]>
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 | 153 | addr->a.v4.sin_family = AF_INET; |
2a6fd78a | 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 | 167 | static 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 | 184 | static 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 | 197 | int sctp_copy_local_addr_list(struct net *net, struct sctp_bind_addr *bp, |
1c662018 | 198 | enum sctp_scope scope, gfp_t gfp, int copy_flags) |
1da177e4 LT |
199 | { |
200 | struct sctp_sockaddr_entry *addr; | |
2e3ce5bc | 201 | union sctp_addr laddr; |
1da177e4 | 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; | |
165f2cf6 XL |
208 | if (!sctp_in_scope(net, &addr->a, scope)) |
209 | continue; | |
210 | ||
211 | /* Now that the address is in scope, check to see if | |
212 | * the address type is really supported by the local | |
213 | * sock as well as the remote peer. | |
214 | */ | |
215 | if (addr->a.sa.sa_family == AF_INET && | |
216 | !(copy_flags & SCTP_ADDR4_PEERSUPP)) | |
217 | continue; | |
218 | if (addr->a.sa.sa_family == AF_INET6 && | |
219 | (!(copy_flags & SCTP_ADDR6_ALLOWED) || | |
220 | !(copy_flags & SCTP_ADDR6_PEERSUPP))) | |
221 | continue; | |
222 | ||
2e3ce5bc XL |
223 | laddr = addr->a; |
224 | /* also works for setting ipv6 address port */ | |
225 | laddr.v4.sin_port = htons(bp->port); | |
226 | if (sctp_bind_addr_state(bp, &laddr) != -1) | |
b8607805 XL |
227 | continue; |
228 | ||
165f2cf6 XL |
229 | error = sctp_add_bind_addr(bp, &addr->a, sizeof(addr->a), |
230 | SCTP_ADDR_SRC, GFP_ATOMIC); | |
231 | if (error) | |
232 | break; | |
1da177e4 LT |
233 | } |
234 | ||
29303547 | 235 | rcu_read_unlock(); |
1da177e4 LT |
236 | return error; |
237 | } | |
238 | ||
239 | /* Initialize a sctp_addr from in incoming skb. */ | |
240 | static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb, | |
241 | int is_saddr) | |
242 | { | |
0630c56e MRL |
243 | /* Always called on head skb, so this is safe */ |
244 | struct sctphdr *sh = sctp_hdr(skb); | |
245 | struct sockaddr_in *sa = &addr->v4; | |
1da177e4 | 246 | |
1da177e4 LT |
247 | addr->v4.sin_family = AF_INET; |
248 | ||
1da177e4 | 249 | if (is_saddr) { |
0630c56e MRL |
250 | sa->sin_port = sh->source; |
251 | sa->sin_addr.s_addr = ip_hdr(skb)->saddr; | |
1da177e4 | 252 | } else { |
0630c56e MRL |
253 | sa->sin_port = sh->dest; |
254 | sa->sin_addr.s_addr = ip_hdr(skb)->daddr; | |
1da177e4 | 255 | } |
4c04b91e | 256 | memset(sa->sin_zero, 0, sizeof(sa->sin_zero)); |
1da177e4 LT |
257 | } |
258 | ||
259 | /* Initialize an sctp_addr from a socket. */ | |
260 | static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk) | |
261 | { | |
262 | addr->v4.sin_family = AF_INET; | |
7dcdbd95 | 263 | addr->v4.sin_port = 0; |
c720c7e8 | 264 | addr->v4.sin_addr.s_addr = inet_sk(sk)->inet_rcv_saddr; |
4c04b91e | 265 | memset(addr->v4.sin_zero, 0, sizeof(addr->v4.sin_zero)); |
1da177e4 LT |
266 | } |
267 | ||
268 | /* Initialize sk->sk_rcv_saddr from sctp_addr. */ | |
269 | static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk) | |
270 | { | |
c720c7e8 | 271 | inet_sk(sk)->inet_rcv_saddr = addr->v4.sin_addr.s_addr; |
1da177e4 LT |
272 | } |
273 | ||
274 | /* Initialize sk->sk_daddr from sctp_addr. */ | |
275 | static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk) | |
276 | { | |
c720c7e8 | 277 | inet_sk(sk)->inet_daddr = addr->v4.sin_addr.s_addr; |
1da177e4 LT |
278 | } |
279 | ||
280 | /* Initialize a sctp_addr from an address parameter. */ | |
281 | static void sctp_v4_from_addr_param(union sctp_addr *addr, | |
282 | union sctp_addr_param *param, | |
dd86d136 | 283 | __be16 port, int iif) |
1da177e4 LT |
284 | { |
285 | addr->v4.sin_family = AF_INET; | |
286 | addr->v4.sin_port = port; | |
287 | addr->v4.sin_addr.s_addr = param->v4.addr.s_addr; | |
4c04b91e | 288 | memset(addr->v4.sin_zero, 0, sizeof(addr->v4.sin_zero)); |
1da177e4 LT |
289 | } |
290 | ||
291 | /* Initialize an address parameter from a sctp_addr and return the length | |
292 | * of the address parameter. | |
293 | */ | |
294 | static int sctp_v4_to_addr_param(const union sctp_addr *addr, | |
295 | union sctp_addr_param *param) | |
296 | { | |
a38905e6 | 297 | int length = sizeof(struct sctp_ipv4addr_param); |
1da177e4 LT |
298 | |
299 | param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS; | |
dbc16db1 | 300 | param->v4.param_hdr.length = htons(length); |
d808ad9a | 301 | param->v4.addr.s_addr = addr->v4.sin_addr.s_addr; |
1da177e4 LT |
302 | |
303 | return length; | |
304 | } | |
305 | ||
306 | /* Initialize a sctp_addr from a dst_entry. */ | |
18a353f4 | 307 | static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct flowi4 *fl4, |
854d43a4 | 308 | __be16 port) |
1da177e4 | 309 | { |
1da177e4 LT |
310 | saddr->v4.sin_family = AF_INET; |
311 | saddr->v4.sin_port = port; | |
18a353f4 | 312 | saddr->v4.sin_addr.s_addr = fl4->saddr; |
