Commit | Line | Data |
---|---|---|
60c778b2 | 1 | /* SCTP kernel implementation |
1da177e4 LT |
2 | * (C) Copyright IBM Corp. 2001, 2004 |
3 | * Copyright (c) 1999-2000 Cisco, Inc. | |
4 | * Copyright (c) 1999-2001 Motorola, Inc. | |
5 | * Copyright (c) 2001 Intel Corp. | |
6 | * Copyright (c) 2001 Nokia, Inc. | |
7 | * Copyright (c) 2001 La Monte H.P. Yarroll | |
8 | * | |
60c778b2 | 9 | * This file is part of the SCTP kernel implementation |
1da177e4 LT |
10 | * |
11 | * Initialization/cleanup for SCTP protocol support. | |
12 | * | |
60c778b2 | 13 | * This SCTP implementation is free software; |
1da177e4 LT |
14 | * you can redistribute it and/or modify it under the terms of |
15 | * the GNU General Public License as published by | |
16 | * the Free Software Foundation; either version 2, or (at your option) | |
17 | * any later version. | |
18 | * | |
60c778b2 | 19 | * This SCTP implementation is distributed in the hope that it |
1da177e4 LT |
20 | * will be useful, but WITHOUT ANY WARRANTY; without even the implied |
21 | * ************************ | |
22 | * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. | |
23 | * See the GNU General Public License for more details. | |
24 | * | |
25 | * You should have received a copy of the GNU General Public License | |
4b2f13a2 JK |
26 | * along with GNU CC; see the file COPYING. If not, see |
27 | * <http://www.gnu.org/licenses/>. | |
1da177e4 LT |
28 | * |
29 | * Please send any bug reports or fixes you make to the | |
30 | * email address(es): | |
91705c61 | 31 | * lksctp developers <linux-sctp@vger.kernel.org> |
1da177e4 | 32 | * |
1da177e4 LT |
33 | * Written or modified by: |
34 | * La Monte H.P. Yarroll <piggy@acm.org> | |
35 | * Karl Knutson <karl@athena.chicago.il.us> | |
36 | * Jon Grimm <jgrimm@us.ibm.com> | |
37 | * Sridhar Samudrala <sri@us.ibm.com> | |
38 | * Daisy Chang <daisyc@us.ibm.com> | |
39 | * Ardelle Fan <ardelle.fan@intel.com> | |
1da177e4 LT |
40 | */ |
41 | ||
145ce502 JP |
42 | #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt |
43 | ||
1da177e4 LT |
44 | #include <linux/module.h> |
45 | #include <linux/init.h> | |
46 | #include <linux/netdevice.h> | |
47 | #include <linux/inetdevice.h> | |
48 | #include <linux/seq_file.h> | |
4d93df0a | 49 | #include <linux/bootmem.h> |
845525a6 VY |
50 | #include <linux/highmem.h> |
51 | #include <linux/swap.h> | |
5a0e3ad6 | 52 | #include <linux/slab.h> |
457c4cbc | 53 | #include <net/net_namespace.h> |
1da177e4 LT |
54 | #include <net/protocol.h> |
55 | #include <net/ip.h> | |
56 | #include <net/ipv6.h> | |
14c85021 | 57 | #include <net/route.h> |
1da177e4 LT |
58 | #include <net/sctp/sctp.h> |
59 | #include <net/addrconf.h> | |
60 | #include <net/inet_common.h> | |
61 | #include <net/inet_ecn.h> | |
62 | ||
d9749fb5 NH |
63 | #define MAX_SCTP_PORT_HASH_ENTRIES (64 * 1024) |
64 | ||
1da177e4 | 65 | /* Global data structures. */ |
4cbf1cae | 66 | struct sctp_globals sctp_globals __read_mostly; |
1da177e4 LT |
67 | |
68 | struct idr sctp_assocs_id; | |
69 | DEFINE_SPINLOCK(sctp_assocs_id_lock); | |
70 | ||
1da177e4 LT |
71 | static struct sctp_pf *sctp_pf_inet6_specific; |
72 | static struct sctp_pf *sctp_pf_inet_specific; | |
73 | static struct sctp_af *sctp_af_v4_specific; | |
74 | static struct sctp_af *sctp_af_v6_specific; | |
75 | ||
e18b890b CL |
76 | struct kmem_cache *sctp_chunk_cachep __read_mostly; |
77 | struct kmem_cache *sctp_bucket_cachep __read_mostly; | |
1da177e4 | 78 | |
8d987e5c | 79 | long sysctl_sctp_mem[3]; |
007e3936 VY |
80 | int sysctl_sctp_rmem[3]; |
81 | int sysctl_sctp_wmem[3]; | |
4d93df0a | 82 | |
1da177e4 | 83 | /* Set up the proc fs entry for the SCTP protocol. */ |
f5f417c0 | 84 | static int __net_init sctp_proc_init(struct net *net) |
1da177e4 | 85 | { |
c4e85f82 | 86 | #ifdef CONFIG_PROC_FS |
13d782f6 EB |
87 | net->sctp.proc_net_sctp = proc_net_mkdir(net, "sctp", net->proc_net); |
88 | if (!net->sctp.proc_net_sctp) | |
89 | goto out_proc_net_sctp; | |
90 | if (sctp_snmp_proc_init(net)) | |
80896a35 | 91 | goto out_snmp_proc_init; |
13d782f6 | 92 | if (sctp_eps_proc_init(net)) |
80896a35 | 93 | goto out_eps_proc_init; |
13d782f6 | 94 | if (sctp_assocs_proc_init(net)) |
80896a35 | 95 | goto out_assocs_proc_init; |
13d782f6 | 96 | if (sctp_remaddr_proc_init(net)) |
caea902f | 97 | goto out_remaddr_proc_init; |
1da177e4 LT |
98 | |
99 | return 0; | |
100 | ||
caea902f | 101 | out_remaddr_proc_init: |
13d782f6 | 102 | sctp_assocs_proc_exit(net); |
80896a35 | 103 | out_assocs_proc_init: |
13d782f6 | 104 | sctp_eps_proc_exit(net); |
80896a35 | 105 | out_eps_proc_init: |
13d782f6 | 106 | sctp_snmp_proc_exit(net); |
80896a35 | 107 | out_snmp_proc_init: |
13d782f6 EB |
108 | remove_proc_entry("sctp", net->proc_net); |
109 | net->sctp.proc_net_sctp = NULL; | |
110 | out_proc_net_sctp: | |
157653ce | 111 | return -ENOMEM; |
13d782f6 EB |
112 | #endif /* CONFIG_PROC_FS */ |
113 | return 0; | |
1da177e4 LT |
114 | } |
115 | ||
d808ad9a | 116 | /* Clean up the proc fs entry for the SCTP protocol. |
1da177e4 LT |
117 | * Note: Do not make this __exit as it is used in the init error |
118 | * path. | |
119 | */ | |
13d782f6 | 120 | static void sctp_proc_exit(struct net *net) |
1da177e4 | 121 | { |
c4e85f82 | 122 | #ifdef CONFIG_PROC_FS |
13d782f6 EB |
123 | sctp_snmp_proc_exit(net); |
124 | sctp_eps_proc_exit(net); | |
125 | sctp_assocs_proc_exit(net); | |
126 | sctp_remaddr_proc_exit(net); | |
127 | ||
128 | remove_proc_entry("sctp", net->proc_net); | |
129 | net->sctp.proc_net_sctp = NULL; | |
c4e85f82 | 130 | #endif |
1da177e4 LT |
131 | } |
132 | ||
133 | /* Private helper to extract ipv4 address and stash them in | |
134 | * the protocol structure. | |
135 | */ | |
136 | static void sctp_v4_copy_addrlist(struct list_head *addrlist, | |
137 | struct net_device *dev) | |
138 | { | |
139 | struct in_device *in_dev; | |
140 | struct in_ifaddr *ifa; | |
141 | struct sctp_sockaddr_entry *addr; | |
142 | ||
143 | rcu_read_lock(); | |
e5ed6399 | 144 | if ((in_dev = __in_dev_get_rcu(dev)) == NULL) { |
1da177e4 LT |
145 | rcu_read_unlock(); |
146 | return; | |
147 | } | |
148 | ||
149 | for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) { | |
150 | /* Add the address to the local list. */ | |
939cfa75 | 151 | addr = kzalloc(sizeof(*addr), GFP_ATOMIC); |
1da177e4 | 152 | if (addr) { |
2a6fd78a AV |
153 | addr->a.v4.sin_family = AF_INET; |
154 | addr->a.v4.sin_port = 0; | |
155 | addr->a.v4.sin_addr.s_addr = ifa->ifa_local; | |
29303547 VY |
156 | addr->valid = 1; |
157 | INIT_LIST_HEAD(&addr->list); | |
1da177e4 LT |
158 | list_add_tail(&addr->list, addrlist); |
159 | } | |
160 | } | |
161 | ||
162 | rcu_read_unlock(); | |
163 | } | |
164 | ||
165 | /* Extract our IP addresses from the system and stash them in the | |
166 | * protocol structure. | |
167 | */ | |
4db67e80 | 168 | static void sctp_get_local_addr_list(struct net *net) |
1da177e4 LT |
169 | { |
170 | struct net_device *dev; | |
171 | struct list_head *pos; | |
172 | struct sctp_af *af; | |
173 | ||
c6d14c84 | 174 | rcu_read_lock(); |
4db67e80 | 175 | for_each_netdev_rcu(net, dev) { |
2c0740e4 | 176 | list_for_each(pos, &sctp_address_families) { |
1da177e4 | 177 | af = list_entry(pos, struct sctp_af, list); |
4db67e80 | 178 | af->copy_addrlist(&net->sctp.local_addr_list, dev); |
1da177e4 LT |
179 | } |
180 | } | |
c6d14c84 | 181 | rcu_read_unlock(); |
1da177e4 LT |
182 | } |
183 | ||
1da177e4 | 184 | /* Free the existing local addresses. */ |
4db67e80 | 185 | static void sctp_free_local_addr_list(struct net *net) |
1da177e4 LT |
186 | { |
187 | struct sctp_sockaddr_entry *addr; | |
188 | struct list_head *pos, *temp; | |
189 | ||
4db67e80 | 190 | list_for_each_safe(pos, temp, &net->sctp.local_addr_list) { |
1da177e4 LT |
191 | addr = list_entry(pos, struct sctp_sockaddr_entry, list); |
192 | list_del(pos); | |
193 | kfree(addr); | |
194 | } | |
195 | } | |
196 | ||
