]> git.proxmox.com Git - mirror_ubuntu-zesty-kernel.git/blame - net/sctp/protocol.c
mm: /proc/pid/pagemap: inspect _PAGE_SOFT_DIRTY only on present pages
[mirror_ubuntu-zesty-kernel.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):
91705c61 32 * lksctp developers <linux-sctp@vger.kernel.org>
1da177e4 33 *
1da177e4
LT
34 * Written or modified by:
35 * La Monte H.P. Yarroll <piggy@acm.org>
36 * Karl Knutson <karl@athena.chicago.il.us>
37 * Jon Grimm <jgrimm@us.ibm.com>
38 * Sridhar Samudrala <sri@us.ibm.com>
39 * Daisy Chang <daisyc@us.ibm.com>
40 * Ardelle Fan <ardelle.fan@intel.com>
1da177e4
LT
41 */
42
145ce502
JP
43#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
44
1da177e4
LT
45#include <linux/module.h>
46#include <linux/init.h>
47#include <linux/netdevice.h>
48#include <linux/inetdevice.h>
49#include <linux/seq_file.h>
4d93df0a 50#include <linux/bootmem.h>
845525a6
VY
51#include <linux/highmem.h>
52#include <linux/swap.h>
5a0e3ad6 53#include <linux/slab.h>
457c4cbc 54#include <net/net_namespace.h>
1da177e4
LT
55#include <net/protocol.h>
56#include <net/ip.h>
57#include <net/ipv6.h>
14c85021 58#include <net/route.h>
1da177e4
LT
59#include <net/sctp/sctp.h>
60#include <net/addrconf.h>
61#include <net/inet_common.h>
62#include <net/inet_ecn.h>
63
64/* Global data structures. */
4cbf1cae 65struct sctp_globals sctp_globals __read_mostly;
1da177e4
LT
66
67struct idr sctp_assocs_id;
68DEFINE_SPINLOCK(sctp_assocs_id_lock);
69
1da177e4
LT
70static struct sctp_pf *sctp_pf_inet6_specific;
71static struct sctp_pf *sctp_pf_inet_specific;
72static struct sctp_af *sctp_af_v4_specific;
73static struct sctp_af *sctp_af_v6_specific;
74
e18b890b
CL
75struct kmem_cache *sctp_chunk_cachep __read_mostly;
76struct kmem_cache *sctp_bucket_cachep __read_mostly;
1da177e4 77
8d987e5c 78long sysctl_sctp_mem[3];
007e3936
VY
79int sysctl_sctp_rmem[3];
80int sysctl_sctp_wmem[3];
4d93df0a 81
1da177e4 82/* Set up the proc fs entry for the SCTP protocol. */
f5f417c0 83static int __net_init sctp_proc_init(struct net *net)
1da177e4 84{
c4e85f82 85#ifdef CONFIG_PROC_FS
13d782f6
EB
86 net->sctp.proc_net_sctp = proc_net_mkdir(net, "sctp", net->proc_net);
87 if (!net->sctp.proc_net_sctp)
88 goto out_proc_net_sctp;
89 if (sctp_snmp_proc_init(net))
80896a35 90 goto out_snmp_proc_init;
13d782f6 91 if (sctp_eps_proc_init(net))
80896a35 92 goto out_eps_proc_init;
13d782f6 93 if (sctp_assocs_proc_init(net))
80896a35 94 goto out_assocs_proc_init;
13d782f6 95 if (sctp_remaddr_proc_init(net))
caea902f 96 goto out_remaddr_proc_init;
1da177e4
LT
97
98 return 0;
99
caea902f 100out_remaddr_proc_init:
13d782f6 101 sctp_assocs_proc_exit(net);
80896a35 102out_assocs_proc_init:
13d782f6 103 sctp_eps_proc_exit(net);
80896a35 104out_eps_proc_init:
13d782f6 105 sctp_snmp_proc_exit(net);
80896a35 106out_snmp_proc_init:
13d782f6
EB
107 remove_proc_entry("sctp", net->proc_net);
108 net->sctp.proc_net_sctp = NULL;
109out_proc_net_sctp:
157653ce 110 return -ENOMEM;
13d782f6
EB
111#endif /* CONFIG_PROC_FS */
112 return 0;
1da177e4
LT
113}
114
d808ad9a 115/* Clean up the proc fs entry for the SCTP protocol.
1da177e4
LT
116 * Note: Do not make this __exit as it is used in the init error
117 * path.
118 */
13d782f6 119static void sctp_proc_exit(struct net *net)
1da177e4 120{
c4e85f82 121#ifdef CONFIG_PROC_FS
13d782f6
EB
122 sctp_snmp_proc_exit(net);
123 sctp_eps_proc_exit(net);
124 sctp_assocs_proc_exit(net);
125 sctp_remaddr_proc_exit(net);
126
127 remove_proc_entry("sctp", net->proc_net);
128 net->sctp.proc_net_sctp = NULL;
c4e85f82 129#endif
1da177e4
LT
130}
131
132/* Private helper to extract ipv4 address and stash them in
133 * the protocol structure.
134 */
135static void sctp_v4_copy_addrlist(struct list_head *addrlist,
136 struct net_device *dev)
137{
138 struct in_device *in_dev;
139 struct in_ifaddr *ifa;
140 struct sctp_sockaddr_entry *addr;
141
142 rcu_read_lock();
e5ed6399 143 if ((in_dev = __in_dev_get_rcu(dev)) == NULL) {
1da177e4
LT
144 rcu_read_unlock();
145 return;
146 }
147
148 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
149 /* Add the address to the local list. */
939cfa75 150 addr = kzalloc(sizeof(*addr), GFP_ATOMIC);
1da177e4 151 if (addr) {
2a6fd78a
AV
152 addr->a.v4.sin_family = AF_INET;
153 addr->a.v4.sin_port = 0;
154 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
29303547
VY
155 addr->valid = 1;
156 INIT_LIST_HEAD(&addr->list);
1da177e4
LT
157 list_add_tail(&addr->list, addrlist);
158 }
159 }
160
161 rcu_read_unlock();
162}
163
164/* Extract our IP addresses from the system and stash them in the
165 * protocol structure.
166 */
4db67e80 167static void sctp_get_local_addr_list(struct net *net)
1da177e4
LT
168{
169 struct net_device *dev;
170 struct list_head *pos;
171 struct sctp_af *af;
172
c6d14c84 173 rcu_read_lock();
4db67e80 174 for_each_netdev_rcu(net, dev) {
2c0740e4 175 list_for_each(pos, &sctp_address_families) {
1da177e4 176 af = list_entry(pos, struct sctp_af, list);
4db67e80 177 af->copy_addrlist(&net->sctp.local_addr_list, dev);
1da177e4
LT
178 }
179 }
c6d14c84 180 rcu_read_unlock();
1da177e4
LT
181}
182
1da177e4 183/* Free the existing local addresses. */
4db67e80 184static void sctp_free_local_addr_list(struct net *net)
1da177e4
LT
185{
186 struct sctp_sockaddr_entry *addr;
187 struct list_head *pos, *temp;
188
4db67e80 189 list_for_each_safe(pos, temp, &net->sctp.local_addr_list) {
1da177e4
LT
190 addr = list_entry(pos, struct sctp_sockaddr_entry, list);
191 list_del(pos);
192 kfree(addr);
193 }
194}
195
1da177e4 196/* Copy the local addresses which are valid for 'scope' into 'bp'. */
4db67e80
EB
197int sctp_copy_local_addr_list(struct net *net, struct sctp_bind_addr *bp,
198 sctp_scope_t scope, gfp_t gfp, int copy_flags)
1da177e4
LT
199{
200 struct sctp_sockaddr_entry *addr;
201 int error = 0;
1da177e4 202
29303547 203 rcu_read_lock();
4db67e80 204 list_for_each_entry_rcu(addr, &net->sctp.local_addr_list, list) {
29303547
VY
205 if (!addr->valid)
206 continue;
e7ff4a70 207 if (sctp_in_scope(net, &addr->a, scope)) {
1da177e4
LT
208 /* Now that the address is in scope, check to see if
209 * the address type is really supported by the local
210 * sock as well as the remote peer.
