packet: minor: remove unused err assignment
[GitHub/mt8127/android_kernel_alcatel_ttab.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
26 * along with GNU CC; see the file COPYING. If not, write to
27 * the Free Software Foundation, 59 Temple Place - Suite 330,
28 * Boston, MA 02111-1307, USA.
29 *
30 * Please send any bug reports or fixes you make to the
31 * email address(es):
32 * lksctp developers <lksctp-developers@lists.sourceforge.net>
33 *
34 * Or submit a bug report through the following website:
35 * http://www.sf.net/projects/lksctp
36 *
37 * Written or modified by:
38 * La Monte H.P. Yarroll <piggy@acm.org>
39 * Karl Knutson <karl@athena.chicago.il.us>
40 * Jon Grimm <jgrimm@us.ibm.com>
41 * Sridhar Samudrala <sri@us.ibm.com>
42 * Daisy Chang <daisyc@us.ibm.com>
43 * Ardelle Fan <ardelle.fan@intel.com>
44 *
45 * Any bugs reported given to us we will try to fix... any fixes shared will
46 * be incorporated into the next SCTP release.
47 */
48
145ce502
JP
49#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
50
1da177e4
LT
51#include <linux/module.h>
52#include <linux/init.h>
53#include <linux/netdevice.h>
54#include <linux/inetdevice.h>
55#include <linux/seq_file.h>
4d93df0a 56#include <linux/bootmem.h>
845525a6
VY
57#include <linux/highmem.h>
58#include <linux/swap.h>
5a0e3ad6 59#include <linux/slab.h>
457c4cbc 60#include <net/net_namespace.h>
1da177e4
LT
61#include <net/protocol.h>
62#include <net/ip.h>
63#include <net/ipv6.h>
14c85021 64#include <net/route.h>
1da177e4
LT
65#include <net/sctp/sctp.h>
66#include <net/addrconf.h>
67#include <net/inet_common.h>
68#include <net/inet_ecn.h>
69
70/* Global data structures. */
4cbf1cae 71struct sctp_globals sctp_globals __read_mostly;
1da177e4
LT
72
73struct idr sctp_assocs_id;
74DEFINE_SPINLOCK(sctp_assocs_id_lock);
75
1da177e4
LT
76static struct sctp_pf *sctp_pf_inet6_specific;
77static struct sctp_pf *sctp_pf_inet_specific;
78static struct sctp_af *sctp_af_v4_specific;
79static struct sctp_af *sctp_af_v6_specific;
80
e18b890b
CL
81struct kmem_cache *sctp_chunk_cachep __read_mostly;
82struct kmem_cache *sctp_bucket_cachep __read_mostly;
1da177e4 83
8d987e5c 84long sysctl_sctp_mem[3];
007e3936
VY
85int sysctl_sctp_rmem[3];
86int sysctl_sctp_wmem[3];
4d93df0a 87
1da177e4 88/* Set up the proc fs entry for the SCTP protocol. */
13d782f6 89static __net_init int sctp_proc_init(struct net *net)
1da177e4 90{
c4e85f82 91#ifdef CONFIG_PROC_FS
13d782f6
EB
92 net->sctp.proc_net_sctp = proc_net_mkdir(net, "sctp", net->proc_net);
93 if (!net->sctp.proc_net_sctp)
94 goto out_proc_net_sctp;
95 if (sctp_snmp_proc_init(net))
80896a35 96 goto out_snmp_proc_init;
13d782f6 97 if (sctp_eps_proc_init(net))
80896a35 98 goto out_eps_proc_init;
13d782f6 99 if (sctp_assocs_proc_init(net))
80896a35 100 goto out_assocs_proc_init;
13d782f6 101 if (sctp_remaddr_proc_init(net))
caea902f 102 goto out_remaddr_proc_init;
1da177e4
LT
103
104 return 0;
105
caea902f 106out_remaddr_proc_init:
13d782f6 107 sctp_assocs_proc_exit(net);
80896a35 108out_assocs_proc_init:
13d782f6 109 sctp_eps_proc_exit(net);
80896a35 110out_eps_proc_init:
13d782f6 111 sctp_snmp_proc_exit(net);
80896a35 112out_snmp_proc_init:
13d782f6
EB
113 remove_proc_entry("sctp", net->proc_net);
114 net->sctp.proc_net_sctp = NULL;
115out_proc_net_sctp:
157653ce 116 return -ENOMEM;
13d782f6
EB
117#endif /* CONFIG_PROC_FS */
118 return 0;
1da177e4
LT
119}
120
d808ad9a 121/* Clean up the proc fs entry for the SCTP protocol.
1da177e4
LT
122 * Note: Do not make this __exit as it is used in the init error
123 * path.
124 */
13d782f6 125static void sctp_proc_exit(struct net *net)
1da177e4 126{
c4e85f82 127#ifdef CONFIG_PROC_FS
13d782f6
EB
128 sctp_snmp_proc_exit(net);
129 sctp_eps_proc_exit(net);
130 sctp_assocs_proc_exit(net);
131 sctp_remaddr_proc_exit(net);
132
133 remove_proc_entry("sctp", net->proc_net);
134 net->sctp.proc_net_sctp = NULL;
c4e85f82 135#endif
1da177e4
LT
136}
137
138/* Private helper to extract ipv4 address and stash them in
139 * the protocol structure.
140 */
141static void sctp_v4_copy_addrlist(struct list_head *addrlist,
142 struct net_device *dev)
143{
144 struct in_device *in_dev;
145 struct in_ifaddr *ifa;
146 struct sctp_sockaddr_entry *addr;
147
148 rcu_read_lock();
e5ed6399 149 if ((in_dev = __in_dev_get_rcu(dev)) == NULL) {
1da177e4
LT
150 rcu_read_unlock();
151 return;
152 }
153
154 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
155 /* Add the address to the local list. */
156 addr = t_new(struct sctp_sockaddr_entry, GFP_ATOMIC);
157 if (addr) {
2a6fd78a
AV
158 addr->a.v4.sin_family = AF_INET;
159 addr->a.v4.sin_port = 0;
160 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
29303547
VY
161 addr->valid = 1;
162 INIT_LIST_HEAD(&addr->list);
1da177e4
LT
163 list_add_tail(&addr->list, addrlist);
164 }
165 }
166
167 rcu_read_unlock();
168}
169
170/* Extract our IP addresses from the system and stash them in the
171 * protocol structure.
172 */
4db67e80 173static void sctp_get_local_addr_list(struct net *net)
1da177e4
LT
174{
175 struct net_device *dev;
176 struct list_head *pos;
177 struct sctp_af *af;
178
c6d14c84 179 rcu_read_lock();
4db67e80 180 for_each_netdev_rcu(net, dev) {
1da177e4
LT
181 __list_for_each(pos, &sctp_address_families) {
182 af = list_entry(pos, struct sctp_af, list);
4db67e80 183 af->copy_addrlist(&net->sctp.local_addr_list, dev);
1da177e4
LT
184 }
185 }
c6d14c84 186 rcu_read_unlock();
1da177e4
LT
187}
188
1da177e4 189/* Free the existing local addresses. */
4db67e80 190static void sctp_free_local_addr_list(struct net *net)
1da177e4
LT
191{
192 struct sctp_sockaddr_entry *addr;
193 struct list_head *pos, *temp;
194
4db67e80 195 list_for_each_safe(pos, temp, &net->sctp.local_addr_list) {
1da177e4
LT
196 addr = list_entry(pos, struct sctp_sockaddr_entry, list);
197 list_del(pos);
198 kfree(addr);
199 }
200}
201
1da177e4 202/* Copy the local addresses which are valid for 'scope' into 'bp'. */
4db67e80
EB
203int sctp_copy_local_addr_list(struct net *net, struct sctp_bind_addr *bp,
204 sctp_scope_t scope, gfp_t gfp, int copy_flags)
1da177e4
LT
205{
206 struct sctp_sockaddr_entry *addr;
207 int error = 0;
1da177e4 208
29303547 209 rcu_read_lock();
4db67e80 210 list_for_each_entry_rcu(addr, &net->sctp.local_addr_list, list) {
29303547
VY
211 if (!addr->valid)
212 continue;
e7ff4a70 213 if (sctp_in_scope(net, &addr->a, scope)) {
1da177e4
LT
214 /* Now that the address is in scope, check to see if
215 * the address type is really supported by the local
216 * sock as well as the remote peer.