1da177e4 | 197 | /* Copy the local addresses which are valid for 'scope' into 'bp'. */ |
4db67e80 EB |
198 | int sctp_copy_local_addr_list(struct net *net, struct sctp_bind_addr *bp, |
199 | sctp_scope_t scope, gfp_t gfp, int copy_flags) | |
1da177e4 LT |
200 | { |
201 | struct sctp_sockaddr_entry *addr; | |
202 | int error = 0; | |
1da177e4 | 203 | |
29303547 | 204 | rcu_read_lock(); |
4db67e80 | 205 | list_for_each_entry_rcu(addr, &net->sctp.local_addr_list, list) { |
29303547 VY |
206 | if (!addr->valid) |
207 | continue; | |
e7ff4a70 | 208 | if (sctp_in_scope(net, &addr->a, scope)) { |
1da177e4 LT |
209 | /* Now that the address is in scope, check to see if |
210 | * the address type is really supported by the local | |
211 | * sock as well as the remote peer. | |
212 | */ | |
6244be4e | 213 | if ((((AF_INET == addr->a.sa.sa_family) && |
1da177e4 | 214 | (copy_flags & SCTP_ADDR4_PEERSUPP))) || |
6244be4e | 215 | (((AF_INET6 == addr->a.sa.sa_family) && |
1da177e4 LT |
216 | (copy_flags & SCTP_ADDR6_ALLOWED) && |
217 | (copy_flags & SCTP_ADDR6_PEERSUPP)))) { | |
f57d96b2 | 218 | error = sctp_add_bind_addr(bp, &addr->a, |
133800d1 | 219 | sizeof(addr->a), |
f57d96b2 | 220 | SCTP_ADDR_SRC, GFP_ATOMIC); |
1da177e4 LT |
221 | if (error) |
222 | goto end_copy; | |
223 | } | |
224 | } | |
225 | } | |
226 | ||
227 | end_copy: | |
29303547 | 228 | rcu_read_unlock(); |
1da177e4 LT |
229 | return error; |
230 | } | |
231 | ||
232 | /* Initialize a sctp_addr from in incoming skb. */ | |
233 | static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb, | |
234 | int is_saddr) | |
235 | { | |
236 | void *from; | |
d55c41b1 | 237 | __be16 *port; |
1da177e4 LT |
238 | struct sctphdr *sh; |
239 | ||
240 | port = &addr->v4.sin_port; | |
241 | addr->v4.sin_family = AF_INET; | |
242 | ||
2c0fd387 | 243 | sh = sctp_hdr(skb); |
1da177e4 | 244 | if (is_saddr) { |
d55c41b1 | 245 | *port = sh->source; |
eddc9ec5 | 246 | from = &ip_hdr(skb)->saddr; |
1da177e4 | 247 | } else { |
d55c41b1 | 248 | *port = sh->dest; |
eddc9ec5 | 249 | from = &ip_hdr(skb)->daddr; |
1da177e4 LT |
250 | } |
251 | memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr)); | |
252 | } | |
253 | ||
254 | /* Initialize an sctp_addr from a socket. */ | |
255 | static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk) | |
256 | { | |
257 | addr->v4.sin_family = AF_INET; | |
7dcdbd95 | 258 | addr->v4.sin_port = 0; |
c720c7e8 | 259 | addr->v4.sin_addr.s_addr = inet_sk(sk)->inet_rcv_saddr; |
1da177e4 LT |
260 | } |
261 | ||
262 | /* Initialize sk->sk_rcv_saddr from sctp_addr. */ | |
263 | static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk) | |
264 | { | |
c720c7e8 | 265 | inet_sk(sk)->inet_rcv_saddr = addr->v4.sin_addr.s_addr; |
1da177e4 LT |
266 | } |
267 | ||
268 | /* Initialize sk->sk_daddr from sctp_addr. */ | |
269 | static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk) | |
270 | { | |
c720c7e8 | 271 | inet_sk(sk)->inet_daddr = addr->v4.sin_addr.s_addr; |
1da177e4 LT |
272 | } |
273 | ||
274 | /* Initialize a sctp_addr from an address parameter. */ | |
275 | static void sctp_v4_from_addr_param(union sctp_addr *addr, | |
276 | union sctp_addr_param *param, | |
dd86d136 | 277 | __be16 port, int iif) |
1da177e4 LT |
278 | { |
279 | addr->v4.sin_family = AF_INET; | |
280 | addr->v4.sin_port = port; | |
281 | addr->v4.sin_addr.s_addr = param->v4.addr.s_addr; | |
282 | } | |
283 | ||
284 | /* Initialize an address parameter from a sctp_addr and return the length | |
285 | * of the address parameter. | |
286 | */ | |
287 | static int sctp_v4_to_addr_param(const union sctp_addr *addr, | |
288 | union sctp_addr_param *param) | |
289 | { | |
290 | int length = sizeof(sctp_ipv4addr_param_t); | |
291 | ||
292 | param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS; | |
dbc16db1 | 293 | param->v4.param_hdr.length = htons(length); |
d808ad9a | 294 | param->v4.addr.s_addr = addr->v4.sin_addr.s_addr; |
1da177e4 LT |
295 | |
296 | return length; | |
297 | } | |
298 | ||
299 | /* Initialize a sctp_addr from a dst_entry. */ | |
18a353f4 | 300 | static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct flowi4 *fl4, |
854d43a4 | 301 | __be16 port) |
1da177e4 | 302 | { |
1da177e4 LT |
303 | saddr->v4.sin_family = AF_INET; |
304 | saddr->v4.sin_port = port; | |
18a353f4 | 305 | saddr->v4.sin_addr.s_addr = fl4->saddr; |
1da177e4 LT |
306 | } |
307 | ||
308 | /* Compare two addresses exactly. */ | |
309 | static int sctp_v4_cmp_addr(const union sctp_addr *addr1, | |
310 | const union sctp_addr *addr2) | |
311 | { | |
312 | if (addr1->sa.sa_family != addr2->sa.sa_family) | |
313 | return 0; | |
314 | if (addr1->v4.sin_port != addr2->v4.sin_port) | |
315 | return 0; | |
316 | if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr) | |
317 | return 0; | |
318 | ||
319 | return 1; | |
320 | } | |
321 | ||
322 | /* Initialize addr struct to INADDR_ANY. */ | |
6fbfa9f9 | 323 | static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port) |
1da177e4 LT |
324 | { |
325 | addr->v4.sin_family = AF_INET; | |
e6f1cebf | 326 | addr->v4.sin_addr.s_addr = htonl(INADDR_ANY); |
1da177e4 LT |
327 | addr->v4.sin_port = port; |
328 | } | |
329 | ||
330 | /* Is this a wildcard address? */ | |
331 | static int sctp_v4_is_any(const union sctp_addr *addr) | |
332 | { | |
e6f1cebf | 333 | return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr; |
1da177e4 LT |
334 | } |
335 | ||
336 | /* This function checks if the address is a valid address to be used for | |
337 | * SCTP binding. | |
338 | * | |
339 | * Output: | |
340 | * Return 0 - If the address is a non-unicast or an illegal address. | |
341 | * Return 1 - If the address is a unicast. | |
342 | */ | |
5636bef7 VY |
343 | static int sctp_v4_addr_valid(union sctp_addr *addr, |
344 | struct sctp_sock *sp, | |
345 | const struct sk_buff *skb) | |
1da177e4 | 346 | { |
7dab83de VY |
347 | /* IPv4 addresses not allowed */ |
348 | if (sp && ipv6_only_sock(sctp_opt2sk(sp))) | |
349 | return 0; | |
350 | ||
1da177e4 | 351 | /* Is this a non-unicast address or a unusable SCTP address? */ |
b5cb2bbc | 352 | if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) |
1da177e4 LT |
353 | return 0; |
354 | ||
d808ad9a | 355 | /* Is this a broadcast address? */ |
511c3f92 | 356 | if (skb && skb_rtable(skb)->rt_flags & RTCF_BROADCAST) |
d808ad9a | 357 | return 0; |
5636bef7 | 358 | |
1da177e4 LT |
359 | return 1; |
360 | } | |
361 | ||
362 | /* Should this be available for binding? */ | |
363 | static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp) | |
364 | { | |
bb2db45b EB |
365 | struct net *net = sock_net(&sp->inet.sk); |
366 | int ret = inet_addr_type(net, addr->v4.sin_addr.s_addr); | |
1da177e4 | 367 | |
1da177e4 | 368 | |
e6f1cebf | 369 | if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) && |
cdac4e07 NH |
370 | ret != RTN_LOCAL && |
371 | !sp->inet.freebind && | |
49a60158 | 372 | !net->ipv4.sysctl_ip_nonlocal_bind) |
1da177e4 | 373 | return 0; |
cdac4e07 | 374 | |
7dab83de VY |
375 | if (ipv6_only_sock(sctp_opt2sk(sp))) |
376 | return 0; | |
377 | ||
1da177e4 LT |
378 | return 1; |
379 | } | |
380 | ||
381 | /* Checking the loopback, private and other address scopes as defined in | |
382 | * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4 | |
383 | * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>. | |
384 | * | |
385 | * Level 0 - unusable SCTP addresses | |
386 | * Level 1 - loopback address | |
387 | * Level 2 - link-local addresses | |
388 | * Level 3 - private addresses. | |
389 | * Level 4 - global addresses | |
390 | * For INIT and INIT-ACK address list, let L be the level of | |
391 | * of requested destination address, sender and receiver | |
392 | * SHOULD include all of its addresses with level greater | |
393 | * than or equal to L. | |
72388433 BD |
394 | * |
395 | * IPv4 scoping can be controlled through sysctl option | |