211 */
6244be4e 212 if ((((AF_INET == addr->a.sa.sa_family) &&
1da177e4 213 (copy_flags & SCTP_ADDR4_PEERSUPP))) ||
6244be4e 214 (((AF_INET6 == addr->a.sa.sa_family) &&
1da177e4
LT
215 (copy_flags & SCTP_ADDR6_ALLOWED) &&
216 (copy_flags & SCTP_ADDR6_PEERSUPP)))) {
f57d96b2
VY
217 error = sctp_add_bind_addr(bp, &addr->a,
218 SCTP_ADDR_SRC, GFP_ATOMIC);
1da177e4
LT
219 if (error)
220 goto end_copy;
221 }
222 }
223 }
224
225end_copy:
29303547 226 rcu_read_unlock();
1da177e4
LT
227 return error;
228}
229
230/* Initialize a sctp_addr from in incoming skb. */
231static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
232 int is_saddr)
233{
234 void *from;
d55c41b1 235 __be16 *port;
1da177e4
LT
236 struct sctphdr *sh;
237
238 port = &addr->v4.sin_port;
239 addr->v4.sin_family = AF_INET;
240
2c0fd387 241 sh = sctp_hdr(skb);
1da177e4 242 if (is_saddr) {
d55c41b1 243 *port = sh->source;
eddc9ec5 244 from = &ip_hdr(skb)->saddr;
1da177e4 245 } else {
d55c41b1 246 *port = sh->dest;
eddc9ec5 247 from = &ip_hdr(skb)->daddr;
1da177e4
LT
248 }
249 memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr));
250}
251
252/* Initialize an sctp_addr from a socket. */
253static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk)
254{
255 addr->v4.sin_family = AF_INET;
7dcdbd95 256 addr->v4.sin_port = 0;
c720c7e8 257 addr->v4.sin_addr.s_addr = inet_sk(sk)->inet_rcv_saddr;
1da177e4
LT
258}
259
260/* Initialize sk->sk_rcv_saddr from sctp_addr. */
261static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk)
262{
c720c7e8 263 inet_sk(sk)->inet_rcv_saddr = addr->v4.sin_addr.s_addr;
1da177e4
LT
264}
265
266/* Initialize sk->sk_daddr from sctp_addr. */
267static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk)
268{
c720c7e8 269 inet_sk(sk)->inet_daddr = addr->v4.sin_addr.s_addr;
1da177e4
LT
270}
271
272/* Initialize a sctp_addr from an address parameter. */
273static void sctp_v4_from_addr_param(union sctp_addr *addr,
274 union sctp_addr_param *param,
dd86d136 275 __be16 port, int iif)
1da177e4
LT
276{
277 addr->v4.sin_family = AF_INET;
278 addr->v4.sin_port = port;
279 addr->v4.sin_addr.s_addr = param->v4.addr.s_addr;
280}
281
282/* Initialize an address parameter from a sctp_addr and return the length
283 * of the address parameter.
284 */
285static int sctp_v4_to_addr_param(const union sctp_addr *addr,
286 union sctp_addr_param *param)
287{
288 int length = sizeof(sctp_ipv4addr_param_t);
289
290 param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS;
dbc16db1 291 param->v4.param_hdr.length = htons(length);
d808ad9a 292 param->v4.addr.s_addr = addr->v4.sin_addr.s_addr;
1da177e4
LT
293
294 return length;
295}
296
297/* Initialize a sctp_addr from a dst_entry. */
18a353f4 298static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct flowi4 *fl4,
854d43a4 299 __be16 port)
1da177e4 300{
1da177e4
LT
301 saddr->v4.sin_family = AF_INET;
302 saddr->v4.sin_port = port;
18a353f4 303 saddr->v4.sin_addr.s_addr = fl4->saddr;
1da177e4
LT
304}
305
306/* Compare two addresses exactly. */
307static int sctp_v4_cmp_addr(const union sctp_addr *addr1,
308 const union sctp_addr *addr2)
309{
310 if (addr1->sa.sa_family != addr2->sa.sa_family)
311 return 0;
312 if (addr1->v4.sin_port != addr2->v4.sin_port)
313 return 0;
314 if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr)
315 return 0;
316
317 return 1;
318}
319
320/* Initialize addr struct to INADDR_ANY. */
6fbfa9f9 321static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port)
1da177e4
LT
322{
323 addr->v4.sin_family = AF_INET;
e6f1cebf 324 addr->v4.sin_addr.s_addr = htonl(INADDR_ANY);
1da177e4
LT
325 addr->v4.sin_port = port;
326}
327
328/* Is this a wildcard address? */
329static int sctp_v4_is_any(const union sctp_addr *addr)
330{
e6f1cebf 331 return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr;
1da177e4
LT
332}
333
334/* This function checks if the address is a valid address to be used for
335 * SCTP binding.
336 *
337 * Output:
338 * Return 0 - If the address is a non-unicast or an illegal address.
339 * Return 1 - If the address is a unicast.
340 */
5636bef7
VY
341static int sctp_v4_addr_valid(union sctp_addr *addr,
342 struct sctp_sock *sp,
343 const struct sk_buff *skb)
1da177e4 344{
7dab83de
VY
345 /* IPv4 addresses not allowed */
346 if (sp && ipv6_only_sock(sctp_opt2sk(sp)))
347 return 0;
348
1da177e4 349 /* Is this a non-unicast address or a unusable SCTP address? */
b5cb2bbc 350 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr))
1da177e4
LT
351 return 0;
352
d808ad9a 353 /* Is this a broadcast address? */
511c3f92 354 if (skb && skb_rtable(skb)->rt_flags & RTCF_BROADCAST)
d808ad9a 355 return 0;
5636bef7 356
1da177e4
LT
357 return 1;
358}
359
360/* Should this be available for binding? */
361static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp)
362{
bb2db45b
EB
363 struct net *net = sock_net(&sp->inet.sk);
364 int ret = inet_addr_type(net, addr->v4.sin_addr.s_addr);
1da177e4 365
1da177e4 366
e6f1cebf 367 if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) &&
cdac4e07
NH
368 ret != RTN_LOCAL &&
369 !sp->inet.freebind &&
370 !sysctl_ip_nonlocal_bind)
1da177e4 371 return 0;
cdac4e07 372
7dab83de
VY
373 if (ipv6_only_sock(sctp_opt2sk(sp)))
374 return 0;
375
1da177e4
LT
376 return 1;
377}
378
379/* Checking the loopback, private and other address scopes as defined in
380 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
381 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
382 *
383 * Level 0 - unusable SCTP addresses
384 * Level 1 - loopback address
385 * Level 2 - link-local addresses
386 * Level 3 - private addresses.
387 * Level 4 - global addresses
388 * For INIT and INIT-ACK address list, let L be the level of
389 * of requested destination address, sender and receiver
390 * SHOULD include all of its addresses with level greater
391 * than or equal to L.
72388433
BD
392 *
393 * IPv4 scoping can be controlled through sysctl option
394 * net.sctp.addr_scope_policy
1da177e4
LT
395 */
396static sctp_scope_t sctp_v4_scope(union sctp_addr *addr)
397{
398 sctp_scope_t retval;
399
1da177e4 400 /* Check for unusable SCTP addresses. */
b5cb2bbc 401 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) {
1da177e4 402 retval = SCTP_SCOPE_UNUSABLE;
b5cb2bbc 403 } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) {
1da177e4 404 retval = SCTP_SCOPE_LOOPBACK;
b5cb2bbc 405 } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) {
1da177e4 406 retval = SCTP_SCOPE_LINK;
b5cb2bbc
JP
407 } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) ||
408 ipv4_is_private_172(addr->v4.sin_addr.s_addr) ||
409 ipv4_is_private_192(addr->v4.sin_addr.s_addr)) {
1da177e4
LT
410 retval = SCTP_SCOPE_PRIVATE;
411 } else {
412 retval = SCTP_SCOPE_GLOBAL;
413 }
414
415 return retval;
416}
417
418/* Returns a valid dst cache entry for the given source and destination ip
419 * addresses. If an association is passed, trys to get a dst entry with a
420 * source address that matches an address in the bind address list.