217 */
6244be4e 218 if ((((AF_INET == addr->a.sa.sa_family) &&
1da177e4 219 (copy_flags & SCTP_ADDR4_PEERSUPP))) ||
6244be4e 220 (((AF_INET6 == addr->a.sa.sa_family) &&
1da177e4
LT
221 (copy_flags & SCTP_ADDR6_ALLOWED) &&
222 (copy_flags & SCTP_ADDR6_PEERSUPP)))) {
f57d96b2
VY
223 error = sctp_add_bind_addr(bp, &addr->a,
224 SCTP_ADDR_SRC, GFP_ATOMIC);
1da177e4
LT
225 if (error)
226 goto end_copy;
227 }
228 }
229 }
230
231end_copy:
29303547 232 rcu_read_unlock();
1da177e4
LT
233 return error;
234}
235
236/* Initialize a sctp_addr from in incoming skb. */
237static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
238 int is_saddr)
239{
240 void *from;
d55c41b1 241 __be16 *port;
1da177e4
LT
242 struct sctphdr *sh;
243
244 port = &addr->v4.sin_port;
245 addr->v4.sin_family = AF_INET;
246
2c0fd387 247 sh = sctp_hdr(skb);
1da177e4 248 if (is_saddr) {
d55c41b1 249 *port = sh->source;
eddc9ec5 250 from = &ip_hdr(skb)->saddr;
1da177e4 251 } else {
d55c41b1 252 *port = sh->dest;
eddc9ec5 253 from = &ip_hdr(skb)->daddr;
1da177e4
LT
254 }
255 memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr));
256}
257
258/* Initialize an sctp_addr from a socket. */
259static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk)
260{
261 addr->v4.sin_family = AF_INET;
7dcdbd95 262 addr->v4.sin_port = 0;
c720c7e8 263 addr->v4.sin_addr.s_addr = inet_sk(sk)->inet_rcv_saddr;
1da177e4
LT
264}
265
266/* Initialize sk->sk_rcv_saddr from sctp_addr. */
267static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk)
268{
c720c7e8 269 inet_sk(sk)->inet_rcv_saddr = addr->v4.sin_addr.s_addr;
1da177e4
LT
270}
271
272/* Initialize sk->sk_daddr from sctp_addr. */
273static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk)
274{
c720c7e8 275 inet_sk(sk)->inet_daddr = addr->v4.sin_addr.s_addr;
1da177e4
LT
276}
277
278/* Initialize a sctp_addr from an address parameter. */
279static void sctp_v4_from_addr_param(union sctp_addr *addr,
280 union sctp_addr_param *param,
dd86d136 281 __be16 port, int iif)
1da177e4
LT
282{
283 addr->v4.sin_family = AF_INET;
284 addr->v4.sin_port = port;
285 addr->v4.sin_addr.s_addr = param->v4.addr.s_addr;
286}
287
288/* Initialize an address parameter from a sctp_addr and return the length
289 * of the address parameter.
290 */
291static int sctp_v4_to_addr_param(const union sctp_addr *addr,
292 union sctp_addr_param *param)
293{
294 int length = sizeof(sctp_ipv4addr_param_t);
295
296 param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS;
dbc16db1 297 param->v4.param_hdr.length = htons(length);
d808ad9a 298 param->v4.addr.s_addr = addr->v4.sin_addr.s_addr;
1da177e4
LT
299
300 return length;
301}
302
303/* Initialize a sctp_addr from a dst_entry. */
18a353f4 304static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct flowi4 *fl4,
854d43a4 305 __be16 port)
1da177e4 306{
1da177e4
LT
307 saddr->v4.sin_family = AF_INET;
308 saddr->v4.sin_port = port;
18a353f4 309 saddr->v4.sin_addr.s_addr = fl4->saddr;
1da177e4
LT
310}
311
312/* Compare two addresses exactly. */
313static int sctp_v4_cmp_addr(const union sctp_addr *addr1,
314 const union sctp_addr *addr2)
315{
316 if (addr1->sa.sa_family != addr2->sa.sa_family)
317 return 0;
318 if (addr1->v4.sin_port != addr2->v4.sin_port)
319 return 0;
320 if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr)
321 return 0;
322
323 return 1;
324}
325
326/* Initialize addr struct to INADDR_ANY. */
6fbfa9f9 327static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port)
1da177e4
LT
328{
329 addr->v4.sin_family = AF_INET;
e6f1cebf 330 addr->v4.sin_addr.s_addr = htonl(INADDR_ANY);
1da177e4
LT
331 addr->v4.sin_port = port;
332}
333
334/* Is this a wildcard address? */
335static int sctp_v4_is_any(const union sctp_addr *addr)
336{
e6f1cebf 337 return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr;
1da177e4
LT
338}
339
340/* This function checks if the address is a valid address to be used for
341 * SCTP binding.
342 *
343 * Output:
344 * Return 0 - If the address is a non-unicast or an illegal address.
345 * Return 1 - If the address is a unicast.
346 */
5636bef7
VY
347static int sctp_v4_addr_valid(union sctp_addr *addr,
348 struct sctp_sock *sp,
349 const struct sk_buff *skb)
1da177e4 350{
7dab83de
VY
351 /* IPv4 addresses not allowed */
352 if (sp && ipv6_only_sock(sctp_opt2sk(sp)))
353 return 0;
354
1da177e4 355 /* Is this a non-unicast address or a unusable SCTP address? */
b5cb2bbc 356 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr))
1da177e4
LT
357 return 0;
358
d808ad9a 359 /* Is this a broadcast address? */
511c3f92 360 if (skb && skb_rtable(skb)->rt_flags & RTCF_BROADCAST)
d808ad9a 361 return 0;
5636bef7 362
1da177e4
LT
363 return 1;
364}
365
366/* Should this be available for binding? */
367static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp)
368{
bb2db45b
EB
369 struct net *net = sock_net(&sp->inet.sk);
370 int ret = inet_addr_type(net, addr->v4.sin_addr.s_addr);
1da177e4 371
1da177e4 372
e6f1cebf 373 if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) &&
cdac4e07
NH
374 ret != RTN_LOCAL &&
375 !sp->inet.freebind &&
376 !sysctl_ip_nonlocal_bind)
1da177e4 377 return 0;
cdac4e07 378
7dab83de
VY
379 if (ipv6_only_sock(sctp_opt2sk(sp)))
380 return 0;
381
1da177e4
LT
382 return 1;
383}
384
385/* Checking the loopback, private and other address scopes as defined in
386 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
387 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
388 *
389 * Level 0 - unusable SCTP addresses
390 * Level 1 - loopback address
391 * Level 2 - link-local addresses
392 * Level 3 - private addresses.
