sk = &tsk->sk;
a6ca1094 1794 struct sk_buff *skb = NULL;
739f5e4e 1795 struct tipc_msg *msg;
301bae56
JPM		 1796 u32 peer_port = tsk_peer_port(tsk);
1797 u32 dnode = tsk_peer_node(tsk);
739f5e4e 1798
d6fb7e9c 1799 if (!tipc_sk_connected(sk))
c7268589 1800 return NULL;
c5898636
JPM		 1801 skb = tipc_msg_create(CONN_MANAGER, CONN_ACK, INT_H_SIZE, 0,
1802 dnode, tsk_own_node(tsk), peer_port,
1803 tsk->portid, TIPC_OK);
a6ca1094 1804 if (!skb)
c7268589 1805 return NULL;
a6ca1094 1806 msg = buf_msg(skb);
10724cc7
JPM		 1807 msg_set_conn_ack(msg, tsk->rcv_unacked);
1808 tsk->rcv_unacked = 0;
1809
1810 /* Adjust to and advertize the correct window limit */
1811 if (tsk->peer_caps & TIPC_BLOCK_FLOWCTL) {
1812 tsk->rcv_win = tsk_adv_blocks(tsk->sk.sk_rcvbuf);
1813 msg_set_adv_win(msg, tsk->rcv_win);
1814 }
c7268589
TL		 1815 return skb;
1816}
1817
1818static void tipc_sk_send_ack(struct tipc_sock *tsk)
1819{
1820 struct sk_buff *skb;
1821
1822 skb = tipc_sk_build_ack(tsk);
1823 if (!skb)
1824 return;
1825
1826 tipc_node_xmit_skb(sock_net(&tsk->sk), skb, tsk_peer_node(tsk),
1827 msg_link_selector(buf_msg(skb)));
739f5e4e
JPM		 1828}
1829
85d3fc94 1830static int tipc_wait_for_rcvmsg(struct socket *sock, long *timeop)
9bbb4ecc
YX		 1831{
1832 struct sock *sk = sock->sk;
48766a58 1833 DEFINE_WAIT_FUNC(wait, woken_wake_function);
85d3fc94 1834 long timeo = *timeop;
4e0df495
PB		 1835 int err = sock_error(sk);
1836
1837 if (err)
1838 return err;
9bbb4ecc
YX		 1839
1840 for (;;) {
fe8e4649 1841 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
6f00089c 1842 if (sk->sk_shutdown & RCV_SHUTDOWN) {
9bbb4ecc
YX		 1843 err = -ENOTCONN;
1844 break;
1845 }
48766a58 1846 add_wait_queue(sk_sleep(sk), &wait);
9bbb4ecc 1847 release_sock(sk);
48766a58
TN		 1848 timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
1849 sched_annotate_sleep();
9bbb4ecc 1850 lock_sock(sk);
48766a58 1851 remove_wait_queue(sk_sleep(sk), &wait);
9bbb4ecc
YX		 1852 }
1853 err = 0;
1854 if (!skb_queue_empty(&sk->sk_receive_queue))
1855 break;
9bbb4ecc
YX		 1856 err = -EAGAIN;
1857 if (!timeo)
1858 break;
143fe22f
EH		 1859 err = sock_intr_errno(timeo);
1860 if (signal_pending(current))
1861 break;
4e0df495
PB		 1862
1863 err = sock_error(sk);
1864 if (err)
1865 break;
9bbb4ecc 1866 }
85d3fc94 1867 *timeop = timeo;
9bbb4ecc
YX		 1868 return err;
1869}
1870
c4307285 1871/**
247f0f3c 1872 * tipc_recvmsg - receive packet-oriented message
f172f4b8 1873 * @sock: network socket
b97bf3fd 1874 * @m: descriptor for message info
e9f8b101 1875 * @buflen: length of user buffer area
b97bf3fd 1876 * @flags: receive flags
c4307285 1877 *
b97bf3fd
PL		 1878 * Used for SOCK_DGRAM, SOCK_RDM, and SOCK_SEQPACKET messages.
1879 * If the complete message doesn't fit in user area, truncate it.
1880 *
637b77fd 1881 * Return: size of returned message data, errno otherwise
b97bf3fd 1882 */
e9f8b101
JPM		 1883static int tipc_recvmsg(struct socket *sock, struct msghdr *m,
1884 size_t buflen, int flags)
b97bf3fd 1885{
0c3141e9 1886 struct sock *sk = sock->sk;
e9f8b101 1887 bool connected = !tipc_sk_type_connectionless(sk);
ae236fb2 1888 struct tipc_sock *tsk = tipc_sk(sk);
e9f8b101 1889 int rc, err, hlen, dlen, copy;
f4919ff5 1890 struct tipc_skb_cb *skb_cb;
b7d42635 1891 struct sk_buff_head xmitq;
ae236fb2
JM		 1892 struct tipc_msg *hdr;
1893 struct sk_buff *skb;
1894 bool grp_evt;
e9f8b101 1895 long timeout;
b97bf3fd 1896
0c3141e9 1897 /* Catch invalid receive requests */
e9f8b101 1898 if (unlikely(!buflen))
b97bf3fd
PL		 1899 return -EINVAL;
1900
0c3141e9 1901 lock_sock(sk);
e9f8b101
JPM		 1902 if (unlikely(connected && sk->sk_state == TIPC_OPEN)) {
1903 rc = -ENOTCONN;
b97bf3fd
PL		 1904 goto exit;
1905 }
e9f8b101 1906 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
b97bf3fd 1907
b7d42635 1908 /* Step rcv queue to first msg with data or error; wait if necessary */
e9f8b101 1909 do {
e9f8b101
JPM		 1910 rc = tipc_wait_for_rcvmsg(sock, &timeout);
1911 if (unlikely(rc))
1912 goto exit;
1913 skb = skb_peek(&sk->sk_receive_queue);
f4919ff5 1914 skb_cb = TIPC_SKB_CB(skb);
e9f8b101
JPM		 1915 hdr = buf_msg(skb);
1916 dlen = msg_data_sz(hdr);
1917 hlen = msg_hdr_sz(hdr);
1918 err = msg_errcode(hdr);
ae236fb2 1919 grp_evt = msg_is_grp_evt(hdr);
e9f8b101
JPM		 1920 if (likely(dlen || err))
1921 break;
2e84c60b 1922 tsk_advance_rx_queue(sk);
e9f8b101 1923 } while (1);
b97bf3fd 1924
e9f8b101 1925 /* Collect msg meta data, including error code and rejected data */
31c82a2d 1926 tipc_sk_set_orig_addr(m, skb);
1c1274a5 1927 rc = tipc_sk_anc_data_recv(m, skb, tsk);
e9f8b101 1928 if (unlikely(rc))
b97bf3fd 1929 goto exit;
1c1274a5 1930 hdr = buf_msg(skb);
b97bf3fd 1931
e9f8b101
JPM		 1932 /* Capture data if non-error msg, otherwise just set return value */
1933 if (likely(!err)) {
f4919ff5
XL		 1934 int offset = skb_cb->bytes_read;
1935
1936 copy = min_t(int, dlen - offset, buflen);
1937 rc = skb_copy_datagram_msg(skb, hlen + offset, m, copy);
1938 if (unlikely(rc))
1939 goto exit;
1940 if (unlikely(offset + copy < dlen)) {
1941 if (flags & MSG_EOR) {
1942 if (!(flags & MSG_PEEK))
1943 skb_cb->bytes_read = offset + copy;
1944 } else {
1945 m->msg_flags |= MSG_TRUNC;
1946 skb_cb->bytes_read = 0;
1947 }
1948 } else {
1949 if (flags & MSG_EOR)
1950 m->msg_flags |= MSG_EOR;
1951 skb_cb->bytes_read = 0;
1952 }
b97bf3fd 1953 } else {
e9f8b101
JPM		 1954 copy = 0;
1955 rc = 0;
f4919ff5 1956 if (err != TIPC_CONN_SHUTDOWN && connected && !m->msg_control) {
e9f8b101 1957 rc = -ECONNRESET;
f4919ff5
XL		 1958 goto exit;
1959 }
b97bf3fd
PL		 1960 }
1961
ae236fb2
JM		 1962 /* Mark message as group event if applicable */
1963 if (unlikely(grp_evt)) {
1964 if (msg_grp_evt(hdr) == TIPC_WITHDRAWN)
1965 m->msg_flags |= MSG_EOR;
1966 m->msg_flags |= MSG_OOB;
1967 copy = 0;
1968 }
1969
e9f8b101 1970 /* Caption of data or error code/rejected data was successful */
10724cc7
JPM		 1971 if (unlikely(flags & MSG_PEEK))
1972 goto exit;
1973
b7d42635
JM		 1974 /* Send group flow control advertisement when applicable */
1975 if (tsk->group && msg_in_group(hdr) && !grp_evt) {
e654f9f5 1976 __skb_queue_head_init(&xmitq);
b7d42635
JM		 1977 tipc_group_update_rcv_win(tsk->group, tsk_blocks(hlen + dlen),
1978 msg_orignode(hdr), msg_origport(hdr),
1979 &xmitq);
1980 tipc_node_distr_xmit(sock_net(sk), &xmitq);
1981 }
1982
cc19862f
XL		 1983 if (skb_cb->bytes_read)
1984 goto exit;
1985
1986 tsk_advance_rx_queue(sk);
ae236fb2 1987
cc19862f 1988 if (likely(!connected))
e9f8b101
JPM		 1989 goto exit;
1990
b7d42635 1991 /* Send connection flow control advertisement when applicable */
e9f8b101
JPM		 1992 tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
1993 if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
1994 tipc_sk_send_ack(tsk);
b97bf3fd 1995exit:
0c3141e9 1996 release_sock(sk);
e9f8b101 1997 return rc ? rc : copy;
b97bf3fd
PL		 1998}
1999
c4307285 2000/**
ec8a09fb 2001 * tipc_recvstream - receive stream-oriented data
f172f4b8 2002 * @sock: network socket
b97bf3fd 2003 * @m: descriptor for message info
ec8a09fb 2004 * @buflen: total size of user buffer area
b97bf3fd 2005 * @flags: receive flags
c4307285
YH		 2006 *
2007 * Used for SOCK_STREAM messages only. If not enough data is available
b97bf3fd
PL		 2008 * will optionally wait for more; never truncates data.
