sctp: reduce indent level on sctp_v4_get_dst
[linux-2.6-block.git] / net / sctp / protocol.c
CommitLineData
60c778b2 1/* SCTP kernel implementation
1da177e4
LT
2 * (C) Copyright IBM Corp. 2001, 2004
3 * Copyright (c) 1999-2000 Cisco, Inc.
4 * Copyright (c) 1999-2001 Motorola, Inc.
5 * Copyright (c) 2001 Intel Corp.
6 * Copyright (c) 2001 Nokia, Inc.
7 * Copyright (c) 2001 La Monte H.P. Yarroll
8 *
60c778b2 9 * This file is part of the SCTP kernel implementation
1da177e4
LT
10 *
11 * Initialization/cleanup for SCTP protocol support.
12 *
60c778b2 13 * This SCTP implementation is free software;
1da177e4
LT
14 * you can redistribute it and/or modify it under the terms of
15 * the GNU General Public License as published by
16 * the Free Software Foundation; either version 2, or (at your option)
17 * any later version.
18 *
60c778b2 19 * This SCTP implementation is distributed in the hope that it
1da177e4
LT
20 * will be useful, but WITHOUT ANY WARRANTY; without even the implied
21 * ************************
22 * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
23 * See the GNU General Public License for more details.
24 *
25 * You should have received a copy of the GNU General Public License
4b2f13a2
JK
26 * along with GNU CC; see the file COPYING. If not, see
27 * <http://www.gnu.org/licenses/>.
1da177e4
LT
28 *
29 * Please send any bug reports or fixes you make to the
30 * email address(es):
91705c61 31 * lksctp developers <linux-sctp@vger.kernel.org>
1da177e4 32 *
1da177e4
LT
33 * Written or modified by:
34 * La Monte H.P. Yarroll <piggy@acm.org>
35 * Karl Knutson <karl@athena.chicago.il.us>
36 * Jon Grimm <jgrimm@us.ibm.com>
37 * Sridhar Samudrala <sri@us.ibm.com>
38 * Daisy Chang <daisyc@us.ibm.com>
39 * Ardelle Fan <ardelle.fan@intel.com>
1da177e4
LT
40 */
41
145ce502
JP
42#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
43
1da177e4
LT
44#include <linux/module.h>
45#include <linux/init.h>
46#include <linux/netdevice.h>
47#include <linux/inetdevice.h>
48#include <linux/seq_file.h>
4d93df0a 49#include <linux/bootmem.h>
845525a6
VY
50#include <linux/highmem.h>
51#include <linux/swap.h>
5a0e3ad6 52#include <linux/slab.h>
457c4cbc 53#include <net/net_namespace.h>
1da177e4
LT
54#include <net/protocol.h>
55#include <net/ip.h>
56#include <net/ipv6.h>
14c85021 57#include <net/route.h>
1da177e4
LT
58#include <net/sctp/sctp.h>
59#include <net/addrconf.h>
60#include <net/inet_common.h>
61#include <net/inet_ecn.h>
62
63/* Global data structures. */
4cbf1cae 64struct sctp_globals sctp_globals __read_mostly;
1da177e4
LT
65
66struct idr sctp_assocs_id;
67DEFINE_SPINLOCK(sctp_assocs_id_lock);
68
1da177e4
LT
69static struct sctp_pf *sctp_pf_inet6_specific;
70static struct sctp_pf *sctp_pf_inet_specific;
71static struct sctp_af *sctp_af_v4_specific;
72static struct sctp_af *sctp_af_v6_specific;
73
e18b890b
CL
74struct kmem_cache *sctp_chunk_cachep __read_mostly;
75struct kmem_cache *sctp_bucket_cachep __read_mostly;
1da177e4 76
8d987e5c 77long sysctl_sctp_mem[3];
007e3936
VY
78int sysctl_sctp_rmem[3];
79int sysctl_sctp_wmem[3];
4d93df0a 80
1da177e4 81/* Set up the proc fs entry for the SCTP protocol. */
f5f417c0 82static int __net_init sctp_proc_init(struct net *net)
1da177e4 83{
c4e85f82 84#ifdef CONFIG_PROC_FS
13d782f6
EB
85 net->sctp.proc_net_sctp = proc_net_mkdir(net, "sctp", net->proc_net);
86 if (!net->sctp.proc_net_sctp)
87 goto out_proc_net_sctp;
88 if (sctp_snmp_proc_init(net))
80896a35 89 goto out_snmp_proc_init;
13d782f6 90 if (sctp_eps_proc_init(net))
80896a35 91 goto out_eps_proc_init;
13d782f6 92 if (sctp_assocs_proc_init(net))
80896a35 93 goto out_assocs_proc_init;
13d782f6 94 if (sctp_remaddr_proc_init(net))
caea902f 95 goto out_remaddr_proc_init;
1da177e4
LT
96
97 return 0;
98
caea902f 99out_remaddr_proc_init:
13d782f6 100 sctp_assocs_proc_exit(net);
80896a35 101out_assocs_proc_init:
13d782f6 102 sctp_eps_proc_exit(net);
80896a35 103out_eps_proc_init:
13d782f6 104 sctp_snmp_proc_exit(net);
80896a35 105out_snmp_proc_init:
13d782f6
EB
106 remove_proc_entry("sctp", net->proc_net);
107 net->sctp.proc_net_sctp = NULL;
108out_proc_net_sctp:
157653ce 109 return -ENOMEM;
13d782f6
EB
110#endif /* CONFIG_PROC_FS */
111 return 0;
1da177e4
LT
112}
113
d808ad9a 114/* Clean up the proc fs entry for the SCTP protocol.
1da177e4
LT
115 * Note: Do not make this __exit as it is used in the init error
116 * path.
117 */
13d782f6 118static void sctp_proc_exit(struct net *net)
1da177e4 119{
c4e85f82 120#ifdef CONFIG_PROC_FS
13d782f6
EB
121 sctp_snmp_proc_exit(net);
122 sctp_eps_proc_exit(net);
123 sctp_assocs_proc_exit(net);
124 sctp_remaddr_proc_exit(net);
125
126 remove_proc_entry("sctp", net->proc_net);
127 net->sctp.proc_net_sctp = NULL;
c4e85f82 128#endif
1da177e4
LT
129}
130
131/* Private helper to extract ipv4 address and stash them in
132 * the protocol structure.
133 */
134static void sctp_v4_copy_addrlist(struct list_head *addrlist,
135 struct net_device *dev)
136{
137 struct in_device *in_dev;
138 struct in_ifaddr *ifa;
139 struct sctp_sockaddr_entry *addr;
140
141 rcu_read_lock();
e5ed6399 142 if ((in_dev = __in_dev_get_rcu(dev)) == NULL) {
1da177e4
LT
143 rcu_read_unlock();
144 return;
145 }
146
147 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
148 /* Add the address to the local list. */
939cfa75 149 addr = kzalloc(sizeof(*addr), GFP_ATOMIC);
1da177e4 150 if (addr) {
2a6fd78a
AV
151 addr->a.v4.sin_family = AF_INET;
152 addr->a.v4.sin_port = 0;
153 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
29303547
VY
154 addr->valid = 1;
155 INIT_LIST_HEAD(&addr->list);
1da177e4
LT
156 list_add_tail(&addr->list, addrlist);
157 }
158 }
159
160 rcu_read_unlock();
161}
162
163/* Extract our IP addresses from the system and stash them in the
164 * protocol structure.