393 * Level 4 - global addresses
394 * For INIT and INIT-ACK address list, let L be the level of
395 * of requested destination address, sender and receiver
396 * SHOULD include all of its addresses with level greater
397 * than or equal to L.
72388433
BD
398 *
399 * IPv4 scoping can be controlled through sysctl option
400 * net.sctp.addr_scope_policy
1da177e4
LT
401 */
402static sctp_scope_t sctp_v4_scope(union sctp_addr *addr)
403{
404 sctp_scope_t retval;
405
1da177e4 406 /* Check for unusable SCTP addresses. */
b5cb2bbc 407 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) {
1da177e4 408 retval = SCTP_SCOPE_UNUSABLE;
b5cb2bbc 409 } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) {
1da177e4 410 retval = SCTP_SCOPE_LOOPBACK;
b5cb2bbc 411 } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) {
1da177e4 412 retval = SCTP_SCOPE_LINK;
b5cb2bbc
JP
413 } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) ||
414 ipv4_is_private_172(addr->v4.sin_addr.s_addr) ||
415 ipv4_is_private_192(addr->v4.sin_addr.s_addr)) {
1da177e4
LT
416 retval = SCTP_SCOPE_PRIVATE;
417 } else {
418 retval = SCTP_SCOPE_GLOBAL;
419 }
420
421 return retval;
422}
423
424/* Returns a valid dst cache entry for the given source and destination ip
425 * addresses. If an association is passed, trys to get a dst entry with a
426 * source address that matches an address in the bind address list.
427 */
da0420be
VY
428static void sctp_v4_get_dst(struct sctp_transport *t, union sctp_addr *saddr,
429 struct flowi *fl, struct sock *sk)
1da177e4 430{
da0420be 431 struct sctp_association *asoc = t->asoc;
1da177e4 432 struct rtable *rt;
9914ae3c 433 struct flowi4 *fl4 = &fl->u.ip4;
1da177e4 434 struct sctp_bind_addr *bp;
1da177e4 435 struct sctp_sockaddr_entry *laddr;
1da177e4 436 struct dst_entry *dst = NULL;
da0420be 437 union sctp_addr *daddr = &t->ipaddr;
1da177e4
LT
438 union sctp_addr dst_saddr;
439
9914ae3c
VY
440 memset(fl4, 0x0, sizeof(struct flowi4));
441 fl4->daddr = daddr->v4.sin_addr.s_addr;
442 fl4->fl4_dport = daddr->v4.sin_port;
443 fl4->flowi4_proto = IPPROTO_SCTP;
1da177e4 444 if (asoc) {
9914ae3c
VY
445 fl4->flowi4_tos = RT_CONN_FLAGS(asoc->base.sk);
446 fl4->flowi4_oif = asoc->base.sk->sk_bound_dev_if;
447 fl4->fl4_sport = htons(asoc->base.bind_addr.port);
1da177e4 448 }
6429d3dc 449 if (saddr) {
9914ae3c
VY
450 fl4->saddr = saddr->v4.sin_addr.s_addr;
451 fl4->fl4_sport = saddr->v4.sin_port;
6429d3dc 452 }
1da177e4 453
21454aaa 454 SCTP_DEBUG_PRINTK("%s: DST:%pI4, SRC:%pI4 - ",
9914ae3c 455 __func__, &fl4->daddr, &fl4->saddr);
1da177e4 456
bb2db45b 457 rt = ip_route_output_key(sock_net(sk), fl4);
b23dd4fe 458 if (!IS_ERR(rt))
d8d1f30b 459 dst = &rt->dst;
1da177e4
LT
460
461 /* If there is no association or if a source address is passed, no
462 * more validation is required.
463 */
464 if (!asoc || saddr)
465 goto out;
466
467 bp = &asoc->base.bind_addr;
1da177e4
LT
468
469 if (dst) {
470 /* Walk through the bind address list and look for a bind
471 * address that matches the source address of the returned dst.
472 */
18a353f4 473 sctp_v4_dst_saddr(&dst_saddr, fl4, htons(bp->port));
559cf710
VY
474 rcu_read_lock();
475 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
8a07eb0a
MH
476 if (!laddr->valid || (laddr->state == SCTP_ADDR_DEL) ||
477 (laddr->state != SCTP_ADDR_SRC &&
478 !asoc->src_out_of_asoc_ok))
dc022a98 479 continue;
854d43a4 480 if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a))
1da177e4
LT
481 goto out_unlock;
482 }
559cf710 483 rcu_read_unlock();
1da177e4
LT
484
485 /* None of the bound addresses match the source address of the
486 * dst. So release it.
487 */
488 dst_release(dst);
489 dst = NULL;
490 }
491
492 /* Walk through the bind address list and try to get a dst that
493 * matches a bind address as the source address.
494 */
559cf710
VY
495 rcu_read_lock();
496 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
497 if (!laddr->valid)
498 continue;
f57d96b2 499 if ((laddr->state == SCTP_ADDR_SRC) &&
6244be4e 500 (AF_INET == laddr->a.sa.sa_family)) {
9914ae3c
VY
501 fl4->saddr = laddr->a.v4.sin_addr.s_addr;
502 fl4->fl4_sport = laddr->a.v4.sin_port;
bb2db45b 503 rt = ip_route_output_key(sock_net(sk), fl4);
b23dd4fe 504 if (!IS_ERR(rt)) {
d8d1f30b 505 dst = &rt->dst;
1da177e4
LT
506 goto out_unlock;
507 }
508 }
509 }
510
511out_unlock:
559cf710 512 rcu_read_unlock();
1da177e4 513out:
da0420be 514 t->dst = dst;
1da177e4 515 if (dst)
21454aaa 516 SCTP_DEBUG_PRINTK("rt_dst:%pI4, rt_src:%pI4\n",
18a353f4 517 &fl4->daddr, &fl4->saddr);
1da177e4
LT
518 else
519 SCTP_DEBUG_PRINTK("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,
6257ff21 555 sk->sk_prot);
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
581/* Map address, empty for v4 family */
582static void sctp_v4_addr_v4map(struct sctp_sock *sp, union sctp_addr *addr)
583{
584 /* Empty */
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
9f7d653b
MH
598void sctp_addr_wq_timeout_handler(unsigned long arg)
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) {
9f7d653b 607 SCTP_DEBUG_PRINTK_IPADDR("sctp_addrwq_timo_handler: the first ent in wq %p is ",
4db67e80 608 " for cmd %d at entry %p\n", &net->sctp.addr_waitq, &addrw->a, addrw->state,
9f7d653b
MH
609 addrw);
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
626 SCTP_DEBUG_PRINTK("sctp_timo_handler: this is on DAD, trying %d sec later\n",
627 SCTP_ADDRESS_TICK_DELAY);
628 timeo_val = jiffies;
629 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
4db67e80 630 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
9f7d653b
MH
631 break;
632 }
633 }
5d0c90cf 634#endif
4db67e80 635 list_for_each_entry(sp, &net->sctp.auto_asconf_splist, auto_asconf_list) {
9f7d653b
MH
636 struct sock *sk;
637
638 sk = sctp_opt2sk(sp);
639 /* ignore bound-specific endpoints */
640 if (!sctp_is_ep_boundall(sk))
641 continue;
642 sctp_bh_lock_sock(sk);
643 if (sctp_asconf_mgmt(sp, addrw) < 0)
644 SCTP_DEBUG_PRINTK("sctp_addrwq_timo_handler: sctp_asconf_mgmt failed\n");
645 sctp_bh_unlock_sock(sk);
646 }
39d84a58 647#if IS_ENABLED(CONFIG_IPV6)
9f7d653b 648free_next:
39d84a58 649#endif
9f7d653b
MH
650 list_del(&addrw->list);
651 kfree(addrw);
652 }
4db67e80 653 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
654}
655
4db67e80 656static void sctp_free_addr_wq(struct net *net)
9f7d653b
MH
657{