396 | * net.sctp.addr_scope_policy | |
1da177e4 LT |
397 | */ |
398 | static sctp_scope_t sctp_v4_scope(union sctp_addr *addr) | |
399 | { | |
400 | sctp_scope_t retval; | |
401 | ||
1da177e4 | 402 | /* Check for unusable SCTP addresses. */ |
b5cb2bbc | 403 | if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) { |
1da177e4 | 404 | retval = SCTP_SCOPE_UNUSABLE; |
b5cb2bbc | 405 | } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) { |
1da177e4 | 406 | retval = SCTP_SCOPE_LOOPBACK; |
b5cb2bbc | 407 | } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) { |
1da177e4 | 408 | retval = SCTP_SCOPE_LINK; |
b5cb2bbc JP |
409 | } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) || |
410 | ipv4_is_private_172(addr->v4.sin_addr.s_addr) || | |
411 | ipv4_is_private_192(addr->v4.sin_addr.s_addr)) { | |
1da177e4 LT |
412 | retval = SCTP_SCOPE_PRIVATE; |
413 | } else { | |
414 | retval = SCTP_SCOPE_GLOBAL; | |
415 | } | |
416 | ||
417 | return retval; | |
418 | } | |
419 | ||
420 | /* Returns a valid dst cache entry for the given source and destination ip | |
421 | * addresses. If an association is passed, trys to get a dst entry with a | |
422 | * source address that matches an address in the bind address list. | |
423 | */ | |
da0420be VY |
424 | static void sctp_v4_get_dst(struct sctp_transport *t, union sctp_addr *saddr, |
425 | struct flowi *fl, struct sock *sk) | |
1da177e4 | 426 | { |
da0420be | 427 | struct sctp_association *asoc = t->asoc; |
1da177e4 | 428 | struct rtable *rt; |
9914ae3c | 429 | struct flowi4 *fl4 = &fl->u.ip4; |
1da177e4 | 430 | struct sctp_bind_addr *bp; |
1da177e4 | 431 | struct sctp_sockaddr_entry *laddr; |
1da177e4 | 432 | struct dst_entry *dst = NULL; |
da0420be | 433 | union sctp_addr *daddr = &t->ipaddr; |
1da177e4 LT |
434 | union sctp_addr dst_saddr; |
435 | ||
9914ae3c VY |
436 | memset(fl4, 0x0, sizeof(struct flowi4)); |
437 | fl4->daddr = daddr->v4.sin_addr.s_addr; | |
438 | fl4->fl4_dport = daddr->v4.sin_port; | |
439 | fl4->flowi4_proto = IPPROTO_SCTP; | |
1da177e4 | 440 | if (asoc) { |
9914ae3c VY |
441 | fl4->flowi4_tos = RT_CONN_FLAGS(asoc->base.sk); |
442 | fl4->flowi4_oif = asoc->base.sk->sk_bound_dev_if; | |
443 | fl4->fl4_sport = htons(asoc->base.bind_addr.port); | |
1da177e4 | 444 | } |
6429d3dc | 445 | if (saddr) { |
9914ae3c VY |
446 | fl4->saddr = saddr->v4.sin_addr.s_addr; |
447 | fl4->fl4_sport = saddr->v4.sin_port; | |
6429d3dc | 448 | } |
1da177e4 | 449 | |
bb33381d DB |
450 | pr_debug("%s: dst:%pI4, src:%pI4 - ", __func__, &fl4->daddr, |
451 | &fl4->saddr); | |
1da177e4 | 452 | |
bb2db45b | 453 | rt = ip_route_output_key(sock_net(sk), fl4); |
b23dd4fe | 454 | if (!IS_ERR(rt)) |
d8d1f30b | 455 | dst = &rt->dst; |
1da177e4 LT |
456 | |
457 | /* If there is no association or if a source address is passed, no | |
458 | * more validation is required. | |
459 | */ | |
460 | if (!asoc || saddr) | |
461 | goto out; | |
462 | ||
463 | bp = &asoc->base.bind_addr; | |
1da177e4 LT |
464 | |
465 | if (dst) { | |
466 | /* Walk through the bind address list and look for a bind | |
467 | * address that matches the source address of the returned dst. | |
468 | */ | |
18a353f4 | 469 | sctp_v4_dst_saddr(&dst_saddr, fl4, htons(bp->port)); |
559cf710 VY |
470 | rcu_read_lock(); |
471 | list_for_each_entry_rcu(laddr, &bp->address_list, list) { | |
8a07eb0a MH |
472 | if (!laddr->valid || (laddr->state == SCTP_ADDR_DEL) || |
473 | (laddr->state != SCTP_ADDR_SRC && | |
474 | !asoc->src_out_of_asoc_ok)) | |
dc022a98 | 475 | continue; |
854d43a4 | 476 | if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a)) |
1da177e4 LT |
477 | goto out_unlock; |
478 | } | |
559cf710 | 479 | rcu_read_unlock(); |
1da177e4 LT |
480 | |
481 | /* None of the bound addresses match the source address of the | |
482 | * dst. So release it. | |
483 | */ | |
484 | dst_release(dst); | |
485 | dst = NULL; | |
486 | } | |
487 | ||
488 | /* Walk through the bind address list and try to get a dst that | |
489 | * matches a bind address as the source address. | |
490 | */ | |
559cf710 VY |
491 | rcu_read_lock(); |
492 | list_for_each_entry_rcu(laddr, &bp->address_list, list) { | |
0ca50d12 MRL |
493 | struct net_device *odev; |
494 | ||
559cf710 VY |
495 | if (!laddr->valid) |
496 | continue; | |
07868284 MRL |
497 | if (laddr->state != SCTP_ADDR_SRC || |
498 | AF_INET != laddr->a.sa.sa_family) | |
499 | continue; | |
500 | ||
501 | fl4->fl4_sport = laddr->a.v4.sin_port; | |
502 | flowi4_update_output(fl4, | |
503 | asoc->base.sk->sk_bound_dev_if, | |
504 | RT_CONN_FLAGS(asoc->base.sk), | |
505 | daddr->v4.sin_addr.s_addr, | |
506 | laddr->a.v4.sin_addr.s_addr); | |
507 | ||
508 | rt = ip_route_output_key(sock_net(sk), fl4); | |
509 | if (IS_ERR(rt)) | |
510 | continue; | |
511 | ||
410f0383 MRL |
512 | if (!dst) |
513 | dst = &rt->dst; | |
514 | ||
0ca50d12 MRL |
515 | /* Ensure the src address belongs to the output |
516 | * interface. | |
517 | */ | |
518 | odev = __ip_dev_find(sock_net(sk), laddr->a.v4.sin_addr.s_addr, | |
519 | false); | |
d82f0f1f | 520 | if (!odev || odev->ifindex != fl4->flowi4_oif) { |
410f0383 MRL |
521 | if (&rt->dst != dst) |
522 | dst_release(&rt->dst); | |
0ca50d12 | 523 | continue; |
d82f0f1f | 524 | } |
0ca50d12 | 525 | |
410f0383 MRL |
526 | if (dst != &rt->dst) |
527 | dst_release(dst); | |
07868284 MRL |
528 | dst = &rt->dst; |
529 | break; | |
1da177e4 LT |
530 | } |
531 | ||
532 | out_unlock: | |
559cf710 | 533 | rcu_read_unlock(); |
1da177e4 | 534 | out: |
da0420be | 535 | t->dst = dst; |
1da177e4 | 536 | if (dst) |
bb33381d DB |
537 | pr_debug("rt_dst:%pI4, rt_src:%pI4\n", |
538 | &fl4->daddr, &fl4->saddr); | |
1da177e4 | 539 | else |
bb33381d | 540 | pr_debug("no route\n"); |
1da177e4 LT |
541 | } |
542 | ||
543 | /* For v4, the source address is cached in the route entry(dst). So no need | |
544 | * to cache it separately and hence this is an empty routine. | |
545 | */ | |
e5117101 | 546 | static void sctp_v4_get_saddr(struct sctp_sock *sk, |
9914ae3c | 547 | struct sctp_transport *t, |
9914ae3c | 548 | struct flowi *fl) |
1da177e4 | 549 | { |
9914ae3c VY |
550 | union sctp_addr *saddr = &t->saddr; |
551 | struct rtable *rt = (struct rtable *)t->dst; | |
23ec47a0 | 552 | |
1da177e4 LT |
553 | if (rt) { |
554 | saddr->v4.sin_family = AF_INET; | |
902ebd3e | 555 | saddr->v4.sin_addr.s_addr = fl->u.ip4.saddr; |
1da177e4 LT |
556 | } |
557 | } | |
558 | ||
559 | /* What interface did this skb arrive on? */ | |
560 | static int sctp_v4_skb_iif(const struct sk_buff *skb) | |
561 | { | |
92101b3b | 562 | return inet_iif(skb); |
1da177e4 LT |
563 | } |
564 | ||
565 | /* Was this packet marked by Explicit Congestion Notification? */ | |
566 | static int sctp_v4_is_ce(const struct sk_buff *skb) | |
567 | { | |
eddc9ec5 | 568 | return INET_ECN_is_ce(ip_hdr(skb)->tos); |
1da177e4 LT |
569 | } |
570 | ||
571 | /* Create and initialize a new sk for the socket returned by accept(). */ | |
572 | static struct sock *sctp_v4_create_accept_sk(struct sock *sk, | |
573 | struct sctp_association *asoc) | |
574 | { | |
3b1e0a65 | 575 | struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL, |
11aa9c28 | 576 | sk->sk_prot, 0); |
914e1c8b | 577 | struct inet_sock *newinet; |
1da177e4 LT |
578 | |
579 | if (!newsk) | |
580 | goto out; | |
581 | ||
582 | sock_init_data(NULL, newsk); | |
583 | ||
914e1c8b | 584 | sctp_copy_sock(newsk, sk, asoc); |
1da177e4 LT |
585 | sock_reset_flag(newsk, SOCK_ZAPPED); |
586 | ||
587 | newinet = inet_sk(newsk); | |
588 | ||
c720c7e8 | 589 | newinet->inet_daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr; |
1da177e4 | 590 | |
e6848976 | 591 | sk_refcnt_debug_inc(newsk); |
1da177e4 LT |
592 | |
593 | if (newsk->sk_prot->init(newsk)) { | |
594 | sk_common_release(newsk); | |
595 | newsk = NULL; | |
596 | } | |
597 | ||
598 | out: | |
599 | return newsk; | |
600 | } | |
601 | ||