2009 *
637b77fd 2010 * Return: size of returned message data, errno otherwise
b97bf3fd 2011 */
ec8a09fb
JPM		 2012static int tipc_recvstream(struct socket *sock, struct msghdr *m,
2013 size_t buflen, int flags)
b97bf3fd 2014{
0c3141e9 2015 struct sock *sk = sock->sk;
58ed9442 2016 struct tipc_sock *tsk = tipc_sk(sk);
ec8a09fb
JPM		 2017 struct sk_buff *skb;
2018 struct tipc_msg *hdr;
2019 struct tipc_skb_cb *skb_cb;
2020 bool peek = flags & MSG_PEEK;
2021 int offset, required, copy, copied = 0;
2022 int hlen, dlen, err, rc;
2023 long timeout;
b97bf3fd 2024
0c3141e9 2025 /* Catch invalid receive attempts */
ec8a09fb 2026 if (unlikely(!buflen))
b97bf3fd
PL		 2027 return -EINVAL;
2028
0c3141e9 2029 lock_sock(sk);
b97bf3fd 2030
438adcaf 2031 if (unlikely(sk->sk_state == TIPC_OPEN)) {
ec8a09fb 2032 rc = -ENOTCONN;
9bbb4ecc 2033 goto exit;
b97bf3fd 2034 }
ec8a09fb
JPM		 2035 required = sock_rcvlowat(sk, flags & MSG_WAITALL, buflen);
2036 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
b97bf3fd 2037
ec8a09fb
JPM		 2038 do {
2039 /* Look at first msg in receive queue; wait if necessary */
2040 rc = tipc_wait_for_rcvmsg(sock, &timeout);
2041 if (unlikely(rc))
2042 break;
2043 skb = skb_peek(&sk->sk_receive_queue);
2044 skb_cb = TIPC_SKB_CB(skb);
2045 hdr = buf_msg(skb);
2046 dlen = msg_data_sz(hdr);
2047 hlen = msg_hdr_sz(hdr);
2048 err = msg_errcode(hdr);
0232fd0a 2049
ec8a09fb
JPM		 2050 /* Discard any empty non-errored (SYN-) message */
2051 if (unlikely(!dlen && !err)) {
2052 tsk_advance_rx_queue(sk);
2053 continue;
2054 }
0232fd0a 2055
ec8a09fb
JPM		 2056 /* Collect msg meta data, incl. error code and rejected data */
2057 if (!copied) {
31c82a2d 2058 tipc_sk_set_orig_addr(m, skb);
1c1274a5 2059 rc = tipc_sk_anc_data_recv(m, skb, tsk);
ec8a09fb
JPM		 2060 if (rc)
2061 break;
1c1274a5 2062 hdr = buf_msg(skb);
ec8a09fb 2063 }
b97bf3fd 2064
ec8a09fb
JPM		 2065 /* Copy data if msg ok, otherwise return error/partial data */
2066 if (likely(!err)) {
2067 offset = skb_cb->bytes_read;
2068 copy = min_t(int, dlen - offset, buflen - copied);
2069 rc = skb_copy_datagram_msg(skb, hlen + offset, m, copy);
2070 if (unlikely(rc))
2071 break;
2072 copied += copy;
2073 offset += copy;
2074 if (unlikely(offset < dlen)) {
2075 if (!peek)
2076 skb_cb->bytes_read = offset;
2077 break;
2078 }
2079 } else {
2080 rc = 0;
2081 if ((err != TIPC_CONN_SHUTDOWN) && !m->msg_control)
2082 rc = -ECONNRESET;
2083 if (copied || rc)
2084 break;
b97bf3fd 2085 }
b97bf3fd 2086
ec8a09fb
JPM		 2087 if (unlikely(peek))
2088 break;
b97bf3fd 2089
ec8a09fb 2090 tsk_advance_rx_queue(sk);
10724cc7 2091
ec8a09fb
JPM		 2092 /* Send connection flow control advertisement when applicable */
2093 tsk->rcv_unacked += tsk_inc(tsk, hlen + dlen);
c7268589 2094 if (tsk->rcv_unacked >= tsk->rcv_win / TIPC_ACK_RATE)
ec8a09fb 2095 tipc_sk_send_ack(tsk);
b97bf3fd 2096
ec8a09fb
JPM		 2097 /* Exit if all requested data or FIN/error received */
2098 if (copied == buflen || err)
2099 break;
b97bf3fd 2100
ec8a09fb 2101 } while (!skb_queue_empty(&sk->sk_receive_queue) || copied < required);
b97bf3fd 2102exit:
0c3141e9 2103 release_sock(sk);
ec8a09fb 2104 return copied ? copied : rc;
b97bf3fd
PL		 2105}
2106
f288bef4
YX		 2107/**
2108 * tipc_write_space - wake up thread if port congestion is released
2109 * @sk: socket
2110 */
2111static void tipc_write_space(struct sock *sk)
2112{
2113 struct socket_wq *wq;
2114
2115 rcu_read_lock();
2116 wq = rcu_dereference(sk->sk_wq);
1ce0bf50 2117 if (skwq_has_sleeper(wq))
a9a08845
LT		 2118 wake_up_interruptible_sync_poll(&wq->wait, EPOLLOUT |
2119 EPOLLWRNORM | EPOLLWRBAND);
f288bef4
YX		 2120 rcu_read_unlock();
2121}
2122
2123/**
2124 * tipc_data_ready - wake up threads to indicate messages have been received
2125 * @sk: socket
f288bef4 2126 */
676d2369 2127static void tipc_data_ready(struct sock *sk)
f288bef4
YX		 2128{
2129 struct socket_wq *wq;
2130
40e0b090
PY		 2131 trace_sk_data_ready(sk);
2132
f288bef4
YX		 2133 rcu_read_lock();
2134 wq = rcu_dereference(sk->sk_wq);
1ce0bf50 2135 if (skwq_has_sleeper(wq))
a9a08845
LT		 2136 wake_up_interruptible_sync_poll(&wq->wait, EPOLLIN |
2137 EPOLLRDNORM | EPOLLRDBAND);
f288bef4
YX		 2138 rcu_read_unlock();
2139}
2140
f4195d1e
YX		 2141static void tipc_sock_destruct(struct sock *sk)
2142{
2143 __skb_queue_purge(&sk->sk_receive_queue);
2144}
2145
64ac5f59
JM		 2146static void tipc_sk_proto_rcv(struct sock *sk,
2147 struct sk_buff_head *inputq,
2148 struct sk_buff_head *xmitq)
2149{
2150 struct sk_buff *skb = __skb_dequeue(inputq);
2151 struct tipc_sock *tsk = tipc_sk(sk);
2152 struct tipc_msg *hdr = buf_msg(skb);
75da2163 2153 struct tipc_group *grp = tsk->group;
b7d42635 2154 bool wakeup = false;
64ac5f59
JM		 2155
2156 switch (msg_user(hdr)) {
2157 case CONN_MANAGER:
e7eb0582 2158 tipc_sk_conn_proto_rcv(tsk, skb, inputq, xmitq);
64ac5f59
JM		 2159 return;
2160 case SOCK_WAKEUP:
a80ae530 2161 tipc_dest_del(&tsk->cong_links, msg_orignode(hdr), 0);
bfd07f3d
TN		 2162 /* coupled with smp_rmb() in tipc_wait_for_cond() */
2163 smp_wmb();
64ac5f59 2164 tsk->cong_link_cnt--;
b7d42635 2165 wakeup = true;
0a3e060f 2166 tipc_sk_push_backlog(tsk, false);
64ac5f59 2167 break;
75da2163 2168 case GROUP_PROTOCOL:
b7d42635 2169 tipc_group_proto_rcv(grp, &wakeup, hdr, inputq, xmitq);
75da2163 2170 break;
64ac5f59 2171 case TOP_SRV:
b7d42635 2172 tipc_group_member_evt(tsk->group, &wakeup, &sk->sk_rcvbuf,
7ad32bcb 2173 hdr, inputq, xmitq);
64ac5f59
JM		 2174 break;
2175 default:
2176 break;
2177 }
2178
b7d42635
JM		 2179 if (wakeup)
2180 sk->sk_write_space(sk);
2181
64ac5f59
JM		 2182 kfree_skb(skb);
2183}
2184
7e6c131e 2185/**
39fdc9c7 2186 * tipc_sk_filter_connect - check incoming message for a connection-based socket
58ed9442 2187 * @tsk: TIPC socket
39fdc9c7 2188 * @skb: pointer to message buffer.
c7268589 2189 * @xmitq: for Nagle ACK if any
637b77fd 2190 * Return: true if message should be added to receive queue, false otherwise
7e6c131e 2191 */
c7268589
TL		 2192static bool tipc_sk_filter_connect(struct tipc_sock *tsk, struct sk_buff *skb,
2193 struct sk_buff_head *xmitq)
7e6c131e 2194{
58ed9442 2195 struct sock *sk = &tsk->sk;
f2f9800d 2196 struct net *net = sock_net(sk);
cda3696d 2197 struct tipc_msg *hdr = buf_msg(skb);
39fdc9c7
JM		 2198 bool con_msg = msg_connected(hdr);
2199 u32 pport = tsk_peer_port(tsk);
2200 u32 pnode = tsk_peer_node(tsk);
2201 u32 oport = msg_origport(hdr);
2202 u32 onode = msg_orignode(hdr);
2203 int err = msg_errcode(hdr);
67879274 2204 unsigned long delay;
7e6c131e 2205
cda3696d
JPM		 2206 if (unlikely(msg_mcast(hdr)))
2207 return false;
c0bceb97 2208 tsk->oneway = 0;
7e6c131e 2209
99a20889
PB		 2210 switch (sk->sk_state) {
2211 case TIPC_CONNECTING:
39fdc9c7
JM		 2212 /* Setup ACK */
2213 if (likely(con_msg)) {
2214 if (err)
2215 break;
2216 tipc_sk_finish_conn(tsk, oport, onode);
2217 msg_set_importance(&tsk->phdr, msg_importance(hdr));
2218 /* ACK+ message with data is added to receive queue */
2219 if (msg_data_sz(hdr))
2220 return true;
2221 /* Empty ACK-, - wake up sleeping connect() and drop */
ff946833 2222 sk->sk_state_change(sk);
39fdc9c7
JM		 2223 msg_set_dest_droppable(hdr, 1);
2224 return false;
584d24b3 2225 }
39fdc9c7
JM		 2226 /* Ignore connectionless message if not from listening socket */
2227 if (oport != pport || onode != pnode)
2228 return false;
584d24b3 2229
67879274
TN		 2230 /* Rejected SYN */
2231 if (err != TIPC_ERR_OVERLOAD)
2232 break;
2233
2234 /* Prepare for new setup attempt if we have a SYN clone */
2235 if (skb_queue_empty(&sk->sk_write_queue))
2236 break;
2237 get_random_bytes(&delay, 2);
2238 delay %= (tsk->conn_timeout / 4);
2239 delay = msecs_to_jiffies(delay + 100);
2240 sk_reset_timer(sk, &sk->sk_timer, jiffies + delay);
2241 return false;
438adcaf 2242 case TIPC_OPEN:
9fd4b070 2243 case TIPC_DISCONNECTING:
39fdc9c7 2244 return false;
438adcaf 2245 case TIPC_LISTEN:
7e6c131e 2246 /* Accept only SYN message */
25b9221b
JM		 2247 if (!msg_is_syn(hdr) &&
2248 tipc_node_get_capabilities(net, onode) & TIPC_SYN_BIT)
2249 return false;
39fdc9c7 2250 if (!con_msg && !err)
cda3696d 2251 return true;
39fdc9c7 2252 return false;
f40acbaf 2253 case TIPC_ESTABLISHED:
c0bceb97 2254 if (!skb_queue_empty(&sk->sk_write_queue))
0a3e060f 2255 tipc_sk_push_backlog(tsk, false);
f40acbaf 2256 /* Accept only connection-based messages sent by peer */
c7268589
TL		 2257 if (likely(con_msg && !err && pport == oport &&
2258 pnode == onode)) {
2259 if (msg_ack_required(hdr)) {
2260 struct sk_buff *skb;
2261
2262 skb = tipc_sk_build_ack(tsk);
0a3e060f
TL		 2263 if (skb) {
2264 msg_set_nagle_ack(buf_msg(skb));
c7268589 2265 __skb_queue_tail(xmitq, skb);
0a3e060f 2266 }
c7268589 2267 }
39fdc9c7 2268 return true;
c7268589 2269 }
39fdc9c7 2270 if (!tsk_peer_msg(tsk, hdr))
f40acbaf 2271 return false;
39fdc9c7
JM		 2272 if (!err)
2273 return true;
2274 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2275 tipc_node_remove_conn(net, pnode, tsk->portid);
2276 sk->sk_state_change(sk);
f40acbaf 2277 return true;
7e6c131e 2278 default:
438adcaf 2279 pr_err("Unknown sk_state %u\n", sk->sk_state);
7e6c131e 2280 }
39fdc9c7
JM		 2281 /* Abort connection setup attempt */
2282 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2283 sk->sk_err = ECONNREFUSED;
2284 sk->sk_state_change(sk);
2285 return true;
7e6c131e
YX		 2286}
2287
aba79f33
YX		 2288/**
2289 * rcvbuf_limit - get proper overload limit of socket receive queue
2290 * @sk: socket
10724cc7 2291 * @skb: message
aba79f33 2292 *
10724cc7
JPM		 2293 * For connection oriented messages, irrespective of importance,
2294 * default queue limit is 2 MB.
aba79f33 2295 *
10724cc7
JPM		 2296 * For connectionless messages, queue limits are based on message
2297 * importance as follows:
aba79f33 2298 *
10724cc7
JPM		 2299 * TIPC_LOW_IMPORTANCE (2 MB)
2300 * TIPC_MEDIUM_IMPORTANCE (4 MB)
2301 * TIPC_HIGH_IMPORTANCE (8 MB)
2302 * TIPC_CRITICAL_IMPORTANCE (16 MB)
aba79f33 2303 *
637b77fd 2304 * Return: overload limit according to corresponding message importance
aba79f33 2305 */
10724cc7 2306static unsigned int rcvbuf_limit(struct sock *sk, struct sk_buff *skb)
aba79f33 2307{
10724cc7
JPM		 2308 struct tipc_sock *tsk = tipc_sk(sk);
2309 struct tipc_msg *hdr = buf_msg(skb);
2310
b7d42635 2311 if (unlikely(msg_in_group(hdr)))
8265792b 2312 return READ_ONCE(sk->sk_rcvbuf);
b7d42635 2313
10724cc7 2314 if (unlikely(!msg_connected(hdr)))
8265792b 2315 return READ_ONCE(sk->sk_rcvbuf) << msg_importance(hdr);
aba79f33 2316
10724cc7 2317 if (likely(tsk->peer_caps & TIPC_BLOCK_FLOWCTL))
8265792b 2318 return READ_ONCE(sk->sk_rcvbuf);
0cee6bbe 2319
10724cc7 2320 return FLOWCTL_MSG_LIM;
aba79f33
YX		 2321}
2322
c4307285 2323/**
64ac5f59 2324 * tipc_sk_filter_rcv - validate incoming message
0c3141e9 2325 * @sk: socket
cda3696d 2326 * @skb: pointer to message.
f172f4b8 2327 * @xmitq: output message area (FIXME)
c4307285 2328 *
0c3141e9
AS		 2329 * Enqueues message on receive queue if acceptable; optionally handles
2330 * disconnect indication for a connected socket.