658 struct sctp_sockaddr_entry *addrw;
659 struct sctp_sockaddr_entry *temp;
660
4db67e80
EB
661 spin_lock_bh(&net->sctp.addr_wq_lock);
662 del_timer(&net->sctp.addr_wq_timer);
663 list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) {
9f7d653b
MH
664 list_del(&addrw->list);
665 kfree(addrw);
666 }
4db67e80 667 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
668}
669
670/* lookup the entry for the same address in the addr_waitq
671 * sctp_addr_wq MUST be locked
672 */
4db67e80
EB
673static struct sctp_sockaddr_entry *sctp_addr_wq_lookup(struct net *net,
674 struct sctp_sockaddr_entry *addr)
9f7d653b
MH
675{
676 struct sctp_sockaddr_entry *addrw;
677
4db67e80 678 list_for_each_entry(addrw, &net->sctp.addr_waitq, list) {
9f7d653b
MH
679 if (addrw->a.sa.sa_family != addr->a.sa.sa_family)
680 continue;
681 if (addrw->a.sa.sa_family == AF_INET) {
682 if (addrw->a.v4.sin_addr.s_addr ==
683 addr->a.v4.sin_addr.s_addr)
684 return addrw;
685 } else if (addrw->a.sa.sa_family == AF_INET6) {
686 if (ipv6_addr_equal(&addrw->a.v6.sin6_addr,
687 &addr->a.v6.sin6_addr))
688 return addrw;
689 }
690 }
691 return NULL;
692}
693
4db67e80 694void sctp_addr_wq_mgmt(struct net *net, struct sctp_sockaddr_entry *addr, int cmd)
9f7d653b
MH
695{
696 struct sctp_sockaddr_entry *addrw;
697 unsigned long timeo_val;
698
699 /* first, we check if an opposite message already exist in the queue.
700 * If we found such message, it is removed.
701 * This operation is a bit stupid, but the DHCP client attaches the
702 * new address after a couple of addition and deletion of that address
703 */
704
4db67e80 705 spin_lock_bh(&net->sctp.addr_wq_lock);
9f7d653b 706 /* Offsets existing events in addr_wq */
4db67e80 707 addrw = sctp_addr_wq_lookup(net, addr);
9f7d653b
MH
708 if (addrw) {
709 if (addrw->state != cmd) {
710 SCTP_DEBUG_PRINTK_IPADDR("sctp_addr_wq_mgmt offsets existing entry for %d ",
711 " in wq %p\n", addrw->state, &addrw->a,
4db67e80 712 &net->sctp.addr_waitq);
9f7d653b
MH
713 list_del(&addrw->list);
714 kfree(addrw);
715 }
4db67e80 716 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
717 return;
718 }
719
720 /* OK, we have to add the new address to the wait queue */
721 addrw = kmemdup(addr, sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
722 if (addrw == NULL) {
4db67e80 723 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
724 return;
725 }
726 addrw->state = cmd;
4db67e80 727 list_add_tail(&addrw->list, &net->sctp.addr_waitq);
9f7d653b 728 SCTP_DEBUG_PRINTK_IPADDR("sctp_addr_wq_mgmt add new entry for cmd:%d ",
4db67e80 729 " in wq %p\n", addrw->state, &addrw->a, &net->sctp.addr_waitq);
9f7d653b 730
4db67e80 731 if (!timer_pending(&net->sctp.addr_wq_timer)) {
9f7d653b
MH
732 timeo_val = jiffies;
733 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
4db67e80 734 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
9f7d653b 735 }
4db67e80 736 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
737}
738
29303547
VY
739/* Event handler for inet address addition/deletion events.
740 * The sctp_local_addr_list needs to be protocted by a spin lock since
741 * multiple notifiers (say IPv4 and IPv6) may be running at the same
742 * time and thus corrupt the list.
743 * The reader side is protected with RCU.
744 */
24123186
AB
745static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev,
746 void *ptr)
1da177e4 747{
29c7cf96 748 struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
29303547
VY
749 struct sctp_sockaddr_entry *addr = NULL;
750 struct sctp_sockaddr_entry *temp;
4db67e80 751 struct net *net = dev_net(ifa->ifa_dev->dev);
22626216 752 int found = 0;
1da177e4 753
29c7cf96
SS
754 switch (ev) {
755 case NETDEV_UP:
756 addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
757 if (addr) {
758 addr->a.v4.sin_family = AF_INET;
759 addr->a.v4.sin_port = 0;
760 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
29303547 761 addr->valid = 1;
4db67e80
EB
762 spin_lock_bh(&net->sctp.local_addr_lock);
763 list_add_tail_rcu(&addr->list, &net->sctp.local_addr_list);
764 sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_NEW);
765 spin_unlock_bh(&net->sctp.local_addr_lock);
29c7cf96
SS
766 }
767 break;
768 case NETDEV_DOWN:
4db67e80 769 spin_lock_bh(&net->sctp.local_addr_lock);
29303547 770 list_for_each_entry_safe(addr, temp,
4db67e80 771 &net->sctp.local_addr_list, list) {
a40a7d15
PE
772 if (addr->a.sa.sa_family == AF_INET &&
773 addr->a.v4.sin_addr.s_addr ==
774 ifa->ifa_local) {
4db67e80 775 sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_DEL);
22626216 776 found = 1;
29303547
VY
777 addr->valid = 0;
778 list_del_rcu(&addr->list);
29c7cf96
SS
779 break;
780 }
781 }
4db67e80 782 spin_unlock_bh(&net->sctp.local_addr_lock);
22626216 783 if (found)
1231f0ba 784 kfree_rcu(addr, rcu);
29c7cf96
SS
785 break;
786 }
1da177e4
LT
787
788 return NOTIFY_DONE;
789}
790
791/*
792 * Initialize the control inode/socket with a control endpoint data
793 * structure. This endpoint is reserved exclusively for the OOTB processing.
794 */
2ce95503 795static int sctp_ctl_sock_init(struct net *net)
1da177e4
LT
796{
797 int err;
fb13d9f9 798 sa_family_t family = PF_INET;
1da177e4
LT
799
800 if (sctp_get_pf_specific(PF_INET6))
801 family = PF_INET6;
1da177e4 802
2ce95503
EB
803 err = inet_ctl_sock_create(&net->sctp.ctl_sock, family,
804 SOCK_SEQPACKET, IPPROTO_SCTP, net);
fb13d9f9
BH
805
806 /* If IPv6 socket could not be created, try the IPv4 socket */
807 if (err < 0 && family == PF_INET6)
2ce95503 808 err = inet_ctl_sock_create(&net->sctp.ctl_sock, AF_INET,
fb13d9f9 809 SOCK_SEQPACKET, IPPROTO_SCTP,
2ce95503 810 net);
fb13d9f9 811
1da177e4 812 if (err < 0) {
145ce502 813 pr_err("Failed to create the SCTP control socket\n");
1da177e4
LT
814 return err;
815 }
1da177e4
LT
816 return 0;
817}
818
819/* Register address family specific functions. */
820int sctp_register_af(struct sctp_af *af)
821{
822 switch (af->sa_family) {
823 case AF_INET:
824 if (sctp_af_v4_specific)
825 return 0;
826 sctp_af_v4_specific = af;
827 break;
828 case AF_INET6:
829 if (sctp_af_v6_specific)
830 return 0;
831 sctp_af_v6_specific = af;
832 break;
833 default:
834 return 0;
835 }
836
837 INIT_LIST_HEAD(&af->list);
838 list_add_tail(&af->list, &sctp_address_families);
839 return 1;
840}
841
842/* Get the table of functions for manipulating a particular address
843 * family.