299ee123 | 602 | static int sctp_v4_addr_to_user(struct sctp_sock *sp, union sctp_addr *addr) |
1da177e4 | 603 | { |
299ee123 JG |
604 | /* No address mapping for V4 sockets */ |
605 | return sizeof(struct sockaddr_in); | |
1da177e4 LT |
606 | } |
607 | ||
608 | /* Dump the v4 addr to the seq file. */ | |
609 | static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr) | |
610 | { | |
21454aaa | 611 | seq_printf(seq, "%pI4 ", &addr->v4.sin_addr); |
1da177e4 LT |
612 | } |
613 | ||
b9031d9d VY |
614 | static void sctp_v4_ecn_capable(struct sock *sk) |
615 | { | |
616 | INET_ECN_xmit(sk); | |
617 | } | |
618 | ||
bd2a13e2 | 619 | static void sctp_addr_wq_timeout_handler(unsigned long arg) |
9f7d653b | 620 | { |
4db67e80 | 621 | struct net *net = (struct net *)arg; |
9f7d653b MH |
622 | struct sctp_sockaddr_entry *addrw, *temp; |
623 | struct sctp_sock *sp; | |
624 | ||
4db67e80 | 625 | spin_lock_bh(&net->sctp.addr_wq_lock); |
9f7d653b | 626 | |
4db67e80 | 627 | list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) { |
bb33381d DB |
628 | pr_debug("%s: the first ent in wq:%p is addr:%pISc for cmd:%d at " |
629 | "entry:%p\n", __func__, &net->sctp.addr_waitq, &addrw->a.sa, | |
630 | addrw->state, addrw); | |
9f7d653b | 631 | |
dfd56b8b | 632 | #if IS_ENABLED(CONFIG_IPV6) |
9f7d653b MH |
633 | /* Now we send an ASCONF for each association */ |
634 | /* Note. we currently don't handle link local IPv6 addressees */ | |
635 | if (addrw->a.sa.sa_family == AF_INET6) { | |
636 | struct in6_addr *in6; | |
637 | ||
638 | if (ipv6_addr_type(&addrw->a.v6.sin6_addr) & | |
639 | IPV6_ADDR_LINKLOCAL) | |
640 | goto free_next; | |
641 | ||
642 | in6 = (struct in6_addr *)&addrw->a.v6.sin6_addr; | |
4db67e80 | 643 | if (ipv6_chk_addr(net, in6, NULL, 0) == 0 && |
9f7d653b MH |
644 | addrw->state == SCTP_ADDR_NEW) { |
645 | unsigned long timeo_val; | |
646 | ||
bb33381d DB |
647 | pr_debug("%s: this is on DAD, trying %d sec " |
648 | "later\n", __func__, | |
649 | SCTP_ADDRESS_TICK_DELAY); | |
650 | ||
9f7d653b MH |
651 | timeo_val = jiffies; |
652 | timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY); | |
4db67e80 | 653 | mod_timer(&net->sctp.addr_wq_timer, timeo_val); |
9f7d653b MH |
654 | break; |
655 | } | |
656 | } | |
5d0c90cf | 657 | #endif |
4db67e80 | 658 | list_for_each_entry(sp, &net->sctp.auto_asconf_splist, auto_asconf_list) { |
9f7d653b MH |
659 | struct sock *sk; |
660 | ||
661 | sk = sctp_opt2sk(sp); | |
662 | /* ignore bound-specific endpoints */ | |
663 | if (!sctp_is_ep_boundall(sk)) | |
664 | continue; | |
5bc1d1b4 | 665 | bh_lock_sock(sk); |
9f7d653b | 666 | if (sctp_asconf_mgmt(sp, addrw) < 0) |
bb33381d | 667 | pr_debug("%s: sctp_asconf_mgmt failed\n", __func__); |
5bc1d1b4 | 668 | bh_unlock_sock(sk); |
9f7d653b | 669 | } |
39d84a58 | 670 | #if IS_ENABLED(CONFIG_IPV6) |
9f7d653b | 671 | free_next: |
39d84a58 | 672 | #endif |
9f7d653b MH |
673 | list_del(&addrw->list); |
674 | kfree(addrw); | |
675 | } | |
4db67e80 | 676 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
677 | } |
678 | ||
4db67e80 | 679 | static void sctp_free_addr_wq(struct net *net) |
9f7d653b MH |
680 | { |
681 | struct sctp_sockaddr_entry *addrw; | |
682 | struct sctp_sockaddr_entry *temp; | |
683 | ||
4db67e80 EB |
684 | spin_lock_bh(&net->sctp.addr_wq_lock); |
685 | del_timer(&net->sctp.addr_wq_timer); | |
686 | list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) { | |
9f7d653b MH |
687 | list_del(&addrw->list); |
688 | kfree(addrw); | |
689 | } | |
4db67e80 | 690 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
691 | } |
692 | ||
693 | /* lookup the entry for the same address in the addr_waitq | |
694 | * sctp_addr_wq MUST be locked | |
695 | */ | |
4db67e80 EB |
696 | static struct sctp_sockaddr_entry *sctp_addr_wq_lookup(struct net *net, |
697 | struct sctp_sockaddr_entry *addr) | |
9f7d653b MH |
698 | { |
699 | struct sctp_sockaddr_entry *addrw; | |
700 | ||
4db67e80 | 701 | list_for_each_entry(addrw, &net->sctp.addr_waitq, list) { |
9f7d653b MH |
702 | if (addrw->a.sa.sa_family != addr->a.sa.sa_family) |
703 | continue; | |
704 | if (addrw->a.sa.sa_family == AF_INET) { | |
705 | if (addrw->a.v4.sin_addr.s_addr == | |
706 | addr->a.v4.sin_addr.s_addr) | |
707 | return addrw; | |
708 | } else if (addrw->a.sa.sa_family == AF_INET6) { | |
709 | if (ipv6_addr_equal(&addrw->a.v6.sin6_addr, | |
710 | &addr->a.v6.sin6_addr)) | |
711 | return addrw; | |
712 | } | |
713 | } | |
714 | return NULL; | |
715 | } | |
716 | ||
4db67e80 | 717 | void sctp_addr_wq_mgmt(struct net *net, struct sctp_sockaddr_entry *addr, int cmd) |
9f7d653b MH |
718 | { |
719 | struct sctp_sockaddr_entry *addrw; | |
720 | unsigned long timeo_val; | |
721 | ||
722 | /* first, we check if an opposite message already exist in the queue. | |
723 | * If we found such message, it is removed. | |
724 | * This operation is a bit stupid, but the DHCP client attaches the | |
725 | * new address after a couple of addition and deletion of that address | |
726 | */ | |
727 | ||
4db67e80 | 728 | spin_lock_bh(&net->sctp.addr_wq_lock); |
9f7d653b | 729 | /* Offsets existing events in addr_wq */ |
4db67e80 | 730 | addrw = sctp_addr_wq_lookup(net, addr); |
9f7d653b MH |
731 | if (addrw) { |
732 | if (addrw->state != cmd) { | |
bb33381d DB |
733 | pr_debug("%s: offsets existing entry for %d, addr:%pISc " |
734 | "in wq:%p\n", __func__, addrw->state, &addrw->a.sa, | |
735 | &net->sctp.addr_waitq); | |
736 | ||
9f7d653b MH |
737 | list_del(&addrw->list); |
738 | kfree(addrw); | |
739 | } | |
4db67e80 | 740 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
741 | return; |
742 | } | |
743 | ||
744 | /* OK, we have to add the new address to the wait queue */ | |
745 | addrw = kmemdup(addr, sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC); | |
746 | if (addrw == NULL) { | |
4db67e80 | 747 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
748 | return; |
749 | } | |
750 | addrw->state = cmd; | |
4db67e80 | 751 | list_add_tail(&addrw->list, &net->sctp.addr_waitq); |
bb33381d DB |
752 | |
753 | pr_debug("%s: add new entry for cmd:%d, addr:%pISc in wq:%p\n", | |
754 | __func__, addrw->state, &addrw->a.sa, &net->sctp.addr_waitq); | |
9f7d653b | 755 | |
4db67e80 | 756 | if (!timer_pending(&net->sctp.addr_wq_timer)) { |
9f7d653b MH |
757 | timeo_val = jiffies; |
758 | timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY); | |
4db67e80 | 759 | mod_timer(&net->sctp.addr_wq_timer, timeo_val); |
9f7d653b | 760 | } |
4db67e80 | 761 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
762 | } |
763 | ||
29303547 VY |
764 | /* Event handler for inet address addition/deletion events. |
765 | * The sctp_local_addr_list needs to be protocted by a spin lock since | |
766 | * multiple notifiers (say IPv4 and IPv6) may be running at the same | |
767 | * time and thus corrupt the list. | |
768 | * The reader side is protected with RCU. | |
769 | */ | |
24123186 AB |
770 | static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev, |
771 | void *ptr) | |
1da177e4 | 772 | { |
29c7cf96 | 773 | struct in_ifaddr *ifa = (struct in_ifaddr *)ptr; |
29303547 VY |
774 | struct sctp_sockaddr_entry *addr = NULL; |
775 | struct sctp_sockaddr_entry *temp; | |
4db67e80 | 776 | struct net *net = dev_net(ifa->ifa_dev->dev); |
22626216 | 777 | int found = 0; |
1da177e4 | 778 | |
29c7cf96 SS |
779 | switch (ev) { |
780 | case NETDEV_UP: | |
781 | addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC); | |
782 | if (addr) { | |
783 | addr->a.v4.sin_family = AF_INET; | |
784 | addr->a.v4.sin_port = 0; | |
785 | addr->a.v4.sin_addr.s_addr = ifa->ifa_local; | |
29303547 | 786 | addr->valid = 1; |
4db67e80 EB |
787 | spin_lock_bh(&net->sctp.local_addr_lock); |
788 | list_add_tail_rcu(&addr->list, &net->sctp.local_addr_list); | |
789 | sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_NEW); | |
790 | spin_unlock_bh(&net->sctp.local_addr_lock); | |
29c7cf96 SS |
791 | } |
792 | break; | |