4c04b91e | 313 | memset(saddr->v4.sin_zero, 0, sizeof(saddr->v4.sin_zero)); |
1da177e4 LT |
314 | } |
315 | ||
316 | /* Compare two addresses exactly. */ | |
317 | static int sctp_v4_cmp_addr(const union sctp_addr *addr1, | |
318 | const union sctp_addr *addr2) | |
319 | { | |
320 | if (addr1->sa.sa_family != addr2->sa.sa_family) | |
321 | return 0; | |
322 | if (addr1->v4.sin_port != addr2->v4.sin_port) | |
323 | return 0; | |
324 | if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr) | |
325 | return 0; | |
326 | ||
327 | return 1; | |
328 | } | |
329 | ||
330 | /* Initialize addr struct to INADDR_ANY. */ | |
6fbfa9f9 | 331 | static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port) |
1da177e4 LT |
332 | { |
333 | addr->v4.sin_family = AF_INET; | |
e6f1cebf | 334 | addr->v4.sin_addr.s_addr = htonl(INADDR_ANY); |
1da177e4 | 335 | addr->v4.sin_port = port; |
4c04b91e | 336 | memset(addr->v4.sin_zero, 0, sizeof(addr->v4.sin_zero)); |
1da177e4 LT |
337 | } |
338 | ||
339 | /* Is this a wildcard address? */ | |
340 | static int sctp_v4_is_any(const union sctp_addr *addr) | |
341 | { | |
e6f1cebf | 342 | return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr; |
1da177e4 LT |
343 | } |
344 | ||
345 | /* This function checks if the address is a valid address to be used for | |
346 | * SCTP binding. | |
347 | * | |
348 | * Output: | |
349 | * Return 0 - If the address is a non-unicast or an illegal address. | |
350 | * Return 1 - If the address is a unicast. | |
351 | */ | |
5636bef7 VY |
352 | static int sctp_v4_addr_valid(union sctp_addr *addr, |
353 | struct sctp_sock *sp, | |
354 | const struct sk_buff *skb) | |
1da177e4 | 355 | { |
7dab83de VY |
356 | /* IPv4 addresses not allowed */ |
357 | if (sp && ipv6_only_sock(sctp_opt2sk(sp))) | |
358 | return 0; | |
359 | ||
1da177e4 | 360 | /* Is this a non-unicast address or a unusable SCTP address? */ |
b5cb2bbc | 361 | if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) |
1da177e4 LT |
362 | return 0; |
363 | ||
d808ad9a | 364 | /* Is this a broadcast address? */ |
511c3f92 | 365 | if (skb && skb_rtable(skb)->rt_flags & RTCF_BROADCAST) |
d808ad9a | 366 | return 0; |
5636bef7 | 367 | |
1da177e4 LT |
368 | return 1; |
369 | } | |
370 | ||
371 | /* Should this be available for binding? */ | |
372 | static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp) | |
373 | { | |
bb2db45b EB |
374 | struct net *net = sock_net(&sp->inet.sk); |
375 | int ret = inet_addr_type(net, addr->v4.sin_addr.s_addr); | |
1da177e4 | 376 | |
1da177e4 | 377 | |
e6f1cebf | 378 | if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) && |
cdac4e07 NH |
379 | ret != RTN_LOCAL && |
380 | !sp->inet.freebind && | |
49a60158 | 381 | !net->ipv4.sysctl_ip_nonlocal_bind) |
1da177e4 | 382 | return 0; |
cdac4e07 | 383 | |
7dab83de VY |
384 | if (ipv6_only_sock(sctp_opt2sk(sp))) |
385 | return 0; | |
386 | ||
1da177e4 LT |
387 | return 1; |
388 | } | |
389 | ||
390 | /* Checking the loopback, private and other address scopes as defined in | |
391 | * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4 | |
392 | * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>. | |
393 | * | |
394 | * Level 0 - unusable SCTP addresses | |
395 | * Level 1 - loopback address | |
396 | * Level 2 - link-local addresses | |
397 | * Level 3 - private addresses. | |
398 | * Level 4 - global addresses | |
399 | * For INIT and INIT-ACK address list, let L be the level of | |
400 | * of requested destination address, sender and receiver | |
401 | * SHOULD include all of its addresses with level greater | |
402 | * than or equal to L. | |
72388433 BD |
403 | * |
404 | * IPv4 scoping can be controlled through sysctl option | |
405 | * net.sctp.addr_scope_policy | |
1da177e4 | 406 | */ |
1c662018 | 407 | static enum sctp_scope sctp_v4_scope(union sctp_addr *addr) |
1da177e4 | 408 | { |
1c662018 | 409 | enum sctp_scope retval; |
1da177e4 | 410 | |
1da177e4 | 411 | /* Check for unusable SCTP addresses. */ |
b5cb2bbc | 412 | if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) { |
1da177e4 | 413 | retval = SCTP_SCOPE_UNUSABLE; |
b5cb2bbc | 414 | } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) { |
1da177e4 | 415 | retval = SCTP_SCOPE_LOOPBACK; |
b5cb2bbc | 416 | } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) { |
1da177e4 | 417 | retval = SCTP_SCOPE_LINK; |
b5cb2bbc JP |
418 | } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) || |
419 | ipv4_is_private_172(addr->v4.sin_addr.s_addr) || | |
420 | ipv4_is_private_192(addr->v4.sin_addr.s_addr)) { | |
1da177e4 LT |
421 | retval = SCTP_SCOPE_PRIVATE; |
422 | } else { | |
423 | retval = SCTP_SCOPE_GLOBAL; | |
424 | } | |
425 | ||
426 | return retval; | |
427 | } | |
428 | ||
429 | /* Returns a valid dst cache entry for the given source and destination ip | |
430 | * addresses. If an association is passed, trys to get a dst entry with a | |
431 | * source address that matches an address in the bind address list. | |