165 */
4db67e80 166static void sctp_get_local_addr_list(struct net *net)
1da177e4
LT
167{
168 struct net_device *dev;
169 struct list_head *pos;
170 struct sctp_af *af;
171
c6d14c84 172 rcu_read_lock();
4db67e80 173 for_each_netdev_rcu(net, dev) {
2c0740e4 174 list_for_each(pos, &sctp_address_families) {
1da177e4 175 af = list_entry(pos, struct sctp_af, list);
4db67e80 176 af->copy_addrlist(&net->sctp.local_addr_list, dev);
1da177e4
LT
177 }
178 }
c6d14c84 179 rcu_read_unlock();
1da177e4
LT
180}
181
1da177e4 182/* Free the existing local addresses. */
4db67e80 183static void sctp_free_local_addr_list(struct net *net)
1da177e4
LT
184{
185 struct sctp_sockaddr_entry *addr;
186 struct list_head *pos, *temp;
187
4db67e80 188 list_for_each_safe(pos, temp, &net->sctp.local_addr_list) {
1da177e4
LT
189 addr = list_entry(pos, struct sctp_sockaddr_entry, list);
190 list_del(pos);
191 kfree(addr);
192 }
193}
194
1da177e4 195/* Copy the local addresses which are valid for 'scope' into 'bp'. */
4db67e80
EB
196int sctp_copy_local_addr_list(struct net *net, struct sctp_bind_addr *bp,
197 sctp_scope_t scope, gfp_t gfp, int copy_flags)
1da177e4
LT
198{
199 struct sctp_sockaddr_entry *addr;
200 int error = 0;
1da177e4 201
29303547 202 rcu_read_lock();
4db67e80 203 list_for_each_entry_rcu(addr, &net->sctp.local_addr_list, list) {
29303547
VY
204 if (!addr->valid)
205 continue;
e7ff4a70 206 if (sctp_in_scope(net, &addr->a, scope)) {
1da177e4
LT
207 /* Now that the address is in scope, check to see if
208 * the address type is really supported by the local
209 * sock as well as the remote peer.
210 */
6244be4e 211 if ((((AF_INET == addr->a.sa.sa_family) &&
1da177e4 212 (copy_flags & SCTP_ADDR4_PEERSUPP))) ||
6244be4e 213 (((AF_INET6 == addr->a.sa.sa_family) &&
1da177e4
LT
214 (copy_flags & SCTP_ADDR6_ALLOWED) &&
215 (copy_flags & SCTP_ADDR6_PEERSUPP)))) {
f57d96b2
VY
216 error = sctp_add_bind_addr(bp, &addr->a,
217 SCTP_ADDR_SRC, GFP_ATOMIC);
1da177e4
LT
218 if (error)
219 goto end_copy;
220 }
221 }
222 }
223
224end_copy:
29303547 225 rcu_read_unlock();
1da177e4
LT
226 return error;
227}
228
229/* Initialize a sctp_addr from in incoming skb. */
230static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
231 int is_saddr)
232{
233 void *from;
d55c41b1 234 __be16 *port;
1da177e4
LT
235 struct sctphdr *sh;
236
237 port = &addr->v4.sin_port;
238 addr->v4.sin_family = AF_INET;
239
2c0fd387 240 sh = sctp_hdr(skb);
1da177e4 241 if (is_saddr) {
d55c41b1 242 *port = sh->source;
eddc9ec5 243 from = &ip_hdr(skb)->saddr;
1da177e4 244 } else {
d55c41b1 245 *port = sh->dest;
eddc9ec5 246 from = &ip_hdr(skb)->daddr;
1da177e4
LT
247 }
248 memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr));
249}
250
251/* Initialize an sctp_addr from a socket. */
252static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk)
253{
254 addr->v4.sin_family = AF_INET;
7dcdbd95 255 addr->v4.sin_port = 0;
c720c7e8 256 addr->v4.sin_addr.s_addr = inet_sk(sk)->inet_rcv_saddr;
1da177e4
LT
257}
258
259/* Initialize sk->sk_rcv_saddr from sctp_addr. */
260static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk)
261{
c720c7e8 262 inet_sk(sk)->inet_rcv_saddr = addr->v4.sin_addr.s_addr;
1da177e4
LT
263}
264
265/* Initialize sk->sk_daddr from sctp_addr. */
266static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk)
267{
c720c7e8 268 inet_sk(sk)->inet_daddr = addr->v4.sin_addr.s_addr;
1da177e4
LT
269}
270
271/* Initialize a sctp_addr from an address parameter. */
272static void sctp_v4_from_addr_param(union sctp_addr *addr,
273 union sctp_addr_param *param,
dd86d136 274 __be16 port, int iif)
1da177e4
LT
275{
276 addr->v4.sin_family = AF_INET;
277 addr->v4.sin_port = port;
278 addr->v4.sin_addr.s_addr = param->v4.addr.s_addr;
279}
280
281/* Initialize an address parameter from a sctp_addr and return the length
282 * of the address parameter.
283 */
284static int sctp_v4_to_addr_param(const union sctp_addr *addr,
285 union sctp_addr_param *param)
286{
287 int length = sizeof(sctp_ipv4addr_param_t);
288
289 param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS;
dbc16db1 290 param->v4.param_hdr.length = htons(length);
d808ad9a 291 param->v4.addr.s_addr = addr->v4.sin_addr.s_addr;
1da177e4
LT
292
293 return length;
294}
295
296/* Initialize a sctp_addr from a dst_entry. */
18a353f4 297static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct flowi4 *fl4,
854d43a4 298 __be16 port)
1da177e4 299{
1da177e4
LT
300 saddr->v4.sin_family = AF_INET;
301 saddr->v4.sin_port = port;
18a353f4 302 saddr->v4.sin_addr.s_addr = fl4->saddr;
1da177e4
LT
303}
304
305/* Compare two addresses exactly. */
306static int sctp_v4_cmp_addr(const union sctp_addr *addr1,
307 const union sctp_addr *addr2)
308{
309 if (addr1->sa.sa_family != addr2->sa.sa_family)
310 return 0;
311 if (addr1->v4.sin_port != addr2->v4.sin_port)
312 return 0;
313 if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr)
314 return 0;
315
316 return 1;
317}
318
319/* Initialize addr struct to INADDR_ANY. */
6fbfa9f9 320static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port)
1da177e4
LT
321{
322 addr->v4.sin_family = AF_INET;
e6f1cebf 323 addr->v4.sin_addr.s_addr = htonl(INADDR_ANY);
1da177e4
LT
324 addr->v4.sin_port = port;
325}
326
327/* Is this a wildcard address? */
328static int sctp_v4_is_any(const union sctp_addr *addr)
329{
e6f1cebf 330 return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr;
1da177e4
LT
331}
332
333/* This function checks if the address is a valid address to be used for
334 * SCTP binding.
335 *
336 * Output:
337 * Return 0 - If the address is a non-unicast or an illegal address.
338 * Return 1 - If the address is a unicast.
339 */
5636bef7
VY
340static int sctp_v4_addr_valid(union sctp_addr *addr,
341 struct sctp_sock *sp,
342 const struct sk_buff *skb)
1da177e4 343{
7dab83de
VY
344 /* IPv4 addresses not allowed */
345 if (sp && ipv6_only_sock(sctp_opt2sk(sp)))
346 return 0;
347
1da177e4 348 /* Is this a non-unicast address or a unusable SCTP address? */
b5cb2bbc 349 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr))
1da177e4
LT
350 return 0;
351
d808ad9a 352 /* Is this a broadcast address? */
511c3f92 353 if (skb && skb_rtable(skb)->rt_flags & RTCF_BROADCAST)
d808ad9a 354 return 0;
5636bef7 355
1da177e4
LT
356 return 1;
357}
358
359/* Should this be available for binding? */
360static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp)
361{
bb2db45b
EB
362 struct net *net = sock_net(&sp->inet.sk);
363 int ret = inet_addr_type(net, addr->v4.sin_addr.s_addr);
1da177e4 364
1da177e4 365
e6f1cebf 366 if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) &&
cdac4e07
NH
367 ret != RTN_LOCAL &&
368 !sp->inet.freebind &&
49a60158 369 !net->ipv4.sysctl_ip_nonlocal_bind)
1da177e4 370 return 0;
cdac4e07 371
7dab83de
VY
372 if (ipv6_only_sock(sctp_opt2sk(sp)))
373 return 0;
374
1da177e4
LT
375 return 1;
376}
377
378/* Checking the loopback, private and other address scopes as defined in
379 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
380 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
381 *
382 * Level 0 - unusable SCTP addresses
383 * Level 1 - loopback address
384 * Level 2 - link-local addresses
385 * Level 3 - private addresses.
386 * Level 4 - global addresses
387 * For INIT and INIT-ACK address list, let L be the level of
388 * of requested destination address, sender and receiver
389 * SHOULD include all of its addresses with level greater
390 * than or equal to L.