793 | case NETDEV_DOWN: | |
4db67e80 | 794 | spin_lock_bh(&net->sctp.local_addr_lock); |
29303547 | 795 | list_for_each_entry_safe(addr, temp, |
4db67e80 | 796 | &net->sctp.local_addr_list, list) { |
a40a7d15 PE |
797 | if (addr->a.sa.sa_family == AF_INET && |
798 | addr->a.v4.sin_addr.s_addr == | |
799 | ifa->ifa_local) { | |
4db67e80 | 800 | sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_DEL); |
22626216 | 801 | found = 1; |
29303547 VY |
802 | addr->valid = 0; |
803 | list_del_rcu(&addr->list); | |
29c7cf96 SS |
804 | break; |
805 | } | |
806 | } | |
4db67e80 | 807 | spin_unlock_bh(&net->sctp.local_addr_lock); |
22626216 | 808 | if (found) |
1231f0ba | 809 | kfree_rcu(addr, rcu); |
29c7cf96 SS |
810 | break; |
811 | } | |
1da177e4 LT |
812 | |
813 | return NOTIFY_DONE; | |
814 | } | |
815 | ||
816 | /* | |
817 | * Initialize the control inode/socket with a control endpoint data | |
818 | * structure. This endpoint is reserved exclusively for the OOTB processing. | |
819 | */ | |
2ce95503 | 820 | static int sctp_ctl_sock_init(struct net *net) |
1da177e4 LT |
821 | { |
822 | int err; | |
fb13d9f9 | 823 | sa_family_t family = PF_INET; |
1da177e4 LT |
824 | |
825 | if (sctp_get_pf_specific(PF_INET6)) | |
826 | family = PF_INET6; | |
1da177e4 | 827 | |
2ce95503 EB |
828 | err = inet_ctl_sock_create(&net->sctp.ctl_sock, family, |
829 | SOCK_SEQPACKET, IPPROTO_SCTP, net); | |
fb13d9f9 BH |
830 | |
831 | /* If IPv6 socket could not be created, try the IPv4 socket */ | |
832 | if (err < 0 && family == PF_INET6) | |
2ce95503 | 833 | err = inet_ctl_sock_create(&net->sctp.ctl_sock, AF_INET, |
fb13d9f9 | 834 | SOCK_SEQPACKET, IPPROTO_SCTP, |
2ce95503 | 835 | net); |
fb13d9f9 | 836 | |
1da177e4 | 837 | if (err < 0) { |
145ce502 | 838 | pr_err("Failed to create the SCTP control socket\n"); |
1da177e4 LT |
839 | return err; |
840 | } | |
1da177e4 LT |
841 | return 0; |
842 | } | |
843 | ||
844 | /* Register address family specific functions. */ | |
845 | int sctp_register_af(struct sctp_af *af) | |
846 | { | |
847 | switch (af->sa_family) { | |
848 | case AF_INET: | |
849 | if (sctp_af_v4_specific) | |
850 | return 0; | |
851 | sctp_af_v4_specific = af; | |
852 | break; | |
853 | case AF_INET6: | |
854 | if (sctp_af_v6_specific) | |
855 | return 0; | |
856 | sctp_af_v6_specific = af; | |
857 | break; | |
858 | default: | |
859 | return 0; | |
860 | } | |
861 | ||
862 | INIT_LIST_HEAD(&af->list); | |
863 | list_add_tail(&af->list, &sctp_address_families); | |
864 | return 1; | |
865 | } | |
866 | ||
867 | /* Get the table of functions for manipulating a particular address | |
868 | * family. | |
869 | */ | |
870 | struct sctp_af *sctp_get_af_specific(sa_family_t family) | |
871 | { | |
872 | switch (family) { | |
873 | case AF_INET: | |
874 | return sctp_af_v4_specific; | |
875 | case AF_INET6: | |
876 | return sctp_af_v6_specific; | |
877 | default: | |
878 | return NULL; | |
879 | } | |
880 | } | |
881 | ||
882 | /* Common code to initialize a AF_INET msg_name. */ | |
883 | static void sctp_inet_msgname(char *msgname, int *addr_len) | |
884 | { | |
885 | struct sockaddr_in *sin; | |
886 | ||
887 | sin = (struct sockaddr_in *)msgname; | |
888 | *addr_len = sizeof(struct sockaddr_in); | |
889 | sin->sin_family = AF_INET; | |
890 | memset(sin->sin_zero, 0, sizeof(sin->sin_zero)); | |
891 | } | |
892 | ||
893 | /* Copy the primary address of the peer primary address as the msg_name. */ | |
894 | static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname, | |
895 | int *addr_len) | |
896 | { | |
897 | struct sockaddr_in *sin, *sinfrom; | |
898 | ||
899 | if (msgname) { | |
900 | struct sctp_association *asoc; | |
901 | ||
902 | asoc = event->asoc; | |
903 | sctp_inet_msgname(msgname, addr_len); | |
904 | sin = (struct sockaddr_in *)msgname; | |
905 | sinfrom = &asoc->peer.primary_addr.v4; | |
906 | sin->sin_port = htons(asoc->peer.port); | |
907 | sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr; | |
908 | } | |
909 | } | |
910 | ||
911 | /* Initialize and copy out a msgname from an inbound skb. */ | |
912 | static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len) | |
913 | { | |
1da177e4 | 914 | if (msgname) { |
2c0fd387 ACM |
915 | struct sctphdr *sh = sctp_hdr(skb); |
916 | struct sockaddr_in *sin = (struct sockaddr_in *)msgname; | |
917 | ||
1da177e4 | 918 | sctp_inet_msgname(msgname, len); |
1da177e4 | 919 | sin->sin_port = sh->source; |
eddc9ec5 | 920 | sin->sin_addr.s_addr = ip_hdr(skb)->saddr; |
1da177e4 LT |
921 | } |
922 | } | |
923 | ||
924 | /* Do we support this AF? */ | |
925 | static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp) | |
926 | { | |
927 | /* PF_INET only supports AF_INET addresses. */ | |
a02cec21 | 928 | return AF_INET == family; |
1da177e4 LT |
929 | } |
930 | ||
931 | /* Address matching with wildcards allowed. */ | |
932 | static int sctp_inet_cmp_addr(const union sctp_addr *addr1, | |
933 | const union sctp_addr *addr2, | |
934 | struct sctp_sock *opt) | |
935 | { | |
936 | /* PF_INET only supports AF_INET addresses. */ | |
937 | if (addr1->sa.sa_family != addr2->sa.sa_family) | |
938 | return 0; | |
e6f1cebf AV |
939 | if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr || |
940 | htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr) | |
1da177e4 LT |
941 | return 1; |
942 | if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr) | |
943 | return 1; | |
944 | ||
945 | return 0; | |
946 | } | |
947 | ||
948 | /* Verify that provided sockaddr looks bindable. Common verification has | |
949 | * already been taken care of. | |
950 | */ | |
951 | static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr) | |
952 | { | |
953 | return sctp_v4_available(addr, opt); | |
954 | } | |
955 | ||
956 | /* Verify that sockaddr looks sendable. Common verification has already | |
957 | * been taken care of. | |
958 | */ | |
959 | static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr) | |
960 | { | |
961 | return 1; | |
962 | } | |
963 | ||
964 | /* Fill in Supported Address Type information for INIT and INIT-ACK | |
965 | * chunks. Returns number of addresses supported. | |
966 | */ | |
967 | static int sctp_inet_supported_addrs(const struct sctp_sock *opt, | |
3dbe8656 | 968 | __be16 *types) |
1da177e4 LT |
969 | { |
970 | types[0] = SCTP_PARAM_IPV4_ADDRESS; | |
971 | return 1; | |
972 | } | |
973 | ||
974 | /* Wrapper routine that calls the ip transmit routine. */ | |
975 | static inline int sctp_v4_xmit(struct sk_buff *skb, | |
f880374c | 976 | struct sctp_transport *transport) |
1da177e4 | 977 | { |
f880374c HX |
978 | struct inet_sock *inet = inet_sk(skb->sk); |
979 | ||
bb33381d DB |
980 | pr_debug("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n", __func__, skb, |
981 | skb->len, &transport->fl.u.ip4.saddr, &transport->fl.u.ip4.daddr); | |
1da177e4 | 982 | |
f880374c HX |
983 | inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ? |
984 | IP_PMTUDISC_DO : IP_PMTUDISC_DONT; | |
985 | ||
b01a2407 | 986 | SCTP_INC_STATS(sock_net(&inet->sk), SCTP_MIB_OUTSCTPPACKS); |
bb33381d | 987 | |
b0270e91 | 988 | return ip_queue_xmit(&inet->sk, skb, &transport->fl); |
1da177e4 LT |
989 | } |
990 | ||
15efbe76 | 991 | static struct sctp_af sctp_af_inet; |
1da177e4 LT |
992 | |
993 | static struct sctp_pf sctp_pf_inet = { | |
994 | .event_msgname = sctp_inet_event_msgname, | |
995 | .skb_msgname = sctp_inet_skb_msgname, | |
996 | .af_supported = sctp_inet_af_supported, | |
997 | .cmp_addr = sctp_inet_cmp_addr, | |
998 | .bind_verify = sctp_inet_bind_verify, | |
999 | .send_verify = sctp_inet_send_verify, | |
1000 | .supported_addrs = sctp_inet_supported_addrs, | |
1001 | .create_accept_sk = sctp_v4_create_accept_sk, | |
299ee123 JG |
1002 | .addr_to_user = sctp_v4_addr_to_user, |
1003 | .to_sk_saddr = sctp_v4_to_sk_saddr, | |
1004 | .to_sk_daddr = sctp_v4_to_sk_daddr, | |
15efbe76 | 1005 | .af = &sctp_af_inet |