844 */
845struct sctp_af *sctp_get_af_specific(sa_family_t family)
846{
847 switch (family) {
848 case AF_INET:
849 return sctp_af_v4_specific;
850 case AF_INET6:
851 return sctp_af_v6_specific;
852 default:
853 return NULL;
854 }
855}
856
857/* Common code to initialize a AF_INET msg_name. */
858static void sctp_inet_msgname(char *msgname, int *addr_len)
859{
860 struct sockaddr_in *sin;
861
862 sin = (struct sockaddr_in *)msgname;
863 *addr_len = sizeof(struct sockaddr_in);
864 sin->sin_family = AF_INET;
865 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
866}
867
868/* Copy the primary address of the peer primary address as the msg_name. */
869static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname,
870 int *addr_len)
871{
872 struct sockaddr_in *sin, *sinfrom;
873
874 if (msgname) {
875 struct sctp_association *asoc;
876
877 asoc = event->asoc;
878 sctp_inet_msgname(msgname, addr_len);
879 sin = (struct sockaddr_in *)msgname;
880 sinfrom = &asoc->peer.primary_addr.v4;
881 sin->sin_port = htons(asoc->peer.port);
882 sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr;
883 }
884}
885
886/* Initialize and copy out a msgname from an inbound skb. */
887static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len)
888{
1da177e4 889 if (msgname) {
2c0fd387
ACM
890 struct sctphdr *sh = sctp_hdr(skb);
891 struct sockaddr_in *sin = (struct sockaddr_in *)msgname;
892
1da177e4 893 sctp_inet_msgname(msgname, len);
1da177e4 894 sin->sin_port = sh->source;
eddc9ec5 895 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
1da177e4
LT
896 }
897}
898
899/* Do we support this AF? */
900static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp)
901{
902 /* PF_INET only supports AF_INET addresses. */
a02cec21 903 return AF_INET == family;
1da177e4
LT
904}
905
906/* Address matching with wildcards allowed. */
907static int sctp_inet_cmp_addr(const union sctp_addr *addr1,
908 const union sctp_addr *addr2,
909 struct sctp_sock *opt)
910{
911 /* PF_INET only supports AF_INET addresses. */
912 if (addr1->sa.sa_family != addr2->sa.sa_family)
913 return 0;
e6f1cebf
AV
914 if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr ||
915 htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr)
1da177e4
LT
916 return 1;
917 if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr)
918 return 1;
919
920 return 0;
921}
922
923/* Verify that provided sockaddr looks bindable. Common verification has
924 * already been taken care of.
925 */
926static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr)
927{
928 return sctp_v4_available(addr, opt);
929}
930
931/* Verify that sockaddr looks sendable. Common verification has already
932 * been taken care of.
933 */
934static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr)
935{
936 return 1;
937}
938
939/* Fill in Supported Address Type information for INIT and INIT-ACK
940 * chunks. Returns number of addresses supported.
941 */
942static int sctp_inet_supported_addrs(const struct sctp_sock *opt,
3dbe8656 943 __be16 *types)
1da177e4
LT
944{
945 types[0] = SCTP_PARAM_IPV4_ADDRESS;
946 return 1;
947}
948
949/* Wrapper routine that calls the ip transmit routine. */
950static inline int sctp_v4_xmit(struct sk_buff *skb,
f880374c 951 struct sctp_transport *transport)
1da177e4 952{
f880374c
HX
953 struct inet_sock *inet = inet_sk(skb->sk);
954
21454aaa 955 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n",
0dc47877 956 __func__, skb, skb->len,
7ef73bca
DM
957 &transport->fl.u.ip4.saddr,
958 &transport->fl.u.ip4.daddr);
1da177e4 959
f880374c
HX
960 inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ?
961 IP_PMTUDISC_DO : IP_PMTUDISC_DONT;
962
b01a2407 963 SCTP_INC_STATS(sock_net(&inet->sk), SCTP_MIB_OUTSCTPPACKS);
d9d8da80 964 return ip_queue_xmit(skb, &transport->fl);
1da177e4
LT
965}
966
15efbe76 967static struct sctp_af sctp_af_inet;
1da177e4
LT
968
969static struct sctp_pf sctp_pf_inet = {
970 .event_msgname = sctp_inet_event_msgname,
971 .skb_msgname = sctp_inet_skb_msgname,
972 .af_supported = sctp_inet_af_supported,
973 .cmp_addr = sctp_inet_cmp_addr,
974 .bind_verify = sctp_inet_bind_verify,
975 .send_verify = sctp_inet_send_verify,
976 .supported_addrs = sctp_inet_supported_addrs,
977 .create_accept_sk = sctp_v4_create_accept_sk,
978 .addr_v4map = sctp_v4_addr_v4map,
15efbe76 979 .af = &sctp_af_inet
1da177e4
LT
980};
981
982/* Notifier for inetaddr addition/deletion events. */
983static struct notifier_block sctp_inetaddr_notifier = {
984 .notifier_call = sctp_inetaddr_event,
985};
986
987/* Socket operations. */
90ddc4f0 988static const struct proto_ops inet_seqpacket_ops = {
543d9cfe
ACM
989 .family = PF_INET,
990 .owner = THIS_MODULE,
991 .release = inet_release, /* Needs to be wrapped... */
992 .bind = inet_bind,
993 .connect = inet_dgram_connect,
994 .socketpair = sock_no_socketpair,
995 .accept = inet_accept,
996 .getname = inet_getname, /* Semantics are different. */
997 .poll = sctp_poll,
998 .ioctl = inet_ioctl,
999 .listen = sctp_inet_listen,
1000 .shutdown = inet_shutdown, /* Looks harmless. */
1001 .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */
1002 .getsockopt = sock_common_getsockopt,
1003 .sendmsg = inet_sendmsg,
1004 .recvmsg = sock_common_recvmsg,
1005 .mmap = sock_no_mmap,
1006 .sendpage = sock_no_sendpage,
3fdadf7d 1007#ifdef CONFIG_COMPAT
543d9cfe
ACM
1008 .compat_setsockopt = compat_sock_common_setsockopt,
1009 .compat_getsockopt = compat_sock_common_getsockopt,
3fdadf7d 1010#endif
1da177e4
LT
1011};
1012
1013/* Registration with AF_INET family. */