72388433
BD
391 *
392 * IPv4 scoping can be controlled through sysctl option
393 * net.sctp.addr_scope_policy
1da177e4
LT
394 */
395static sctp_scope_t sctp_v4_scope(union sctp_addr *addr)
396{
397 sctp_scope_t retval;
398
1da177e4 399 /* Check for unusable SCTP addresses. */
b5cb2bbc 400 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) {
1da177e4 401 retval = SCTP_SCOPE_UNUSABLE;
b5cb2bbc 402 } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) {
1da177e4 403 retval = SCTP_SCOPE_LOOPBACK;
b5cb2bbc 404 } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) {
1da177e4 405 retval = SCTP_SCOPE_LINK;
b5cb2bbc
JP
406 } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) ||
407 ipv4_is_private_172(addr->v4.sin_addr.s_addr) ||
408 ipv4_is_private_192(addr->v4.sin_addr.s_addr)) {
1da177e4
LT
409 retval = SCTP_SCOPE_PRIVATE;
410 } else {
411 retval = SCTP_SCOPE_GLOBAL;
412 }
413
414 return retval;
415}
416
417/* Returns a valid dst cache entry for the given source and destination ip
418 * addresses. If an association is passed, trys to get a dst entry with a
419 * source address that matches an address in the bind address list.
420 */
da0420be
VY
421static void sctp_v4_get_dst(struct sctp_transport *t, union sctp_addr *saddr,
422 struct flowi *fl, struct sock *sk)
1da177e4 423{
da0420be 424 struct sctp_association *asoc = t->asoc;
1da177e4 425 struct rtable *rt;
9914ae3c 426 struct flowi4 *fl4 = &fl->u.ip4;
1da177e4 427 struct sctp_bind_addr *bp;
1da177e4 428 struct sctp_sockaddr_entry *laddr;
1da177e4 429 struct dst_entry *dst = NULL;
da0420be 430 union sctp_addr *daddr = &t->ipaddr;
1da177e4
LT
431 union sctp_addr dst_saddr;
432
9914ae3c
VY
433 memset(fl4, 0x0, sizeof(struct flowi4));
434 fl4->daddr = daddr->v4.sin_addr.s_addr;
435 fl4->fl4_dport = daddr->v4.sin_port;
436 fl4->flowi4_proto = IPPROTO_SCTP;
1da177e4 437 if (asoc) {
9914ae3c
VY
438 fl4->flowi4_tos = RT_CONN_FLAGS(asoc->base.sk);
439 fl4->flowi4_oif = asoc->base.sk->sk_bound_dev_if;
440 fl4->fl4_sport = htons(asoc->base.bind_addr.port);
1da177e4 441 }
6429d3dc 442 if (saddr) {
9914ae3c
VY
443 fl4->saddr = saddr->v4.sin_addr.s_addr;
444 fl4->fl4_sport = saddr->v4.sin_port;
6429d3dc 445 }
1da177e4 446
bb33381d
DB
447 pr_debug("%s: dst:%pI4, src:%pI4 - ", __func__, &fl4->daddr,
448 &fl4->saddr);
1da177e4 449
bb2db45b 450 rt = ip_route_output_key(sock_net(sk), fl4);
b23dd4fe 451 if (!IS_ERR(rt))
d8d1f30b 452 dst = &rt->dst;
1da177e4
LT
453
454 /* If there is no association or if a source address is passed, no
455 * more validation is required.
456 */
457 if (!asoc || saddr)
458 goto out;
459
460 bp = &asoc->base.bind_addr;
1da177e4
LT
461
462 if (dst) {
463 /* Walk through the bind address list and look for a bind
464 * address that matches the source address of the returned dst.
465 */
18a353f4 466 sctp_v4_dst_saddr(&dst_saddr, fl4, htons(bp->port));
559cf710
VY
467 rcu_read_lock();
468 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
8a07eb0a
MH
469 if (!laddr->valid || (laddr->state == SCTP_ADDR_DEL) ||
470 (laddr->state != SCTP_ADDR_SRC &&
471 !asoc->src_out_of_asoc_ok))
dc022a98 472 continue;
854d43a4 473 if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a))
1da177e4
LT
474 goto out_unlock;
475 }
559cf710 476 rcu_read_unlock();
1da177e4
LT
477
478 /* None of the bound addresses match the source address of the
479 * dst. So release it.
480 */
481 dst_release(dst);
482 dst = NULL;
483 }
484
485 /* Walk through the bind address list and try to get a dst that
486 * matches a bind address as the source address.
487 */
559cf710
VY
488 rcu_read_lock();
489 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
490 if (!laddr->valid)
491 continue;
07868284
MRL
492 if (laddr->state != SCTP_ADDR_SRC ||
493 AF_INET != laddr->a.sa.sa_family)
494 continue;
495
496 fl4->fl4_sport = laddr->a.v4.sin_port;
497 flowi4_update_output(fl4,
498 asoc->base.sk->sk_bound_dev_if,
499 RT_CONN_FLAGS(asoc->base.sk),
500 daddr->v4.sin_addr.s_addr,
501 laddr->a.v4.sin_addr.s_addr);
502
503 rt = ip_route_output_key(sock_net(sk), fl4);
504 if (IS_ERR(rt))
505 continue;
506
507 dst = &rt->dst;
508 break;
1da177e4
LT
509 }
510
511out_unlock:
559cf710 512 rcu_read_unlock();
1da177e4 513out:
da0420be 514 t->dst = dst;
1da177e4 515 if (dst)
bb33381d
DB
516 pr_debug("rt_dst:%pI4, rt_src:%pI4\n",
517 &fl4->daddr, &fl4->saddr);
1da177e4 518 else
bb33381d 519 pr_debug("no route\n");
1da177e4
LT
520}
521
522/* For v4, the source address is cached in the route entry(dst). So no need
523 * to cache it separately and hence this is an empty routine.