432 | */ | |
da0420be VY |
433 | static void sctp_v4_get_dst(struct sctp_transport *t, union sctp_addr *saddr, |
434 | struct flowi *fl, struct sock *sk) | |
1da177e4 | 435 | { |
da0420be | 436 | struct sctp_association *asoc = t->asoc; |
1da177e4 | 437 | struct rtable *rt; |
9914ae3c | 438 | struct flowi4 *fl4 = &fl->u.ip4; |
1da177e4 | 439 | struct sctp_bind_addr *bp; |
1da177e4 | 440 | struct sctp_sockaddr_entry *laddr; |
1da177e4 | 441 | struct dst_entry *dst = NULL; |
da0420be | 442 | union sctp_addr *daddr = &t->ipaddr; |
1da177e4 LT |
443 | union sctp_addr dst_saddr; |
444 | ||
9914ae3c VY |
445 | memset(fl4, 0x0, sizeof(struct flowi4)); |
446 | fl4->daddr = daddr->v4.sin_addr.s_addr; | |
447 | fl4->fl4_dport = daddr->v4.sin_port; | |
448 | fl4->flowi4_proto = IPPROTO_SCTP; | |
1da177e4 | 449 | if (asoc) { |
9914ae3c VY |
450 | fl4->flowi4_tos = RT_CONN_FLAGS(asoc->base.sk); |
451 | fl4->flowi4_oif = asoc->base.sk->sk_bound_dev_if; | |
452 | fl4->fl4_sport = htons(asoc->base.bind_addr.port); | |
1da177e4 | 453 | } |
6429d3dc | 454 | if (saddr) { |
9914ae3c | 455 | fl4->saddr = saddr->v4.sin_addr.s_addr; |
f3bcd953 XL |
456 | if (!fl4->fl4_sport) |
457 | fl4->fl4_sport = saddr->v4.sin_port; | |
6429d3dc | 458 | } |
1da177e4 | 459 | |
bb33381d DB |
460 | pr_debug("%s: dst:%pI4, src:%pI4 - ", __func__, &fl4->daddr, |
461 | &fl4->saddr); | |
1da177e4 | 462 | |
bb2db45b | 463 | rt = ip_route_output_key(sock_net(sk), fl4); |
b23dd4fe | 464 | if (!IS_ERR(rt)) |
d8d1f30b | 465 | dst = &rt->dst; |
1da177e4 LT |
466 | |
467 | /* If there is no association or if a source address is passed, no | |
468 | * more validation is required. | |
469 | */ | |
470 | if (!asoc || saddr) | |
471 | goto out; | |
472 | ||
473 | bp = &asoc->base.bind_addr; | |
1da177e4 LT |
474 | |
475 | if (dst) { | |
476 | /* Walk through the bind address list and look for a bind | |
477 | * address that matches the source address of the returned dst. | |
478 | */ | |
18a353f4 | 479 | sctp_v4_dst_saddr(&dst_saddr, fl4, htons(bp->port)); |
559cf710 VY |
480 | rcu_read_lock(); |
481 | list_for_each_entry_rcu(laddr, &bp->address_list, list) { | |
8a07eb0a MH |
482 | if (!laddr->valid || (laddr->state == SCTP_ADDR_DEL) || |
483 | (laddr->state != SCTP_ADDR_SRC && | |
484 | !asoc->src_out_of_asoc_ok)) | |
dc022a98 | 485 | continue; |
854d43a4 | 486 | if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a)) |
1da177e4 LT |
487 | goto out_unlock; |
488 | } | |
559cf710 | 489 | rcu_read_unlock(); |
1da177e4 LT |
490 | |
491 | /* None of the bound addresses match the source address of the | |
492 | * dst. So release it. | |
493 | */ | |
494 | dst_release(dst); | |
495 | dst = NULL; | |
496 | } | |
497 | ||
498 | /* Walk through the bind address list and try to get a dst that | |
499 | * matches a bind address as the source address. | |
500 | */ | |
559cf710 VY |
501 | rcu_read_lock(); |
502 | list_for_each_entry_rcu(laddr, &bp->address_list, list) { | |
0ca50d12 MRL |
503 | struct net_device *odev; |
504 | ||
559cf710 VY |
505 | if (!laddr->valid) |
506 | continue; | |
07868284 MRL |
507 | if (laddr->state != SCTP_ADDR_SRC || |
508 | AF_INET != laddr->a.sa.sa_family) | |
509 | continue; | |
510 | ||
511 | fl4->fl4_sport = laddr->a.v4.sin_port; | |
512 | flowi4_update_output(fl4, | |
513 | asoc->base.sk->sk_bound_dev_if, | |
514 | RT_CONN_FLAGS(asoc->base.sk), | |
515 | daddr->v4.sin_addr.s_addr, | |
516 | laddr->a.v4.sin_addr.s_addr); | |
517 | ||
518 | rt = ip_route_output_key(sock_net(sk), fl4); | |
519 | if (IS_ERR(rt)) | |
520 | continue; | |
521 | ||
0ca50d12 MRL |
522 | /* Ensure the src address belongs to the output |
523 | * interface. | |
524 | */ | |
525 | odev = __ip_dev_find(sock_net(sk), laddr->a.v4.sin_addr.s_addr, | |
526 | false); | |
d82f0f1f | 527 | if (!odev || odev->ifindex != fl4->flowi4_oif) { |
8a6d278d TR |
528 | if (!dst) |
529 | dst = &rt->dst; | |
530 | else | |
410f0383 | 531 | dst_release(&rt->dst); |
0ca50d12 | 532 | continue; |
d82f0f1f | 533 | } |
0ca50d12 | 534 | |
8a6d278d | 535 | dst_release(dst); |
07868284 MRL |
536 | dst = &rt->dst; |
537 | break; | |
1da177e4 LT |
538 | } |
539 | ||
540 | out_unlock: | |
559cf710 | 541 | rcu_read_unlock(); |
1da177e4 | 542 | out: |
da0420be | 543 | t->dst = dst; |
1da177e4 | 544 | if (dst) |
bb33381d DB |
545 | pr_debug("rt_dst:%pI4, rt_src:%pI4\n", |
546 | &fl4->daddr, &fl4->saddr); | |
1da177e4 | 547 | else |
bb33381d | 548 | pr_debug("no route\n"); |
1da177e4 LT |
549 | } |
550 | ||
551 | /* For v4, the source address is cached in the route entry(dst). So no need | |
552 | * to cache it separately and hence this is an empty routine. | |
553 | */ | |
e5117101 | 554 | static void sctp_v4_get_saddr(struct sctp_sock *sk, |
9914ae3c | 555 | struct sctp_transport *t, |
9914ae3c | 556 | struct flowi *fl) |
1da177e4 | 557 | { |
9914ae3c VY |
558 | union sctp_addr *saddr = &t->saddr; |