421 */
da0420be
VY
422static void sctp_v4_get_dst(struct sctp_transport *t, union sctp_addr *saddr,
423 struct flowi *fl, struct sock *sk)
1da177e4 424{
da0420be 425 struct sctp_association *asoc = t->asoc;
1da177e4 426 struct rtable *rt;
9914ae3c 427 struct flowi4 *fl4 = &fl->u.ip4;
1da177e4 428 struct sctp_bind_addr *bp;
1da177e4 429 struct sctp_sockaddr_entry *laddr;
1da177e4 430 struct dst_entry *dst = NULL;
da0420be 431 union sctp_addr *daddr = &t->ipaddr;
1da177e4
LT
432 union sctp_addr dst_saddr;
433
9914ae3c
VY
434 memset(fl4, 0x0, sizeof(struct flowi4));
435 fl4->daddr = daddr->v4.sin_addr.s_addr;
436 fl4->fl4_dport = daddr->v4.sin_port;
437 fl4->flowi4_proto = IPPROTO_SCTP;
1da177e4 438 if (asoc) {
9914ae3c
VY
439 fl4->flowi4_tos = RT_CONN_FLAGS(asoc->base.sk);
440 fl4->flowi4_oif = asoc->base.sk->sk_bound_dev_if;
441 fl4->fl4_sport = htons(asoc->base.bind_addr.port);
1da177e4 442 }
6429d3dc 443 if (saddr) {
9914ae3c
VY
444 fl4->saddr = saddr->v4.sin_addr.s_addr;
445 fl4->fl4_sport = saddr->v4.sin_port;
6429d3dc 446 }
1da177e4 447
bb33381d
DB
448 pr_debug("%s: dst:%pI4, src:%pI4 - ", __func__, &fl4->daddr,
449 &fl4->saddr);
1da177e4 450
bb2db45b 451 rt = ip_route_output_key(sock_net(sk), fl4);
b23dd4fe 452 if (!IS_ERR(rt))
d8d1f30b 453 dst = &rt->dst;
1da177e4
LT
454
455 /* If there is no association or if a source address is passed, no
456 * more validation is required.
457 */
458 if (!asoc || saddr)
459 goto out;
460
461 bp = &asoc->base.bind_addr;
1da177e4
LT
462
463 if (dst) {
464 /* Walk through the bind address list and look for a bind
465 * address that matches the source address of the returned dst.
466 */
18a353f4 467 sctp_v4_dst_saddr(&dst_saddr, fl4, htons(bp->port));
559cf710
VY
468 rcu_read_lock();
469 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
8a07eb0a
MH
470 if (!laddr->valid || (laddr->state == SCTP_ADDR_DEL) ||
471 (laddr->state != SCTP_ADDR_SRC &&
472 !asoc->src_out_of_asoc_ok))
dc022a98 473 continue;
854d43a4 474 if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a))
1da177e4
LT
475 goto out_unlock;
476 }
559cf710 477 rcu_read_unlock();
1da177e4
LT
478
479 /* None of the bound addresses match the source address of the
480 * dst. So release it.
481 */
482 dst_release(dst);
483 dst = NULL;
484 }
485
486 /* Walk through the bind address list and try to get a dst that
487 * matches a bind address as the source address.
488 */
559cf710
VY
489 rcu_read_lock();
490 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
491 if (!laddr->valid)
492 continue;
f57d96b2 493 if ((laddr->state == SCTP_ADDR_SRC) &&
6244be4e 494 (AF_INET == laddr->a.sa.sa_family)) {
9914ae3c
VY
495 fl4->saddr = laddr->a.v4.sin_addr.s_addr;
496 fl4->fl4_sport = laddr->a.v4.sin_port;
bb2db45b 497 rt = ip_route_output_key(sock_net(sk), fl4);
b23dd4fe 498 if (!IS_ERR(rt)) {
d8d1f30b 499 dst = &rt->dst;
1da177e4
LT
500 goto out_unlock;
501 }
502 }
503 }
504
505out_unlock:
559cf710 506 rcu_read_unlock();
1da177e4 507out:
da0420be 508 t->dst = dst;
1da177e4 509 if (dst)
bb33381d
DB
510 pr_debug("rt_dst:%pI4, rt_src:%pI4\n",
511 &fl4->daddr, &fl4->saddr);
1da177e4 512 else
bb33381d 513 pr_debug("no route\n");
1da177e4
LT
514}
515
516/* For v4, the source address is cached in the route entry(dst). So no need
517 * to cache it separately and hence this is an empty routine.
518 */
e5117101 519static void sctp_v4_get_saddr(struct sctp_sock *sk,
9914ae3c 520 struct sctp_transport *t,
9914ae3c 521 struct flowi *fl)
1da177e4 522{
9914ae3c
VY
523 union sctp_addr *saddr = &t->saddr;
524 struct rtable *rt = (struct rtable *)t->dst;
23ec47a0 525
1da177e4
LT
526 if (rt) {
527 saddr->v4.sin_family = AF_INET;
902ebd3e 528 saddr->v4.sin_addr.s_addr = fl->u.ip4.saddr;
1da177e4
LT
529 }
530}
531
532/* What interface did this skb arrive on? */
533static int sctp_v4_skb_iif(const struct sk_buff *skb)
534{
92101b3b 535 return inet_iif(skb);
1da177e4
LT
536}
537
538/* Was this packet marked by Explicit Congestion Notification? */
539static int sctp_v4_is_ce(const struct sk_buff *skb)
540{
eddc9ec5 541 return INET_ECN_is_ce(ip_hdr(skb)->tos);
1da177e4
LT
542}
543
544/* Create and initialize a new sk for the socket returned by accept(). */
545static struct sock *sctp_v4_create_accept_sk(struct sock *sk,
546 struct sctp_association *asoc)
547{
3b1e0a65 548 struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL,
6257ff21 549 sk->sk_prot);
914e1c8b 550 struct inet_sock *newinet;
1da177e4
LT
551
552 if (!newsk)
553 goto out;
554
555 sock_init_data(NULL, newsk);
556
914e1c8b 557 sctp_copy_sock(newsk, sk, asoc);
1da177e4
LT
558 sock_reset_flag(newsk, SOCK_ZAPPED);
559
560 newinet = inet_sk(newsk);
561
c720c7e8 562 newinet->inet_daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr;
1da177e4 563
e6848976 564 sk_refcnt_debug_inc(newsk);
1da177e4
LT
565
566 if (newsk->sk_prot->init(newsk)) {
567 sk_common_release(newsk);
568 newsk = NULL;
569 }
570
571out:
572 return newsk;
573}
574
575/* Map address, empty for v4 family */
576static void sctp_v4_addr_v4map(struct sctp_sock *sp, union sctp_addr *addr)
577{
578 /* Empty */
579}
580
581/* Dump the v4 addr to the seq file. */
582static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr)
583{
21454aaa 584 seq_printf(seq, "%pI4 ", &addr->v4.sin_addr);
1da177e4
LT
585}
586
b9031d9d
VY
587static void sctp_v4_ecn_capable(struct sock *sk)
588{
589 INET_ECN_xmit(sk);
590}
591
bd2a13e2 592static void sctp_addr_wq_timeout_handler(unsigned long arg)
9f7d653b 593{
4db67e80 594 struct net *net = (struct net *)arg;
9f7d653b
MH
595 struct sctp_sockaddr_entry *addrw, *temp;
596 struct sctp_sock *sp;
597
4db67e80 598 spin_lock_bh(&net->sctp.addr_wq_lock);
9f7d653b 599
4db67e80 600 list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) {
bb33381d
DB
601 pr_debug("%s: the first ent in wq:%p is addr:%pISc for cmd:%d at "
602 "entry:%p\n", __func__, &net->sctp.addr_waitq, &addrw->a.sa,
603 addrw->state, addrw);
9f7d653b 604
dfd56b8b 605#if IS_ENABLED(CONFIG_IPV6)
9f7d653b
MH
606 /* Now we send an ASCONF for each association */
607 /* Note. we currently don't handle link local IPv6 addressees */
608 if (addrw->a.sa.sa_family == AF_INET6) {
609 struct in6_addr *in6;
610
611 if (ipv6_addr_type(&addrw->a.v6.sin6_addr) &
612 IPV6_ADDR_LINKLOCAL)
613 goto free_next;