524 */
e5117101 525static void sctp_v4_get_saddr(struct sctp_sock *sk,
9914ae3c 526 struct sctp_transport *t,
9914ae3c 527 struct flowi *fl)
1da177e4 528{
9914ae3c
VY
529 union sctp_addr *saddr = &t->saddr;
530 struct rtable *rt = (struct rtable *)t->dst;
23ec47a0 531
1da177e4
LT
532 if (rt) {
533 saddr->v4.sin_family = AF_INET;
902ebd3e 534 saddr->v4.sin_addr.s_addr = fl->u.ip4.saddr;
1da177e4
LT
535 }
536}
537
538/* What interface did this skb arrive on? */
539static int sctp_v4_skb_iif(const struct sk_buff *skb)
540{
92101b3b 541 return inet_iif(skb);
1da177e4
LT
542}
543
544/* Was this packet marked by Explicit Congestion Notification? */
545static int sctp_v4_is_ce(const struct sk_buff *skb)
546{
eddc9ec5 547 return INET_ECN_is_ce(ip_hdr(skb)->tos);
1da177e4
LT
548}
549
550/* Create and initialize a new sk for the socket returned by accept(). */
551static struct sock *sctp_v4_create_accept_sk(struct sock *sk,
552 struct sctp_association *asoc)
553{
3b1e0a65 554 struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL,
11aa9c28 555 sk->sk_prot, 0);
914e1c8b 556 struct inet_sock *newinet;
1da177e4
LT
557
558 if (!newsk)
559 goto out;
560
561 sock_init_data(NULL, newsk);
562
914e1c8b 563 sctp_copy_sock(newsk, sk, asoc);
1da177e4
LT
564 sock_reset_flag(newsk, SOCK_ZAPPED);
565
566 newinet = inet_sk(newsk);
567
c720c7e8 568 newinet->inet_daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr;
1da177e4 569
e6848976 570 sk_refcnt_debug_inc(newsk);
1da177e4
LT
571
572 if (newsk->sk_prot->init(newsk)) {
573 sk_common_release(newsk);
574 newsk = NULL;
575 }
576
577out:
578 return newsk;
579}
580
299ee123 581static int sctp_v4_addr_to_user(struct sctp_sock *sp, union sctp_addr *addr)
1da177e4 582{
299ee123
JG
583 /* No address mapping for V4 sockets */
584 return sizeof(struct sockaddr_in);
1da177e4
LT
585}
586
587/* Dump the v4 addr to the seq file. */
588static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr)
589{
21454aaa 590 seq_printf(seq, "%pI4 ", &addr->v4.sin_addr);
1da177e4
LT
591}
592
b9031d9d
VY
593static void sctp_v4_ecn_capable(struct sock *sk)
594{
595 INET_ECN_xmit(sk);
596}
597
bd2a13e2 598static void sctp_addr_wq_timeout_handler(unsigned long arg)
9f7d653b 599{
4db67e80 600 struct net *net = (struct net *)arg;
9f7d653b
MH
601 struct sctp_sockaddr_entry *addrw, *temp;
602 struct sctp_sock *sp;
603
4db67e80 604 spin_lock_bh(&net->sctp.addr_wq_lock);
9f7d653b 605
4db67e80 606 list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) {
bb33381d
DB
607 pr_debug("%s: the first ent in wq:%p is addr:%pISc for cmd:%d at "
608 "entry:%p\n", __func__, &net->sctp.addr_waitq, &addrw->a.sa,
609 addrw->state, addrw);
9f7d653b 610
dfd56b8b 611#if IS_ENABLED(CONFIG_IPV6)
9f7d653b
MH
612 /* Now we send an ASCONF for each association */
613 /* Note. we currently don't handle link local IPv6 addressees */
614 if (addrw->a.sa.sa_family == AF_INET6) {
615 struct in6_addr *in6;
616
617 if (ipv6_addr_type(&addrw->a.v6.sin6_addr) &
618 IPV6_ADDR_LINKLOCAL)
619 goto free_next;
620
621 in6 = (struct in6_addr *)&addrw->a.v6.sin6_addr;
4db67e80 622 if (ipv6_chk_addr(net, in6, NULL, 0) == 0 &&
9f7d653b
MH
623 addrw->state == SCTP_ADDR_NEW) {
624 unsigned long timeo_val;
625
bb33381d
DB
626 pr_debug("%s: this is on DAD, trying %d sec "
627 "later\n", __func__,
628 SCTP_ADDRESS_TICK_DELAY);
629
9f7d653b
MH
630 timeo_val = jiffies;
631 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
4db67e80 632 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
9f7d653b
MH
633 break;
634 }
635 }
5d0c90cf 636#endif
4db67e80 637 list_for_each_entry(sp, &net->sctp.auto_asconf_splist, auto_asconf_list) {
9f7d653b
MH
638 struct sock *sk;
639
640 sk = sctp_opt2sk(sp);
641 /* ignore bound-specific endpoints */
642 if (!sctp_is_ep_boundall(sk))
643 continue;
5bc1d1b4 644 bh_lock_sock(sk);
9f7d653b 645 if (sctp_asconf_mgmt(sp, addrw) < 0)
bb33381d 646 pr_debug("%s: sctp_asconf_mgmt failed\n", __func__);
5bc1d1b4 647 bh_unlock_sock(sk);
9f7d653b 648 }
39d84a58 649#if IS_ENABLED(CONFIG_IPV6)
9f7d653b 650free_next:
39d84a58 651#endif
9f7d653b
MH
652 list_del(&addrw->list);
653 kfree(addrw);
654 }
4db67e80 655 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
656}
657
4db67e80 658static void sctp_free_addr_wq(struct net *net)
9f7d653b
MH
659{
660 struct sctp_sockaddr_entry *addrw;
661 struct sctp_sockaddr_entry *temp;
662
4db67e80
EB
663 spin_lock_bh(&net->sctp.addr_wq_lock);
664 del_timer(&net->sctp.addr_wq_timer);
665 list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) {
9f7d653b
MH
666 list_del(&addrw->list);
667 kfree(addrw);
668 }
4db67e80 669 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
670}
671
672/* lookup the entry for the same address in the addr_waitq
673 * sctp_addr_wq MUST be locked
674 */
4db67e80
EB
675static struct sctp_sockaddr_entry *sctp_addr_wq_lookup(struct net *net,
676 struct sctp_sockaddr_entry *addr)
9f7d653b
MH
677{
678 struct sctp_sockaddr_entry *addrw;
679
4db67e80 680 list_for_each_entry(addrw, &net->sctp.addr_waitq, list) {
9f7d653b
MH
681 if (addrw->a.sa.sa_family != addr->a.sa.sa_family)
682 continue;
683 if (addrw->a.sa.sa_family == AF_INET) {
684 if (addrw->a.v4.sin_addr.s_addr ==
685 addr->a.v4.sin_addr.s_addr)
686 return addrw;
687 } else if (addrw->a.sa.sa_family == AF_INET6) {
688 if (ipv6_addr_equal(&addrw->a.v6.sin6_addr,
689 &addr->a.v6.sin6_addr))
690 return addrw;
691 }
692 }
693 return NULL;
694}
695
4db67e80 696void sctp_addr_wq_mgmt(struct net *net, struct sctp_sockaddr_entry *addr, int cmd)
9f7d653b
MH
697{
698 struct sctp_sockaddr_entry *addrw;
699 unsigned long timeo_val;
700
701 /* first, we check if an opposite message already exist in the queue.
702 * If we found such message, it is removed.
703 * This operation is a bit stupid, but the DHCP client attaches the
704 * new address after a couple of addition and deletion of that address
705 */
706
4db67e80 707 spin_lock_bh(&net->sctp.addr_wq_lock);
9f7d653b 708 /* Offsets existing events in addr_wq */
4db67e80 709 addrw = sctp_addr_wq_lookup(net, addr);
9f7d653b
MH
710 if (addrw) {
711 if (addrw->state != cmd) {
bb33381d
DB
712 pr_debug("%s: offsets existing entry for %d, addr:%pISc "
713 "in wq:%p\n", __func__, addrw->state, &addrw->a.sa,
714 &net->sctp.addr_waitq);
715
9f7d653b
MH
716 list_del(&addrw->list);
717 kfree(addrw);
718 }
4db67e80 719 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
720 return;
721 }
722
723 /* OK, we have to add the new address to the wait queue */
724 addrw = kmemdup(addr, sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
725 if (addrw == NULL) {
4db67e80 726 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
727 return;
728 }
729 addrw->state = cmd;
4db67e80 730 list_add_tail(&addrw->list, &net->sctp.addr_waitq);
bb33381d
DB
731
732 pr_debug("%s: add new entry for cmd:%d, addr:%pISc in wq:%p\n",
733 __func__, addrw->state, &addrw->a.sa, &net->sctp.addr_waitq);
9f7d653b 734
4db67e80 735 if (!timer_pending(&net->sctp.addr_wq_timer)) {
9f7d653b
MH
736 timeo_val = jiffies;
737 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
4db67e80 738 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
9f7d653b 739 }
4db67e80 740 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
741}
742
29303547
VY
743/* Event handler for inet address addition/deletion events.
744 * The sctp_local_addr_list needs to be protocted by a spin lock since
745 * multiple notifiers (say IPv4 and IPv6) may be running at the same
746 * time and thus corrupt the list.
747 * The reader side is protected with RCU.