559 | struct rtable *rt = (struct rtable *)t->dst; | |
23ec47a0 | 560 | |
1da177e4 LT |
561 | if (rt) { |
562 | saddr->v4.sin_family = AF_INET; | |
902ebd3e | 563 | saddr->v4.sin_addr.s_addr = fl->u.ip4.saddr; |
1da177e4 LT |
564 | } |
565 | } | |
566 | ||
567 | /* What interface did this skb arrive on? */ | |
568 | static int sctp_v4_skb_iif(const struct sk_buff *skb) | |
569 | { | |
92101b3b | 570 | return inet_iif(skb); |
1da177e4 LT |
571 | } |
572 | ||
573 | /* Was this packet marked by Explicit Congestion Notification? */ | |
574 | static int sctp_v4_is_ce(const struct sk_buff *skb) | |
575 | { | |
eddc9ec5 | 576 | return INET_ECN_is_ce(ip_hdr(skb)->tos); |
1da177e4 LT |
577 | } |
578 | ||
579 | /* Create and initialize a new sk for the socket returned by accept(). */ | |
580 | static struct sock *sctp_v4_create_accept_sk(struct sock *sk, | |
cdfbabfb DH |
581 | struct sctp_association *asoc, |
582 | bool kern) | |
1da177e4 | 583 | { |
3b1e0a65 | 584 | struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL, |
cdfbabfb | 585 | sk->sk_prot, kern); |
914e1c8b | 586 | struct inet_sock *newinet; |
1da177e4 LT |
587 | |
588 | if (!newsk) | |
589 | goto out; | |
590 | ||
591 | sock_init_data(NULL, newsk); | |
592 | ||
914e1c8b | 593 | sctp_copy_sock(newsk, sk, asoc); |
1da177e4 LT |
594 | sock_reset_flag(newsk, SOCK_ZAPPED); |
595 | ||
596 | newinet = inet_sk(newsk); | |
597 | ||
c720c7e8 | 598 | newinet->inet_daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr; |
1da177e4 | 599 | |
e6848976 | 600 | sk_refcnt_debug_inc(newsk); |
1da177e4 LT |
601 | |
602 | if (newsk->sk_prot->init(newsk)) { | |
603 | sk_common_release(newsk); | |
604 | newsk = NULL; | |
605 | } | |
606 | ||
607 | out: | |
608 | return newsk; | |
609 | } | |
610 | ||
299ee123 | 611 | static int sctp_v4_addr_to_user(struct sctp_sock *sp, union sctp_addr *addr) |
1da177e4 | 612 | { |
299ee123 | 613 | /* No address mapping for V4 sockets */ |
58b4357e | 614 | memset(addr->v4.sin_zero, 0, sizeof(addr->v4.sin_zero)); |
299ee123 | 615 | return sizeof(struct sockaddr_in); |
1da177e4 LT |
616 | } |
617 | ||
618 | /* Dump the v4 addr to the seq file. */ | |
619 | static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr) | |
620 | { | |
21454aaa | 621 | seq_printf(seq, "%pI4 ", &addr->v4.sin_addr); |
1da177e4 LT |
622 | } |
623 | ||
b9031d9d VY |
624 | static void sctp_v4_ecn_capable(struct sock *sk) |
625 | { | |
626 | INET_ECN_xmit(sk); | |
627 | } | |
628 | ||
9c3b5751 | 629 | static void sctp_addr_wq_timeout_handler(struct timer_list *t) |
9f7d653b | 630 | { |
9c3b5751 | 631 | struct net *net = from_timer(net, t, sctp.addr_wq_timer); |
9f7d653b MH |
632 | struct sctp_sockaddr_entry *addrw, *temp; |
633 | struct sctp_sock *sp; | |
634 | ||
4db67e80 | 635 | spin_lock_bh(&net->sctp.addr_wq_lock); |
9f7d653b | 636 | |
4db67e80 | 637 | list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) { |
bb33381d DB |
638 | pr_debug("%s: the first ent in wq:%p is addr:%pISc for cmd:%d at " |
639 | "entry:%p\n", __func__, &net->sctp.addr_waitq, &addrw->a.sa, | |
640 | addrw->state, addrw); | |
9f7d653b | 641 | |
dfd56b8b | 642 | #if IS_ENABLED(CONFIG_IPV6) |
9f7d653b MH |
643 | /* Now we send an ASCONF for each association */ |
644 | /* Note. we currently don't handle link local IPv6 addressees */ | |
645 | if (addrw->a.sa.sa_family == AF_INET6) { | |
646 | struct in6_addr *in6; | |
647 | ||
648 | if (ipv6_addr_type(&addrw->a.v6.sin6_addr) & | |
649 | IPV6_ADDR_LINKLOCAL) | |
650 | goto free_next; | |
651 | ||
652 | in6 = (struct in6_addr *)&addrw->a.v6.sin6_addr; | |
4db67e80 | 653 | if (ipv6_chk_addr(net, in6, NULL, 0) == 0 && |
9f7d653b MH |
654 | addrw->state == SCTP_ADDR_NEW) { |
655 | unsigned long timeo_val; | |
656 | ||
bb33381d DB |
657 | pr_debug("%s: this is on DAD, trying %d sec " |
658 | "later\n", __func__, | |
659 | SCTP_ADDRESS_TICK_DELAY); | |
660 | ||
9f7d653b MH |
661 | timeo_val = jiffies; |
662 | timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY); | |
4db67e80 | 663 | mod_timer(&net->sctp.addr_wq_timer, timeo_val); |
9f7d653b MH |
664 | break; |
665 | } | |
666 | } | |
5d0c90cf | 667 | #endif |
4db67e80 | 668 | list_for_each_entry(sp, &net->sctp.auto_asconf_splist, auto_asconf_list) { |
9f7d653b MH |
669 | struct sock *sk; |
670 | ||
671 | sk = sctp_opt2sk(sp); | |
672 | /* ignore bound-specific endpoints */ | |
673 | if (!sctp_is_ep_boundall(sk)) | |
674 | continue; | |
5bc1d1b4 | 675 | bh_lock_sock(sk); |
9f7d653b | 676 | if (sctp_asconf_mgmt(sp, addrw) < 0) |
bb33381d | 677 | pr_debug("%s: sctp_asconf_mgmt failed\n", __func__); |
5bc1d1b4 | 678 | bh_unlock_sock(sk); |
9f7d653b | 679 | } |
39d84a58 | 680 | #if IS_ENABLED(CONFIG_IPV6) |
9f7d653b | 681 | free_next: |
39d84a58 | 682 | #endif |
9f7d653b MH |
683 | list_del(&addrw->list); |
684 | kfree(addrw); | |
685 | } | |
4db67e80 | 686 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