2331 *
1186adf7 2332 * Called with socket lock already taken
b97bf3fd 2333 */
64ac5f59
JM		 2334static void tipc_sk_filter_rcv(struct sock *sk, struct sk_buff *skb,
2335 struct sk_buff_head *xmitq)
b97bf3fd 2336{
64ac5f59 2337 bool sk_conn = !tipc_sk_type_connectionless(sk);
58ed9442 2338 struct tipc_sock *tsk = tipc_sk(sk);
75da2163 2339 struct tipc_group *grp = tsk->group;
cda3696d 2340 struct tipc_msg *hdr = buf_msg(skb);
64ac5f59
JM		 2341 struct net *net = sock_net(sk);
2342 struct sk_buff_head inputq;
77d5ad40 2343 int mtyp = msg_type(hdr);
64ac5f59 2344 int limit, err = TIPC_OK;
ec8a2e56 2345
01e661eb 2346 trace_tipc_sk_filter_rcv(sk, skb, TIPC_DUMP_ALL, " ");
64ac5f59
JM		 2347 TIPC_SKB_CB(skb)->bytes_read = 0;
2348 __skb_queue_head_init(&inputq);
2349 __skb_queue_tail(&inputq, skb);
50100a5e 2350
64ac5f59
JM		 2351 if (unlikely(!msg_isdata(hdr)))
2352 tipc_sk_proto_rcv(sk, &inputq, xmitq);
0c3141e9 2353
75da2163
JM		 2354 if (unlikely(grp))
2355 tipc_group_filter_msg(grp, &inputq, xmitq);
2356
77d5ad40 2357 if (unlikely(!grp) && mtyp == TIPC_MCAST_MSG)
08e046c8 2358 tipc_mcast_filter_msg(net, &tsk->mc_method.deferredq, &inputq);
c55c8eda 2359
64ac5f59
JM		 2360 /* Validate and add to receive buffer if there is space */
2361 while ((skb = __skb_dequeue(&inputq))) {
2362 hdr = buf_msg(skb);
2363 limit = rcvbuf_limit(sk, skb);
c7268589 2364 if ((sk_conn && !tipc_sk_filter_connect(tsk, skb, xmitq)) ||
75da2163
JM		 2365 (!sk_conn && msg_connected(hdr)) ||
2366 (!grp && msg_in_group(hdr)))
cda3696d 2367 err = TIPC_ERR_NO_PORT;
872619d8 2368 else if (sk_rmem_alloc_get(sk) + skb->truesize >= limit) {
01e661eb
TL		 2369 trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL,
2370 "err_overload2!");
872619d8 2371 atomic_inc(&sk->sk_drops);
64ac5f59 2372 err = TIPC_ERR_OVERLOAD;
872619d8 2373 }
b97bf3fd 2374
64ac5f59 2375 if (unlikely(err)) {
01e661eb
TL		 2376 if (tipc_msg_reverse(tipc_own_addr(net), &skb, err)) {
2377 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_NONE,
2378 "@filter_rcv!");
2379 __skb_queue_tail(xmitq, skb);
2380 }
64ac5f59
JM		 2381 err = TIPC_OK;
2382 continue;
2383 }
2384 __skb_queue_tail(&sk->sk_receive_queue, skb);
2385 skb_set_owner_r(skb, sk);
01e661eb
TL		 2386 trace_tipc_sk_overlimit2(sk, skb, TIPC_DUMP_ALL,
2387 "rcvq >90% allocated!");
64ac5f59 2388 sk->sk_data_ready(sk);
cda3696d 2389 }
0c3141e9 2390}
b97bf3fd 2391
0c3141e9 2392/**
64ac5f59 2393 * tipc_sk_backlog_rcv - handle incoming message from backlog queue
0c3141e9 2394 * @sk: socket
a6ca1094 2395 * @skb: message
0c3141e9 2396 *
e3a77561 2397 * Caller must hold socket lock
0c3141e9 2398 */
64ac5f59 2399static int tipc_sk_backlog_rcv(struct sock *sk, struct sk_buff *skb)
0c3141e9 2400{
64ac5f59 2401 unsigned int before = sk_rmem_alloc_get(sk);
f1d048f2 2402 struct sk_buff_head xmitq;
64ac5f59 2403 unsigned int added;
0c3141e9 2404
f1d048f2
JPM		 2405 __skb_queue_head_init(&xmitq);
2406
64ac5f59
JM		 2407 tipc_sk_filter_rcv(sk, skb, &xmitq);
2408 added = sk_rmem_alloc_get(sk) - before;
2409 atomic_add(added, &tipc_sk(sk)->dupl_rcvcnt);
f1d048f2 2410
64ac5f59 2411 /* Send pending response/rejected messages, if any */
f70d37b7 2412 tipc_node_distr_xmit(sock_net(sk), &xmitq);
0c3141e9
AS		 2413 return 0;
2414}
2415
d570d864 2416/**
c637c103
JPM		 2417 * tipc_sk_enqueue - extract all buffers with destination 'dport' from
2418 * inputq and try adding them to socket or backlog queue
2419 * @inputq: list of incoming buffers with potentially different destinations
2420 * @sk: socket where the buffers should be enqueued
2421 * @dport: port number for the socket
f172f4b8 2422 * @xmitq: output queue
d570d864
JPM		 2423 *
2424 * Caller must hold socket lock
d570d864 2425 */
cda3696d 2426static void tipc_sk_enqueue(struct sk_buff_head *inputq, struct sock *sk,
f1d048f2 2427 u32 dport, struct sk_buff_head *xmitq)
d570d864 2428{
f4bb62e6 2429 unsigned long time_limit = jiffies + usecs_to_jiffies(20000);
f1d048f2 2430 struct sk_buff *skb;
d570d864
JPM		 2431 unsigned int lim;
2432 atomic_t *dcnt;
f1d048f2 2433 u32 onode;
c637c103
JPM		 2434
2435 while (skb_queue_len(inputq)) {
51a00daf 2436 if (unlikely(time_after_eq(jiffies, time_limit)))
cda3696d
JPM		 2437 return;
2438
c637c103
JPM		 2439 skb = tipc_skb_dequeue(inputq, dport);
2440 if (unlikely(!skb))
cda3696d
JPM		 2441 return;
2442
2443 /* Add message directly to receive queue if possible */
c637c103 2444 if (!sock_owned_by_user(sk)) {
64ac5f59 2445 tipc_sk_filter_rcv(sk, skb, xmitq);
cda3696d 2446 continue;
c637c103 2447 }
cda3696d
JPM		 2448
2449 /* Try backlog, compensating for double-counted bytes */
c637c103 2450 dcnt = &tipc_sk(sk)->dupl_rcvcnt;
7c8bcfb1 2451 if (!sk->sk_backlog.len)
c637c103
JPM		 2452 atomic_set(dcnt, 0);
2453 lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
01e661eb
TL		 2454 if (likely(!sk_add_backlog(sk, skb, lim))) {
2455 trace_tipc_sk_overlimit1(sk, skb, TIPC_DUMP_ALL,
2456 "bklg & rcvq >90% allocated!");
c637c103 2457 continue;
01e661eb 2458 }
cda3696d 2459
01e661eb 2460 trace_tipc_sk_dump(sk, skb, TIPC_DUMP_ALL, "err_overload!");
cda3696d 2461 /* Overload => reject message back to sender */
f1d048f2 2462 onode = tipc_own_addr(sock_net(sk));
872619d8 2463 atomic_inc(&sk->sk_drops);
01e661eb
TL		 2464 if (tipc_msg_reverse(onode, &skb, TIPC_ERR_OVERLOAD)) {
2465 trace_tipc_sk_rej_msg(sk, skb, TIPC_DUMP_ALL,
2466 "@sk_enqueue!");
f1d048f2 2467 __skb_queue_tail(xmitq, skb);
01e661eb 2468 }
cda3696d 2469 break;
c637c103 2470 }
d570d864
JPM		 2471}
2472
0c3141e9 2473/**
c637c103 2474 * tipc_sk_rcv - handle a chain of incoming buffers
f172f4b8 2475 * @net: the associated network namespace
c637c103
JPM		 2476 * @inputq: buffer list containing the buffers
2477 * Consumes all buffers in list until inputq is empty
2478 * Note: may be called in multiple threads referring to the same queue
0c3141e9 2479 */
cda3696d 2480void tipc_sk_rcv(struct net *net, struct sk_buff_head *inputq)
0c3141e9 2481{
f1d048f2 2482 struct sk_buff_head xmitq;
c637c103 2483 u32 dnode, dport = 0;
9871b27f 2484 int err;
9816f061 2485 struct tipc_sock *tsk;
9816f061 2486 struct sock *sk;
cda3696d 2487 struct sk_buff *skb;
9816f061 2488
f1d048f2 2489 __skb_queue_head_init(&xmitq);
c637c103 2490 while (skb_queue_len(inputq)) {
c637c103
JPM		 2491 dport = tipc_skb_peek_port(inputq, dport);
2492 tsk = tipc_sk_lookup(net, dport);
cda3696d 2493
c637c103
JPM		 2494 if (likely(tsk)) {
2495 sk = &tsk->sk;
2496 if (likely(spin_trylock_bh(&sk->sk_lock.slock))) {
f1d048f2 2497 tipc_sk_enqueue(inputq, sk, dport, &xmitq);
c637c103 2498 spin_unlock_bh(&sk->sk_lock.slock);
c637c103 2499 }
f1d048f2 2500 /* Send pending response/rejected messages, if any */
f70d37b7 2501 tipc_node_distr_xmit(sock_net(sk), &xmitq);
c637c103 2502 sock_put(sk);
c637c103 2503 continue;
c637c103 2504 }
cda3696d
JPM		 2505 /* No destination socket => dequeue skb if still there */
2506 skb = tipc_skb_dequeue(inputq, dport);
2507 if (!skb)
2508 return;
2509
2510 /* Try secondary lookup if unresolved named message */
2511 err = TIPC_ERR_NO_PORT;
2512 if (tipc_msg_lookup_dest(net, skb, &err))
2513 goto xmit;
2514
2515 /* Prepare for message rejection */
2516 if (!tipc_msg_reverse(tipc_own_addr(net), &skb, err))
c637c103 2517 continue;
01e661eb
TL		 2518
2519 trace_tipc_sk_rej_msg(NULL, skb, TIPC_DUMP_NONE, "@sk_rcv!");
e3a77561 2520xmit:
cda3696d 2521 dnode = msg_destnode(buf_msg(skb));
af9b028e 2522 tipc_node_xmit_skb(net, skb, dnode, dport);
c637c103 2523 }
b97bf3fd
PL		 2524}
2525
78eb3a53
YX		 2526static int tipc_wait_for_connect(struct socket *sock, long *timeo_p)
2527{
d9dc8b0f 2528 DEFINE_WAIT_FUNC(wait, woken_wake_function);
78eb3a53 2529 struct sock *sk = sock->sk;
78eb3a53
YX		 2530 int done;
2531
2532 do {
2533 int err = sock_error(sk);
2534 if (err)
2535 return err;
2536 if (!*timeo_p)
2537 return -ETIMEDOUT;
2538 if (signal_pending(current))
2539 return sock_intr_errno(*timeo_p);
5391a877
TL		 2540 if (sk->sk_state == TIPC_DISCONNECTING)
2541 break;
78eb3a53 2542
d9dc8b0f 2543 add_wait_queue(sk_sleep(sk), &wait);
9546a0b7
TL		 2544 done = sk_wait_event(sk, timeo_p, tipc_sk_connected(sk),
2545 &wait);
d9dc8b0f 2546 remove_wait_queue(sk_sleep(sk), &wait);
78eb3a53
YX		 2547 } while (!done);
2548 return 0;
2549}
2550
ea239314
EH		 2551static bool tipc_sockaddr_is_sane(struct sockaddr_tipc *addr)
2552{
2553 if (addr->family != AF_TIPC)
2554 return false;
2555 if (addr->addrtype == TIPC_SERVICE_RANGE)
2556 return (addr->addr.nameseq.lower <= addr->addr.nameseq.upper);
2557 return (addr->addrtype == TIPC_SERVICE_ADDR ||
2558 addr->addrtype == TIPC_SOCKET_ADDR);
2559}
2560
b97bf3fd 2561/**
247f0f3c 2562 * tipc_connect - establish a connection to another TIPC port
b97bf3fd
PL		 2563 * @sock: socket structure
2564 * @dest: socket address for destination port
2565 * @destlen: size of socket address data structure
0c3141e9 2566 * @flags: file-related flags associated with socket
b97bf3fd 2567 *
637b77fd 2568 * Return: 0 on success, errno otherwise
b97bf3fd 2569 */
247f0f3c
YX		 2570static int tipc_connect(struct socket *sock, struct sockaddr *dest,
2571 int destlen, int flags)
b97bf3fd 2572{
0c3141e9 2573 struct sock *sk = sock->sk;
f2f8036e 2574 struct tipc_sock *tsk = tipc_sk(sk);
b89741a0
AS		 2575 struct sockaddr_tipc *dst = (struct sockaddr_tipc *)dest;
2576 struct msghdr m = {NULL,};
f2f8036e 2577 long timeout = (flags & O_NONBLOCK) ? 0 : tsk->conn_timeout;
99a20889 2578 int previous;
f2f8036e 2579 int res = 0;
b89741a0 2580
23998835
JM		 2581 if (destlen != sizeof(struct sockaddr_tipc))
2582 return -EINVAL;
2583
0c3141e9
AS		 2584 lock_sock(sk);
2585
75da2163
JM		 2586 if (tsk->group) {
2587 res = -EINVAL;
2588 goto exit;
2589 }
2590
23998835
JM		 2591 if (dst->family == AF_UNSPEC) {
2592 memset(&tsk->peer, 0, sizeof(struct sockaddr_tipc));
2593 if (!tipc_sk_type_connectionless(sk))
610600c8 2594 res = -EINVAL;
0c3141e9
AS		 2595 goto exit;
2596 }
ea239314 2597 if (!tipc_sockaddr_is_sane(dst)) {
0c3141e9 2598 res = -EINVAL;
23998835 2599 goto exit;
ea239314 2600 }
23998835
JM		 2601 /* DGRAM/RDM connect(), just save the destaddr */
2602 if (tipc_sk_type_connectionless(sk)) {
2603 memcpy(&tsk->peer, dest, destlen);
0c3141e9 2604 goto exit;
ea239314
EH		 2605 } else if (dst->addrtype == TIPC_SERVICE_RANGE) {
2606 res = -EINVAL;
2607 goto exit;
0c3141e9
AS		 2608 }
2609
99a20889 2610 previous = sk->sk_state;
438adcaf
PB		 2611
2612 switch (sk->sk_state) {
2613 case TIPC_OPEN:
584d24b3
YX		 2614 /* Send a 'SYN-' to destination */
2615 m.msg_name = dest;
2616 m.msg_namelen = destlen;
11a4d6f6 2617 iov_iter_kvec(&m.msg_iter, ITER_SOURCE, NULL, 0, 0);
584d24b3
YX		 2618
2619 /* If connect is in non-blocking case, set MSG_DONTWAIT to
2620 * indicate send_msg() is never blocked.