614
615 in6 = (struct in6_addr *)&addrw->a.v6.sin6_addr;
4db67e80 616 if (ipv6_chk_addr(net, in6, NULL, 0) == 0 &&
9f7d653b
MH
617 addrw->state == SCTP_ADDR_NEW) {
618 unsigned long timeo_val;
619
bb33381d
DB
620 pr_debug("%s: this is on DAD, trying %d sec "
621 "later\n", __func__,
622 SCTP_ADDRESS_TICK_DELAY);
623
9f7d653b
MH
624 timeo_val = jiffies;
625 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
4db67e80 626 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
9f7d653b
MH
627 break;
628 }
629 }
5d0c90cf 630#endif
4db67e80 631 list_for_each_entry(sp, &net->sctp.auto_asconf_splist, auto_asconf_list) {
9f7d653b
MH
632 struct sock *sk;
633
634 sk = sctp_opt2sk(sp);
635 /* ignore bound-specific endpoints */
636 if (!sctp_is_ep_boundall(sk))
637 continue;
638 sctp_bh_lock_sock(sk);
639 if (sctp_asconf_mgmt(sp, addrw) < 0)
bb33381d 640 pr_debug("%s: sctp_asconf_mgmt failed\n", __func__);
9f7d653b
MH
641 sctp_bh_unlock_sock(sk);
642 }
39d84a58 643#if IS_ENABLED(CONFIG_IPV6)
9f7d653b 644free_next:
39d84a58 645#endif
9f7d653b
MH
646 list_del(&addrw->list);
647 kfree(addrw);
648 }
4db67e80 649 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
650}
651
4db67e80 652static void sctp_free_addr_wq(struct net *net)
9f7d653b
MH
653{
654 struct sctp_sockaddr_entry *addrw;
655 struct sctp_sockaddr_entry *temp;
656
4db67e80
EB
657 spin_lock_bh(&net->sctp.addr_wq_lock);
658 del_timer(&net->sctp.addr_wq_timer);
659 list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) {
9f7d653b
MH
660 list_del(&addrw->list);
661 kfree(addrw);
662 }
4db67e80 663 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
664}
665
666/* lookup the entry for the same address in the addr_waitq
667 * sctp_addr_wq MUST be locked
668 */
4db67e80
EB
669static struct sctp_sockaddr_entry *sctp_addr_wq_lookup(struct net *net,
670 struct sctp_sockaddr_entry *addr)
9f7d653b
MH
671{
672 struct sctp_sockaddr_entry *addrw;
673
4db67e80 674 list_for_each_entry(addrw, &net->sctp.addr_waitq, list) {
9f7d653b
MH
675 if (addrw->a.sa.sa_family != addr->a.sa.sa_family)
676 continue;
677 if (addrw->a.sa.sa_family == AF_INET) {
678 if (addrw->a.v4.sin_addr.s_addr ==
679 addr->a.v4.sin_addr.s_addr)
680 return addrw;
681 } else if (addrw->a.sa.sa_family == AF_INET6) {
682 if (ipv6_addr_equal(&addrw->a.v6.sin6_addr,
683 &addr->a.v6.sin6_addr))
684 return addrw;
685 }
686 }
687 return NULL;
688}
689
4db67e80 690void sctp_addr_wq_mgmt(struct net *net, struct sctp_sockaddr_entry *addr, int cmd)
9f7d653b
MH
691{
692 struct sctp_sockaddr_entry *addrw;
693 unsigned long timeo_val;
694
695 /* first, we check if an opposite message already exist in the queue.
696 * If we found such message, it is removed.
697 * This operation is a bit stupid, but the DHCP client attaches the
698 * new address after a couple of addition and deletion of that address
699 */
700
4db67e80 701 spin_lock_bh(&net->sctp.addr_wq_lock);
9f7d653b 702 /* Offsets existing events in addr_wq */
4db67e80 703 addrw = sctp_addr_wq_lookup(net, addr);
9f7d653b
MH
704 if (addrw) {
705 if (addrw->state != cmd) {
bb33381d
DB
706 pr_debug("%s: offsets existing entry for %d, addr:%pISc "
707 "in wq:%p\n", __func__, addrw->state, &addrw->a.sa,
708 &net->sctp.addr_waitq);
709
9f7d653b
MH
710 list_del(&addrw->list);
711 kfree(addrw);
712 }
4db67e80 713 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
714 return;
715 }
716
717 /* OK, we have to add the new address to the wait queue */
718 addrw = kmemdup(addr, sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
719 if (addrw == NULL) {
4db67e80 720 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
721 return;
722 }
723 addrw->state = cmd;
4db67e80 724 list_add_tail(&addrw->list, &net->sctp.addr_waitq);
bb33381d
DB
725
726 pr_debug("%s: add new entry for cmd:%d, addr:%pISc in wq:%p\n",
727 __func__, addrw->state, &addrw->a.sa, &net->sctp.addr_waitq);
9f7d653b 728
4db67e80 729 if (!timer_pending(&net->sctp.addr_wq_timer)) {
9f7d653b
MH
730 timeo_val = jiffies;
731 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
4db67e80 732 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
9f7d653b 733 }
4db67e80 734 spin_unlock_bh(&net->sctp.addr_wq_lock);
9f7d653b
MH
735}
736
29303547
VY
737/* Event handler for inet address addition/deletion events.
738 * The sctp_local_addr_list needs to be protocted by a spin lock since
739 * multiple notifiers (say IPv4 and IPv6) may be running at the same
740 * time and thus corrupt the list.
741 * The reader side is protected with RCU.
742 */
24123186
AB
743static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev,
744 void *ptr)
1da177e4 745{
29c7cf96 746 struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
29303547
VY
747 struct sctp_sockaddr_entry *addr = NULL;
748 struct sctp_sockaddr_entry *temp;
4db67e80 749 struct net *net = dev_net(ifa->ifa_dev->dev);
22626216 750 int found = 0;
1da177e4 751
29c7cf96
SS
752 switch (ev) {
753 case NETDEV_UP:
754 addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
755 if (addr) {
756 addr->a.v4.sin_family = AF_INET;
757 addr->a.v4.sin_port = 0;
758 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
29303547 759 addr->valid = 1;
4db67e80
EB
760 spin_lock_bh(&net->sctp.local_addr_lock);
761 list_add_tail_rcu(&addr->list, &net->sctp.local_addr_list);
762 sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_NEW);
763 spin_unlock_bh(&net->sctp.local_addr_lock);
29c7cf96
SS
764 }
765 break;
766 case NETDEV_DOWN:
4db67e80 767 spin_lock_bh(&net->sctp.local_addr_lock);
29303547 768 list_for_each_entry_safe(addr, temp,
4db67e80 769 &net->sctp.local_addr_list, list) {
a40a7d15
PE
770 if (addr->a.sa.sa_family == AF_INET &&
771 addr->a.v4.sin_addr.s_addr ==
772 ifa->ifa_local) {
4db67e80 773 sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_DEL);
22626216 774 found = 1;
29303547
VY
775 addr->valid = 0;
776 list_del_rcu(&addr->list);
29c7cf96
SS
777 break;
778 }
779 }
4db67e80 780 spin_unlock_bh(&net->sctp.local_addr_lock);
22626216 781 if (found)
1231f0ba 782 kfree_rcu(addr, rcu);
29c7cf96
SS
783 break;
784 }
1da177e4
LT
785
786 return NOTIFY_DONE;
787}
788
789/*
790 * Initialize the control inode/socket with a control endpoint data
791 * structure. This endpoint is reserved exclusively for the OOTB processing.