687 | } |
688 | ||
4db67e80 | 689 | static void sctp_free_addr_wq(struct net *net) |
9f7d653b MH |
690 | { |
691 | struct sctp_sockaddr_entry *addrw; | |
692 | struct sctp_sockaddr_entry *temp; | |
693 | ||
4db67e80 EB |
694 | spin_lock_bh(&net->sctp.addr_wq_lock); |
695 | del_timer(&net->sctp.addr_wq_timer); | |
696 | list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) { | |
9f7d653b MH |
697 | list_del(&addrw->list); |
698 | kfree(addrw); | |
699 | } | |
4db67e80 | 700 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
701 | } |
702 | ||
703 | /* lookup the entry for the same address in the addr_waitq | |
704 | * sctp_addr_wq MUST be locked | |
705 | */ | |
4db67e80 EB |
706 | static struct sctp_sockaddr_entry *sctp_addr_wq_lookup(struct net *net, |
707 | struct sctp_sockaddr_entry *addr) | |
9f7d653b MH |
708 | { |
709 | struct sctp_sockaddr_entry *addrw; | |
710 | ||
4db67e80 | 711 | list_for_each_entry(addrw, &net->sctp.addr_waitq, list) { |
9f7d653b MH |
712 | if (addrw->a.sa.sa_family != addr->a.sa.sa_family) |
713 | continue; | |
714 | if (addrw->a.sa.sa_family == AF_INET) { | |
715 | if (addrw->a.v4.sin_addr.s_addr == | |
716 | addr->a.v4.sin_addr.s_addr) | |
717 | return addrw; | |
718 | } else if (addrw->a.sa.sa_family == AF_INET6) { | |
719 | if (ipv6_addr_equal(&addrw->a.v6.sin6_addr, | |
720 | &addr->a.v6.sin6_addr)) | |
721 | return addrw; | |
722 | } | |
723 | } | |
724 | return NULL; | |
725 | } | |
726 | ||
4db67e80 | 727 | void sctp_addr_wq_mgmt(struct net *net, struct sctp_sockaddr_entry *addr, int cmd) |
9f7d653b MH |
728 | { |
729 | struct sctp_sockaddr_entry *addrw; | |
730 | unsigned long timeo_val; | |
731 | ||
732 | /* first, we check if an opposite message already exist in the queue. | |
733 | * If we found such message, it is removed. | |
734 | * This operation is a bit stupid, but the DHCP client attaches the | |
735 | * new address after a couple of addition and deletion of that address | |
736 | */ | |
737 | ||
4db67e80 | 738 | spin_lock_bh(&net->sctp.addr_wq_lock); |
9f7d653b | 739 | /* Offsets existing events in addr_wq */ |
4db67e80 | 740 | addrw = sctp_addr_wq_lookup(net, addr); |
9f7d653b MH |
741 | if (addrw) { |
742 | if (addrw->state != cmd) { | |
bb33381d DB |
743 | pr_debug("%s: offsets existing entry for %d, addr:%pISc " |
744 | "in wq:%p\n", __func__, addrw->state, &addrw->a.sa, | |
745 | &net->sctp.addr_waitq); | |
746 | ||
9f7d653b MH |
747 | list_del(&addrw->list); |
748 | kfree(addrw); | |
749 | } | |
4db67e80 | 750 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
751 | return; |
752 | } | |
753 | ||
754 | /* OK, we have to add the new address to the wait queue */ | |
755 | addrw = kmemdup(addr, sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC); | |
756 | if (addrw == NULL) { | |
4db67e80 | 757 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
758 | return; |
759 | } | |
760 | addrw->state = cmd; | |
4db67e80 | 761 | list_add_tail(&addrw->list, &net->sctp.addr_waitq); |
bb33381d DB |
762 | |
763 | pr_debug("%s: add new entry for cmd:%d, addr:%pISc in wq:%p\n", | |
764 | __func__, addrw->state, &addrw->a.sa, &net->sctp.addr_waitq); | |
9f7d653b | 765 | |
4db67e80 | 766 | if (!timer_pending(&net->sctp.addr_wq_timer)) { |
9f7d653b MH |
767 | timeo_val = jiffies; |
768 | timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY); | |
4db67e80 | 769 | mod_timer(&net->sctp.addr_wq_timer, timeo_val); |
9f7d653b | 770 | } |
4db67e80 | 771 | spin_unlock_bh(&net->sctp.addr_wq_lock); |
9f7d653b MH |
772 | } |
773 | ||
29303547 VY |
774 | /* Event handler for inet address addition/deletion events. |
775 | * The sctp_local_addr_list needs to be protocted by a spin lock since | |
776 | * multiple notifiers (say IPv4 and IPv6) may be running at the same | |
777 | * time and thus corrupt the list. | |
778 | * The reader side is protected with RCU. | |
779 | */ | |
24123186 AB |
780 | static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev, |
781 | void *ptr) | |
1da177e4 | 782 | { |
29c7cf96 | 783 | struct in_ifaddr *ifa = (struct in_ifaddr *)ptr; |
29303547 VY |
784 | struct sctp_sockaddr_entry *addr = NULL; |
785 | struct sctp_sockaddr_entry *temp; | |
4db67e80 | 786 | struct net *net = dev_net(ifa->ifa_dev->dev); |
22626216 | 787 | int found = 0; |
1da177e4 | 788 | |
29c7cf96 SS |
789 | switch (ev) { |
790 | case NETDEV_UP: | |
ef495abd | 791 | addr = kzalloc(sizeof(*addr), GFP_ATOMIC); |
29c7cf96 SS |
792 | if (addr) { |
793 | addr->a.v4.sin_family = AF_INET; | |
29c7cf96 | 794 | addr->a.v4.sin_addr.s_addr = ifa->ifa_local; |
29303547 | 795 | addr->valid = 1; |
4db67e80 EB |
796 | spin_lock_bh(&net->sctp.local_addr_lock); |
797 | list_add_tail_rcu(&addr->list, &net->sctp.local_addr_list); | |
798 | sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_NEW); | |
799 | spin_unlock_bh(&net->sctp.local_addr_lock); | |
29c7cf96 SS |
800 | } |
801 | break; | |