2621 */
2622 if (!timeout)
2623 m.msg_flags = MSG_DONTWAIT;
2624
39a0295f 2625 res = __tipc_sendmsg(sock, &m, 0);
584d24b3
YX		 2626 if ((res < 0) && (res != -EWOULDBLOCK))
2627 goto exit;
2628
99a20889 2629 /* Just entered TIPC_CONNECTING state; the only
584d24b3
YX		 2630 * difference is that return value in non-blocking
2631 * case is EINPROGRESS, rather than EALREADY.
2632 */
2633 res = -EINPROGRESS;
7f8901b7 2634 fallthrough;
99a20889
PB		 2635 case TIPC_CONNECTING:
2636 if (!timeout) {
2637 if (previous == TIPC_CONNECTING)
2638 res = -EALREADY;
78eb3a53 2639 goto exit;
99a20889 2640 }
78eb3a53
YX		 2641 timeout = msecs_to_jiffies(timeout);
2642 /* Wait until an 'ACK' or 'RST' arrives, or a timeout occurs */
2643 res = tipc_wait_for_connect(sock, &timeout);
f40acbaf
PB		 2644 break;
2645 case TIPC_ESTABLISHED:
584d24b3 2646 res = -EISCONN;
f40acbaf
PB		 2647 break;
2648 default:
584d24b3 2649 res = -EINVAL;
f40acbaf 2650 }
99a20889 2651
0c3141e9
AS		 2652exit:
2653 release_sock(sk);
b89741a0 2654 return res;
b97bf3fd
PL		 2655}
2656
c4307285 2657/**
247f0f3c 2658 * tipc_listen - allow socket to listen for incoming connections
b97bf3fd
PL		 2659 * @sock: socket structure
2660 * @len: (unused)
c4307285 2661 *
637b77fd 2662 * Return: 0 on success, errno otherwise
b97bf3fd 2663 */
247f0f3c 2664static int tipc_listen(struct socket *sock, int len)
b97bf3fd 2665{
0c3141e9
AS		 2666 struct sock *sk = sock->sk;
2667 int res;
2668
2669 lock_sock(sk);
0c288c86 2670 res = tipc_set_sk_state(sk, TIPC_LISTEN);
0c3141e9 2671 release_sock(sk);
0c288c86 2672
0c3141e9 2673 return res;
b97bf3fd
PL		 2674}
2675
6398e23c
YX		 2676static int tipc_wait_for_accept(struct socket *sock, long timeo)
2677{
2678 struct sock *sk = sock->sk;
d237a7f1 2679 DEFINE_WAIT_FUNC(wait, woken_wake_function);
6398e23c
YX		 2680 int err;
2681
2682 /* True wake-one mechanism for incoming connections: only
2683 * one process gets woken up, not the 'whole herd'.
2684 * Since we do not 'race & poll' for established sockets
2685 * anymore, the common case will execute the loop only once.
2686 */
2687 for (;;) {
fe8e4649 2688 if (timeo && skb_queue_empty(&sk->sk_receive_queue)) {
d237a7f1 2689 add_wait_queue(sk_sleep(sk), &wait);
6398e23c 2690 release_sock(sk);
d237a7f1 2691 timeo = wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
6398e23c 2692 lock_sock(sk);
d237a7f1 2693 remove_wait_queue(sk_sleep(sk), &wait);
6398e23c
YX		 2694 }
2695 err = 0;
2696 if (!skb_queue_empty(&sk->sk_receive_queue))
2697 break;
6398e23c
YX		 2698 err = -EAGAIN;
2699 if (!timeo)
2700 break;
143fe22f
EH		 2701 err = sock_intr_errno(timeo);
2702 if (signal_pending(current))
2703 break;
6398e23c 2704 }
6398e23c
YX		 2705 return err;
2706}
2707
c4307285 2708/**
247f0f3c 2709 * tipc_accept - wait for connection request
b97bf3fd 2710 * @sock: listening socket
d8141208 2711 * @new_sock: new socket that is to be connected
92ef0fd5 2712 * @arg: arguments for accept
c4307285 2713 *
637b77fd 2714 * Return: 0 on success, errno otherwise
b97bf3fd 2715 */
92ef0fd5
JA		 2716static int tipc_accept(struct socket *sock, struct socket *new_sock,
2717 struct proto_accept_arg *arg)
b97bf3fd 2718{
0fef8f20 2719 struct sock *new_sk, *sk = sock->sk;
301bae56 2720 struct tipc_sock *new_tsock;
f8dd60de 2721 struct msghdr m = {NULL,};
0fef8f20 2722 struct tipc_msg *msg;
f8dd60de 2723 struct sk_buff *buf;
6398e23c 2724 long timeo;
0c3141e9 2725 int res;
b97bf3fd 2726
0c3141e9 2727 lock_sock(sk);
b97bf3fd 2728
0c288c86 2729 if (sk->sk_state != TIPC_LISTEN) {
0c3141e9 2730 res = -EINVAL;
b97bf3fd
PL		 2731 goto exit;
2732 }
92ef0fd5 2733 timeo = sock_rcvtimeo(sk, arg->flags & O_NONBLOCK);
6398e23c
YX		 2734 res = tipc_wait_for_accept(sock, timeo);
2735 if (res)
2736 goto exit;
0c3141e9
AS		 2737
2738 buf = skb_peek(&sk->sk_receive_queue);
2739
92ef0fd5 2740 res = tipc_sk_create(sock_net(sock->sk), new_sock, 0, arg->kern);
0fef8f20
PG		 2741 if (res)
2742 goto exit;
fdd75ea8 2743 security_sk_clone(sock->sk, new_sock->sk);
b97bf3fd 2744
0fef8f20 2745 new_sk = new_sock->sk;
301bae56 2746 new_tsock = tipc_sk(new_sk);
0fef8f20 2747 msg = buf_msg(buf);
b97bf3fd 2748
0fef8f20
PG		 2749 /* we lock on new_sk; but lockdep sees the lock on sk */
2750 lock_sock_nested(new_sk, SINGLE_DEPTH_NESTING);
2751
2752 /*
2753 * Reject any stray messages received by new socket
2754 * before the socket lock was taken (very, very unlikely)
2755 */
49afb806 2756 tsk_rej_rx_queue(new_sk, TIPC_ERR_NO_PORT);
0fef8f20
PG		 2757
2758 /* Connect new socket to it's peer */
301bae56 2759 tipc_sk_finish_conn(new_tsock, msg_origport(msg), msg_orignode(msg));
0fef8f20 2760
095ae612 2761 tsk_set_importance(new_sk, msg_importance(msg));
0fef8f20 2762 if (msg_named(msg)) {
14623e00
JM		 2763 new_tsock->conn_addrtype = TIPC_SERVICE_ADDR;
2764 msg_set_nametype(&new_tsock->phdr, msg_nametype(msg));
2765 msg_set_nameinst(&new_tsock->phdr, msg_nameinst(msg));
b97bf3fd 2766 }
0fef8f20
PG		 2767
2768 /*
f8dd60de
XL		 2769 * Respond to 'SYN-' by discarding it & returning 'ACK'.
2770 * Respond to 'SYN+' by queuing it on new socket & returning 'ACK'.
0fef8f20
PG		 2771 */
2772 if (!msg_data_sz(msg)) {
2e84c60b 2773 tsk_advance_rx_queue(sk);
0fef8f20
PG		 2774 } else {
2775 __skb_dequeue(&sk->sk_receive_queue);
2776 __skb_queue_head(&new_sk->sk_receive_queue, buf);
aba79f33 2777 skb_set_owner_r(buf, new_sk);
0fef8f20 2778 }
11a4d6f6 2779 iov_iter_kvec(&m.msg_iter, ITER_SOURCE, NULL, 0, 0);
f8dd60de 2780 __tipc_sendstream(new_sock, &m, 0);
0fef8f20 2781 release_sock(new_sk);
b97bf3fd 2782exit:
0c3141e9 2783 release_sock(sk);
b97bf3fd
PL		 2784 return res;
2785}
2786
2787/**
247f0f3c 2788 * tipc_shutdown - shutdown socket connection
b97bf3fd 2789 * @sock: socket structure
e247a8f5 2790 * @how: direction to close (must be SHUT_RDWR)
b97bf3fd
PL		 2791 *
2792 * Terminates connection (if necessary), then purges socket's receive queue.
c4307285 2793 *
637b77fd 2794 * Return: 0 on success, errno otherwise
b97bf3fd 2795 */
247f0f3c 2796static int tipc_shutdown(struct socket *sock, int how)
b97bf3fd 2797{
0c3141e9 2798 struct sock *sk = sock->sk;
b97bf3fd
PL		 2799 int res;
2800
e247a8f5
AS		 2801 if (how != SHUT_RDWR)
2802 return -EINVAL;
b97bf3fd 2803
0c3141e9 2804 lock_sock(sk);
b97bf3fd 2805
01e661eb 2806 trace_tipc_sk_shutdown(sk, NULL, TIPC_DUMP_ALL, " ");
6f00089c 2807 __tipc_shutdown(sock, TIPC_CONN_SHUTDOWN);
a4b5cc9e 2808 sk->sk_shutdown = SHUTDOWN_MASK;
b97bf3fd 2809
6f00089c 2810 if (sk->sk_state == TIPC_DISCONNECTING) {
75031151 2811 /* Discard any unreceived messages */
57467e56 2812 __skb_queue_purge(&sk->sk_receive_queue);
75031151 2813
b97bf3fd 2814 res = 0;
6f00089c 2815 } else {
b97bf3fd
PL		 2816 res = -ENOTCONN;
2817 }
2a63866c
TH		 2818 /* Wake up anyone sleeping in poll. */
2819 sk->sk_state_change(sk);
b97bf3fd 2820
0c3141e9 2821 release_sock(sk);
b97bf3fd
PL		 2822 return res;
2823}
2824
afe8792f
JM		 2825static void tipc_sk_check_probing_state(struct sock *sk,
2826 struct sk_buff_head *list)
2827{
2828 struct tipc_sock *tsk = tipc_sk(sk);
2829 u32 pnode = tsk_peer_node(tsk);
2830 u32 pport = tsk_peer_port(tsk);
2831 u32 self = tsk_own_node(tsk);
2832 u32 oport = tsk->portid;
2833 struct sk_buff *skb;
2834
2835 if (tsk->probe_unacked) {
2836 tipc_set_sk_state(sk, TIPC_DISCONNECTING);
2837 sk->sk_err = ECONNABORTED;
2838 tipc_node_remove_conn(sock_net(sk), pnode, pport);
2839 sk->sk_state_change(sk);
2840 return;
2841 }
2842 /* Prepare new probe */
2843 skb = tipc_msg_create(CONN_MANAGER, CONN_PROBE, INT_H_SIZE, 0,
2844 pnode, self, pport, oport, TIPC_OK);
2845 if (skb)
2846 __skb_queue_tail(list, skb);
2847 tsk->probe_unacked = true;
2848 sk_reset_timer(sk, &sk->sk_timer, jiffies + CONN_PROBING_INTV);
2849}
2850
67879274
TN		 2851static void tipc_sk_retry_connect(struct sock *sk, struct sk_buff_head *list)
2852{
2853 struct tipc_sock *tsk = tipc_sk(sk);
2854
2855 /* Try again later if dest link is congested */
2856 if (tsk->cong_link_cnt) {
6a7d8cff
HL		 2857 sk_reset_timer(sk, &sk->sk_timer,
2858 jiffies + msecs_to_jiffies(100));
67879274
TN		 2859 return;
2860 }
2861 /* Prepare SYN for retransmit */
2862 tipc_msg_skb_clone(&sk->sk_write_queue, list);
2863}
2864
31b102bb 2865static void tipc_sk_timeout(struct timer_list *t)
57289015 2866{
31b102bb
KC		 2867 struct sock *sk = from_timer(sk, t, sk_timer);
2868 struct tipc_sock *tsk = tipc_sk(sk);
afe8792f
JM		 2869 u32 pnode = tsk_peer_node(tsk);
2870 struct sk_buff_head list;
67879274 2871 int rc = 0;
57289015 2872
e654f9f5 2873 __skb_queue_head_init(&list);
6c9808ce 2874 bh_lock_sock(sk);
0d5fcebf
JM		 2875
2876 /* Try again later if socket is busy */
2877 if (sock_owned_by_user(sk)) {
2878 sk_reset_timer(sk, &sk->sk_timer, jiffies + HZ / 20);
afe8792f 2879 bh_unlock_sock(sk);
91a4a3eb 2880 sock_put(sk);
afe8792f 2881 return;
57289015 2882 }
57289015 2883
afe8792f
JM		 2884 if (sk->sk_state == TIPC_ESTABLISHED)
2885 tipc_sk_check_probing_state(sk, &list);
67879274
TN		 2886 else if (sk->sk_state == TIPC_CONNECTING)
2887 tipc_sk_retry_connect(sk, &list);
afe8792f 2888
57289015 2889 bh_unlock_sock(sk);
afe8792f
JM		 2890
2891 if (!skb_queue_empty(&list))
67879274 2892 rc = tipc_node_xmit(sock_net(sk), &list, pnode, tsk->portid);
afe8792f 2893
67879274
TN		 2894 /* SYN messages may cause link congestion */
2895 if (rc == -ELINKCONG) {
2896 tipc_dest_push(&tsk->cong_links, pnode, 0);
2897 tsk->cong_link_cnt = 1;
2898 }
07f6c4bc 2899 sock_put(sk);
57289015
JPM		 2900}
2901
50a3499a 2902static int tipc_sk_publish(struct tipc_sock *tsk, struct tipc_uaddr *ua)
0fc87aae 2903{
d6fb7e9c
PB		 2904 struct sock *sk = &tsk->sk;
2905 struct net *net = sock_net(sk);
50a3499a
JM		 2906 struct tipc_socket_addr skaddr;
2907 struct publication *p;
0fc87aae
JPM		 2908 u32 key;
2909
d6fb7e9c 2910 if (tipc_sk_connected(sk))
0fc87aae 2911 return -EINVAL;
07f6c4bc
YX		 2912 key = tsk->portid + tsk->pub_count + 1;
2913 if (key == tsk->portid)
0fc87aae 2914 return -EADDRINUSE;
50a3499a
JM		 2915 skaddr.ref = tsk->portid;
2916 skaddr.node = tipc_own_addr(net);
2917 p = tipc_nametbl_publish(net, ua, &skaddr, key);
2918 if (unlikely(!p))
0fc87aae
JPM		 2919 return -EINVAL;
2920
50a3499a 2921 list_add(&p->binding_sock, &tsk->publications);
301bae56 2922 tsk->pub_count++;
50a3499a 2923 tsk->published = true;
0fc87aae
JPM		 2924 return 0;
2925}
2926
2c98da07 2927static int tipc_sk_withdraw(struct tipc_sock *tsk, struct tipc_uaddr *ua)
0fc87aae 2928{
f2f9800d 2929 struct net *net = sock_net(&tsk->sk);
2c98da07
JM		 2930 struct publication *safe, *p;
2931 struct tipc_uaddr _ua;
0fc87aae
JPM		 2932 int rc = -EINVAL;
2933
998d3907 2934 list_for_each_entry_safe(p, safe, &tsk->publications, binding_sock) {
2c98da07
JM		 2935 if (!ua) {
2936 tipc_uaddr(&_ua, TIPC_SERVICE_RANGE, p->scope,
2937 p->sr.type, p->sr.lower, p->sr.upper);
2938 tipc_nametbl_withdraw(net, &_ua, &p->sk, p->key);
2939 continue;
0fc87aae 2940 }
2c98da07
JM		 2941 /* Unbind specific publication */
2942 if (p->scope != ua->scope)
2943 continue;
2944 if (p->sr.type != ua->sr.type)
2945 continue;
2946 if (p->sr.lower != ua->sr.lower)
2947 continue;
2948 if (p->sr.upper != ua->sr.upper)
2949 break;
2950 tipc_nametbl_withdraw(net, ua, &p->sk, p->key);
0fc87aae 2951 rc = 0;
2c98da07 2952 break;
0fc87aae 2953 }
2c98da07 2954 if (list_empty(&tsk->publications)) {
301bae56 2955 tsk->published = 0;
2c98da07
JM		 2956 rc = 0;
2957 }
0fc87aae
JPM		 2958 return rc;
2959}
2960
5a9ee0be
JPM		 2961/* tipc_sk_reinit: set non-zero address in all existing sockets
2962 * when we go from standalone to network mode.