1da177e4 LT |
1006 | }; |
1007 | ||
1008 | /* Notifier for inetaddr addition/deletion events. */ | |
1009 | static struct notifier_block sctp_inetaddr_notifier = { | |
1010 | .notifier_call = sctp_inetaddr_event, | |
1011 | }; | |
1012 | ||
1013 | /* Socket operations. */ | |
90ddc4f0 | 1014 | static const struct proto_ops inet_seqpacket_ops = { |
543d9cfe ACM |
1015 | .family = PF_INET, |
1016 | .owner = THIS_MODULE, | |
1017 | .release = inet_release, /* Needs to be wrapped... */ | |
1018 | .bind = inet_bind, | |
1019 | .connect = inet_dgram_connect, | |
1020 | .socketpair = sock_no_socketpair, | |
1021 | .accept = inet_accept, | |
1022 | .getname = inet_getname, /* Semantics are different. */ | |
1023 | .poll = sctp_poll, | |
1024 | .ioctl = inet_ioctl, | |
1025 | .listen = sctp_inet_listen, | |
1026 | .shutdown = inet_shutdown, /* Looks harmless. */ | |
1027 | .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */ | |
1028 | .getsockopt = sock_common_getsockopt, | |
1029 | .sendmsg = inet_sendmsg, | |
1030 | .recvmsg = sock_common_recvmsg, | |
1031 | .mmap = sock_no_mmap, | |
1032 | .sendpage = sock_no_sendpage, | |
3fdadf7d | 1033 | #ifdef CONFIG_COMPAT |
543d9cfe ACM |
1034 | .compat_setsockopt = compat_sock_common_setsockopt, |
1035 | .compat_getsockopt = compat_sock_common_getsockopt, | |
3fdadf7d | 1036 | #endif |
1da177e4 LT |
1037 | }; |
1038 | ||
1039 | /* Registration with AF_INET family. */ | |
1040 | static struct inet_protosw sctp_seqpacket_protosw = { | |
1041 | .type = SOCK_SEQPACKET, | |
1042 | .protocol = IPPROTO_SCTP, | |
1043 | .prot = &sctp_prot, | |
1044 | .ops = &inet_seqpacket_ops, | |
1da177e4 LT |
1045 | .flags = SCTP_PROTOSW_FLAG |
1046 | }; | |
1047 | static struct inet_protosw sctp_stream_protosw = { | |
1048 | .type = SOCK_STREAM, | |
1049 | .protocol = IPPROTO_SCTP, | |
1050 | .prot = &sctp_prot, | |
1051 | .ops = &inet_seqpacket_ops, | |
1da177e4 LT |
1052 | .flags = SCTP_PROTOSW_FLAG |
1053 | }; | |
1054 | ||
1055 | /* Register with IP layer. */ | |
32613090 | 1056 | static const struct net_protocol sctp_protocol = { |
1da177e4 LT |
1057 | .handler = sctp_rcv, |
1058 | .err_handler = sctp_v4_err, | |
1059 | .no_policy = 1, | |
bb2db45b | 1060 | .netns_ok = 1, |
8ed1dc44 | 1061 | .icmp_strict_tag_validation = 1, |
1da177e4 LT |
1062 | }; |
1063 | ||
1064 | /* IPv4 address related functions. */ | |
15efbe76 | 1065 | static struct sctp_af sctp_af_inet = { |
543d9cfe ACM |
1066 | .sa_family = AF_INET, |
1067 | .sctp_xmit = sctp_v4_xmit, | |
1068 | .setsockopt = ip_setsockopt, | |
1069 | .getsockopt = ip_getsockopt, | |
1070 | .get_dst = sctp_v4_get_dst, | |
1071 | .get_saddr = sctp_v4_get_saddr, | |
1072 | .copy_addrlist = sctp_v4_copy_addrlist, | |
1073 | .from_skb = sctp_v4_from_skb, | |
1074 | .from_sk = sctp_v4_from_sk, | |
543d9cfe ACM |
1075 | .from_addr_param = sctp_v4_from_addr_param, |
1076 | .to_addr_param = sctp_v4_to_addr_param, | |
543d9cfe ACM |
1077 | .cmp_addr = sctp_v4_cmp_addr, |
1078 | .addr_valid = sctp_v4_addr_valid, | |
1079 | .inaddr_any = sctp_v4_inaddr_any, | |
1080 | .is_any = sctp_v4_is_any, | |
1081 | .available = sctp_v4_available, | |
1082 | .scope = sctp_v4_scope, | |
1083 | .skb_iif = sctp_v4_skb_iif, | |
1084 | .is_ce = sctp_v4_is_ce, | |
1085 | .seq_dump_addr = sctp_v4_seq_dump_addr, | |
b9031d9d | 1086 | .ecn_capable = sctp_v4_ecn_capable, |
543d9cfe ACM |
1087 | .net_header_len = sizeof(struct iphdr), |
1088 | .sockaddr_len = sizeof(struct sockaddr_in), | |
3fdadf7d | 1089 | #ifdef CONFIG_COMPAT |
543d9cfe ACM |
1090 | .compat_setsockopt = compat_ip_setsockopt, |
1091 | .compat_getsockopt = compat_ip_getsockopt, | |
3fdadf7d | 1092 | #endif |
1da177e4 LT |
1093 | }; |
1094 | ||
8d72651d | 1095 | struct sctp_pf *sctp_get_pf_specific(sa_family_t family) |
1096 | { | |
1da177e4 LT |
1097 | switch (family) { |
1098 | case PF_INET: | |
1099 | return sctp_pf_inet_specific; | |
1100 | case PF_INET6: | |
1101 | return sctp_pf_inet6_specific; | |
1102 | default: | |
1103 | return NULL; | |
1104 | } | |
1105 | } | |
1106 | ||
1107 | /* Register the PF specific function table. */ | |
1108 | int sctp_register_pf(struct sctp_pf *pf, sa_family_t family) | |
1109 | { | |
1110 | switch (family) { | |
1111 | case PF_INET: | |
1112 | if (sctp_pf_inet_specific) | |
1113 | return 0; | |
1114 | sctp_pf_inet_specific = pf; | |
1115 | break; | |
1116 | case PF_INET6: | |
1117 | if (sctp_pf_inet6_specific) | |
1118 | return 0; | |
1119 | sctp_pf_inet6_specific = pf; | |
1120 | break; | |
1121 | default: | |
1122 | return 0; | |
1123 | } | |
1124 | return 1; | |
1125 | } | |
1126 | ||
b01a2407 | 1127 | static inline int init_sctp_mibs(struct net *net) |
996b1dba | 1128 | { |
698365fa WC |
1129 | net->sctp.sctp_statistics = alloc_percpu(struct sctp_mib); |
1130 | if (!net->sctp.sctp_statistics) | |
1131 | return -ENOMEM; | |
1132 | return 0; | |
1da177e4 LT |
1133 | } |
1134 | ||
b01a2407 | 1135 | static inline void cleanup_sctp_mibs(struct net *net) |
1da177e4 | 1136 | { |
698365fa | 1137 | free_percpu(net->sctp.sctp_statistics); |
1da177e4 LT |
1138 | } |
1139 | ||
270637ab VY |
1140 | static void sctp_v4_pf_init(void) |
1141 | { | |
1142 | /* Initialize the SCTP specific PF functions. */ | |
1143 | sctp_register_pf(&sctp_pf_inet, PF_INET); | |
1144 | sctp_register_af(&sctp_af_inet); | |
1145 | } | |
1146 | ||
1147 | static void sctp_v4_pf_exit(void) | |
1148 | { | |
1149 | list_del(&sctp_af_inet.list); | |
1150 | } | |
1151 | ||
1152 | static int sctp_v4_protosw_init(void) | |
1153 | { | |
1154 | int rc; | |
1155 | ||
1156 | rc = proto_register(&sctp_prot, 1); | |
1157 | if (rc) | |
1158 | return rc; | |
1159 | ||
1160 | /* Register SCTP(UDP and TCP style) with socket layer. */ | |
1161 | inet_register_protosw(&sctp_seqpacket_protosw); | |
1162 | inet_register_protosw(&sctp_stream_protosw); | |
1163 | ||
1164 | return 0; | |
1165 | } | |
1166 | ||
1167 | static void sctp_v4_protosw_exit(void) | |
1168 | { | |
1169 | inet_unregister_protosw(&sctp_stream_protosw); | |
1170 | inet_unregister_protosw(&sctp_seqpacket_protosw); | |
1171 | proto_unregister(&sctp_prot); | |
1172 | } | |
1173 | ||
1174 | static int sctp_v4_add_protocol(void) | |
1175 | { | |
1176 | /* Register notifier for inet address additions/deletions. */ | |
1177 | register_inetaddr_notifier(&sctp_inetaddr_notifier); | |
1178 | ||
1179 | /* Register SCTP with inet layer. */ | |
1180 | if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0) | |
1181 | return -EAGAIN; | |
1182 | ||
1183 | return 0; | |
1184 | } | |
1185 | ||
1186 | static void sctp_v4_del_protocol(void) | |
1187 | { | |
1188 | inet_del_protocol(&sctp_protocol, IPPROTO_SCTP); | |
1189 | unregister_inetaddr_notifier(&sctp_inetaddr_notifier); | |
1190 | } | |
1191 | ||
8e2d61e0 | 1192 | static int __net_init sctp_defaults_init(struct net *net) |
4db67e80 | 1193 | { |
2ce95503 EB |
1194 | int status; |
1195 | ||
e1fc3b14 EB |
1196 | /* |
1197 | * 14. Suggested SCTP Protocol Parameter Values | |
1198 | */ | |
1199 | /* The following protocol parameters are RECOMMENDED: */ | |
1200 | /* RTO.Initial - 3 seconds */ | |
1201 | net->sctp.rto_initial = SCTP_RTO_INITIAL; | |
1202 | /* RTO.Min - 1 second */ | |
1203 | net->sctp.rto_min = SCTP_RTO_MIN; | |
1204 | /* RTO.Max - 60 seconds */ | |
1205 | net->sctp.rto_max = SCTP_RTO_MAX; | |
1206 | /* RTO.Alpha - 1/8 */ | |
1207 | net->sctp.rto_alpha = SCTP_RTO_ALPHA; | |
1208 | /* RTO.Beta - 1/4 */ | |
1209 | net->sctp.rto_beta = SCTP_RTO_BETA; | |
1210 | ||
1211 | /* Valid.Cookie.Life - 60 seconds */ | |
1212 | net->sctp.valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE; | |
1213 | ||
1214 | /* Whether Cookie Preservative is enabled(1) or not(0) */ | |
1215 | net->sctp.cookie_preserve_enable = 1; | |
1216 | ||
3c68198e | 1217 | /* Default sctp sockets to use md5 as their hmac alg */ |
0d0863b0 | 1218 | #if defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_MD5) |