1014static struct inet_protosw sctp_seqpacket_protosw = {
1015 .type = SOCK_SEQPACKET,
1016 .protocol = IPPROTO_SCTP,
1017 .prot = &sctp_prot,
1018 .ops = &inet_seqpacket_ops,
1da177e4
LT
1019 .no_check = 0,
1020 .flags = SCTP_PROTOSW_FLAG
1021};
1022static struct inet_protosw sctp_stream_protosw = {
1023 .type = SOCK_STREAM,
1024 .protocol = IPPROTO_SCTP,
1025 .prot = &sctp_prot,
1026 .ops = &inet_seqpacket_ops,
1da177e4
LT
1027 .no_check = 0,
1028 .flags = SCTP_PROTOSW_FLAG
1029};
1030
1031/* Register with IP layer. */
32613090 1032static const struct net_protocol sctp_protocol = {
1da177e4
LT
1033 .handler = sctp_rcv,
1034 .err_handler = sctp_v4_err,
1035 .no_policy = 1,
bb2db45b 1036 .netns_ok = 1,
1da177e4
LT
1037};
1038
1039/* IPv4 address related functions. */
15efbe76 1040static struct sctp_af sctp_af_inet = {
543d9cfe
ACM
1041 .sa_family = AF_INET,
1042 .sctp_xmit = sctp_v4_xmit,
1043 .setsockopt = ip_setsockopt,
1044 .getsockopt = ip_getsockopt,
1045 .get_dst = sctp_v4_get_dst,
1046 .get_saddr = sctp_v4_get_saddr,
1047 .copy_addrlist = sctp_v4_copy_addrlist,
1048 .from_skb = sctp_v4_from_skb,
1049 .from_sk = sctp_v4_from_sk,
1050 .to_sk_saddr = sctp_v4_to_sk_saddr,
1051 .to_sk_daddr = sctp_v4_to_sk_daddr,
1052 .from_addr_param = sctp_v4_from_addr_param,
1053 .to_addr_param = sctp_v4_to_addr_param,
543d9cfe
ACM
1054 .cmp_addr = sctp_v4_cmp_addr,
1055 .addr_valid = sctp_v4_addr_valid,
1056 .inaddr_any = sctp_v4_inaddr_any,
1057 .is_any = sctp_v4_is_any,
1058 .available = sctp_v4_available,
1059 .scope = sctp_v4_scope,
1060 .skb_iif = sctp_v4_skb_iif,
1061 .is_ce = sctp_v4_is_ce,
1062 .seq_dump_addr = sctp_v4_seq_dump_addr,
b9031d9d 1063 .ecn_capable = sctp_v4_ecn_capable,
543d9cfe
ACM
1064 .net_header_len = sizeof(struct iphdr),
1065 .sockaddr_len = sizeof(struct sockaddr_in),
3fdadf7d 1066#ifdef CONFIG_COMPAT
543d9cfe
ACM
1067 .compat_setsockopt = compat_ip_setsockopt,
1068 .compat_getsockopt = compat_ip_getsockopt,
3fdadf7d 1069#endif
1da177e4
LT
1070};
1071
1072struct sctp_pf *sctp_get_pf_specific(sa_family_t family) {
1073
1074 switch (family) {
1075 case PF_INET:
1076 return sctp_pf_inet_specific;
1077 case PF_INET6:
1078 return sctp_pf_inet6_specific;
1079 default:
1080 return NULL;
1081 }
1082}
1083
1084/* Register the PF specific function table. */
1085int sctp_register_pf(struct sctp_pf *pf, sa_family_t family)
1086{
1087 switch (family) {
1088 case PF_INET:
1089 if (sctp_pf_inet_specific)
1090 return 0;
1091 sctp_pf_inet_specific = pf;
1092 break;
1093 case PF_INET6:
1094 if (sctp_pf_inet6_specific)
1095 return 0;
1096 sctp_pf_inet6_specific = pf;
1097 break;
1098 default:
1099 return 0;
1100 }
1101 return 1;
1102}
1103
b01a2407 1104static inline int init_sctp_mibs(struct net *net)
996b1dba 1105{
b01a2407 1106 return snmp_mib_init((void __percpu **)net->sctp.sctp_statistics,
1823e4c8
ED
1107 sizeof(struct sctp_mib),
1108 __alignof__(struct sctp_mib));
1da177e4
LT
1109}
1110
b01a2407 1111static inline void cleanup_sctp_mibs(struct net *net)
1da177e4 1112{
b01a2407 1113 snmp_mib_free((void __percpu **)net->sctp.sctp_statistics);
1da177e4
LT
1114}
1115
270637ab
VY
1116static void sctp_v4_pf_init(void)
1117{
1118 /* Initialize the SCTP specific PF functions. */
1119 sctp_register_pf(&sctp_pf_inet, PF_INET);
1120 sctp_register_af(&sctp_af_inet);
1121}
1122
1123static void sctp_v4_pf_exit(void)
1124{
1125 list_del(&sctp_af_inet.list);
1126}
1127
1128static int sctp_v4_protosw_init(void)
1129{
1130 int rc;
1131
1132 rc = proto_register(&sctp_prot, 1);
1133 if (rc)
1134 return rc;
1135
1136 /* Register SCTP(UDP and TCP style) with socket layer. */
1137 inet_register_protosw(&sctp_seqpacket_protosw);
1138 inet_register_protosw(&sctp_stream_protosw);
1139
1140 return 0;
1141}
1142
1143static void sctp_v4_protosw_exit(void)
1144{
1145 inet_unregister_protosw(&sctp_stream_protosw);
1146 inet_unregister_protosw(&sctp_seqpacket_protosw);
1147 proto_unregister(&sctp_prot);
1148}
1149
1150static int sctp_v4_add_protocol(void)
1151{
1152 /* Register notifier for inet address additions/deletions. */
1153 register_inetaddr_notifier(&sctp_inetaddr_notifier);
1154
1155 /* Register SCTP with inet layer. */
1156 if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0)
1157 return -EAGAIN;
1158
1159 return 0;
1160}
1161
1162static void sctp_v4_del_protocol(void)
1163{
1164 inet_del_protocol(&sctp_protocol, IPPROTO_SCTP);
1165 unregister_inetaddr_notifier(&sctp_inetaddr_notifier);
1166}
1167
4db67e80
EB
1168static int sctp_net_init(struct net *net)
1169{
2ce95503
EB
1170 int status;
1171
e1fc3b14
EB
1172 /*
1173 * 14. Suggested SCTP Protocol Parameter Values
1174 */
1175 /* The following protocol parameters are RECOMMENDED: */
1176 /* RTO.Initial - 3 seconds */
1177 net->sctp.rto_initial = SCTP_RTO_INITIAL;
1178 /* RTO.Min - 1 second */
1179 net->sctp.rto_min = SCTP_RTO_MIN;
1180 /* RTO.Max - 60 seconds */
1181 net->sctp.rto_max = SCTP_RTO_MAX;
1182 /* RTO.Alpha - 1/8 */
1183 net->sctp.rto_alpha = SCTP_RTO_ALPHA;
1184 /* RTO.Beta - 1/4 */
1185 net->sctp.rto_beta = SCTP_RTO_BETA;
1186
1187 /* Valid.Cookie.Life - 60 seconds */
1188 net->sctp.valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE;
1189
1190 /* Whether Cookie Preservative is enabled(1) or not(0) */
1191 net->sctp.cookie_preserve_enable = 1;
1192
1193 /* Max.Burst - 4 */
1194 net->sctp.max_burst = SCTP_DEFAULT_MAX_BURST;
1195
1196 /* Association.Max.Retrans - 10 attempts
1197 * Path.Max.Retrans - 5 attempts (per destination address)
1198 * Max.Init.Retransmits - 8 attempts