2963 */
e05b31f4 2964void tipc_sk_reinit(struct net *net)
5a9ee0be 2965{
e05b31f4 2966 struct tipc_net *tn = net_generic(net, tipc_net_id);
40f9f439 2967 struct rhashtable_iter iter;
07f6c4bc 2968 struct tipc_sock *tsk;
5a9ee0be 2969 struct tipc_msg *msg;
5a9ee0be 2970
40f9f439
HX		 2971 rhashtable_walk_enter(&tn->sk_rht, &iter);
2972
2973 do {
97a6ec4a 2974 rhashtable_walk_start(&iter);
40f9f439
HX		 2975
2976 while ((tsk = rhashtable_walk_next(&iter)) && !IS_ERR(tsk)) {
15ef70e2
CW		 2977 sock_hold(&tsk->sk);
2978 rhashtable_walk_stop(&iter);
2979 lock_sock(&tsk->sk);
07f6c4bc 2980 msg = &tsk->phdr;
23fd3eac
JM		 2981 msg_set_prevnode(msg, tipc_own_addr(net));
2982 msg_set_orignode(msg, tipc_own_addr(net));
15ef70e2
CW		 2983 release_sock(&tsk->sk);
2984 rhashtable_walk_start(&iter);
2985 sock_put(&tsk->sk);
07f6c4bc 2986 }
97a6ec4a 2987
40f9f439
HX		 2988 rhashtable_walk_stop(&iter);
2989 } while (tsk == ERR_PTR(-EAGAIN));
bd583fe3
CW		 2990
2991 rhashtable_walk_exit(&iter);
5a9ee0be
JPM		 2992}
2993
e05b31f4 2994static struct tipc_sock *tipc_sk_lookup(struct net *net, u32 portid)
808d90f9 2995{
e05b31f4 2996 struct tipc_net *tn = net_generic(net, tipc_net_id);
07f6c4bc 2997 struct tipc_sock *tsk;
808d90f9 2998
07f6c4bc 2999 rcu_read_lock();
ab818362 3000 tsk = rhashtable_lookup(&tn->sk_rht, &portid, tsk_rht_params);
07f6c4bc
YX		 3001 if (tsk)
3002 sock_hold(&tsk->sk);
3003 rcu_read_unlock();
808d90f9 3004
07f6c4bc 3005 return tsk;
808d90f9
JPM		 3006}
3007
07f6c4bc 3008static int tipc_sk_insert(struct tipc_sock *tsk)
808d90f9 3009{
e05b31f4
YX		 3010 struct sock *sk = &tsk->sk;
3011 struct net *net = sock_net(sk);
3012 struct tipc_net *tn = net_generic(net, tipc_net_id);
07f6c4bc 3013 u32 remaining = (TIPC_MAX_PORT - TIPC_MIN_PORT) + 1;
8032bf12 3014 u32 portid = get_random_u32_below(remaining) + TIPC_MIN_PORT;
808d90f9 3015
07f6c4bc
YX		 3016 while (remaining--) {
3017 portid++;
3018 if ((portid < TIPC_MIN_PORT) || (portid > TIPC_MAX_PORT))
3019 portid = TIPC_MIN_PORT;
3020 tsk->portid = portid;
3021 sock_hold(&tsk->sk);
6cca7289
HX		 3022 if (!rhashtable_lookup_insert_fast(&tn->sk_rht, &tsk->node,
3023 tsk_rht_params))
07f6c4bc
YX		 3024 return 0;
3025 sock_put(&tsk->sk);
808d90f9
JPM		 3026 }
3027
07f6c4bc 3028 return -1;
808d90f9
JPM		 3029}
3030
07f6c4bc 3031static void tipc_sk_remove(struct tipc_sock *tsk)
808d90f9 3032{
07f6c4bc 3033 struct sock *sk = &tsk->sk;
e05b31f4 3034 struct tipc_net *tn = net_generic(sock_net(sk), tipc_net_id);
808d90f9 3035
6cca7289 3036 if (!rhashtable_remove_fast(&tn->sk_rht, &tsk->node, tsk_rht_params)) {
41c6d650 3037 WARN_ON(refcount_read(&sk->sk_refcnt) == 1);
07f6c4bc 3038 __sock_put(sk);
808d90f9 3039 }
808d90f9
JPM		 3040}
3041
6cca7289
HX		 3042static const struct rhashtable_params tsk_rht_params = {
3043 .nelem_hint = 192,
3044 .head_offset = offsetof(struct tipc_sock, node),
3045 .key_offset = offsetof(struct tipc_sock, portid),
3046 .key_len = sizeof(u32), /* portid */
6cca7289
HX		 3047 .max_size = 1048576,
3048 .min_size = 256,
b5e2c150 3049 .automatic_shrinking = true,
6cca7289
HX		 3050};
3051
e05b31f4 3052int tipc_sk_rht_init(struct net *net)
808d90f9 3053{
e05b31f4 3054 struct tipc_net *tn = net_generic(net, tipc_net_id);
6cca7289
HX		 3055
3056 return rhashtable_init(&tn->sk_rht, &tsk_rht_params);
808d90f9
JPM		 3057}
3058
e05b31f4 3059void tipc_sk_rht_destroy(struct net *net)
808d90f9 3060{
e05b31f4
YX		 3061 struct tipc_net *tn = net_generic(net, tipc_net_id);
3062
07f6c4bc
YX		 3063 /* Wait for socket readers to complete */
3064 synchronize_net();
808d90f9 3065
e05b31f4 3066 rhashtable_destroy(&tn->sk_rht);
808d90f9
JPM		 3067}
3068
75da2163
JM		 3069static int tipc_sk_join(struct tipc_sock *tsk, struct tipc_group_req *mreq)
3070{
3071 struct net *net = sock_net(&tsk->sk);
75da2163
JM		 3072 struct tipc_group *grp = tsk->group;
3073 struct tipc_msg *hdr = &tsk->phdr;
50a3499a 3074 struct tipc_uaddr ua;
75da2163
JM		 3075 int rc;
3076
3077 if (mreq->type < TIPC_RESERVED_TYPES)
3078 return -EACCES;
232d07b7
JM		 3079 if (mreq->scope > TIPC_NODE_SCOPE)
3080 return -EINVAL;
50a3499a
JM		 3081 if (mreq->scope != TIPC_NODE_SCOPE)
3082 mreq->scope = TIPC_CLUSTER_SCOPE;
75da2163
JM		 3083 if (grp)
3084 return -EACCES;
60c25306 3085 grp = tipc_group_create(net, tsk->portid, mreq, &tsk->group_is_open);
75da2163
JM		 3086 if (!grp)
3087 return -ENOMEM;
3088 tsk->group = grp;
3089 msg_set_lookup_scope(hdr, mreq->scope);
3090 msg_set_nametype(hdr, mreq->type);
3091 msg_set_dest_droppable(hdr, true);
50a3499a
JM		 3092 tipc_uaddr(&ua, TIPC_SERVICE_RANGE, mreq->scope,
3093 mreq->type, mreq->instance, mreq->instance);
6e44867b 3094 tipc_nametbl_build_group(net, grp, &ua);
50a3499a 3095 rc = tipc_sk_publish(tsk, &ua);
e233df01 3096 if (rc) {
75da2163 3097 tipc_group_delete(net, grp);
e233df01 3098 tsk->group = NULL;
febafc84 3099 return rc;
e233df01 3100 }
d12d2e12 3101 /* Eliminate any risk that a broadcast overtakes sent JOINs */
399574d4
JM		 3102 tsk->mc_method.rcast = true;
3103 tsk->mc_method.mandatory = true;
d12d2e12 3104 tipc_group_join(net, grp, &tsk->sk.sk_rcvbuf);
75da2163
JM		 3105 return rc;
3106}
3107
3108static int tipc_sk_leave(struct tipc_sock *tsk)
3109{
3110 struct net *net = sock_net(&tsk->sk);
3111 struct tipc_group *grp = tsk->group;
2c98da07 3112 struct tipc_uaddr ua;
75da2163
JM		 3113 int scope;
3114
3115 if (!grp)
3116 return -EINVAL;
2c98da07
JM		 3117 ua.addrtype = TIPC_SERVICE_RANGE;
3118 tipc_group_self(grp, &ua.sr, &scope);
3119 ua.scope = scope;
75da2163
JM		 3120 tipc_group_delete(net, grp);
3121 tsk->group = NULL;
2c98da07 3122 tipc_sk_withdraw(tsk, &ua);
75da2163
JM		 3123 return 0;
3124}
3125
b97bf3fd 3126/**
247f0f3c 3127 * tipc_setsockopt - set socket option
b97bf3fd
PL		 3128 * @sock: socket structure
3129 * @lvl: option level
3130 * @opt: option identifier
3131 * @ov: pointer to new option value
3132 * @ol: length of option value
c4307285
YH		 3133 *
3134 * For stream sockets only, accepts and ignores all IPPROTO_TCP options
b97bf3fd 3135 * (to ease compatibility).