792 */
2ce95503 793static int sctp_ctl_sock_init(struct net *net)
1da177e4
LT
794{
795 int err;
fb13d9f9 796 sa_family_t family = PF_INET;
1da177e4
LT
797
798 if (sctp_get_pf_specific(PF_INET6))
799 family = PF_INET6;
1da177e4 800
2ce95503
EB
801 err = inet_ctl_sock_create(&net->sctp.ctl_sock, family,
802 SOCK_SEQPACKET, IPPROTO_SCTP, net);
fb13d9f9
BH
803
804 /* If IPv6 socket could not be created, try the IPv4 socket */
805 if (err < 0 && family == PF_INET6)
2ce95503 806 err = inet_ctl_sock_create(&net->sctp.ctl_sock, AF_INET,
fb13d9f9 807 SOCK_SEQPACKET, IPPROTO_SCTP,
2ce95503 808 net);
fb13d9f9 809
1da177e4 810 if (err < 0) {
145ce502 811 pr_err("Failed to create the SCTP control socket\n");
1da177e4
LT
812 return err;
813 }
1da177e4
LT
814 return 0;
815}
816
817/* Register address family specific functions. */
818int sctp_register_af(struct sctp_af *af)
819{
820 switch (af->sa_family) {
821 case AF_INET:
822 if (sctp_af_v4_specific)
823 return 0;
824 sctp_af_v4_specific = af;
825 break;
826 case AF_INET6:
827 if (sctp_af_v6_specific)
828 return 0;
829 sctp_af_v6_specific = af;
830 break;
831 default:
832 return 0;
833 }
834
835 INIT_LIST_HEAD(&af->list);
836 list_add_tail(&af->list, &sctp_address_families);
837 return 1;
838}
839
840/* Get the table of functions for manipulating a particular address
841 * family.
842 */
843struct sctp_af *sctp_get_af_specific(sa_family_t family)
844{
845 switch (family) {
846 case AF_INET:
847 return sctp_af_v4_specific;
848 case AF_INET6:
849 return sctp_af_v6_specific;
850 default:
851 return NULL;
852 }
853}
854
855/* Common code to initialize a AF_INET msg_name. */
856static void sctp_inet_msgname(char *msgname, int *addr_len)
857{
858 struct sockaddr_in *sin;
859
860 sin = (struct sockaddr_in *)msgname;
861 *addr_len = sizeof(struct sockaddr_in);
862 sin->sin_family = AF_INET;
863 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
864}
865
866/* Copy the primary address of the peer primary address as the msg_name. */
867static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname,
868 int *addr_len)
869{
870 struct sockaddr_in *sin, *sinfrom;
871
872 if (msgname) {
873 struct sctp_association *asoc;
874
875 asoc = event->asoc;
876 sctp_inet_msgname(msgname, addr_len);
877 sin = (struct sockaddr_in *)msgname;
878 sinfrom = &asoc->peer.primary_addr.v4;
879 sin->sin_port = htons(asoc->peer.port);
880 sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr;
881 }
882}
883
884/* Initialize and copy out a msgname from an inbound skb. */
885static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len)
886{
1da177e4 887 if (msgname) {
2c0fd387
ACM
888 struct sctphdr *sh = sctp_hdr(skb);
889 struct sockaddr_in *sin = (struct sockaddr_in *)msgname;
890
1da177e4 891 sctp_inet_msgname(msgname, len);
1da177e4 892 sin->sin_port = sh->source;
eddc9ec5 893 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
1da177e4
LT
894 }
895}
896
897/* Do we support this AF? */
898static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp)
899{
900 /* PF_INET only supports AF_INET addresses. */
a02cec21 901 return AF_INET == family;
1da177e4
LT
902}
903
904/* Address matching with wildcards allowed. */
905static int sctp_inet_cmp_addr(const union sctp_addr *addr1,
906 const union sctp_addr *addr2,
907 struct sctp_sock *opt)
908{
909 /* PF_INET only supports AF_INET addresses. */
910 if (addr1->sa.sa_family != addr2->sa.sa_family)
911 return 0;
e6f1cebf
AV
912 if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr ||
913 htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr)
1da177e4
LT
914 return 1;
915 if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr)
916 return 1;
917
918 return 0;
919}
920
921/* Verify that provided sockaddr looks bindable. Common verification has
922 * already been taken care of.
923 */
924static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr)
925{
926 return sctp_v4_available(addr, opt);
927}
928
929/* Verify that sockaddr looks sendable. Common verification has already
930 * been taken care of.
931 */
932static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr)
933{
934 return 1;
935}
936
937/* Fill in Supported Address Type information for INIT and INIT-ACK
938 * chunks. Returns number of addresses supported.
939 */
940static int sctp_inet_supported_addrs(const struct sctp_sock *opt,
3dbe8656 941 __be16 *types)
1da177e4
LT
942{
943 types[0] = SCTP_PARAM_IPV4_ADDRESS;
944 return 1;
945}
946
947/* Wrapper routine that calls the ip transmit routine. */
948static inline int sctp_v4_xmit(struct sk_buff *skb,
f880374c 949 struct sctp_transport *transport)
1da177e4 950{
f880374c
HX
951 struct inet_sock *inet = inet_sk(skb->sk);
952
bb33381d
DB
953 pr_debug("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n", __func__, skb,
954 skb->len, &transport->fl.u.ip4.saddr, &transport->fl.u.ip4.daddr);
1da177e4 955
f880374c
HX
956 inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ?