748 */
24123186
AB
749static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev,
750 void *ptr)
1da177e4 751{
29c7cf96 752 struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
29303547
VY
753 struct sctp_sockaddr_entry *addr = NULL;
754 struct sctp_sockaddr_entry *temp;
4db67e80 755 struct net *net = dev_net(ifa->ifa_dev->dev);
22626216 756 int found = 0;
1da177e4 757
29c7cf96
SS
758 switch (ev) {
759 case NETDEV_UP:
760 addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
761 if (addr) {
762 addr->a.v4.sin_family = AF_INET;
763 addr->a.v4.sin_port = 0;
764 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
29303547 765 addr->valid = 1;
4db67e80
EB
766 spin_lock_bh(&net->sctp.local_addr_lock);
767 list_add_tail_rcu(&addr->list, &net->sctp.local_addr_list);
768 sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_NEW);
769 spin_unlock_bh(&net->sctp.local_addr_lock);
29c7cf96
SS
770 }
771 break;
772 case NETDEV_DOWN:
4db67e80 773 spin_lock_bh(&net->sctp.local_addr_lock);
29303547 774 list_for_each_entry_safe(addr, temp,
4db67e80 775 &net->sctp.local_addr_list, list) {
a40a7d15
PE
776 if (addr->a.sa.sa_family == AF_INET &&
777 addr->a.v4.sin_addr.s_addr ==
778 ifa->ifa_local) {
4db67e80 779 sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_DEL);
22626216 780 found = 1;
29303547
VY
781 addr->valid = 0;
782 list_del_rcu(&addr->list);
29c7cf96
SS
783 break;
784 }
785 }
4db67e80 786 spin_unlock_bh(&net->sctp.local_addr_lock);
22626216 787 if (found)
1231f0ba 788 kfree_rcu(addr, rcu);
29c7cf96
SS
789 break;
790 }
1da177e4
LT
791
792 return NOTIFY_DONE;
793}
794
795/*
796 * Initialize the control inode/socket with a control endpoint data
797 * structure. This endpoint is reserved exclusively for the OOTB processing.
798 */
2ce95503 799static int sctp_ctl_sock_init(struct net *net)
1da177e4
LT
800{
801 int err;
fb13d9f9 802 sa_family_t family = PF_INET;
1da177e4
LT
803
804 if (sctp_get_pf_specific(PF_INET6))
805 family = PF_INET6;
1da177e4 806
2ce95503
EB
807 err = inet_ctl_sock_create(&net->sctp.ctl_sock, family,
808 SOCK_SEQPACKET, IPPROTO_SCTP, net);
fb13d9f9
BH
809
810 /* If IPv6 socket could not be created, try the IPv4 socket */
811 if (err < 0 && family == PF_INET6)
2ce95503 812 err = inet_ctl_sock_create(&net->sctp.ctl_sock, AF_INET,
fb13d9f9 813 SOCK_SEQPACKET, IPPROTO_SCTP,
2ce95503 814 net);
fb13d9f9 815
1da177e4 816 if (err < 0) {
145ce502 817 pr_err("Failed to create the SCTP control socket\n");
1da177e4
LT
818 return err;
819 }
1da177e4
LT
820 return 0;
821}
822
823/* Register address family specific functions. */
824int sctp_register_af(struct sctp_af *af)
825{
826 switch (af->sa_family) {
827 case AF_INET:
828 if (sctp_af_v4_specific)
829 return 0;
830 sctp_af_v4_specific = af;
831 break;
832 case AF_INET6:
833 if (sctp_af_v6_specific)
834 return 0;
835 sctp_af_v6_specific = af;
836 break;
837 default:
838 return 0;
839 }
840
841 INIT_LIST_HEAD(&af->list);
842 list_add_tail(&af->list, &sctp_address_families);
843 return 1;
844}
845
846/* Get the table of functions for manipulating a particular address
847 * family.
848 */
849struct sctp_af *sctp_get_af_specific(sa_family_t family)
850{
851 switch (family) {
852 case AF_INET:
853 return sctp_af_v4_specific;
854 case AF_INET6:
855 return sctp_af_v6_specific;
856 default:
857 return NULL;
858 }
859}
860
861/* Common code to initialize a AF_INET msg_name. */
862static void sctp_inet_msgname(char *msgname, int *addr_len)
863{
864 struct sockaddr_in *sin;
865
866 sin = (struct sockaddr_in *)msgname;
867 *addr_len = sizeof(struct sockaddr_in);
868 sin->sin_family = AF_INET;
869 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
870}
871
872/* Copy the primary address of the peer primary address as the msg_name. */
873static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname,
874 int *addr_len)
875{
876 struct sockaddr_in *sin, *sinfrom;
877
878 if (msgname) {
879 struct sctp_association *asoc;
880
881 asoc = event->asoc;
882 sctp_inet_msgname(msgname, addr_len);
883 sin = (struct sockaddr_in *)msgname;
884 sinfrom = &asoc->peer.primary_addr.v4;
885 sin->sin_port = htons(asoc->peer.port);
886 sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr;
887 }
888}
889
890/* Initialize and copy out a msgname from an inbound skb. */
891static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len)
892{
1da177e4 893 if (msgname) {
2c0fd387
ACM
894 struct sctphdr *sh = sctp_hdr(skb);
895 struct sockaddr_in *sin = (struct sockaddr_in *)msgname;
896
1da177e4 897 sctp_inet_msgname(msgname, len);
1da177e4 898 sin->sin_port = sh->source;
eddc9ec5 899 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
1da177e4
LT
900 }
901}
902
903/* Do we support this AF? */
904static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp)
905{
906 /* PF_INET only supports AF_INET addresses. */
a02cec21 907 return AF_INET == family;
1da177e4
LT
908}
909
910/* Address matching with wildcards allowed. */
911static int sctp_inet_cmp_addr(const union sctp_addr *addr1,
912 const union sctp_addr *addr2,
913 struct sctp_sock *opt)
914{
915 /* PF_INET only supports AF_INET addresses. */
916 if (addr1->sa.sa_family != addr2->sa.sa_family)
917 return 0;
e6f1cebf
AV
918 if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr ||
919 htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr)
1da177e4
LT
920 return 1;
921 if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr)
922 return 1;
923
924 return 0;
925}
926
927/* Verify that provided sockaddr looks bindable. Common verification has
928 * already been taken care of.
929 */
930static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr)
931{
932 return sctp_v4_available(addr, opt);
933}
934
935/* Verify that sockaddr looks sendable. Common verification has already
936 * been taken care of.
937 */
938static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr)
939{
940 return 1;
941}
942
943/* Fill in Supported Address Type information for INIT and INIT-ACK
944 * chunks. Returns number of addresses supported.
945 */
946static int sctp_inet_supported_addrs(const struct sctp_sock *opt,
3dbe8656 947 __be16 *types)
1da177e4
LT
948{
949 types[0] = SCTP_PARAM_IPV4_ADDRESS;
950 return 1;
951}
952
953/* Wrapper routine that calls the ip transmit routine. */
954static inline int sctp_v4_xmit(struct sk_buff *skb,
f880374c 955 struct sctp_transport *transport)
1da177e4 956{
f880374c
HX
957 struct inet_sock *inet = inet_sk(skb->sk);
958
bb33381d
DB
959 pr_debug("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n", __func__, skb,
960 skb->len, &transport->fl.u.ip4.saddr, &transport->fl.u.ip4.daddr);
1da177e4 961
f880374c
HX
962 inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ?