802 | case NETDEV_DOWN: | |
4db67e80 | 803 | spin_lock_bh(&net->sctp.local_addr_lock); |
29303547 | 804 | list_for_each_entry_safe(addr, temp, |
4db67e80 | 805 | &net->sctp.local_addr_list, list) { |
a40a7d15 PE |
806 | if (addr->a.sa.sa_family == AF_INET && |
807 | addr->a.v4.sin_addr.s_addr == | |
808 | ifa->ifa_local) { | |
4db67e80 | 809 | sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_DEL); |
22626216 | 810 | found = 1; |
29303547 VY |
811 | addr->valid = 0; |
812 | list_del_rcu(&addr->list); | |
29c7cf96 SS |
813 | break; |
814 | } | |
815 | } | |
4db67e80 | 816 | spin_unlock_bh(&net->sctp.local_addr_lock); |
22626216 | 817 | if (found) |
1231f0ba | 818 | kfree_rcu(addr, rcu); |
29c7cf96 SS |
819 | break; |
820 | } | |
1da177e4 LT |
821 | |
822 | return NOTIFY_DONE; | |
823 | } | |
824 | ||
825 | /* | |
826 | * Initialize the control inode/socket with a control endpoint data | |
827 | * structure. This endpoint is reserved exclusively for the OOTB processing. | |
828 | */ | |
2ce95503 | 829 | static int sctp_ctl_sock_init(struct net *net) |
1da177e4 LT |
830 | { |
831 | int err; | |
fb13d9f9 | 832 | sa_family_t family = PF_INET; |
1da177e4 LT |
833 | |
834 | if (sctp_get_pf_specific(PF_INET6)) | |
835 | family = PF_INET6; | |
1da177e4 | 836 | |
2ce95503 EB |
837 | err = inet_ctl_sock_create(&net->sctp.ctl_sock, family, |
838 | SOCK_SEQPACKET, IPPROTO_SCTP, net); | |
fb13d9f9 BH |
839 | |
840 | /* If IPv6 socket could not be created, try the IPv4 socket */ | |
841 | if (err < 0 && family == PF_INET6) | |
2ce95503 | 842 | err = inet_ctl_sock_create(&net->sctp.ctl_sock, AF_INET, |
fb13d9f9 | 843 | SOCK_SEQPACKET, IPPROTO_SCTP, |
2ce95503 | 844 | net); |
fb13d9f9 | 845 | |
1da177e4 | 846 | if (err < 0) { |
145ce502 | 847 | pr_err("Failed to create the SCTP control socket\n"); |
1da177e4 LT |
848 | return err; |
849 | } | |
1da177e4 LT |
850 | return 0; |
851 | } | |
852 | ||
853 | /* Register address family specific functions. */ | |
854 | int sctp_register_af(struct sctp_af *af) | |
855 | { | |
856 | switch (af->sa_family) { | |
857 | case AF_INET: | |
858 | if (sctp_af_v4_specific) | |
859 | return 0; | |
860 | sctp_af_v4_specific = af; | |
861 | break; | |
862 | case AF_INET6: | |
863 | if (sctp_af_v6_specific) | |
864 | return 0; | |
865 | sctp_af_v6_specific = af; | |
866 | break; | |
867 | default: | |
868 | return 0; | |
869 | } | |
870 | ||
871 | INIT_LIST_HEAD(&af->list); | |
872 | list_add_tail(&af->list, &sctp_address_families); | |
873 | return 1; | |
874 | } | |
875 | ||
876 | /* Get the table of functions for manipulating a particular address | |
877 | * family. | |
878 | */ | |
879 | struct sctp_af *sctp_get_af_specific(sa_family_t family) | |
880 | { | |
881 | switch (family) { | |
882 | case AF_INET: | |
883 | return sctp_af_v4_specific; | |
884 | case AF_INET6: | |
885 | return sctp_af_v6_specific; | |
886 | default: | |
887 | return NULL; | |
888 | } | |
889 | } | |
890 | ||
891 | /* Common code to initialize a AF_INET msg_name. */ | |
892 | static void sctp_inet_msgname(char *msgname, int *addr_len) | |
893 | { | |
894 | struct sockaddr_in *sin; | |
895 | ||
896 | sin = (struct sockaddr_in *)msgname; | |
897 | *addr_len = sizeof(struct sockaddr_in); | |
898 | sin->sin_family = AF_INET; | |
899 | memset(sin->sin_zero, 0, sizeof(sin->sin_zero)); | |
900 | } | |
901 | ||
902 | /* Copy the primary address of the peer primary address as the msg_name. */ | |
903 | static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname, | |
904 | int *addr_len) | |
905 | { | |
906 | struct sockaddr_in *sin, *sinfrom; | |
907 | ||
908 | if (msgname) { | |
909 | struct sctp_association *asoc; | |
910 | ||
911 | asoc = event->asoc; | |
912 | sctp_inet_msgname(msgname, addr_len); | |
913 | sin = (struct sockaddr_in *)msgname; | |
914 | sinfrom = &asoc->peer.primary_addr.v4; | |
915 | sin->sin_port = htons(asoc->peer.port); | |
916 | sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr; | |
917 | } | |
918 | } | |
919 | ||
920 | /* Initialize and copy out a msgname from an inbound skb. */ | |
921 | static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len) | |
922 | { | |
1da177e4 | 923 | if (msgname) { |
2c0fd387 ACM |
924 | struct sctphdr *sh = sctp_hdr(skb); |
925 | struct sockaddr_in *sin = (struct sockaddr_in *)msgname; | |
926 | ||
1da177e4 | 927 | sctp_inet_msgname(msgname, len); |
1da177e4 | 928 | sin->sin_port = sh->source; |
eddc9ec5 | 929 | sin->sin_addr.s_addr = ip_hdr(skb)->saddr; |
1da177e4 LT |
930 | } |
931 | } | |
932 | ||
933 | /* Do we support this AF? */ | |
934 | static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp) | |
935 | { | |
936 | /* PF_INET only supports AF_INET addresses. */ | |
a02cec21 | 937 | return AF_INET == family; |
1da177e4 LT |
938 | } |
939 | ||
940 | /* Address matching with wildcards allowed. */ | |
941 | static int sctp_inet_cmp_addr(const union sctp_addr *addr1, | |