957 IP_PMTUDISC_DO : IP_PMTUDISC_DONT;
958
b01a2407 959 SCTP_INC_STATS(sock_net(&inet->sk), SCTP_MIB_OUTSCTPPACKS);
bb33381d 960
d9d8da80 961 return ip_queue_xmit(skb, &transport->fl);
1da177e4
LT
962}
963
15efbe76 964static struct sctp_af sctp_af_inet;
1da177e4
LT
965
966static struct sctp_pf sctp_pf_inet = {
967 .event_msgname = sctp_inet_event_msgname,
968 .skb_msgname = sctp_inet_skb_msgname,
969 .af_supported = sctp_inet_af_supported,
970 .cmp_addr = sctp_inet_cmp_addr,
971 .bind_verify = sctp_inet_bind_verify,
972 .send_verify = sctp_inet_send_verify,
973 .supported_addrs = sctp_inet_supported_addrs,
974 .create_accept_sk = sctp_v4_create_accept_sk,
975 .addr_v4map = sctp_v4_addr_v4map,
15efbe76 976 .af = &sctp_af_inet
1da177e4
LT
977};
978
979/* Notifier for inetaddr addition/deletion events. */
980static struct notifier_block sctp_inetaddr_notifier = {
981 .notifier_call = sctp_inetaddr_event,
982};
983
984/* Socket operations. */
90ddc4f0 985static const struct proto_ops inet_seqpacket_ops = {
543d9cfe
ACM
986 .family = PF_INET,
987 .owner = THIS_MODULE,
988 .release = inet_release, /* Needs to be wrapped... */
989 .bind = inet_bind,
990 .connect = inet_dgram_connect,
991 .socketpair = sock_no_socketpair,
992 .accept = inet_accept,
993 .getname = inet_getname, /* Semantics are different. */
994 .poll = sctp_poll,
995 .ioctl = inet_ioctl,
996 .listen = sctp_inet_listen,
997 .shutdown = inet_shutdown, /* Looks harmless. */
998 .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */
999 .getsockopt = sock_common_getsockopt,
1000 .sendmsg = inet_sendmsg,
1001 .recvmsg = sock_common_recvmsg,
1002 .mmap = sock_no_mmap,
1003 .sendpage = sock_no_sendpage,
3fdadf7d 1004#ifdef CONFIG_COMPAT
543d9cfe
ACM
1005 .compat_setsockopt = compat_sock_common_setsockopt,
1006 .compat_getsockopt = compat_sock_common_getsockopt,
3fdadf7d 1007#endif
1da177e4
LT
1008};
1009
1010/* Registration with AF_INET family. */
1011static struct inet_protosw sctp_seqpacket_protosw = {
1012 .type = SOCK_SEQPACKET,
1013 .protocol = IPPROTO_SCTP,
1014 .prot = &sctp_prot,
1015 .ops = &inet_seqpacket_ops,
1da177e4
LT
1016 .no_check = 0,
1017 .flags = SCTP_PROTOSW_FLAG
1018};
1019static struct inet_protosw sctp_stream_protosw = {
1020 .type = SOCK_STREAM,
1021 .protocol = IPPROTO_SCTP,
1022 .prot = &sctp_prot,
1023 .ops = &inet_seqpacket_ops,
1da177e4
LT
1024 .no_check = 0,
1025 .flags = SCTP_PROTOSW_FLAG
1026};
1027
1028/* Register with IP layer. */
32613090 1029static const struct net_protocol sctp_protocol = {
1da177e4
LT
1030 .handler = sctp_rcv,
1031 .err_handler = sctp_v4_err,
1032 .no_policy = 1,
bb2db45b 1033 .netns_ok = 1,
1da177e4
LT
1034};
1035
1036/* IPv4 address related functions. */
15efbe76 1037static struct sctp_af sctp_af_inet = {
543d9cfe
ACM
1038 .sa_family = AF_INET,
1039 .sctp_xmit = sctp_v4_xmit,
1040 .setsockopt = ip_setsockopt,
1041 .getsockopt = ip_getsockopt,
1042 .get_dst = sctp_v4_get_dst,
1043 .get_saddr = sctp_v4_get_saddr,
1044 .copy_addrlist = sctp_v4_copy_addrlist,
1045 .from_skb = sctp_v4_from_skb,
1046 .from_sk = sctp_v4_from_sk,
1047 .to_sk_saddr = sctp_v4_to_sk_saddr,
1048 .to_sk_daddr = sctp_v4_to_sk_daddr,
1049 .from_addr_param = sctp_v4_from_addr_param,
1050 .to_addr_param = sctp_v4_to_addr_param,
543d9cfe
ACM
1051 .cmp_addr = sctp_v4_cmp_addr,
1052 .addr_valid = sctp_v4_addr_valid,
1053 .inaddr_any = sctp_v4_inaddr_any,
1054 .is_any = sctp_v4_is_any,
1055 .available = sctp_v4_available,
1056 .scope = sctp_v4_scope,
1057 .skb_iif = sctp_v4_skb_iif,
1058 .is_ce = sctp_v4_is_ce,
1059 .seq_dump_addr = sctp_v4_seq_dump_addr,
b9031d9d 1060 .ecn_capable = sctp_v4_ecn_capable,
543d9cfe
ACM
1061 .net_header_len = sizeof(struct iphdr),
1062 .sockaddr_len = sizeof(struct sockaddr_in),
3fdadf7d 1063#ifdef CONFIG_COMPAT
543d9cfe
ACM
1064 .compat_setsockopt = compat_ip_setsockopt,
1065 .compat_getsockopt = compat_ip_getsockopt,
3fdadf7d 1066#endif
1da177e4
LT
1067};
1068
1069struct sctp_pf *sctp_get_pf_specific(sa_family_t family) {
1070
1071 switch (family) {
1072 case PF_INET:
1073 return sctp_pf_inet_specific;
1074 case PF_INET6:
1075 return sctp_pf_inet6_specific;
1076 default:
1077 return NULL;
1078 }
1079}
1080
1081/* Register the PF specific function table. */
1082int sctp_register_pf(struct sctp_pf *pf, sa_family_t family)
1083{
1084 switch (family) {
1085 case PF_INET:
1086 if (sctp_pf_inet_specific)
1087 return 0;
1088 sctp_pf_inet_specific = pf;
1089 break;
1090 case PF_INET6:
1091 if (sctp_pf_inet6_specific)
1092 return 0;
1093 sctp_pf_inet6_specific = pf;
1094 break;
1095 default:
1096 return 0;
1097 }
1098 return 1;
1099}
1100
b01a2407 1101static inline int init_sctp_mibs(struct net *net)
996b1dba 1102{
b01a2407 1103 return snmp_mib_init((void __percpu **)net->sctp.sctp_statistics,
1823e4c8
ED
1104 sizeof(struct sctp_mib),
1105 __alignof__(struct sctp_mib));
1da177e4
LT
1106}
1107
b01a2407 1108static inline void cleanup_sctp_mibs(struct net *net)
1da177e4 1109{
b01a2407 1110 snmp_mib_free((void __percpu **)net->sctp.sctp_statistics);
1da177e4
LT
1111}
1112
270637ab
VY
1113static void sctp_v4_pf_init(void)
1114{
1115 /* Initialize the SCTP specific PF functions. */
1116 sctp_register_pf(&sctp_pf_inet, PF_INET);
1117 sctp_register_af(&sctp_af_inet);
1118}
1119
1120static void sctp_v4_pf_exit(void)
1121{
1122 list_del(&sctp_af_inet.list);
1123}
1124
1125static int sctp_v4_protosw_init(void)
1126{
1127 int rc;
1128
1129 rc = proto_register(&sctp_prot, 1);
1130 if (rc)
1131 return rc;
1132
1133 /* Register SCTP(UDP and TCP style) with socket layer. */
1134 inet_register_protosw(&sctp_seqpacket_protosw);
1135 inet_register_protosw(&sctp_stream_protosw);
1136
1137 return 0;
1138}
1139
1140static void sctp_v4_protosw_exit(void)
1141{
1142 inet_unregister_protosw(&sctp_stream_protosw);
1143 inet_unregister_protosw(&sctp_seqpacket_protosw);
1144 proto_unregister(&sctp_prot);
1145}
1146
1147static int sctp_v4_add_protocol(void)
1148{
1149 /* Register notifier for inet address additions/deletions. */
1150 register_inetaddr_notifier(&sctp_inetaddr_notifier);
1151
1152 /* Register SCTP with inet layer. */
1153 if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0)
1154 return -EAGAIN;
1155
1156 return 0;
1157}
1158
1159static void sctp_v4_del_protocol(void)
1160{
1161 inet_del_protocol(&sctp_protocol, IPPROTO_SCTP);
1162 unregister_inetaddr_notifier(&sctp_inetaddr_notifier);
1163}
1164
f5f417c0 1165static int __net_init sctp_net_init(struct net *net)
4db67e80 1166{
2ce95503
EB
1167 int status;
1168
e1fc3b14
EB
1169 /*
1170 * 14. Suggested SCTP Protocol Parameter Values
1171 */
1172 /* The following protocol parameters are RECOMMENDED: */
1173 /* RTO.Initial - 3 seconds */
1174 net->sctp.rto_initial = SCTP_RTO_INITIAL;