3c68198e | 1219 | net->sctp.sctp_hmac_alg = "md5"; |
0d0863b0 | 1220 | #elif defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_SHA1) |
3c68198e NH |
1221 | net->sctp.sctp_hmac_alg = "sha1"; |
1222 | #else | |
1223 | net->sctp.sctp_hmac_alg = NULL; | |
1224 | #endif | |
1225 | ||
e1fc3b14 EB |
1226 | /* Max.Burst - 4 */ |
1227 | net->sctp.max_burst = SCTP_DEFAULT_MAX_BURST; | |
1228 | ||
566178f8 ZY |
1229 | /* Enable pf state by default */ |
1230 | net->sctp.pf_enable = 1; | |
1231 | ||
e1fc3b14 EB |
1232 | /* Association.Max.Retrans - 10 attempts |
1233 | * Path.Max.Retrans - 5 attempts (per destination address) | |
1234 | * Max.Init.Retransmits - 8 attempts | |
1235 | */ | |
1236 | net->sctp.max_retrans_association = 10; | |
1237 | net->sctp.max_retrans_path = 5; | |
1238 | net->sctp.max_retrans_init = 8; | |
1239 | ||
1240 | /* Sendbuffer growth - do per-socket accounting */ | |
1241 | net->sctp.sndbuf_policy = 0; | |
1242 | ||
1243 | /* Rcvbuffer growth - do per-socket accounting */ | |
1244 | net->sctp.rcvbuf_policy = 0; | |
1245 | ||
1246 | /* HB.interval - 30 seconds */ | |
1247 | net->sctp.hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT; | |
1248 | ||
1249 | /* delayed SACK timeout */ | |
1250 | net->sctp.sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK; | |
1251 | ||
1252 | /* Disable ADDIP by default. */ | |
1253 | net->sctp.addip_enable = 0; | |
1254 | net->sctp.addip_noauth = 0; | |
1255 | net->sctp.default_auto_asconf = 0; | |
1256 | ||
1257 | /* Enable PR-SCTP by default. */ | |
1258 | net->sctp.prsctp_enable = 1; | |
1259 | ||
1260 | /* Disable AUTH by default. */ | |
1261 | net->sctp.auth_enable = 0; | |
1262 | ||
1263 | /* Set SCOPE policy to enabled */ | |
1264 | net->sctp.scope_policy = SCTP_SCOPE_POLICY_ENABLE; | |
1265 | ||
1266 | /* Set the default rwnd update threshold */ | |
1267 | net->sctp.rwnd_upd_shift = SCTP_DEFAULT_RWND_SHIFT; | |
1268 | ||
1269 | /* Initialize maximum autoclose timeout. */ | |
1270 | net->sctp.max_autoclose = INT_MAX / HZ; | |
1271 | ||
ebb7e95d EB |
1272 | status = sctp_sysctl_net_register(net); |
1273 | if (status) | |
1274 | goto err_sysctl_register; | |
1275 | ||
b01a2407 EB |
1276 | /* Allocate and initialise sctp mibs. */ |
1277 | status = init_sctp_mibs(net); | |
1278 | if (status) | |
1279 | goto err_init_mibs; | |
1280 | ||
13d782f6 EB |
1281 | /* Initialize proc fs directory. */ |
1282 | status = sctp_proc_init(net); | |
1283 | if (status) | |
1284 | goto err_init_proc; | |
1285 | ||
1286 | sctp_dbg_objcnt_init(net); | |
1287 | ||
4db67e80 EB |
1288 | /* Initialize the local address list. */ |
1289 | INIT_LIST_HEAD(&net->sctp.local_addr_list); | |
1290 | spin_lock_init(&net->sctp.local_addr_lock); | |
1291 | sctp_get_local_addr_list(net); | |
1292 | ||
1293 | /* Initialize the address event list */ | |
1294 | INIT_LIST_HEAD(&net->sctp.addr_waitq); | |
1295 | INIT_LIST_HEAD(&net->sctp.auto_asconf_splist); | |
1296 | spin_lock_init(&net->sctp.addr_wq_lock); | |
1297 | net->sctp.addr_wq_timer.expires = 0; | |
1298 | setup_timer(&net->sctp.addr_wq_timer, sctp_addr_wq_timeout_handler, | |
1299 | (unsigned long)net); | |
1300 | ||
1301 | return 0; | |
2ce95503 | 1302 | |
13d782f6 | 1303 | err_init_proc: |
b01a2407 EB |
1304 | cleanup_sctp_mibs(net); |
1305 | err_init_mibs: | |
ebb7e95d EB |
1306 | sctp_sysctl_net_unregister(net); |
1307 | err_sysctl_register: | |
2ce95503 | 1308 | return status; |
4db67e80 EB |
1309 | } |
1310 | ||
8e2d61e0 | 1311 | static void __net_exit sctp_defaults_exit(struct net *net) |
4db67e80 EB |
1312 | { |
1313 | /* Free the local address list */ | |
1314 | sctp_free_addr_wq(net); | |
1315 | sctp_free_local_addr_list(net); | |
2ce95503 | 1316 | |
13d782f6 EB |
1317 | sctp_dbg_objcnt_exit(net); |
1318 | ||
1319 | sctp_proc_exit(net); | |
b01a2407 | 1320 | cleanup_sctp_mibs(net); |
ebb7e95d | 1321 | sctp_sysctl_net_unregister(net); |
4db67e80 EB |
1322 | } |
1323 | ||
8e2d61e0 MRL |
1324 | static struct pernet_operations sctp_defaults_ops = { |
1325 | .init = sctp_defaults_init, | |
1326 | .exit = sctp_defaults_exit, | |
1327 | }; | |
1328 | ||
1329 | static int __net_init sctp_ctrlsock_init(struct net *net) | |
1330 | { | |
1331 | int status; | |
1332 | ||
1333 | /* Initialize the control inode/socket for handling OOTB packets. */ | |
1334 | status = sctp_ctl_sock_init(net); | |
1335 | if (status) | |
1336 | pr_err("Failed to initialize the SCTP control sock\n"); | |
1337 | ||
1338 | return status; | |
1339 | } | |
1340 | ||
1341 | static void __net_init sctp_ctrlsock_exit(struct net *net) | |
1342 | { | |
1343 | /* Free the control endpoint. */ | |
1344 | inet_ctl_sock_destroy(net->sctp.ctl_sock); | |
1345 | } | |
1346 | ||
1347 | static struct pernet_operations sctp_ctrlsock_ops = { | |
1348 | .init = sctp_ctrlsock_init, | |
1349 | .exit = sctp_ctrlsock_exit, | |
4db67e80 EB |
1350 | }; |
1351 | ||
1da177e4 | 1352 | /* Initialize the universe into something sensible. */ |
dda91928 | 1353 | static __init int sctp_init(void) |
1da177e4 LT |
1354 | { |
1355 | int i; | |
1356 | int status = -EINVAL; | |
1357 | unsigned long goal; | |
4d93df0a NH |
1358 | unsigned long limit; |
1359 | int max_share; | |
1da177e4 | 1360 | int order; |
d9749fb5 NH |
1361 | int num_entries; |
1362 | int max_entry_order; | |
1da177e4 | 1363 | |
b4772ef8 | 1364 | sock_skb_cb_check_size(sizeof(struct sctp_ulpevent)); |
1da177e4 | 1365 | |
827bf122 | 1366 | /* Allocate bind_bucket and chunk caches. */ |
1da177e4 LT |
1367 | status = -ENOBUFS; |
1368 | sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket", | |
1369 | sizeof(struct sctp_bind_bucket), | |
1370 | 0, SLAB_HWCACHE_ALIGN, | |
20c2df83 | 1371 | NULL); |
1da177e4 | 1372 | if (!sctp_bucket_cachep) |
827bf122 | 1373 | goto out; |
1da177e4 LT |
1374 | |
1375 | sctp_chunk_cachep = kmem_cache_create("sctp_chunk", | |
1376 | sizeof(struct sctp_chunk), | |
1377 | 0, SLAB_HWCACHE_ALIGN, | |
20c2df83 | 1378 | NULL); |
1da177e4 LT |
1379 | if (!sctp_chunk_cachep) |
1380 | goto err_chunk_cachep; | |
1381 | ||
908c7f19 | 1382 | status = percpu_counter_init(&sctp_sockets_allocated, 0, GFP_KERNEL); |
632c928a EB |
1383 | if (status) |
1384 | goto err_percpu_counter_init; | |
1385 | ||
1da177e4 LT |
1386 | /* Implementation specific variables. */ |
1387 | ||
1388 | /* Initialize default stream count setup information. */ | |
1389 | sctp_max_instreams = SCTP_DEFAULT_INSTREAMS; | |
1390 | sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS; | |
1391 | ||
1392 | /* Initialize handle used for association ids. */ | |
1393 | idr_init(&sctp_assocs_id); | |
1394 | ||
f03d78db | 1395 | limit = nr_free_buffer_pages() / 8; |
4d93df0a NH |
1396 | limit = max(limit, 128UL); |
1397 | sysctl_sctp_mem[0] = limit / 4 * 3; | |
1398 | sysctl_sctp_mem[1] = limit; | |
1399 | sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2; | |
1400 | ||
1401 | /* Set per-socket limits to no more than 1/128 the pressure threshold*/ | |
1402 | limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7); | |
1403 | max_share = min(4UL*1024*1024, limit); | |
1404 | ||
845525a6 | 1405 | sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */ |
87fb4b7b | 1406 | sysctl_sctp_rmem[1] = 1500 * SKB_TRUESIZE(1); |
4d93df0a NH |
1407 | sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share); |
1408 | ||
3ab224be | 1409 | sysctl_sctp_wmem[0] = SK_MEM_QUANTUM; |
4d93df0a NH |
1410 | sysctl_sctp_wmem[1] = 16*1024; |
1411 | sysctl_sctp_wmem[2] = max(64*1024, max_share); | |
1412 | ||
1da177e4 LT |
1413 | /* Size and allocate the association hash table. |
1414 | * The methodology is similar to that of the tcp hash tables. | |
d9749fb5 | 1415 | * Though not identical. Start by getting a goal size |
1da177e4 | 1416 | */ |
4481374c JB |
1417 | if (totalram_pages >= (128 * 1024)) |
1418 | goal = totalram_pages >> (22 - PAGE_SHIFT); | |
1da177e4 | 1419 | else |