c4307285 3136 *
637b77fd 3137 * Return: 0 on success, errno otherwise
b97bf3fd 3138 */
247f0f3c 3139static int tipc_setsockopt(struct socket *sock, int lvl, int opt,
a7b75c5a 3140 sockptr_t ov, unsigned int ol)
b97bf3fd 3141{
0c3141e9 3142 struct sock *sk = sock->sk;
58ed9442 3143 struct tipc_sock *tsk = tipc_sk(sk);
75da2163 3144 struct tipc_group_req mreq;
01fd12bb 3145 u32 value = 0;
a08ef476 3146 int res = 0;
b97bf3fd 3147
c4307285
YH		 3148 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3149 return 0;
b97bf3fd
PL		 3150 if (lvl != SOL_TIPC)
3151 return -ENOPROTOOPT;
01fd12bb
JPM		 3152
3153 switch (opt) {
3154 case TIPC_IMPORTANCE:
3155 case TIPC_SRC_DROPPABLE:
3156 case TIPC_DEST_DROPPABLE:
3157 case TIPC_CONN_TIMEOUT:
c0bceb97 3158 case TIPC_NODELAY:
01fd12bb
JPM		 3159 if (ol < sizeof(value))
3160 return -EINVAL;
a7b75c5a 3161 if (copy_from_sockptr(&value, ov, sizeof(u32)))
75da2163
JM		 3162 return -EFAULT;
3163 break;
3164 case TIPC_GROUP_JOIN:
3165 if (ol < sizeof(mreq))
3166 return -EINVAL;
a7b75c5a 3167 if (copy_from_sockptr(&mreq, ov, sizeof(mreq)))
75da2163 3168 return -EFAULT;
01fd12bb
JPM		 3169 break;
3170 default:
a7b75c5a 3171 if (!sockptr_is_null(ov) || ol)
01fd12bb
JPM		 3172 return -EINVAL;
3173 }
b97bf3fd 3174
0c3141e9 3175 lock_sock(sk);
c4307285 3176
b97bf3fd
PL		 3177 switch (opt) {
3178 case TIPC_IMPORTANCE:
095ae612 3179 res = tsk_set_importance(sk, value);
b97bf3fd
PL		 3180 break;
3181 case TIPC_SRC_DROPPABLE:
3182 if (sock->type != SOCK_STREAM)
301bae56 3183 tsk_set_unreliable(tsk, value);
c4307285 3184 else
b97bf3fd
PL		 3185 res = -ENOPROTOOPT;
3186 break;
3187 case TIPC_DEST_DROPPABLE:
301bae56 3188 tsk_set_unreturnable(tsk, value);
b97bf3fd
PL		 3189 break;
3190 case TIPC_CONN_TIMEOUT:
a0f40f02 3191 tipc_sk(sk)->conn_timeout = value;
b97bf3fd 3192 break;
01fd12bb
JPM		 3193 case TIPC_MCAST_BROADCAST:
3194 tsk->mc_method.rcast = false;
3195 tsk->mc_method.mandatory = true;
3196 break;
3197 case TIPC_MCAST_REPLICAST:
3198 tsk->mc_method.rcast = true;
3199 tsk->mc_method.mandatory = true;
3200 break;
75da2163
JM		 3201 case TIPC_GROUP_JOIN:
3202 res = tipc_sk_join(tsk, &mreq);
3203 break;
3204 case TIPC_GROUP_LEAVE:
3205 res = tipc_sk_leave(tsk);
3206 break;
c0bceb97
JM		 3207 case TIPC_NODELAY:
3208 tsk->nodelay = !!value;
3209 tsk_set_nagle(tsk);
3210 break;
b97bf3fd
PL		 3211 default:
3212 res = -EINVAL;
3213 }
3214
0c3141e9
AS		 3215 release_sock(sk);
3216
b97bf3fd
PL		 3217 return res;
3218}
3219
3220/**
247f0f3c 3221 * tipc_getsockopt - get socket option
b97bf3fd
PL		 3222 * @sock: socket structure
3223 * @lvl: option level
3224 * @opt: option identifier
3225 * @ov: receptacle for option value
3226 * @ol: receptacle for length of option value
c4307285
YH		 3227 *
3228 * For stream sockets only, returns 0 length result for all IPPROTO_TCP options
b97bf3fd 3229 * (to ease compatibility).
c4307285 3230 *
637b77fd 3231 * Return: 0 on success, errno otherwise
b97bf3fd 3232 */
247f0f3c
YX		 3233static int tipc_getsockopt(struct socket *sock, int lvl, int opt,
3234 char __user *ov, int __user *ol)
b97bf3fd 3235{
0c3141e9 3236 struct sock *sk = sock->sk;
58ed9442 3237 struct tipc_sock *tsk = tipc_sk(sk);
b6f88d9c 3238 struct tipc_service_range seq;
75da2163 3239 int len, scope;
b97bf3fd 3240 u32 value;
c4307285 3241 int res;
b97bf3fd 3242
c4307285
YH		 3243 if ((lvl == IPPROTO_TCP) && (sock->type == SOCK_STREAM))
3244 return put_user(0, ol);
b97bf3fd
PL		 3245 if (lvl != SOL_TIPC)
3246 return -ENOPROTOOPT;
2db9983a
AS		 3247 res = get_user(len, ol);
3248 if (res)
c4307285 3249 return res;
b97bf3fd 3250
0c3141e9 3251 lock_sock(sk);
b97bf3fd
PL		 3252
3253 switch (opt) {
3254 case TIPC_IMPORTANCE:
301bae56 3255 value = tsk_importance(tsk);
b97bf3fd
PL		 3256 break;
3257 case TIPC_SRC_DROPPABLE:
301bae56 3258 value = tsk_unreliable(tsk);
b97bf3fd
PL		 3259 break;
3260 case TIPC_DEST_DROPPABLE:
301bae56 3261 value = tsk_unreturnable(tsk);
b97bf3fd
PL		 3262 break;
3263 case TIPC_CONN_TIMEOUT:
301bae56 3264 value = tsk->conn_timeout;
0c3141e9 3265 /* no need to set "res", since already 0 at this point */
b97bf3fd 3266 break;
0e65967e 3267 case TIPC_NODE_RECVQ_DEPTH:
9da3d475 3268 value = 0; /* was tipc_queue_size, now obsolete */
6650613d 3269 break;
0e65967e 3270 case TIPC_SOCK_RECVQ_DEPTH:
6650613d 3271 value = skb_queue_len(&sk->sk_receive_queue);
3272 break;
42e5425a
TN		 3273 case TIPC_SOCK_RECVQ_USED:
3274 value = sk_rmem_alloc_get(sk);
3275 break;
75da2163
JM		 3276 case TIPC_GROUP_JOIN:
3277 seq.type = 0;
3278 if (tsk->group)
3279 tipc_group_self(tsk->group, &seq, &scope);
3280 value = seq.type;
3281 break;
b97bf3fd
PL		 3282 default:
3283 res = -EINVAL;
3284 }
3285
0c3141e9
AS		 3286 release_sock(sk);
3287
25860c3b
PG		 3288 if (res)
3289 return res; /* "get" failed */
b97bf3fd 3290
25860c3b
PG		 3291 if (len < sizeof(value))
3292 return -EINVAL;
3293
3294 if (copy_to_user(ov, &value, sizeof(value)))
3295 return -EFAULT;
3296
3297 return put_user(sizeof(value), ol);
b97bf3fd
PL		 3298}
3299
f2f9800d 3300static int tipc_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
78acb1f9 3301{
3e5cf362
JM		 3302 struct net *net = sock_net(sock->sk);
3303 struct tipc_sioc_nodeid_req nr = {0};
78acb1f9
EH		 3304 struct tipc_sioc_ln_req lnr;
3305 void __user *argp = (void __user *)arg;
3306
3307 switch (cmd) {
3308 case SIOCGETLINKNAME:
3309 if (copy_from_user(&lnr, argp, sizeof(lnr)))
3310 return -EFAULT;
3e5cf362 3311 if (!tipc_node_get_linkname(net,
f2f9800d 3312 lnr.bearer_id & 0xffff, lnr.peer,
78acb1f9
EH		 3313 lnr.linkname, TIPC_MAX_LINK_NAME)) {
3314 if (copy_to_user(argp, &lnr, sizeof(lnr)))
3315 return -EFAULT;
3316 return 0;
3317 }
3318 return -EADDRNOTAVAIL;
3e5cf362
JM		 3319 case SIOCGETNODEID:
3320 if (copy_from_user(&nr, argp, sizeof(nr)))
3321 return -EFAULT;
3322 if (!tipc_node_get_id(net, nr.peer, nr.node_id))
3323 return -EADDRNOTAVAIL;
3324 if (copy_to_user(argp, &nr, sizeof(nr)))
3325 return -EFAULT;
3326 return 0;
78acb1f9
EH		 3327 default:
3328 return -ENOIOCTLCMD;
3329 }
3330}
3331
70b03759
EH		 3332static int tipc_socketpair(struct socket *sock1, struct socket *sock2)
3333{
3334 struct tipc_sock *tsk2 = tipc_sk(sock2->sk);
3335 struct tipc_sock *tsk1 = tipc_sk(sock1->sk);
66bc1e8d
EH		 3336 u32 onode = tipc_own_addr(sock_net(sock1->sk));
3337
3338 tsk1->peer.family = AF_TIPC;
b6f88d9c 3339 tsk1->peer.addrtype = TIPC_SOCKET_ADDR;
66bc1e8d
EH		 3340 tsk1->peer.scope = TIPC_NODE_SCOPE;
3341 tsk1->peer.addr.id.ref = tsk2->portid;
3342 tsk1->peer.addr.id.node = onode;
3343 tsk2->peer.family = AF_TIPC;
b6f88d9c 3344 tsk2->peer.addrtype = TIPC_SOCKET_ADDR;
66bc1e8d
EH		 3345 tsk2->peer.scope = TIPC_NODE_SCOPE;
3346 tsk2->peer.addr.id.ref = tsk1->portid;
3347 tsk2->peer.addr.id.node = onode;
3348
3349 tipc_sk_finish_conn(tsk1, tsk2->portid, onode);
3350 tipc_sk_finish_conn(tsk2, tsk1->portid, onode);
70b03759
EH		 3351 return 0;
3352}
3353
ae86b9e3
BH		 3354/* Protocol switches for the various types of TIPC sockets */
3355
bca65eae 3356static const struct proto_ops msg_ops = {
0e65967e 3357 .owner = THIS_MODULE,
b97bf3fd 3358 .family = AF_TIPC,
247f0f3c
YX		 3359 .release = tipc_release,
3360 .bind = tipc_bind,
3361 .connect = tipc_connect,
66bc1e8d 3362 .socketpair = tipc_socketpair,
245f3d34 3363 .accept = sock_no_accept,
247f0f3c 3364 .getname = tipc_getname,
a11e1d43 3365 .poll = tipc_poll,
78acb1f9 3366 .ioctl = tipc_ioctl,
245f3d34 3367 .listen = sock_no_listen,
247f0f3c
YX		 3368 .shutdown = tipc_shutdown,
3369 .setsockopt = tipc_setsockopt,
3370 .getsockopt = tipc_getsockopt,
3371 .sendmsg = tipc_sendmsg,
3372 .recvmsg = tipc_recvmsg,
8238745a 3373 .mmap = sock_no_mmap,
b97bf3fd
PL		 3374};
3375
bca65eae 3376static const struct proto_ops packet_ops = {
0e65967e 3377 .owner = THIS_MODULE,
b97bf3fd 3378 .family = AF_TIPC,
247f0f3c
YX		 3379 .release = tipc_release,
3380 .bind = tipc_bind,
3381 .connect = tipc_connect,
70b03759 3382 .socketpair = tipc_socketpair,
247f0f3c
YX		 3383 .accept = tipc_accept,
3384 .getname = tipc_getname,
a11e1d43 3385 .poll = tipc_poll,
78acb1f9 3386 .ioctl = tipc_ioctl,
247f0f3c
YX		 3387 .listen = tipc_listen,
3388 .shutdown = tipc_shutdown,
3389 .setsockopt = tipc_setsockopt,
3390 .getsockopt = tipc_getsockopt,
3391 .sendmsg = tipc_send_packet,
3392 .recvmsg = tipc_recvmsg,
8238745a 3393 .mmap = sock_no_mmap,
b97bf3fd
PL		 3394};
3395
bca65eae 3396static const struct proto_ops stream_ops = {
0e65967e 3397 .owner = THIS_MODULE,
b97bf3fd 3398 .family = AF_TIPC,
247f0f3c
YX		 3399 .release = tipc_release,
3400 .bind = tipc_bind,
3401 .connect = tipc_connect,
70b03759 3402 .socketpair = tipc_socketpair,
247f0f3c
YX		 3403 .accept = tipc_accept,
3404 .getname = tipc_getname,
a11e1d43 3405 .poll = tipc_poll,
78acb1f9 3406 .ioctl = tipc_ioctl,
247f0f3c
YX		 3407 .listen = tipc_listen,
3408 .shutdown = tipc_shutdown,
3409 .setsockopt = tipc_setsockopt,
3410 .getsockopt = tipc_getsockopt,
365ad353 3411 .sendmsg = tipc_sendstream,
ec8a09fb 3412 .recvmsg = tipc_recvstream,
8238745a 3413 .mmap = sock_no_mmap,
b97bf3fd
PL		 3414};
3415
bca65eae 3416static const struct net_proto_family tipc_family_ops = {
0e65967e 3417 .owner = THIS_MODULE,
b97bf3fd 3418 .family = AF_TIPC,
c5fa7b3c 3419 .create = tipc_sk_create
b97bf3fd
PL		 3420};
3421
3422static struct proto tipc_proto = {
3423 .name = "TIPC",
3424 .owner = THIS_MODULE,
cc79dd1b
YX		 3425 .obj_size = sizeof(struct tipc_sock),
3426 .sysctl_rmem = sysctl_tipc_rmem
b97bf3fd
PL		 3427};
3428
3429/**
4323add6 3430 * tipc_socket_init - initialize TIPC socket interface
c4307285 3431 *
637b77fd 3432 * Return: 0 on success, errno otherwise
b97bf3fd 3433 */
4323add6 3434int tipc_socket_init(void)
b97bf3fd
PL		 3435{
3436 int res;
3437
c4307285 3438 res = proto_register(&tipc_proto, 1);
b97bf3fd 3439 if (res) {
2cf8aa19 3440 pr_err("Failed to register TIPC protocol type\n");
b97bf3fd
PL		 3441 goto out;
3442 }
3443