963 IP_PMTUDISC_DO : IP_PMTUDISC_DONT;
964
b01a2407 965 SCTP_INC_STATS(sock_net(&inet->sk), SCTP_MIB_OUTSCTPPACKS);
bb33381d 966
b0270e91 967 return ip_queue_xmit(&inet->sk, skb, &transport->fl);
1da177e4
LT
968}
969
15efbe76 970static struct sctp_af sctp_af_inet;
1da177e4
LT
971
972static struct sctp_pf sctp_pf_inet = {
973 .event_msgname = sctp_inet_event_msgname,
974 .skb_msgname = sctp_inet_skb_msgname,
975 .af_supported = sctp_inet_af_supported,
976 .cmp_addr = sctp_inet_cmp_addr,
977 .bind_verify = sctp_inet_bind_verify,
978 .send_verify = sctp_inet_send_verify,
979 .supported_addrs = sctp_inet_supported_addrs,
980 .create_accept_sk = sctp_v4_create_accept_sk,
299ee123
JG
981 .addr_to_user = sctp_v4_addr_to_user,
982 .to_sk_saddr = sctp_v4_to_sk_saddr,
983 .to_sk_daddr = sctp_v4_to_sk_daddr,
15efbe76 984 .af = &sctp_af_inet
1da177e4
LT
985};
986
987/* Notifier for inetaddr addition/deletion events. */
988static struct notifier_block sctp_inetaddr_notifier = {
989 .notifier_call = sctp_inetaddr_event,
990};
991
992/* Socket operations. */
90ddc4f0 993static const struct proto_ops inet_seqpacket_ops = {
543d9cfe
ACM
994 .family = PF_INET,
995 .owner = THIS_MODULE,
996 .release = inet_release, /* Needs to be wrapped... */
997 .bind = inet_bind,
998 .connect = inet_dgram_connect,
999 .socketpair = sock_no_socketpair,
1000 .accept = inet_accept,
1001 .getname = inet_getname, /* Semantics are different. */
1002 .poll = sctp_poll,
1003 .ioctl = inet_ioctl,
1004 .listen = sctp_inet_listen,
1005 .shutdown = inet_shutdown, /* Looks harmless. */
1006 .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */
1007 .getsockopt = sock_common_getsockopt,
1008 .sendmsg = inet_sendmsg,
1009 .recvmsg = sock_common_recvmsg,
1010 .mmap = sock_no_mmap,
1011 .sendpage = sock_no_sendpage,
3fdadf7d 1012#ifdef CONFIG_COMPAT
543d9cfe
ACM
1013 .compat_setsockopt = compat_sock_common_setsockopt,
1014 .compat_getsockopt = compat_sock_common_getsockopt,
3fdadf7d 1015#endif
1da177e4
LT
1016};
1017
1018/* Registration with AF_INET family. */
1019static struct inet_protosw sctp_seqpacket_protosw = {
1020 .type = SOCK_SEQPACKET,
1021 .protocol = IPPROTO_SCTP,
1022 .prot = &sctp_prot,
1023 .ops = &inet_seqpacket_ops,
1da177e4
LT
1024 .flags = SCTP_PROTOSW_FLAG
1025};
1026static struct inet_protosw sctp_stream_protosw = {
1027 .type = SOCK_STREAM,
1028 .protocol = IPPROTO_SCTP,
1029 .prot = &sctp_prot,
1030 .ops = &inet_seqpacket_ops,
1da177e4
LT
1031 .flags = SCTP_PROTOSW_FLAG
1032};
1033
1034/* Register with IP layer. */
32613090 1035static const struct net_protocol sctp_protocol = {
1da177e4
LT
1036 .handler = sctp_rcv,
1037 .err_handler = sctp_v4_err,
1038 .no_policy = 1,
bb2db45b 1039 .netns_ok = 1,
8ed1dc44 1040 .icmp_strict_tag_validation = 1,
1da177e4
LT
1041};
1042
1043/* IPv4 address related functions. */
15efbe76 1044static struct sctp_af sctp_af_inet = {
543d9cfe
ACM
1045 .sa_family = AF_INET,
1046 .sctp_xmit = sctp_v4_xmit,
1047 .setsockopt = ip_setsockopt,
1048 .getsockopt = ip_getsockopt,
1049 .get_dst = sctp_v4_get_dst,
1050 .get_saddr = sctp_v4_get_saddr,
1051 .copy_addrlist = sctp_v4_copy_addrlist,
1052 .from_skb = sctp_v4_from_skb,
1053 .from_sk = sctp_v4_from_sk,
543d9cfe
ACM
1054 .from_addr_param = sctp_v4_from_addr_param,
1055 .to_addr_param = sctp_v4_to_addr_param,
543d9cfe
ACM
1056 .cmp_addr = sctp_v4_cmp_addr,
1057 .addr_valid = sctp_v4_addr_valid,
1058 .inaddr_any = sctp_v4_inaddr_any,
1059 .is_any = sctp_v4_is_any,
1060 .available = sctp_v4_available,
1061 .scope = sctp_v4_scope,
1062 .skb_iif = sctp_v4_skb_iif,
1063 .is_ce = sctp_v4_is_ce,
1064 .seq_dump_addr = sctp_v4_seq_dump_addr,
b9031d9d 1065 .ecn_capable = sctp_v4_ecn_capable,
543d9cfe
ACM
1066 .net_header_len = sizeof(struct iphdr),
1067 .sockaddr_len = sizeof(struct sockaddr_in),
3fdadf7d 1068#ifdef CONFIG_COMPAT
543d9cfe
ACM
1069 .compat_setsockopt = compat_ip_setsockopt,
1070 .compat_getsockopt = compat_ip_getsockopt,
3fdadf7d 1071#endif
1da177e4
LT
1072};
1073
8d72651d 1074struct sctp_pf *sctp_get_pf_specific(sa_family_t family)
1075{
1da177e4
LT
1076 switch (family) {
1077 case PF_INET:
1078 return sctp_pf_inet_specific;
1079 case PF_INET6:
1080 return sctp_pf_inet6_specific;
1081 default:
1082 return NULL;
1083 }
1084}
1085
1086/* Register the PF specific function table. */
1087int sctp_register_pf(struct sctp_pf *pf, sa_family_t family)
1088{
1089 switch (family) {
1090 case PF_INET:
1091 if (sctp_pf_inet_specific)
1092 return 0;
1093 sctp_pf_inet_specific = pf;
1094 break;
1095 case PF_INET6:
1096 if (sctp_pf_inet6_specific)
1097 return 0;
1098 sctp_pf_inet6_specific = pf;
1099 break;
1100 default:
1101 return 0;
1102 }
1103 return 1;
1104}
1105
b01a2407 1106static inline int init_sctp_mibs(struct net *net)
996b1dba 1107{
698365fa
WC
1108 net->sctp.sctp_statistics = alloc_percpu(struct sctp_mib);
1109 if (!net->sctp.sctp_statistics)
1110 return -ENOMEM;
1111 return 0;
1da177e4
LT
1112}
1113
b01a2407 1114static inline void cleanup_sctp_mibs(struct net *net)
1da177e4 1115{
698365fa 1116 free_percpu(net->sctp.sctp_statistics);
1da177e4
LT
1117}
1118
270637ab
VY
1119static void sctp_v4_pf_init(void)
1120{
1121 /* Initialize the SCTP specific PF functions. */
1122 sctp_register_pf(&sctp_pf_inet, PF_INET);
1123 sctp_register_af(&sctp_af_inet);
1124}
1125
1126static void sctp_v4_pf_exit(void)
1127{
1128 list_del(&sctp_af_inet.list);
1129}
1130
1131static int sctp_v4_protosw_init(void)
1132{
1133 int rc;
1134
1135 rc = proto_register(&sctp_prot, 1);
1136 if (rc)
1137 return rc;
1138
1139 /* Register SCTP(UDP and TCP style) with socket layer. */
1140 inet_register_protosw(&sctp_seqpacket_protosw);
1141 inet_register_protosw(&sctp_stream_protosw);
1142
1143 return 0;
1144}
1145
1146static void sctp_v4_protosw_exit(void)
1147{
1148 inet_unregister_protosw(&sctp_stream_protosw);
1149 inet_unregister_protosw(&sctp_seqpacket_protosw);
1150 proto_unregister(&sctp_prot);
1151}
1152
1153static int sctp_v4_add_protocol(void)
1154{
1155 /* Register notifier for inet address additions/deletions. */
1156 register_inetaddr_notifier(&sctp_inetaddr_notifier);
1157
1158 /* Register SCTP with inet layer. */
1159 if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0)
1160 return -EAGAIN;
1161
1162 return 0;
1163}
1164
1165static void sctp_v4_del_protocol(void)
1166{
1167 inet_del_protocol(&sctp_protocol, IPPROTO_SCTP);
1168 unregister_inetaddr_notifier(&sctp_inetaddr_notifier);
1169}
1170
f5f417c0 1171static int __net_init sctp_net_init(struct net *net)
4db67e80 1172{
2ce95503
EB
1173 int status;
1174
e1fc3b14
EB
1175 /*
1176 * 14. Suggested SCTP Protocol Parameter Values
1177 */
1178 /* The following protocol parameters are RECOMMENDED: */
1179 /* RTO.Initial - 3 seconds */