4481374c | 1420 | goal = totalram_pages >> (24 - PAGE_SHIFT); |
1da177e4 | 1421 | |
d9749fb5 NH |
1422 | /* Then compute the page order for said goal */ |
1423 | order = get_order(goal); | |
1424 | ||
1425 | /* Now compute the required page order for the maximum sized table we | |
1426 | * want to create | |
1427 | */ | |
1428 | max_entry_order = get_order(MAX_SCTP_PORT_HASH_ENTRIES * | |
1429 | sizeof(struct sctp_bind_hashbucket)); | |
1430 | ||
1431 | /* Limit the page order by that maximum hash table size */ | |
1432 | order = min(order, max_entry_order); | |
1da177e4 | 1433 | |
1da177e4 LT |
1434 | /* Allocate and initialize the endpoint hash table. */ |
1435 | sctp_ep_hashsize = 64; | |
3b77d661 | 1436 | sctp_ep_hashtable = |
1da177e4 LT |
1437 | kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL); |
1438 | if (!sctp_ep_hashtable) { | |
145ce502 | 1439 | pr_err("Failed endpoint_hash alloc\n"); |
1da177e4 LT |
1440 | status = -ENOMEM; |
1441 | goto err_ehash_alloc; | |
1442 | } | |
1443 | for (i = 0; i < sctp_ep_hashsize; i++) { | |
1444 | rwlock_init(&sctp_ep_hashtable[i].lock); | |
d970dbf8 | 1445 | INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain); |
1da177e4 LT |
1446 | } |
1447 | ||
d9749fb5 NH |
1448 | /* Allocate and initialize the SCTP port hash table. |
1449 | * Note that order is initalized to start at the max sized | |
1450 | * table we want to support. If we can't get that many pages | |
1451 | * reduce the order and try again | |
1452 | */ | |
1da177e4 | 1453 | do { |
1da177e4 | 1454 | sctp_port_hashtable = (struct sctp_bind_hashbucket *) |
6857a02a | 1455 | __get_free_pages(GFP_KERNEL | __GFP_NOWARN, order); |
1da177e4 | 1456 | } while (!sctp_port_hashtable && --order > 0); |
d9749fb5 | 1457 | |
1da177e4 | 1458 | if (!sctp_port_hashtable) { |
145ce502 | 1459 | pr_err("Failed bind hash alloc\n"); |
1da177e4 LT |
1460 | status = -ENOMEM; |
1461 | goto err_bhash_alloc; | |
1462 | } | |
d9749fb5 NH |
1463 | |
1464 | /* Now compute the number of entries that will fit in the | |
1465 | * port hash space we allocated | |
1466 | */ | |
1467 | num_entries = (1UL << order) * PAGE_SIZE / | |
1468 | sizeof(struct sctp_bind_hashbucket); | |
1469 | ||
1470 | /* And finish by rounding it down to the nearest power of two | |
1471 | * this wastes some memory of course, but its needed because | |
1472 | * the hash function operates based on the assumption that | |
1473 | * that the number of entries is a power of two | |
1474 | */ | |
1475 | sctp_port_hashsize = rounddown_pow_of_two(num_entries); | |
1476 | ||
1da177e4 LT |
1477 | for (i = 0; i < sctp_port_hashsize; i++) { |
1478 | spin_lock_init(&sctp_port_hashtable[i].lock); | |
d970dbf8 | 1479 | INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain); |
1da177e4 LT |
1480 | } |
1481 | ||
4f008781 XL |
1482 | if (sctp_transport_hashtable_init()) |
1483 | goto err_thash_alloc; | |
1484 | ||
d9749fb5 NH |
1485 | pr_info("Hash tables configured (bind %d/%d)\n", sctp_port_hashsize, |
1486 | num_entries); | |
1da177e4 | 1487 | |
1da177e4 LT |
1488 | sctp_sysctl_register(); |
1489 | ||
1490 | INIT_LIST_HEAD(&sctp_address_families); | |
270637ab VY |
1491 | sctp_v4_pf_init(); |
1492 | sctp_v6_pf_init(); | |
1da177e4 | 1493 | |
8e2d61e0 MRL |
1494 | status = register_pernet_subsys(&sctp_defaults_ops); |
1495 | if (status) | |
1496 | goto err_register_defaults; | |
827bf122 | 1497 | |
8e2d61e0 | 1498 | status = sctp_v4_protosw_init(); |
1da177e4 | 1499 | if (status) |
270637ab VY |
1500 | goto err_protosw_init; |
1501 | ||
1502 | status = sctp_v6_protosw_init(); | |
1503 | if (status) | |
1504 | goto err_v6_protosw_init; | |
1da177e4 | 1505 | |
8e2d61e0 | 1506 | status = register_pernet_subsys(&sctp_ctrlsock_ops); |
4db67e80 | 1507 | if (status) |
8e2d61e0 | 1508 | goto err_register_ctrlsock; |
4db67e80 | 1509 | |
270637ab VY |
1510 | status = sctp_v4_add_protocol(); |
1511 | if (status) | |
827bf122 | 1512 | goto err_add_protocol; |
827bf122 SS |
1513 | |
1514 | /* Register SCTP with inet6 layer. */ | |
1515 | status = sctp_v6_add_protocol(); | |
1516 | if (status) | |
1517 | goto err_v6_add_protocol; | |
1518 | ||
90017acc MRL |
1519 | if (sctp_offload_init() < 0) |
1520 | pr_crit("%s: Cannot add SCTP protocol offload\n", __func__); | |
1521 | ||
1da177e4 LT |
1522 | out: |
1523 | return status; | |
827bf122 | 1524 | err_v6_add_protocol: |
270637ab | 1525 | sctp_v4_del_protocol(); |
d1dd5247 | 1526 | err_add_protocol: |
8e2d61e0 MRL |
1527 | unregister_pernet_subsys(&sctp_ctrlsock_ops); |
1528 | err_register_ctrlsock: | |
270637ab VY |
1529 | sctp_v6_protosw_exit(); |
1530 | err_v6_protosw_init: | |
1531 | sctp_v4_protosw_exit(); | |
1532 | err_protosw_init: | |
8e2d61e0 MRL |
1533 | unregister_pernet_subsys(&sctp_defaults_ops); |
1534 | err_register_defaults: | |
270637ab VY |
1535 | sctp_v4_pf_exit(); |
1536 | sctp_v6_pf_exit(); | |
1da177e4 | 1537 | sctp_sysctl_unregister(); |
1da177e4 LT |
1538 | free_pages((unsigned long)sctp_port_hashtable, |
1539 | get_order(sctp_port_hashsize * | |
1540 | sizeof(struct sctp_bind_hashbucket))); | |
1541 | err_bhash_alloc: | |
4f008781 XL |
1542 | sctp_transport_hashtable_destroy(); |
1543 | err_thash_alloc: | |
1da177e4 LT |
1544 | kfree(sctp_ep_hashtable); |
1545 | err_ehash_alloc: | |
632c928a EB |
1546 | percpu_counter_destroy(&sctp_sockets_allocated); |
1547 | err_percpu_counter_init: | |
1da177e4 LT |
1548 | kmem_cache_destroy(sctp_chunk_cachep); |
1549 | err_chunk_cachep: | |
1550 | kmem_cache_destroy(sctp_bucket_cachep); | |
1da177e4 LT |
1551 | goto out; |
1552 | } | |
1553 | ||
1554 | /* Exit handler for the SCTP protocol. */ | |
dda91928 | 1555 | static __exit void sctp_exit(void) |
1da177e4 LT |
1556 | { |
1557 | /* BUG. This should probably do something useful like clean | |
1558 | * up all the remaining associations and all that memory. | |
1559 | */ | |
1560 | ||
827bf122 SS |
1561 | /* Unregister with inet6/inet layers. */ |
1562 | sctp_v6_del_protocol(); | |
270637ab | 1563 | sctp_v4_del_protocol(); |
1da177e4 | 1564 | |
8e2d61e0 | 1565 | unregister_pernet_subsys(&sctp_ctrlsock_ops); |
4db67e80 | 1566 | |
270637ab VY |
1567 | /* Free protosw registrations */ |
1568 | sctp_v6_protosw_exit(); | |
1569 | sctp_v4_protosw_exit(); | |
1570 | ||
8e2d61e0 MRL |
1571 | unregister_pernet_subsys(&sctp_defaults_ops); |
1572 | ||
827bf122 | 1573 | /* Unregister with socket layer. */ |
270637ab VY |
1574 | sctp_v6_pf_exit(); |
1575 | sctp_v4_pf_exit(); | |
827bf122 | 1576 | |
1da177e4 | 1577 | sctp_sysctl_unregister(); |
1da177e4 | 1578 | |
1da177e4 LT |
1579 | free_pages((unsigned long)sctp_port_hashtable, |
1580 | get_order(sctp_port_hashsize * | |
1581 | sizeof(struct sctp_bind_hashbucket))); | |
b5eff712 | 1582 | kfree(sctp_ep_hashtable); |
4f008781 | 1583 | sctp_transport_hashtable_destroy(); |
1da177e4 | 1584 | |
632c928a | 1585 | percpu_counter_destroy(&sctp_sockets_allocated); |
1da177e4 | 1586 | |
eaa184a1 JDB |
1587 | rcu_barrier(); /* Wait for completion of call_rcu()'s */ |
1588 | ||
827bf122 SS |
1589 | kmem_cache_destroy(sctp_chunk_cachep); |
1590 | kmem_cache_destroy(sctp_bucket_cachep); | |
1da177e4 LT |
1591 | } |
1592 | ||
1593 | module_init(sctp_init); | |
1594 | module_exit(sctp_exit); | |
1595 | ||
bb97d31f ACM |
1596 | /* |
1597 | * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly. | |
1598 | */ | |
1599 | MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132"); | |
882a382c | 1600 | MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132"); |
91705c61 | 1601 | MODULE_AUTHOR("Linux Kernel SCTP developers <linux-sctp@vger.kernel.org>"); |
1da177e4 | 1602 | MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)"); |
71acc0dd DM |
1603 | module_param_named(no_checksums, sctp_checksum_disable, bool, 0644); |
1604 | MODULE_PARM_DESC(no_checksums, "Disable checksums computing and verification"); | |
1da177e4 | 1605 | MODULE_LICENSE("GPL"); |