1175 /* RTO.Min - 1 second */
1176 net->sctp.rto_min = SCTP_RTO_MIN;
1177 /* RTO.Max - 60 seconds */
1178 net->sctp.rto_max = SCTP_RTO_MAX;
1179 /* RTO.Alpha - 1/8 */
1180 net->sctp.rto_alpha = SCTP_RTO_ALPHA;
1181 /* RTO.Beta - 1/4 */
1182 net->sctp.rto_beta = SCTP_RTO_BETA;
1183
1184 /* Valid.Cookie.Life - 60 seconds */
1185 net->sctp.valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE;
1186
1187 /* Whether Cookie Preservative is enabled(1) or not(0) */
1188 net->sctp.cookie_preserve_enable = 1;
1189
3c68198e 1190 /* Default sctp sockets to use md5 as their hmac alg */
0d0863b0 1191#if defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_MD5)
3c68198e 1192 net->sctp.sctp_hmac_alg = "md5";
0d0863b0 1193#elif defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_SHA1)
3c68198e
NH
1194 net->sctp.sctp_hmac_alg = "sha1";
1195#else
1196 net->sctp.sctp_hmac_alg = NULL;
1197#endif
1198
e1fc3b14
EB
1199 /* Max.Burst - 4 */
1200 net->sctp.max_burst = SCTP_DEFAULT_MAX_BURST;
1201
1202 /* Association.Max.Retrans - 10 attempts
1203 * Path.Max.Retrans - 5 attempts (per destination address)
1204 * Max.Init.Retransmits - 8 attempts
1205 */
1206 net->sctp.max_retrans_association = 10;
1207 net->sctp.max_retrans_path = 5;
1208 net->sctp.max_retrans_init = 8;
1209
1210 /* Sendbuffer growth - do per-socket accounting */
1211 net->sctp.sndbuf_policy = 0;
1212
1213 /* Rcvbuffer growth - do per-socket accounting */
1214 net->sctp.rcvbuf_policy = 0;
1215
1216 /* HB.interval - 30 seconds */
1217 net->sctp.hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT;
1218
1219 /* delayed SACK timeout */
1220 net->sctp.sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK;
1221
1222 /* Disable ADDIP by default. */
1223 net->sctp.addip_enable = 0;
1224 net->sctp.addip_noauth = 0;
1225 net->sctp.default_auto_asconf = 0;
1226
1227 /* Enable PR-SCTP by default. */
1228 net->sctp.prsctp_enable = 1;
1229
1230 /* Disable AUTH by default. */
1231 net->sctp.auth_enable = 0;
1232
1233 /* Set SCOPE policy to enabled */
1234 net->sctp.scope_policy = SCTP_SCOPE_POLICY_ENABLE;
1235
1236 /* Set the default rwnd update threshold */
1237 net->sctp.rwnd_upd_shift = SCTP_DEFAULT_RWND_SHIFT;
1238
1239 /* Initialize maximum autoclose timeout. */
1240 net->sctp.max_autoclose = INT_MAX / HZ;
1241
ebb7e95d
EB
1242 status = sctp_sysctl_net_register(net);
1243 if (status)
1244 goto err_sysctl_register;
1245
b01a2407
EB
1246 /* Allocate and initialise sctp mibs. */
1247 status = init_sctp_mibs(net);
1248 if (status)
1249 goto err_init_mibs;
1250
13d782f6
EB
1251 /* Initialize proc fs directory. */
1252 status = sctp_proc_init(net);
1253 if (status)
1254 goto err_init_proc;
1255
1256 sctp_dbg_objcnt_init(net);
1257
2ce95503
EB
1258 /* Initialize the control inode/socket for handling OOTB packets. */
1259 if ((status = sctp_ctl_sock_init(net))) {
1260 pr_err("Failed to initialize the SCTP control sock\n");
1261 goto err_ctl_sock_init;
1262 }
1263
4db67e80
EB
1264 /* Initialize the local address list. */
1265 INIT_LIST_HEAD(&net->sctp.local_addr_list);
1266 spin_lock_init(&net->sctp.local_addr_lock);
1267 sctp_get_local_addr_list(net);
1268
1269 /* Initialize the address event list */
1270 INIT_LIST_HEAD(&net->sctp.addr_waitq);
1271 INIT_LIST_HEAD(&net->sctp.auto_asconf_splist);
1272 spin_lock_init(&net->sctp.addr_wq_lock);
1273 net->sctp.addr_wq_timer.expires = 0;
1274 setup_timer(&net->sctp.addr_wq_timer, sctp_addr_wq_timeout_handler,
1275 (unsigned long)net);
1276
1277 return 0;
2ce95503
EB
1278
1279err_ctl_sock_init:
13d782f6
EB
1280 sctp_dbg_objcnt_exit(net);
1281 sctp_proc_exit(net);
1282err_init_proc:
b01a2407
EB
1283 cleanup_sctp_mibs(net);
1284err_init_mibs:
ebb7e95d
EB
1285 sctp_sysctl_net_unregister(net);
1286err_sysctl_register:
2ce95503 1287 return status;
4db67e80
EB
1288}
1289
f5f417c0 1290static void __net_exit sctp_net_exit(struct net *net)
4db67e80
EB
1291{
1292 /* Free the local address list */
1293 sctp_free_addr_wq(net);
1294 sctp_free_local_addr_list(net);
2ce95503
EB
1295
1296 /* Free the control endpoint. */
1297 inet_ctl_sock_destroy(net->sctp.ctl_sock);
13d782f6
EB
1298
1299 sctp_dbg_objcnt_exit(net);
1300
1301 sctp_proc_exit(net);
b01a2407 1302 cleanup_sctp_mibs(net);
ebb7e95d 1303 sctp_sysctl_net_unregister(net);
4db67e80
EB
1304}
1305
1306static struct pernet_operations sctp_net_ops = {
1307 .init = sctp_net_init,
1308 .exit = sctp_net_exit,
1309};
1310
1da177e4 1311/* Initialize the universe into something sensible. */
dda91928 1312static __init int sctp_init(void)
1da177e4
LT
1313{
1314 int i;
1315 int status = -EINVAL;
1316 unsigned long goal;
4d93df0a
NH
1317 unsigned long limit;
1318 int max_share;
1da177e4
LT
1319 int order;
1320
bb33381d
DB
1321 BUILD_BUG_ON(sizeof(struct sctp_ulpevent) >
1322 sizeof(((struct sk_buff *) 0)->cb));
1da177e4 1323
827bf122 1324 /* Allocate bind_bucket and chunk caches. */
1da177e4
LT
1325 status = -ENOBUFS;
1326 sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket",
1327 sizeof(struct sctp_bind_bucket),
1328 0, SLAB_HWCACHE_ALIGN,
20c2df83 1329 NULL);
1da177e4 1330 if (!sctp_bucket_cachep)
827bf122 1331 goto out;
1da177e4
LT
1332
1333 sctp_chunk_cachep = kmem_cache_create("sctp_chunk",
1334 sizeof(struct sctp_chunk),
1335 0, SLAB_HWCACHE_ALIGN,
20c2df83 1336 NULL);
1da177e4
LT
1337 if (!sctp_chunk_cachep)
1338 goto err_chunk_cachep;
1339
632c928a
EB
1340 status = percpu_counter_init(&sctp_sockets_allocated, 0);
1341 if (status)
1342 goto err_percpu_counter_init;
1343
1da177e4
LT
1344 /* Implementation specific variables. */
1345
1346 /* Initialize default stream count setup information. */
1347 sctp_max_instreams = SCTP_DEFAULT_INSTREAMS;
1348 sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS;
1349
1350 /* Initialize handle used for association ids. */
1351 idr_init(&sctp_assocs_id);
1352
f03d78db 1353 limit = nr_free_buffer_pages() / 8;
4d93df0a
NH
1354 limit = max(limit, 128UL);
1355 sysctl_sctp_mem[0] = limit / 4 * 3;
1356 sysctl_sctp_mem[1] = limit;
1357 sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2;
1358
1359 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1360 limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7);
1361 max_share = min(4UL*1024*1024, limit);
1362
845525a6 1363 sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */
87fb4b7b 1364 sysctl_sctp_rmem[1] = 1500 * SKB_TRUESIZE(1);
4d93df0a
NH
1365 sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share);
1366
3ab224be 1367 sysctl_sctp_wmem[0] = SK_MEM_QUANTUM;
4d93df0a
NH
1368 sysctl_sctp_wmem[1] = 16*1024;
1369 sysctl_sctp_wmem[2] = max(64*1024, max_share);
1370
1da177e4
LT
1371 /* Size and allocate the association hash table.
1372 * The methodology is similar to that of the tcp hash tables.