1180 net->sctp.rto_initial = SCTP_RTO_INITIAL;
1181 /* RTO.Min - 1 second */
1182 net->sctp.rto_min = SCTP_RTO_MIN;
1183 /* RTO.Max - 60 seconds */
1184 net->sctp.rto_max = SCTP_RTO_MAX;
1185 /* RTO.Alpha - 1/8 */
1186 net->sctp.rto_alpha = SCTP_RTO_ALPHA;
1187 /* RTO.Beta - 1/4 */
1188 net->sctp.rto_beta = SCTP_RTO_BETA;
1189
1190 /* Valid.Cookie.Life - 60 seconds */
1191 net->sctp.valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE;
1192
1193 /* Whether Cookie Preservative is enabled(1) or not(0) */
1194 net->sctp.cookie_preserve_enable = 1;
1195
3c68198e 1196 /* Default sctp sockets to use md5 as their hmac alg */
0d0863b0 1197#if defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_MD5)
3c68198e 1198 net->sctp.sctp_hmac_alg = "md5";
0d0863b0 1199#elif defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_SHA1)
3c68198e
NH
1200 net->sctp.sctp_hmac_alg = "sha1";
1201#else
1202 net->sctp.sctp_hmac_alg = NULL;
1203#endif
1204
e1fc3b14
EB
1205 /* Max.Burst - 4 */
1206 net->sctp.max_burst = SCTP_DEFAULT_MAX_BURST;
1207
1208 /* Association.Max.Retrans - 10 attempts
1209 * Path.Max.Retrans - 5 attempts (per destination address)
1210 * Max.Init.Retransmits - 8 attempts
1211 */
1212 net->sctp.max_retrans_association = 10;
1213 net->sctp.max_retrans_path = 5;
1214 net->sctp.max_retrans_init = 8;
1215
1216 /* Sendbuffer growth - do per-socket accounting */
1217 net->sctp.sndbuf_policy = 0;
1218
1219 /* Rcvbuffer growth - do per-socket accounting */
1220 net->sctp.rcvbuf_policy = 0;
1221
1222 /* HB.interval - 30 seconds */
1223 net->sctp.hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT;
1224
1225 /* delayed SACK timeout */
1226 net->sctp.sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK;
1227
1228 /* Disable ADDIP by default. */
1229 net->sctp.addip_enable = 0;
1230 net->sctp.addip_noauth = 0;
1231 net->sctp.default_auto_asconf = 0;
1232
1233 /* Enable PR-SCTP by default. */
1234 net->sctp.prsctp_enable = 1;
1235
1236 /* Disable AUTH by default. */
1237 net->sctp.auth_enable = 0;
1238
1239 /* Set SCOPE policy to enabled */
1240 net->sctp.scope_policy = SCTP_SCOPE_POLICY_ENABLE;
1241
1242 /* Set the default rwnd update threshold */
1243 net->sctp.rwnd_upd_shift = SCTP_DEFAULT_RWND_SHIFT;
1244
1245 /* Initialize maximum autoclose timeout. */
1246 net->sctp.max_autoclose = INT_MAX / HZ;
1247
ebb7e95d
EB
1248 status = sctp_sysctl_net_register(net);
1249 if (status)
1250 goto err_sysctl_register;
1251
b01a2407
EB
1252 /* Allocate and initialise sctp mibs. */
1253 status = init_sctp_mibs(net);
1254 if (status)
1255 goto err_init_mibs;
1256
13d782f6
EB
1257 /* Initialize proc fs directory. */
1258 status = sctp_proc_init(net);
1259 if (status)
1260 goto err_init_proc;
1261
1262 sctp_dbg_objcnt_init(net);
1263
2ce95503
EB
1264 /* Initialize the control inode/socket for handling OOTB packets. */
1265 if ((status = sctp_ctl_sock_init(net))) {
1266 pr_err("Failed to initialize the SCTP control sock\n");
1267 goto err_ctl_sock_init;
1268 }
1269
4db67e80
EB
1270 /* Initialize the local address list. */
1271 INIT_LIST_HEAD(&net->sctp.local_addr_list);
1272 spin_lock_init(&net->sctp.local_addr_lock);
1273 sctp_get_local_addr_list(net);
1274
1275 /* Initialize the address event list */
1276 INIT_LIST_HEAD(&net->sctp.addr_waitq);
1277 INIT_LIST_HEAD(&net->sctp.auto_asconf_splist);
1278 spin_lock_init(&net->sctp.addr_wq_lock);
1279 net->sctp.addr_wq_timer.expires = 0;
1280 setup_timer(&net->sctp.addr_wq_timer, sctp_addr_wq_timeout_handler,
1281 (unsigned long)net);
1282
1283 return 0;
2ce95503
EB
1284
1285err_ctl_sock_init:
13d782f6
EB
1286 sctp_dbg_objcnt_exit(net);
1287 sctp_proc_exit(net);
1288err_init_proc:
b01a2407
EB
1289 cleanup_sctp_mibs(net);
1290err_init_mibs:
ebb7e95d
EB
1291 sctp_sysctl_net_unregister(net);
1292err_sysctl_register:
2ce95503 1293 return status;
4db67e80
EB
1294}
1295
f5f417c0 1296static void __net_exit sctp_net_exit(struct net *net)
4db67e80
EB
1297{
1298 /* Free the local address list */
1299 sctp_free_addr_wq(net);
1300 sctp_free_local_addr_list(net);
2ce95503
EB
1301
1302 /* Free the control endpoint. */
1303 inet_ctl_sock_destroy(net->sctp.ctl_sock);
13d782f6
EB
1304
1305 sctp_dbg_objcnt_exit(net);
1306
1307 sctp_proc_exit(net);
b01a2407 1308 cleanup_sctp_mibs(net);
ebb7e95d 1309 sctp_sysctl_net_unregister(net);
4db67e80
EB
1310}
1311
1312static struct pernet_operations sctp_net_ops = {
1313 .init = sctp_net_init,
1314 .exit = sctp_net_exit,
1315};
1316
1da177e4 1317/* Initialize the universe into something sensible. */
dda91928 1318static __init int sctp_init(void)
1da177e4
LT
1319{
1320 int i;
1321 int status = -EINVAL;
1322 unsigned long goal;
4d93df0a
NH
1323 unsigned long limit;
1324 int max_share;
1da177e4
LT
1325 int order;
1326
b4772ef8 1327 sock_skb_cb_check_size(sizeof(struct sctp_ulpevent));
1da177e4 1328
827bf122 1329 /* Allocate bind_bucket and chunk caches. */
1da177e4
LT
1330 status = -ENOBUFS;
1331 sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket",
1332 sizeof(struct sctp_bind_bucket),
1333 0, SLAB_HWCACHE_ALIGN,
20c2df83 1334 NULL);
1da177e4 1335 if (!sctp_bucket_cachep)
827bf122 1336 goto out;
1da177e4
LT
1337
1338 sctp_chunk_cachep = kmem_cache_create("sctp_chunk",
1339 sizeof(struct sctp_chunk),
1340 0, SLAB_HWCACHE_ALIGN,
20c2df83 1341 NULL);
1da177e4
LT
1342 if (!sctp_chunk_cachep)
1343 goto err_chunk_cachep;
1344
908c7f19 1345 status = percpu_counter_init(&sctp_sockets_allocated, 0, GFP_KERNEL);
632c928a
EB
1346 if (status)
1347 goto err_percpu_counter_init;
1348
1da177e4
LT
1349 /* Implementation specific variables. */
1350
1351 /* Initialize default stream count setup information. */
1352 sctp_max_instreams = SCTP_DEFAULT_INSTREAMS;
1353 sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS;
1354
1355 /* Initialize handle used for association ids. */
1356 idr_init(&sctp_assocs_id);
1357
f03d78db 1358 limit = nr_free_buffer_pages() / 8;
4d93df0a
NH
1359 limit = max(limit, 128UL);
1360 sysctl_sctp_mem[0] = limit / 4 * 3;
1361 sysctl_sctp_mem[1] = limit;
1362 sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2;
1363
1364 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1365 limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7);
1366 max_share = min(4UL*1024*1024, limit);
1367
845525a6 1368 sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */
87fb4b7b 1369 sysctl_sctp_rmem[1] = 1500 * SKB_TRUESIZE(1);
4d93df0a
NH
1370 sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share);
1371
3ab224be 1372 sysctl_sctp_wmem[0] = SK_MEM_QUANTUM;
4d93df0a
NH
1373 sysctl_sctp_wmem[1] = 16*1024;
1374 sysctl_sctp_wmem[2] = max(64*1024, max_share);
1375
1da177e4
LT
1376 /* Size and allocate the association hash table.