942 | const union sctp_addr *addr2, | |
943 | struct sctp_sock *opt) | |
944 | { | |
945 | /* PF_INET only supports AF_INET addresses. */ | |
946 | if (addr1->sa.sa_family != addr2->sa.sa_family) | |
947 | return 0; | |
e6f1cebf AV |
948 | if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr || |
949 | htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr) | |
1da177e4 LT |
950 | return 1; |
951 | if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr) | |
952 | return 1; | |
953 | ||
954 | return 0; | |
955 | } | |
956 | ||
957 | /* Verify that provided sockaddr looks bindable. Common verification has | |
958 | * already been taken care of. | |
959 | */ | |
960 | static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr) | |
961 | { | |
962 | return sctp_v4_available(addr, opt); | |
963 | } | |
964 | ||
965 | /* Verify that sockaddr looks sendable. Common verification has already | |
966 | * been taken care of. | |
967 | */ | |
968 | static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr) | |
969 | { | |
970 | return 1; | |
971 | } | |
972 | ||
973 | /* Fill in Supported Address Type information for INIT and INIT-ACK | |
974 | * chunks. Returns number of addresses supported. | |
975 | */ | |
976 | static int sctp_inet_supported_addrs(const struct sctp_sock *opt, | |
3dbe8656 | 977 | __be16 *types) |
1da177e4 LT |
978 | { |
979 | types[0] = SCTP_PARAM_IPV4_ADDRESS; | |
980 | return 1; | |
981 | } | |
982 | ||
983 | /* Wrapper routine that calls the ip transmit routine. */ | |
984 | static inline int sctp_v4_xmit(struct sk_buff *skb, | |
f880374c | 985 | struct sctp_transport *transport) |
1da177e4 | 986 | { |
f880374c HX |
987 | struct inet_sock *inet = inet_sk(skb->sk); |
988 | ||
bb33381d DB |
989 | pr_debug("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n", __func__, skb, |
990 | skb->len, &transport->fl.u.ip4.saddr, &transport->fl.u.ip4.daddr); | |
1da177e4 | 991 | |
f880374c HX |
992 | inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ? |
993 | IP_PMTUDISC_DO : IP_PMTUDISC_DONT; | |
994 | ||
b01a2407 | 995 | SCTP_INC_STATS(sock_net(&inet->sk), SCTP_MIB_OUTSCTPPACKS); |
bb33381d | 996 | |
b0270e91 | 997 | return ip_queue_xmit(&inet->sk, skb, &transport->fl); |
1da177e4 LT |
998 | } |
999 | ||
15efbe76 | 1000 | static struct sctp_af sctp_af_inet; |
1da177e4 LT |
1001 | |
1002 | static struct sctp_pf sctp_pf_inet = { | |
1003 | .event_msgname = sctp_inet_event_msgname, | |
1004 | .skb_msgname = sctp_inet_skb_msgname, | |
1005 | .af_supported = sctp_inet_af_supported, | |
1006 | .cmp_addr = sctp_inet_cmp_addr, | |
1007 | .bind_verify = sctp_inet_bind_verify, | |
1008 | .send_verify = sctp_inet_send_verify, | |
1009 | .supported_addrs = sctp_inet_supported_addrs, | |
1010 | .create_accept_sk = sctp_v4_create_accept_sk, | |
299ee123 JG |
1011 | .addr_to_user = sctp_v4_addr_to_user, |
1012 | .to_sk_saddr = sctp_v4_to_sk_saddr, | |
1013 | .to_sk_daddr = sctp_v4_to_sk_daddr, | |
15efbe76 | 1014 | .af = &sctp_af_inet |
1da177e4 LT |
1015 | }; |
1016 | ||
1017 | /* Notifier for inetaddr addition/deletion events. */ | |
1018 | static struct notifier_block sctp_inetaddr_notifier = { | |
1019 | .notifier_call = sctp_inetaddr_event, | |
1020 | }; | |
1021 | ||
1022 | /* Socket operations. */ | |
90ddc4f0 | 1023 | static const struct proto_ops inet_seqpacket_ops = { |
543d9cfe ACM |
1024 | .family = PF_INET, |
1025 | .owner = THIS_MODULE, | |
1026 | .release = inet_release, /* Needs to be wrapped... */ | |
1027 | .bind = inet_bind, | |
acbfe3cb | 1028 | .connect = sctp_inet_connect, |
543d9cfe ACM |
1029 | .socketpair = sock_no_socketpair, |
1030 | .accept = inet_accept, | |
1031 | .getname = inet_getname, /* Semantics are different. */ | |
1032 | .poll = sctp_poll, | |
1033 | .ioctl = inet_ioctl, | |
1034 | .listen = sctp_inet_listen, | |
1035 | .shutdown = inet_shutdown, /* Looks harmless. */ | |
1036 | .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */ | |
1037 | .getsockopt = sock_common_getsockopt, | |
1038 | .sendmsg = inet_sendmsg, | |
fd2d180a | 1039 | .recvmsg = inet_recvmsg, |
543d9cfe ACM |
1040 | .mmap = sock_no_mmap, |
1041 | .sendpage = sock_no_sendpage, | |
3fdadf7d | 1042 | #ifdef CONFIG_COMPAT |
543d9cfe ACM |
1043 | .compat_setsockopt = compat_sock_common_setsockopt, |
1044 | .compat_getsockopt = compat_sock_common_getsockopt, | |
3fdadf7d | 1045 | #endif |
1da177e4 LT |
1046 | }; |
1047 | ||
1048 | /* Registration with AF_INET family. */ | |
1049 | static struct inet_protosw sctp_seqpacket_protosw = { | |
1050 | .type = SOCK_SEQPACKET, | |
1051 | .protocol = IPPROTO_SCTP, | |
1052 | .prot = &sctp_prot, | |
1053 | .ops = &inet_seqpacket_ops, | |
1da177e4 LT |
1054 | .flags = SCTP_PROTOSW_FLAG |
1055 | }; | |
1056 | static struct inet_protosw sctp_stream_protosw = { | |
1057 | .type = SOCK_STREAM, | |
1058 | .protocol = IPPROTO_SCTP, | |
1059 | .prot = &sctp_prot, | |