3444 res = sock_register(&tipc_family_ops);
3445 if (res) {
2cf8aa19 3446 pr_err("Failed to register TIPC socket type\n");
b97bf3fd
PL		 3447 proto_unregister(&tipc_proto);
3448 goto out;
3449 }
b97bf3fd
PL		 3450 out:
3451 return res;
3452}
3453
3454/**
4323add6 3455 * tipc_socket_stop - stop TIPC socket interface
b97bf3fd 3456 */
4323add6 3457void tipc_socket_stop(void)
b97bf3fd 3458{
b97bf3fd
PL		 3459 sock_unregister(tipc_family_ops.family);
3460 proto_unregister(&tipc_proto);
3461}
34b78a12
RA		 3462
3463/* Caller should hold socket lock for the passed tipc socket. */
d8182804 3464static int __tipc_nl_add_sk_con(struct sk_buff *skb, struct tipc_sock *tsk)
34b78a12 3465{
14623e00
JM		 3466 u32 peer_node, peer_port;
3467 u32 conn_type, conn_instance;
34b78a12
RA		 3468 struct nlattr *nest;
3469
3470 peer_node = tsk_peer_node(tsk);
3471 peer_port = tsk_peer_port(tsk);
14623e00
JM		 3472 conn_type = msg_nametype(&tsk->phdr);
3473 conn_instance = msg_nameinst(&tsk->phdr);
ae0be8de 3474 nest = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_CON);
517ccc2a
KL		 3475 if (!nest)
3476 return -EMSGSIZE;
34b78a12
RA		 3477
3478 if (nla_put_u32(skb, TIPC_NLA_CON_NODE, peer_node))
3479 goto msg_full;
3480 if (nla_put_u32(skb, TIPC_NLA_CON_SOCK, peer_port))
3481 goto msg_full;
3482
14623e00 3483 if (tsk->conn_addrtype != 0) {
34b78a12
RA		 3484 if (nla_put_flag(skb, TIPC_NLA_CON_FLAG))
3485 goto msg_full;
14623e00 3486 if (nla_put_u32(skb, TIPC_NLA_CON_TYPE, conn_type))
34b78a12 3487 goto msg_full;
14623e00 3488 if (nla_put_u32(skb, TIPC_NLA_CON_INST, conn_instance))
34b78a12
RA		 3489 goto msg_full;
3490 }
3491 nla_nest_end(skb, nest);
3492
3493 return 0;
3494
3495msg_full:
3496 nla_nest_cancel(skb, nest);
3497
3498 return -EMSGSIZE;
3499}
3500
dfde331e
GM		 3501static int __tipc_nl_add_sk_info(struct sk_buff *skb, struct tipc_sock
3502 *tsk)
3503{
3504 struct net *net = sock_net(skb->sk);
dfde331e
GM		 3505 struct sock *sk = &tsk->sk;
3506
3507 if (nla_put_u32(skb, TIPC_NLA_SOCK_REF, tsk->portid) ||
23fd3eac 3508 nla_put_u32(skb, TIPC_NLA_SOCK_ADDR, tipc_own_addr(net)))
dfde331e
GM		 3509 return -EMSGSIZE;
3510
3511 if (tipc_sk_connected(sk)) {
3512 if (__tipc_nl_add_sk_con(skb, tsk))
3513 return -EMSGSIZE;
3514 } else if (!list_empty(&tsk->publications)) {
3515 if (nla_put_flag(skb, TIPC_NLA_SOCK_HAS_PUBL))
3516 return -EMSGSIZE;
3517 }
3518 return 0;
3519}
3520
34b78a12 3521/* Caller should hold socket lock for the passed tipc socket. */
d8182804
RA		 3522static int __tipc_nl_add_sk(struct sk_buff *skb, struct netlink_callback *cb,
3523 struct tipc_sock *tsk)
34b78a12 3524{
34b78a12 3525 struct nlattr *attrs;
dfde331e 3526 void *hdr;
34b78a12
RA		 3527
3528 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
bfb3e5dd 3529 &tipc_genl_family, NLM_F_MULTI, TIPC_NL_SOCK_GET);
34b78a12
RA		 3530 if (!hdr)
3531 goto msg_cancel;
3532
ae0be8de 3533 attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK);
34b78a12
RA		 3534 if (!attrs)
3535 goto genlmsg_cancel;
dfde331e
GM		 3536
3537 if (__tipc_nl_add_sk_info(skb, tsk))
34b78a12
RA		 3538 goto attr_msg_cancel;
3539
34b78a12
RA		 3540 nla_nest_end(skb, attrs);
3541 genlmsg_end(skb, hdr);
3542
3543 return 0;
3544
3545attr_msg_cancel:
3546 nla_nest_cancel(skb, attrs);
3547genlmsg_cancel:
3548 genlmsg_cancel(skb, hdr);
3549msg_cancel:
3550 return -EMSGSIZE;
3551}
3552
c30b70de
GM		 3553int tipc_nl_sk_walk(struct sk_buff *skb, struct netlink_callback *cb,
3554 int (*skb_handler)(struct sk_buff *skb,
3555 struct netlink_callback *cb,
3556 struct tipc_sock *tsk))
34b78a12 3557{
8f5c5fcf 3558 struct rhashtable_iter *iter = (void *)cb->args[4];
dfde331e
GM		 3559 struct tipc_sock *tsk;
3560 int err;
34b78a12 3561
9a07efa9
CW		 3562 rhashtable_walk_start(iter);
3563 while ((tsk = rhashtable_walk_next(iter)) != NULL) {
3564 if (IS_ERR(tsk)) {
195b7fc5 3565 if (PTR_ERR(tsk) == -EAGAIN)
d6e164e3 3566 continue;
9a07efa9
CW		 3567 break;
3568 }
d6e164e3 3569
9a07efa9
CW		 3570 sock_hold(&tsk->sk);
3571 rhashtable_walk_stop(iter);
3572 lock_sock(&tsk->sk);
3573 err = skb_handler(skb, cb, tsk);
3574 if (err) {
3575 release_sock(&tsk->sk);
3576 sock_put(&tsk->sk);
3577 goto out;
07f6c4bc 3578 }
9a07efa9
CW		 3579 release_sock(&tsk->sk);
3580 rhashtable_walk_start(iter);
3581 sock_put(&tsk->sk);
34b78a12 3582 }
9a07efa9 3583 rhashtable_walk_stop(iter);
d6e164e3 3584out:
34b78a12
RA		 3585 return skb->len;
3586}
c30b70de
GM		 3587EXPORT_SYMBOL(tipc_nl_sk_walk);
3588
9a07efa9
CW		 3589int tipc_dump_start(struct netlink_callback *cb)
3590{
8f5c5fcf
CW		 3591 return __tipc_dump_start(cb, sock_net(cb->skb->sk));
3592}
3593EXPORT_SYMBOL(tipc_dump_start);
3594
3595int __tipc_dump_start(struct netlink_callback *cb, struct net *net)
3596{
3597 /* tipc_nl_name_table_dump() uses cb->args[0...3]. */
3598 struct rhashtable_iter *iter = (void *)cb->args[4];
9a07efa9
CW		 3599 struct tipc_net *tn = tipc_net(net);
3600
3601 if (!iter) {
3602 iter = kmalloc(sizeof(*iter), GFP_KERNEL);
3603 if (!iter)
3604 return -ENOMEM;
3605
8f5c5fcf 3606 cb->args[4] = (long)iter;
9a07efa9
CW		 3607 }
3608
3609 rhashtable_walk_enter(&tn->sk_rht, iter);
3610 return 0;
3611}
9a07efa9
CW		 3612
3613int tipc_dump_done(struct netlink_callback *cb)
3614{
8f5c5fcf 3615 struct rhashtable_iter *hti = (void *)cb->args[4];
9a07efa9
CW		 3616
3617 rhashtable_walk_exit(hti);
3618 kfree(hti);
3619 return 0;
3620}
3621EXPORT_SYMBOL(tipc_dump_done);
3622
e41f0548
CW		 3623int tipc_sk_fill_sock_diag(struct sk_buff *skb, struct netlink_callback *cb,
3624 struct tipc_sock *tsk, u32 sk_filter_state,
c30b70de
GM		 3625 u64 (*tipc_diag_gen_cookie)(struct sock *sk))
3626{
3627 struct sock *sk = &tsk->sk;
3628 struct nlattr *attrs;
3629 struct nlattr *stat;
3630
3631 /*filter response w.r.t sk_state*/
3632 if (!(sk_filter_state & (1 << sk->sk_state)))
3633 return 0;
3634
ae0be8de 3635 attrs = nla_nest_start_noflag(skb, TIPC_NLA_SOCK);
c30b70de
GM		 3636 if (!attrs)
3637 goto msg_cancel;
3638
3639 if (__tipc_nl_add_sk_info(skb, tsk))
3640 goto attr_msg_cancel;
3641
3642 if (nla_put_u32(skb, TIPC_NLA_SOCK_TYPE, (u32)sk->sk_type) ||
3643 nla_put_u32(skb, TIPC_NLA_SOCK_TIPC_STATE, (u32)sk->sk_state) ||
3644 nla_put_u32(skb, TIPC_NLA_SOCK_INO, sock_i_ino(sk)) ||
3645 nla_put_u32(skb, TIPC_NLA_SOCK_UID,
e41f0548 3646 from_kuid_munged(sk_user_ns(NETLINK_CB(cb->skb).sk),
4b2e6877 3647 sock_i_uid(sk))) ||
c30b70de
GM		 3648 nla_put_u64_64bit(skb, TIPC_NLA_SOCK_COOKIE,
3649 tipc_diag_gen_cookie(sk),
3650 TIPC_NLA_SOCK_PAD))
3651 goto attr_msg_cancel;
3652
ae0be8de 3653 stat = nla_nest_start_noflag(skb, TIPC_NLA_SOCK_STAT);
c30b70de
GM		 3654 if (!stat)
3655 goto attr_msg_cancel;
3656
3657 if (nla_put_u32(skb, TIPC_NLA_SOCK_STAT_RCVQ,
3658 skb_queue_len(&sk->sk_receive_queue)) ||
3659 nla_put_u32(skb, TIPC_NLA_SOCK_STAT_SENDQ,
872619d8
GM		 3660 skb_queue_len(&sk->sk_write_queue)) ||
3661 nla_put_u32(skb, TIPC_NLA_SOCK_STAT_DROP,
3662 atomic_read(&sk->sk_drops)))
c30b70de
GM		 3663 goto stat_msg_cancel;
3664
3665 if (tsk->cong_link_cnt &&
3666 nla_put_flag(skb, TIPC_NLA_SOCK_STAT_LINK_CONG))
3667 goto stat_msg_cancel;
3668
3669 if (tsk_conn_cong(tsk) &&
3670 nla_put_flag(skb, TIPC_NLA_SOCK_STAT_CONN_CONG))
3671 goto stat_msg_cancel;
3672
3673 nla_nest_end(skb, stat);
a1be5a20
GM		 3674
3675 if (tsk->group)
3676 if (tipc_group_fill_sock_diag(tsk->group, skb))
3677 goto stat_msg_cancel;
3678
c30b70de
GM		 3679 nla_nest_end(skb, attrs);
3680
3681 return 0;
3682
3683stat_msg_cancel:
3684 nla_nest_cancel(skb, stat);
3685attr_msg_cancel:
3686 nla_nest_cancel(skb, attrs);
3687msg_cancel:
3688 return -EMSGSIZE;
3689}
3690EXPORT_SYMBOL(tipc_sk_fill_sock_diag);
1a1a143d 3691
dfde331e
GM		 3692int tipc_nl_sk_dump(struct sk_buff *skb, struct netlink_callback *cb)
3693{
c30b70de 3694 return tipc_nl_sk_walk(skb, cb, __tipc_nl_add_sk);
dfde331e
GM		 3695}
3696
1a1a143d 3697/* Caller should hold socket lock for the passed tipc socket. */
d8182804
RA		 3698static int __tipc_nl_add_sk_publ(struct sk_buff *skb,
3699 struct netlink_callback *cb,
3700 struct publication *publ)
1a1a143d
RA		 3701{
3702 void *hdr;
3703 struct nlattr *attrs;
3704
3705 hdr = genlmsg_put(skb, NETLINK_CB(cb->skb).portid, cb->nlh->nlmsg_seq,
bfb3e5dd 3706 &tipc_genl_family, NLM_F_MULTI, TIPC_NL_PUBL_GET);
1a1a143d
RA		 3707 if (!hdr)
3708 goto msg_cancel;
3709
ae0be8de 3710 attrs = nla_nest_start_noflag(skb, TIPC_NLA_PUBL);
1a1a143d
RA		 3711 if (!attrs)
3712 goto genlmsg_cancel;
3713
3714 if (nla_put_u32(skb, TIPC_NLA_PUBL_KEY, publ->key))
3715 goto attr_msg_cancel;
998d3907 3716 if (nla_put_u32(skb, TIPC_NLA_PUBL_TYPE, publ->sr.type))
1a1a143d 3717 goto attr_msg_cancel;
998d3907 3718 if (nla_put_u32(skb, TIPC_NLA_PUBL_LOWER, publ->sr.lower))
1a1a143d 3719 goto attr_msg_cancel;
998d3907 3720 if (nla_put_u32(skb, TIPC_NLA_PUBL_UPPER, publ->sr.upper))
1a1a143d
RA		 3721 goto attr_msg_cancel;
3722
3723 nla_nest_end(skb, attrs);
3724 genlmsg_end(skb, hdr);
3725
3726 return 0;
3727
3728attr_msg_cancel:
3729 nla_nest_cancel(skb, attrs);
3730genlmsg_cancel:
3731 genlmsg_cancel(skb, hdr);
3732msg_cancel:
3733 return -EMSGSIZE;
3734}
3735
3736/* Caller should hold socket lock for the passed tipc socket. */
d8182804
RA		 3737static int __tipc_nl_list_sk_publ(struct sk_buff *skb,
3738 struct netlink_callback *cb,
3739 struct tipc_sock *tsk, u32 *last_publ)
1a1a143d
RA		 3740{
3741 int err;
3742 struct publication *p;
3743
3744 if (*last_publ) {
e50e73e1 3745 list_for_each_entry(p, &tsk->publications, binding_sock) {
1a1a143d
RA		 3746 if (p->key == *last_publ)
3747 break;
3748 }
a1f8fec4 3749 if (list_entry_is_head(p, &tsk->publications, binding_sock)) {
1a1a143d
RA		 3750 /* We never set seq or call nl_dump_check_consistent()
3751 * this means that setting prev_seq here will cause the
3752 * consistence check to fail in the netlink callback
3753 * handler. Resulting in the last NLMSG_DONE message
3754 * having the NLM_F_DUMP_INTR flag set.