1199 */
1200 net->sctp.max_retrans_association = 10;
1201 net->sctp.max_retrans_path = 5;
1202 net->sctp.max_retrans_init = 8;
1203
1204 /* Sendbuffer growth - do per-socket accounting */
1205 net->sctp.sndbuf_policy = 0;
1206
1207 /* Rcvbuffer growth - do per-socket accounting */
1208 net->sctp.rcvbuf_policy = 0;
1209
1210 /* HB.interval - 30 seconds */
1211 net->sctp.hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT;
1212
1213 /* delayed SACK timeout */
1214 net->sctp.sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK;
1215
1216 /* Disable ADDIP by default. */
1217 net->sctp.addip_enable = 0;
1218 net->sctp.addip_noauth = 0;
1219 net->sctp.default_auto_asconf = 0;
1220
1221 /* Enable PR-SCTP by default. */
1222 net->sctp.prsctp_enable = 1;
1223
1224 /* Disable AUTH by default. */
1225 net->sctp.auth_enable = 0;
1226
1227 /* Set SCOPE policy to enabled */
1228 net->sctp.scope_policy = SCTP_SCOPE_POLICY_ENABLE;
1229
1230 /* Set the default rwnd update threshold */
1231 net->sctp.rwnd_upd_shift = SCTP_DEFAULT_RWND_SHIFT;
1232
1233 /* Initialize maximum autoclose timeout. */
1234 net->sctp.max_autoclose = INT_MAX / HZ;
1235
ebb7e95d
EB
1236 status = sctp_sysctl_net_register(net);
1237 if (status)
1238 goto err_sysctl_register;
1239
b01a2407
EB
1240 /* Allocate and initialise sctp mibs. */
1241 status = init_sctp_mibs(net);
1242 if (status)
1243 goto err_init_mibs;
1244
13d782f6
EB
1245 /* Initialize proc fs directory. */
1246 status = sctp_proc_init(net);
1247 if (status)
1248 goto err_init_proc;
1249
1250 sctp_dbg_objcnt_init(net);
1251
2ce95503
EB
1252 /* Initialize the control inode/socket for handling OOTB packets. */
1253 if ((status = sctp_ctl_sock_init(net))) {
1254 pr_err("Failed to initialize the SCTP control sock\n");
1255 goto err_ctl_sock_init;
1256 }
1257
4db67e80
EB
1258 /* Initialize the local address list. */
1259 INIT_LIST_HEAD(&net->sctp.local_addr_list);
1260 spin_lock_init(&net->sctp.local_addr_lock);
1261 sctp_get_local_addr_list(net);
1262
1263 /* Initialize the address event list */
1264 INIT_LIST_HEAD(&net->sctp.addr_waitq);
1265 INIT_LIST_HEAD(&net->sctp.auto_asconf_splist);
1266 spin_lock_init(&net->sctp.addr_wq_lock);
1267 net->sctp.addr_wq_timer.expires = 0;
1268 setup_timer(&net->sctp.addr_wq_timer, sctp_addr_wq_timeout_handler,
1269 (unsigned long)net);
1270
1271 return 0;
2ce95503
EB
1272
1273err_ctl_sock_init:
13d782f6
EB
1274 sctp_dbg_objcnt_exit(net);
1275 sctp_proc_exit(net);
1276err_init_proc:
b01a2407
EB
1277 cleanup_sctp_mibs(net);
1278err_init_mibs:
ebb7e95d
EB
1279 sctp_sysctl_net_unregister(net);
1280err_sysctl_register:
2ce95503 1281 return status;
4db67e80
EB
1282}
1283
1284static void sctp_net_exit(struct net *net)
1285{
1286 /* Free the local address list */
1287 sctp_free_addr_wq(net);
1288 sctp_free_local_addr_list(net);
2ce95503
EB
1289
1290 /* Free the control endpoint. */
1291 inet_ctl_sock_destroy(net->sctp.ctl_sock);
13d782f6
EB
1292
1293 sctp_dbg_objcnt_exit(net);
1294
1295 sctp_proc_exit(net);
b01a2407 1296 cleanup_sctp_mibs(net);
ebb7e95d 1297 sctp_sysctl_net_unregister(net);
4db67e80
EB
1298}
1299
1300static struct pernet_operations sctp_net_ops = {
1301 .init = sctp_net_init,
1302 .exit = sctp_net_exit,
1303};
1304
1da177e4
LT
1305/* Initialize the universe into something sensible. */
1306SCTP_STATIC __init int sctp_init(void)
1307{
1308 int i;
1309 int status = -EINVAL;
1310 unsigned long goal;
4d93df0a
NH
1311 unsigned long limit;
1312 int max_share;
1da177e4
LT
1313 int order;
1314
1315 /* SCTP_DEBUG sanity check. */
1316 if (!sctp_sanity_check())
1317 goto out;
1318
827bf122 1319 /* Allocate bind_bucket and chunk caches. */
1da177e4
LT
1320 status = -ENOBUFS;
1321 sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket",
1322 sizeof(struct sctp_bind_bucket),
1323 0, SLAB_HWCACHE_ALIGN,
20c2df83 1324 NULL);
1da177e4 1325 if (!sctp_bucket_cachep)
827bf122 1326 goto out;
1da177e4
LT
1327
1328 sctp_chunk_cachep = kmem_cache_create("sctp_chunk",
1329 sizeof(struct sctp_chunk),
1330 0, SLAB_HWCACHE_ALIGN,
20c2df83 1331 NULL);
1da177e4
LT
1332 if (!sctp_chunk_cachep)
1333 goto err_chunk_cachep;
1334
632c928a
EB
1335 status = percpu_counter_init(&sctp_sockets_allocated, 0);
1336 if (status)
1337 goto err_percpu_counter_init;
1338
1da177e4
LT
1339 /* Implementation specific variables. */
1340
1341 /* Initialize default stream count setup information. */
1342 sctp_max_instreams = SCTP_DEFAULT_INSTREAMS;
1343 sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS;
1344
1345 /* Initialize handle used for association ids. */
1346 idr_init(&sctp_assocs_id);
1347
f03d78db 1348 limit = nr_free_buffer_pages() / 8;
4d93df0a
NH
1349 limit = max(limit, 128UL);
1350 sysctl_sctp_mem[0] = limit / 4 * 3;
1351 sysctl_sctp_mem[1] = limit;
1352 sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2;
1353
1354 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1355 limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7);
1356 max_share = min(4UL*1024*1024, limit);
1357
845525a6 1358 sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */
87fb4b7b 1359 sysctl_sctp_rmem[1] = 1500 * SKB_TRUESIZE(1);
4d93df0a
NH
1360 sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share);
1361
3ab224be 1362 sysctl_sctp_wmem[0] = SK_MEM_QUANTUM;
4d93df0a
NH
1363 sysctl_sctp_wmem[1] = 16*1024;
1364 sysctl_sctp_wmem[2] = max(64*1024, max_share);
1365
1da177e4
LT
1366 /* Size and allocate the association hash table.
1367 * The methodology is similar to that of the tcp hash tables.