1373 */
4481374c
JB
1374 if (totalram_pages >= (128 * 1024))
1375 goal = totalram_pages >> (22 - PAGE_SHIFT);
1da177e4 1376 else
4481374c 1377 goal = totalram_pages >> (24 - PAGE_SHIFT);
1da177e4
LT
1378
1379 for (order = 0; (1UL << order) < goal; order++)
1380 ;
1381
1382 do {
1383 sctp_assoc_hashsize = (1UL << order) * PAGE_SIZE /
1384 sizeof(struct sctp_hashbucket);
1385 if ((sctp_assoc_hashsize > (64 * 1024)) && order > 0)
1386 continue;
1387 sctp_assoc_hashtable = (struct sctp_hashbucket *)
a84b50ce 1388 __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, order);
1da177e4
LT
1389 } while (!sctp_assoc_hashtable && --order > 0);
1390 if (!sctp_assoc_hashtable) {
145ce502 1391 pr_err("Failed association hash alloc\n");
1da177e4
LT
1392 status = -ENOMEM;
1393 goto err_ahash_alloc;
1394 }
1395 for (i = 0; i < sctp_assoc_hashsize; i++) {
1396 rwlock_init(&sctp_assoc_hashtable[i].lock);
d970dbf8 1397 INIT_HLIST_HEAD(&sctp_assoc_hashtable[i].chain);
1da177e4
LT
1398 }
1399
1400 /* Allocate and initialize the endpoint hash table. */
1401 sctp_ep_hashsize = 64;
3b77d661 1402 sctp_ep_hashtable =
1da177e4
LT
1403 kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL);
1404 if (!sctp_ep_hashtable) {
145ce502 1405 pr_err("Failed endpoint_hash alloc\n");
1da177e4
LT
1406 status = -ENOMEM;
1407 goto err_ehash_alloc;
1408 }
1409 for (i = 0; i < sctp_ep_hashsize; i++) {
1410 rwlock_init(&sctp_ep_hashtable[i].lock);
d970dbf8 1411 INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain);
1da177e4
LT
1412 }
1413
1414 /* Allocate and initialize the SCTP port hash table. */
1415 do {
1416 sctp_port_hashsize = (1UL << order) * PAGE_SIZE /
1417 sizeof(struct sctp_bind_hashbucket);
1418 if ((sctp_port_hashsize > (64 * 1024)) && order > 0)
1419 continue;
1420 sctp_port_hashtable = (struct sctp_bind_hashbucket *)
a84b50ce 1421 __get_free_pages(GFP_ATOMIC|__GFP_NOWARN, order);
1da177e4
LT
1422 } while (!sctp_port_hashtable && --order > 0);
1423 if (!sctp_port_hashtable) {
145ce502 1424 pr_err("Failed bind hash alloc\n");
1da177e4
LT
1425 status = -ENOMEM;
1426 goto err_bhash_alloc;
1427 }
1428 for (i = 0; i < sctp_port_hashsize; i++) {
1429 spin_lock_init(&sctp_port_hashtable[i].lock);
d970dbf8 1430 INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain);
1da177e4
LT
1431 }
1432
145ce502 1433 pr_info("Hash tables configured (established %d bind %d)\n",
1da177e4
LT
1434 sctp_assoc_hashsize, sctp_port_hashsize);
1435
1da177e4
LT
1436 sctp_sysctl_register();
1437
1438 INIT_LIST_HEAD(&sctp_address_families);
270637ab
VY
1439 sctp_v4_pf_init();
1440 sctp_v6_pf_init();
1da177e4 1441
270637ab 1442 status = sctp_v4_protosw_init();
827bf122 1443
1da177e4 1444 if (status)
270637ab
VY
1445 goto err_protosw_init;
1446
1447 status = sctp_v6_protosw_init();
1448 if (status)
1449 goto err_v6_protosw_init;
1da177e4 1450
4db67e80
EB
1451 status = register_pernet_subsys(&sctp_net_ops);
1452 if (status)
1453 goto err_register_pernet_subsys;
1454
270637ab
VY
1455 status = sctp_v4_add_protocol();
1456 if (status)
827bf122 1457 goto err_add_protocol;
827bf122
SS
1458
1459 /* Register SCTP with inet6 layer. */
1460 status = sctp_v6_add_protocol();
1461 if (status)
1462 goto err_v6_add_protocol;
1463
1da177e4
LT
1464 status = 0;
1465out:
1466 return status;
827bf122 1467err_v6_add_protocol:
270637ab 1468 sctp_v4_del_protocol();
d1dd5247 1469err_add_protocol:
4db67e80
EB
1470 unregister_pernet_subsys(&sctp_net_ops);
1471err_register_pernet_subsys:
270637ab
VY
1472 sctp_v6_protosw_exit();
1473err_v6_protosw_init:
1474 sctp_v4_protosw_exit();
1475err_protosw_init:
270637ab
VY
1476 sctp_v4_pf_exit();
1477 sctp_v6_pf_exit();
1da177e4 1478 sctp_sysctl_unregister();
1da177e4
LT
1479 free_pages((unsigned long)sctp_port_hashtable,
1480 get_order(sctp_port_hashsize *
1481 sizeof(struct sctp_bind_hashbucket)));
1482err_bhash_alloc:
1483 kfree(sctp_ep_hashtable);
1484err_ehash_alloc:
1485 free_pages((unsigned long)sctp_assoc_hashtable,
1486 get_order(sctp_assoc_hashsize *
1487 sizeof(struct sctp_hashbucket)));
1488err_ahash_alloc:
632c928a
EB
1489 percpu_counter_destroy(&sctp_sockets_allocated);
1490err_percpu_counter_init:
1da177e4
LT
1491 kmem_cache_destroy(sctp_chunk_cachep);
1492err_chunk_cachep:
1493 kmem_cache_destroy(sctp_bucket_cachep);
1da177e4
LT
1494 goto out;
1495}
1496
1497/* Exit handler for the SCTP protocol. */
dda91928 1498static __exit void sctp_exit(void)
1da177e4
LT
1499{
1500 /* BUG. This should probably do something useful like clean
1501 * up all the remaining associations and all that memory.
1502 */
1503
827bf122
SS
1504 /* Unregister with inet6/inet layers. */
1505 sctp_v6_del_protocol();
270637ab 1506 sctp_v4_del_protocol();
1da177e4 1507
4db67e80
EB
1508 unregister_pernet_subsys(&sctp_net_ops);
1509
270637ab
VY
1510 /* Free protosw registrations */
1511 sctp_v6_protosw_exit();
1512 sctp_v4_protosw_exit();
1513
827bf122 1514 /* Unregister with socket layer. */
270637ab
VY
1515 sctp_v6_pf_exit();
1516 sctp_v4_pf_exit();
827bf122 1517
1da177e4 1518 sctp_sysctl_unregister();
1da177e4
LT
1519
1520 free_pages((unsigned long)sctp_assoc_hashtable,
1521 get_order(sctp_assoc_hashsize *
1522 sizeof(struct sctp_hashbucket)));
1523 kfree(sctp_ep_hashtable);
1524 free_pages((unsigned long)sctp_port_hashtable,
1525 get_order(sctp_port_hashsize *
1526 sizeof(struct sctp_bind_hashbucket)));
1527
632c928a 1528 percpu_counter_destroy(&sctp_sockets_allocated);
1da177e4 1529
eaa184a1
JDB
1530 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1531
827bf122
SS
1532 kmem_cache_destroy(sctp_chunk_cachep);
1533 kmem_cache_destroy(sctp_bucket_cachep);
1da177e4
LT
1534}
1535
1536module_init(sctp_init);
1537module_exit(sctp_exit);
1538
bb97d31f
ACM
1539/*
1540 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1541 */
1542MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132");
882a382c 1543MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132");
91705c61 1544MODULE_AUTHOR("Linux Kernel SCTP developers <linux-sctp@vger.kernel.org>");
1da177e4 1545MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
71acc0dd
DM
1546module_param_named(no_checksums, sctp_checksum_disable, bool, 0644);
1547MODULE_PARM_DESC(no_checksums, "Disable checksums computing and verification");
1da177e4 1548MODULE_LICENSE("GPL");