1377 * The methodology is similar to that of the tcp hash tables.
1378 */
4481374c
JB
1379 if (totalram_pages >= (128 * 1024))
1380 goal = totalram_pages >> (22 - PAGE_SHIFT);
1da177e4 1381 else
4481374c 1382 goal = totalram_pages >> (24 - PAGE_SHIFT);
1da177e4
LT
1383
1384 for (order = 0; (1UL << order) < goal; order++)
1385 ;
1386
1387 do {
1388 sctp_assoc_hashsize = (1UL << order) * PAGE_SIZE /
1389 sizeof(struct sctp_hashbucket);
1390 if ((sctp_assoc_hashsize > (64 * 1024)) && order > 0)
1391 continue;
1392 sctp_assoc_hashtable = (struct sctp_hashbucket *)
a84b50ce 1393 __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, order);
1da177e4
LT
1394 } while (!sctp_assoc_hashtable && --order > 0);
1395 if (!sctp_assoc_hashtable) {
145ce502 1396 pr_err("Failed association hash alloc\n");
1da177e4
LT
1397 status = -ENOMEM;
1398 goto err_ahash_alloc;
1399 }
1400 for (i = 0; i < sctp_assoc_hashsize; i++) {
1401 rwlock_init(&sctp_assoc_hashtable[i].lock);
d970dbf8 1402 INIT_HLIST_HEAD(&sctp_assoc_hashtable[i].chain);
1da177e4
LT
1403 }
1404
1405 /* Allocate and initialize the endpoint hash table. */
1406 sctp_ep_hashsize = 64;
3b77d661 1407 sctp_ep_hashtable =
1da177e4
LT
1408 kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL);
1409 if (!sctp_ep_hashtable) {
145ce502 1410 pr_err("Failed endpoint_hash alloc\n");
1da177e4
LT
1411 status = -ENOMEM;
1412 goto err_ehash_alloc;
1413 }
1414 for (i = 0; i < sctp_ep_hashsize; i++) {
1415 rwlock_init(&sctp_ep_hashtable[i].lock);
d970dbf8 1416 INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain);
1da177e4
LT
1417 }
1418
1419 /* Allocate and initialize the SCTP port hash table. */
1420 do {
1421 sctp_port_hashsize = (1UL << order) * PAGE_SIZE /
1422 sizeof(struct sctp_bind_hashbucket);
1423 if ((sctp_port_hashsize > (64 * 1024)) && order > 0)
1424 continue;
1425 sctp_port_hashtable = (struct sctp_bind_hashbucket *)
a84b50ce 1426 __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, order);
1da177e4
LT
1427 } while (!sctp_port_hashtable && --order > 0);
1428 if (!sctp_port_hashtable) {
145ce502 1429 pr_err("Failed bind hash alloc\n");
1da177e4
LT
1430 status = -ENOMEM;
1431 goto err_bhash_alloc;
1432 }
1433 for (i = 0; i < sctp_port_hashsize; i++) {
1434 spin_lock_init(&sctp_port_hashtable[i].lock);
d970dbf8 1435 INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain);
1da177e4
LT
1436 }
1437
145ce502 1438 pr_info("Hash tables configured (established %d bind %d)\n",
1da177e4
LT
1439 sctp_assoc_hashsize, sctp_port_hashsize);
1440
1da177e4
LT
1441 sctp_sysctl_register();
1442
1443 INIT_LIST_HEAD(&sctp_address_families);
270637ab
VY
1444 sctp_v4_pf_init();
1445 sctp_v6_pf_init();
1da177e4 1446
270637ab 1447 status = sctp_v4_protosw_init();
827bf122 1448
1da177e4 1449 if (status)
270637ab
VY
1450 goto err_protosw_init;
1451
1452 status = sctp_v6_protosw_init();
1453 if (status)
1454 goto err_v6_protosw_init;
1da177e4 1455
4db67e80
EB
1456 status = register_pernet_subsys(&sctp_net_ops);
1457 if (status)
1458 goto err_register_pernet_subsys;
1459
270637ab
VY
1460 status = sctp_v4_add_protocol();
1461 if (status)
827bf122 1462 goto err_add_protocol;
827bf122
SS
1463
1464 /* Register SCTP with inet6 layer. */
1465 status = sctp_v6_add_protocol();
1466 if (status)
1467 goto err_v6_add_protocol;
1468
1da177e4
LT
1469out:
1470 return status;
827bf122 1471err_v6_add_protocol:
270637ab 1472 sctp_v4_del_protocol();
d1dd5247 1473err_add_protocol:
4db67e80
EB
1474 unregister_pernet_subsys(&sctp_net_ops);
1475err_register_pernet_subsys:
270637ab
VY
1476 sctp_v6_protosw_exit();
1477err_v6_protosw_init:
1478 sctp_v4_protosw_exit();
1479err_protosw_init:
270637ab
VY
1480 sctp_v4_pf_exit();
1481 sctp_v6_pf_exit();
1da177e4 1482 sctp_sysctl_unregister();
1da177e4
LT
1483 free_pages((unsigned long)sctp_port_hashtable,
1484 get_order(sctp_port_hashsize *
1485 sizeof(struct sctp_bind_hashbucket)));
1486err_bhash_alloc:
1487 kfree(sctp_ep_hashtable);
1488err_ehash_alloc:
1489 free_pages((unsigned long)sctp_assoc_hashtable,
1490 get_order(sctp_assoc_hashsize *
1491 sizeof(struct sctp_hashbucket)));
1492err_ahash_alloc:
632c928a
EB
1493 percpu_counter_destroy(&sctp_sockets_allocated);
1494err_percpu_counter_init:
1da177e4
LT
1495 kmem_cache_destroy(sctp_chunk_cachep);
1496err_chunk_cachep:
1497 kmem_cache_destroy(sctp_bucket_cachep);
1da177e4
LT
1498 goto out;
1499}
1500
1501/* Exit handler for the SCTP protocol. */
dda91928 1502static __exit void sctp_exit(void)
1da177e4
LT
1503{
1504 /* BUG. This should probably do something useful like clean
1505 * up all the remaining associations and all that memory.
1506 */
1507
827bf122
SS
1508 /* Unregister with inet6/inet layers. */
1509 sctp_v6_del_protocol();
270637ab 1510 sctp_v4_del_protocol();
1da177e4 1511
4db67e80
EB
1512 unregister_pernet_subsys(&sctp_net_ops);
1513
270637ab
VY
1514 /* Free protosw registrations */
1515 sctp_v6_protosw_exit();
1516 sctp_v4_protosw_exit();
1517
827bf122 1518 /* Unregister with socket layer. */
270637ab
VY
1519 sctp_v6_pf_exit();
1520 sctp_v4_pf_exit();
827bf122 1521
1da177e4 1522 sctp_sysctl_unregister();
1da177e4
LT
1523
1524 free_pages((unsigned long)sctp_assoc_hashtable,
1525 get_order(sctp_assoc_hashsize *
1526 sizeof(struct sctp_hashbucket)));
1527 kfree(sctp_ep_hashtable);
1528 free_pages((unsigned long)sctp_port_hashtable,
1529 get_order(sctp_port_hashsize *
1530 sizeof(struct sctp_bind_hashbucket)));
1531
632c928a 1532 percpu_counter_destroy(&sctp_sockets_allocated);
1da177e4 1533
eaa184a1
JDB
1534 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1535
827bf122
SS
1536 kmem_cache_destroy(sctp_chunk_cachep);
1537 kmem_cache_destroy(sctp_bucket_cachep);
1da177e4
LT
1538}
1539
1540module_init(sctp_init);
1541module_exit(sctp_exit);
1542
bb97d31f
ACM
1543/*
1544 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1545 */
1546MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132");
882a382c 1547MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132");
91705c61 1548MODULE_AUTHOR("Linux Kernel SCTP developers <linux-sctp@vger.kernel.org>");
1da177e4 1549MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
71acc0dd
DM
1550module_param_named(no_checksums, sctp_checksum_disable, bool, 0644);
1551MODULE_PARM_DESC(no_checksums, "Disable checksums computing and verification");
1da177e4 1552MODULE_LICENSE("GPL");