1368 */
4481374c
JB
1369 if (totalram_pages >= (128 * 1024))
1370 goal = totalram_pages >> (22 - PAGE_SHIFT);
1da177e4 1371 else
4481374c 1372 goal = totalram_pages >> (24 - PAGE_SHIFT);
1da177e4
LT
1373
1374 for (order = 0; (1UL << order) < goal; order++)
1375 ;
1376
1377 do {
1378 sctp_assoc_hashsize = (1UL << order) * PAGE_SIZE /
1379 sizeof(struct sctp_hashbucket);
1380 if ((sctp_assoc_hashsize > (64 * 1024)) && order > 0)
1381 continue;
1382 sctp_assoc_hashtable = (struct sctp_hashbucket *)
a84b50ce 1383 __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, order);
1da177e4
LT
1384 } while (!sctp_assoc_hashtable && --order > 0);
1385 if (!sctp_assoc_hashtable) {
145ce502 1386 pr_err("Failed association hash alloc\n");
1da177e4
LT
1387 status = -ENOMEM;
1388 goto err_ahash_alloc;
1389 }
1390 for (i = 0; i < sctp_assoc_hashsize; i++) {
1391 rwlock_init(&sctp_assoc_hashtable[i].lock);
d970dbf8 1392 INIT_HLIST_HEAD(&sctp_assoc_hashtable[i].chain);
1da177e4
LT
1393 }
1394
1395 /* Allocate and initialize the endpoint hash table. */
1396 sctp_ep_hashsize = 64;
1397 sctp_ep_hashtable = (struct sctp_hashbucket *)
1398 kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL);
1399 if (!sctp_ep_hashtable) {
145ce502 1400 pr_err("Failed endpoint_hash alloc\n");
1da177e4
LT
1401 status = -ENOMEM;
1402 goto err_ehash_alloc;
1403 }
1404 for (i = 0; i < sctp_ep_hashsize; i++) {
1405 rwlock_init(&sctp_ep_hashtable[i].lock);
d970dbf8 1406 INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain);
1da177e4
LT
1407 }
1408
1409 /* Allocate and initialize the SCTP port hash table. */
1410 do {
1411 sctp_port_hashsize = (1UL << order) * PAGE_SIZE /
1412 sizeof(struct sctp_bind_hashbucket);
1413 if ((sctp_port_hashsize > (64 * 1024)) && order > 0)
1414 continue;
1415 sctp_port_hashtable = (struct sctp_bind_hashbucket *)
a84b50ce 1416 __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, order);
1da177e4
LT
1417 } while (!sctp_port_hashtable && --order > 0);
1418 if (!sctp_port_hashtable) {
145ce502 1419 pr_err("Failed bind hash alloc\n");
1da177e4
LT
1420 status = -ENOMEM;
1421 goto err_bhash_alloc;
1422 }
1423 for (i = 0; i < sctp_port_hashsize; i++) {
1424 spin_lock_init(&sctp_port_hashtable[i].lock);
d970dbf8 1425 INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain);
1da177e4
LT
1426 }
1427
145ce502 1428 pr_info("Hash tables configured (established %d bind %d)\n",
1da177e4
LT
1429 sctp_assoc_hashsize, sctp_port_hashsize);
1430
1da177e4
LT
1431 sctp_sysctl_register();
1432
1433 INIT_LIST_HEAD(&sctp_address_families);
270637ab
VY
1434 sctp_v4_pf_init();
1435 sctp_v6_pf_init();
1da177e4 1436
270637ab 1437 status = sctp_v4_protosw_init();
827bf122 1438
1da177e4 1439 if (status)
270637ab
VY
1440 goto err_protosw_init;
1441
1442 status = sctp_v6_protosw_init();
1443 if (status)
1444 goto err_v6_protosw_init;
1da177e4 1445
4db67e80
EB
1446 status = register_pernet_subsys(&sctp_net_ops);
1447 if (status)
1448 goto err_register_pernet_subsys;
1449
270637ab
VY
1450 status = sctp_v4_add_protocol();
1451 if (status)
827bf122 1452 goto err_add_protocol;
827bf122
SS
1453
1454 /* Register SCTP with inet6 layer. */
1455 status = sctp_v6_add_protocol();
1456 if (status)
1457 goto err_v6_add_protocol;
1458
1da177e4
LT
1459 status = 0;
1460out:
1461 return status;
827bf122 1462err_v6_add_protocol:
270637ab 1463 sctp_v4_del_protocol();
d1dd5247 1464err_add_protocol:
4db67e80
EB
1465 unregister_pernet_subsys(&sctp_net_ops);
1466err_register_pernet_subsys:
270637ab
VY
1467 sctp_v6_protosw_exit();
1468err_v6_protosw_init:
1469 sctp_v4_protosw_exit();
1470err_protosw_init:
270637ab
VY
1471 sctp_v4_pf_exit();
1472 sctp_v6_pf_exit();
1da177e4 1473 sctp_sysctl_unregister();
1da177e4
LT
1474 free_pages((unsigned long)sctp_port_hashtable,
1475 get_order(sctp_port_hashsize *
1476 sizeof(struct sctp_bind_hashbucket)));
1477err_bhash_alloc:
1478 kfree(sctp_ep_hashtable);
1479err_ehash_alloc:
1480 free_pages((unsigned long)sctp_assoc_hashtable,
1481 get_order(sctp_assoc_hashsize *
1482 sizeof(struct sctp_hashbucket)));
1483err_ahash_alloc:
632c928a
EB
1484 percpu_counter_destroy(&sctp_sockets_allocated);
1485err_percpu_counter_init:
1da177e4
LT
1486 kmem_cache_destroy(sctp_chunk_cachep);
1487err_chunk_cachep:
1488 kmem_cache_destroy(sctp_bucket_cachep);
1da177e4
LT
1489 goto out;
1490}
1491
1492/* Exit handler for the SCTP protocol. */
1493SCTP_STATIC __exit void sctp_exit(void)
1494{
1495 /* BUG. This should probably do something useful like clean
1496 * up all the remaining associations and all that memory.
1497 */
1498
827bf122
SS
1499 /* Unregister with inet6/inet layers. */
1500 sctp_v6_del_protocol();
270637ab 1501 sctp_v4_del_protocol();
1da177e4 1502
4db67e80
EB
1503 unregister_pernet_subsys(&sctp_net_ops);
1504
270637ab
VY
1505 /* Free protosw registrations */
1506 sctp_v6_protosw_exit();
1507 sctp_v4_protosw_exit();
1508
827bf122 1509 /* Unregister with socket layer. */
270637ab
VY
1510 sctp_v6_pf_exit();
1511 sctp_v4_pf_exit();
827bf122 1512
1da177e4 1513 sctp_sysctl_unregister();
1da177e4
LT
1514
1515 free_pages((unsigned long)sctp_assoc_hashtable,
1516 get_order(sctp_assoc_hashsize *
1517 sizeof(struct sctp_hashbucket)));
1518 kfree(sctp_ep_hashtable);
1519 free_pages((unsigned long)sctp_port_hashtable,
1520 get_order(sctp_port_hashsize *
1521 sizeof(struct sctp_bind_hashbucket)));
1522
632c928a 1523 percpu_counter_destroy(&sctp_sockets_allocated);
1da177e4 1524
eaa184a1
JDB
1525 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1526
827bf122
SS
1527 kmem_cache_destroy(sctp_chunk_cachep);
1528 kmem_cache_destroy(sctp_bucket_cachep);
1da177e4
LT
1529}
1530
1531module_init(sctp_init);
1532module_exit(sctp_exit);
1533
bb97d31f
ACM
1534/*
1535 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1536 */
1537MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132");
882a382c 1538MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132");
1da177e4
LT
1539MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1540MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
06e86806
LN
1541module_param_named(no_checksums, sctp_checksum_disable, bool, 0644);
1542MODULE_PARM_DESC(no_checksums, "Disable checksums computing and verification");
1da177e4 1543MODULE_LICENSE("GPL");