3755 */
3756 cb->prev_seq = 1;
3757 *last_publ = 0;
3758 return -EPIPE;
3759 }
3760 } else {
3761 p = list_first_entry(&tsk->publications, struct publication,
e50e73e1 3762 binding_sock);
1a1a143d
RA		 3763 }
3764
e50e73e1 3765 list_for_each_entry_from(p, &tsk->publications, binding_sock) {
1a1a143d
RA		 3766 err = __tipc_nl_add_sk_publ(skb, cb, p);
3767 if (err) {
3768 *last_publ = p->key;
3769 return err;
3770 }
3771 }
3772 *last_publ = 0;
3773
3774 return 0;
3775}
3776
3777int tipc_nl_publ_dump(struct sk_buff *skb, struct netlink_callback *cb)
3778{
3779 int err;
07f6c4bc 3780 u32 tsk_portid = cb->args[0];
1a1a143d
RA		 3781 u32 last_publ = cb->args[1];
3782 u32 done = cb->args[2];
e05b31f4 3783 struct net *net = sock_net(skb->sk);
1a1a143d
RA		 3784 struct tipc_sock *tsk;
3785
07f6c4bc 3786 if (!tsk_portid) {
7288dd2f 3787 struct nlattr **attrs = genl_dumpit_info(cb)->info.attrs;
1a1a143d
RA		 3788 struct nlattr *sock[TIPC_NLA_SOCK_MAX + 1];
3789
45e093ae
RA		 3790 if (!attrs[TIPC_NLA_SOCK])
3791 return -EINVAL;
3792
8cb08174
JB		 3793 err = nla_parse_nested_deprecated(sock, TIPC_NLA_SOCK_MAX,
3794 attrs[TIPC_NLA_SOCK],
3795 tipc_nl_sock_policy, NULL);
1a1a143d
RA		 3796 if (err)
3797 return err;
3798
3799 if (!sock[TIPC_NLA_SOCK_REF])
3800 return -EINVAL;
3801
07f6c4bc 3802 tsk_portid = nla_get_u32(sock[TIPC_NLA_SOCK_REF]);
1a1a143d
RA		 3803 }
3804
3805 if (done)
3806 return 0;
3807
e05b31f4 3808 tsk = tipc_sk_lookup(net, tsk_portid);
1a1a143d
RA		 3809 if (!tsk)
3810 return -EINVAL;
3811
3812 lock_sock(&tsk->sk);
3813 err = __tipc_nl_list_sk_publ(skb, cb, tsk, &last_publ);
3814 if (!err)
3815 done = 1;
3816 release_sock(&tsk->sk);
07f6c4bc 3817 sock_put(&tsk->sk);
1a1a143d 3818
07f6c4bc 3819 cb->args[0] = tsk_portid;
1a1a143d
RA		 3820 cb->args[1] = last_publ;
3821 cb->args[2] = done;
3822
3823 return skb->len;
3824}
b4b9771b 3825
01e661eb
TL		 3826/**
3827 * tipc_sk_filtering - check if a socket should be traced
3828 * @sk: the socket to be examined
f172f4b8
RD		 3829 *
3830 * @sysctl_tipc_sk_filter is used as the socket tuple for filtering:
5fcb7d47 3831 * (portid, sock type, name type, name lower, name upper)
01e661eb 3832 *
637b77fd 3833 * Return: true if the socket meets the socket tuple data
01e661eb
TL		 3834 * (value 0 = 'any') or when there is no tuple set (all = 0),
3835 * otherwise false
3836 */
3837bool tipc_sk_filtering(struct sock *sk)
3838{
3839 struct tipc_sock *tsk;
3840 struct publication *p;
3841 u32 _port, _sktype, _type, _lower, _upper;
3842 u32 type = 0, lower = 0, upper = 0;
3843
3844 if (!sk)
3845 return true;
3846
3847 tsk = tipc_sk(sk);
3848
3849 _port = sysctl_tipc_sk_filter[0];
3850 _sktype = sysctl_tipc_sk_filter[1];
3851 _type = sysctl_tipc_sk_filter[2];
3852 _lower = sysctl_tipc_sk_filter[3];
3853 _upper = sysctl_tipc_sk_filter[4];
3854
3855 if (!_port && !_sktype && !_type && !_lower && !_upper)
3856 return true;
3857
3858 if (_port)
3859 return (_port == tsk->portid);
3860
3861 if (_sktype && _sktype != sk->sk_type)
3862 return false;
3863
3864 if (tsk->published) {
3865 p = list_first_entry_or_null(&tsk->publications,
3866 struct publication, binding_sock);
3867 if (p) {
998d3907
JM		 3868 type = p->sr.type;
3869 lower = p->sr.lower;
3870 upper = p->sr.upper;
01e661eb
TL		 3871 }
3872 }
3873
3874 if (!tipc_sk_type_connectionless(sk)) {
14623e00
JM		 3875 type = msg_nametype(&tsk->phdr);
3876 lower = msg_nameinst(&tsk->phdr);
3877 upper = lower;
01e661eb
TL		 3878 }
3879
3880 if ((_type && _type != type) || (_lower && _lower != lower) ||
3881 (_upper && _upper != upper))
3882 return false;
3883
3884 return true;
3885}
3886
b4b9771b
TL		 3887u32 tipc_sock_get_portid(struct sock *sk)
3888{
3889 return (sk) ? (tipc_sk(sk))->portid : 0;
3890}
3891
01e661eb
TL		 3892/**
3893 * tipc_sk_overlimit1 - check if socket rx queue is about to be overloaded,
3894 * both the rcv and backlog queues are considered
3895 * @sk: tipc sk to be checked
3896 * @skb: tipc msg to be checked
3897 *
637b77fd 3898 * Return: true if the socket rx queue allocation is > 90%, otherwise false
01e661eb
TL		 3899 */
3900
3901bool tipc_sk_overlimit1(struct sock *sk, struct sk_buff *skb)
3902{
3903 atomic_t *dcnt = &tipc_sk(sk)->dupl_rcvcnt;
3904 unsigned int lim = rcvbuf_limit(sk, skb) + atomic_read(dcnt);
3905 unsigned int qsize = sk->sk_backlog.len + sk_rmem_alloc_get(sk);
3906
3907 return (qsize > lim * 90 / 100);
3908}
3909
3910/**
3911 * tipc_sk_overlimit2 - check if socket rx queue is about to be overloaded,
3912 * only the rcv queue is considered
3913 * @sk: tipc sk to be checked
3914 * @skb: tipc msg to be checked
3915 *
637b77fd 3916 * Return: true if the socket rx queue allocation is > 90%, otherwise false
01e661eb
TL		 3917 */
3918
3919bool tipc_sk_overlimit2(struct sock *sk, struct sk_buff *skb)
3920{
3921 unsigned int lim = rcvbuf_limit(sk, skb);
3922 unsigned int qsize = sk_rmem_alloc_get(sk);
3923
3924 return (qsize > lim * 90 / 100);
3925}
3926
b4b9771b
TL		 3927/**
3928 * tipc_sk_dump - dump TIPC socket
3929 * @sk: tipc sk to be dumped
3930 * @dqueues: bitmask to decide if any socket queue to be dumped?
3931 * - TIPC_DUMP_NONE: don't dump socket queues
3932 * - TIPC_DUMP_SK_SNDQ: dump socket send queue
3933 * - TIPC_DUMP_SK_RCVQ: dump socket rcv queue
3934 * - TIPC_DUMP_SK_BKLGQ: dump socket backlog queue
3935 * - TIPC_DUMP_ALL: dump all the socket queues above
3936 * @buf: returned buffer of dump data in format
3937 */
3938int tipc_sk_dump(struct sock *sk, u16 dqueues, char *buf)
3939{
3940 int i = 0;
3941 size_t sz = (dqueues) ? SK_LMAX : SK_LMIN;
14623e00 3942 u32 conn_type, conn_instance;
b4b9771b
TL		 3943 struct tipc_sock *tsk;
3944 struct publication *p;
3945 bool tsk_connected;
3946
3947 if (!sk) {
3948 i += scnprintf(buf, sz, "sk data: (null)\n");
3949 return i;
3950 }
3951
3952 tsk = tipc_sk(sk);
3953 tsk_connected = !tipc_sk_type_connectionless(sk);
3954
3955 i += scnprintf(buf, sz, "sk data: %u", sk->sk_type);
3956 i += scnprintf(buf + i, sz - i, " %d", sk->sk_state);
3957 i += scnprintf(buf + i, sz - i, " %x", tsk_own_node(tsk));
3958 i += scnprintf(buf + i, sz - i, " %u", tsk->portid);
3959 i += scnprintf(buf + i, sz - i, " | %u", tsk_connected);
3960 if (tsk_connected) {
3961 i += scnprintf(buf + i, sz - i, " %x", tsk_peer_node(tsk));
3962 i += scnprintf(buf + i, sz - i, " %u", tsk_peer_port(tsk));
14623e00
JM		 3963 conn_type = msg_nametype(&tsk->phdr);
3964 conn_instance = msg_nameinst(&tsk->phdr);
3965 i += scnprintf(buf + i, sz - i, " %u", conn_type);
3966 i += scnprintf(buf + i, sz - i, " %u", conn_instance);
b4b9771b
TL		 3967 }
3968 i += scnprintf(buf + i, sz - i, " | %u", tsk->published);
3969 if (tsk->published) {
3970 p = list_first_entry_or_null(&tsk->publications,
3971 struct publication, binding_sock);
998d3907
JM		 3972 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->sr.type : 0);
3973 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->sr.lower : 0);
3974 i += scnprintf(buf + i, sz - i, " %u", (p) ? p->sr.upper : 0);
b4b9771b
TL		 3975 }
3976 i += scnprintf(buf + i, sz - i, " | %u", tsk->snd_win);
3977 i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_win);
3978 i += scnprintf(buf + i, sz - i, " %u", tsk->max_pkt);
3979 i += scnprintf(buf + i, sz - i, " %x", tsk->peer_caps);
3980 i += scnprintf(buf + i, sz - i, " %u", tsk->cong_link_cnt);
3981 i += scnprintf(buf + i, sz - i, " %u", tsk->snt_unacked);
3982 i += scnprintf(buf + i, sz - i, " %u", tsk->rcv_unacked);
3983 i += scnprintf(buf + i, sz - i, " %u", atomic_read(&tsk->dupl_rcvcnt));
3984 i += scnprintf(buf + i, sz - i, " %u", sk->sk_shutdown);
3985 i += scnprintf(buf + i, sz - i, " | %d", sk_wmem_alloc_get(sk));
3986 i += scnprintf(buf + i, sz - i, " %d", sk->sk_sndbuf);
3987 i += scnprintf(buf + i, sz - i, " | %d", sk_rmem_alloc_get(sk));
3988 i += scnprintf(buf + i, sz - i, " %d", sk->sk_rcvbuf);
70c26558 3989 i += scnprintf(buf + i, sz - i, " | %d\n", READ_ONCE(sk->sk_backlog.len));
b4b9771b
TL		 3990
3991 if (dqueues & TIPC_DUMP_SK_SNDQ) {
3992 i += scnprintf(buf + i, sz - i, "sk_write_queue: ");
3993 i += tipc_list_dump(&sk->sk_write_queue, false, buf + i);
3994 }
3995
3996 if (dqueues & TIPC_DUMP_SK_RCVQ) {
3997 i += scnprintf(buf + i, sz - i, "sk_receive_queue: ");
3998 i += tipc_list_dump(&sk->sk_receive_queue, false, buf + i);
3999 }
4000
4001 if (dqueues & TIPC_DUMP_SK_BKLGQ) {
4002 i += scnprintf(buf + i, sz - i, "sk_backlog:\n head ");
4003 i += tipc_skb_dump(sk->sk_backlog.head, false, buf + i);
4004 if (sk->sk_backlog.tail != sk->sk_backlog.head) {
4005 i += scnprintf(buf + i, sz - i, " tail ");
4006 i += tipc_skb_dump(sk->sk_backlog.tail, false,
4007 buf + i);
4008 }
4009 }
4010
4011 return i;
4012}
Linux 6.x block layer and io_uring tree(s)