1060 | .ops = &inet_seqpacket_ops, | |
1da177e4 LT |
1061 | .flags = SCTP_PROTOSW_FLAG |
1062 | }; | |
1063 | ||
1064 | /* Register with IP layer. */ | |
32613090 | 1065 | static const struct net_protocol sctp_protocol = { |
1da177e4 LT |
1066 | .handler = sctp_rcv, |
1067 | .err_handler = sctp_v4_err, | |
1068 | .no_policy = 1, | |
bb2db45b | 1069 | .netns_ok = 1, |
8ed1dc44 | 1070 | .icmp_strict_tag_validation = 1, |
1da177e4 LT |
1071 | }; |
1072 | ||
1073 | /* IPv4 address related functions. */ | |
15efbe76 | 1074 | static struct sctp_af sctp_af_inet = { |
543d9cfe ACM |
1075 | .sa_family = AF_INET, |
1076 | .sctp_xmit = sctp_v4_xmit, | |
1077 | .setsockopt = ip_setsockopt, | |
1078 | .getsockopt = ip_getsockopt, | |
1079 | .get_dst = sctp_v4_get_dst, | |
1080 | .get_saddr = sctp_v4_get_saddr, | |
1081 | .copy_addrlist = sctp_v4_copy_addrlist, | |
1082 | .from_skb = sctp_v4_from_skb, | |
1083 | .from_sk = sctp_v4_from_sk, | |
543d9cfe ACM |
1084 | .from_addr_param = sctp_v4_from_addr_param, |
1085 | .to_addr_param = sctp_v4_to_addr_param, | |
543d9cfe ACM |
1086 | .cmp_addr = sctp_v4_cmp_addr, |
1087 | .addr_valid = sctp_v4_addr_valid, | |
1088 | .inaddr_any = sctp_v4_inaddr_any, | |
1089 | .is_any = sctp_v4_is_any, | |
1090 | .available = sctp_v4_available, | |
1091 | .scope = sctp_v4_scope, | |
1092 | .skb_iif = sctp_v4_skb_iif, | |
1093 | .is_ce = sctp_v4_is_ce, | |
1094 | .seq_dump_addr = sctp_v4_seq_dump_addr, | |
b9031d9d | 1095 | .ecn_capable = sctp_v4_ecn_capable, |
543d9cfe ACM |
1096 | .net_header_len = sizeof(struct iphdr), |
1097 | .sockaddr_len = sizeof(struct sockaddr_in), | |
3fdadf7d | 1098 | #ifdef CONFIG_COMPAT |
543d9cfe ACM |
1099 | .compat_setsockopt = compat_ip_setsockopt, |
1100 | .compat_getsockopt = compat_ip_getsockopt, | |
3fdadf7d | 1101 | #endif |
1da177e4 LT |
1102 | }; |
1103 | ||
8d72651d | 1104 | struct sctp_pf *sctp_get_pf_specific(sa_family_t family) |
1105 | { | |
1da177e4 LT |
1106 | switch (family) { |
1107 | case PF_INET: | |
1108 | return sctp_pf_inet_specific; | |
1109 | case PF_INET6: | |
1110 | return sctp_pf_inet6_specific; | |
1111 | default: | |
1112 | return NULL; | |
1113 | } | |
1114 | } | |
1115 | ||
1116 | /* Register the PF specific function table. */ | |
1117 | int sctp_register_pf(struct sctp_pf *pf, sa_family_t family) | |
1118 | { | |
1119 | switch (family) { | |
1120 | case PF_INET: | |
1121 | if (sctp_pf_inet_specific) | |
1122 | return 0; | |
1123 | sctp_pf_inet_specific = pf; | |
1124 | break; | |
1125 | case PF_INET6: | |
1126 | if (sctp_pf_inet6_specific) | |
1127 | return 0; | |
1128 | sctp_pf_inet6_specific = pf; | |
1129 | break; | |
1130 | default: | |
1131 | return 0; | |
1132 | } | |
1133 | return 1; | |
1134 | } | |
1135 | ||
b01a2407 | 1136 | static inline int init_sctp_mibs(struct net *net) |
996b1dba | 1137 | { |
698365fa WC |
1138 | net->sctp.sctp_statistics = alloc_percpu(struct sctp_mib); |
1139 | if (!net->sctp.sctp_statistics) | |
1140 | return -ENOMEM; | |
1141 | return 0; | |
1da177e4 LT |
1142 | } |
1143 | ||
b01a2407 | 1144 | static inline void cleanup_sctp_mibs(struct net *net) |
1da177e4 | 1145 | { |
698365fa | 1146 | free_percpu(net->sctp.sctp_statistics); |
1da177e4 LT |
1147 | } |
1148 | ||
270637ab VY |
1149 | static void sctp_v4_pf_init(void) |
1150 | { | |
1151 | /* Initialize the SCTP specific PF functions. */ | |
1152 | sctp_register_pf(&sctp_pf_inet, PF_INET); | |
1153 | sctp_register_af(&sctp_af_inet); | |
1154 | } | |
1155 | ||
1156 | static void sctp_v4_pf_exit(void) | |
1157 | { | |
1158 | list_del(&sctp_af_inet.list); | |
1159 | } | |
1160 | ||
1161 | static int sctp_v4_protosw_init(void) | |
1162 | { | |
1163 | int rc; | |
1164 | ||
1165 | rc = proto_register(&sctp_prot, 1); | |
1166 | if (rc) | |
1167 | return rc; | |
1168 | ||
1169 | /* Register SCTP(UDP and TCP style) with socket layer. */ | |
1170 | inet_register_protosw(&sctp_seqpacket_protosw); | |
1171 | inet_register_protosw(&sctp_stream_protosw); | |
1172 | ||
1173 | return 0; | |
1174 | } | |
1175 | ||
1176 | static void sctp_v4_protosw_exit(void) | |
1177 | { | |
1178 | inet_unregister_protosw(&sctp_stream_protosw); | |
1179 | inet_unregister_protosw(&sctp_seqpacket_protosw); | |
1180 | proto_unregister(&sctp_prot); | |
1181 | } | |
1182 | ||
1183 | static int sctp_v4_add_protocol(void) | |
1184 | { | |
1185 | /* Register notifier for inet address additions/deletions. */ | |
1186 | register_inetaddr_notifier(&sctp_inetaddr_notifier); | |
1187 | ||
1188 | /* Register SCTP with inet layer. */ | |
1189 | if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0) | |
1190 | return -EAGAIN; | |
1191 | ||
1192 | return 0; | |
1193 | } | |
1194 | ||
1195 | static void sctp_v4_del_protocol(void) | |
1196 | { | |
1197 | inet_del_protocol(&sctp_protocol, IPPROTO_SCTP); | |
1198 | unregister_inetaddr_notifier(&sctp_inetaddr_notifier); | |
1199 | } | |
1200 | ||
8e2d61e0 | 1201 | static int __net_init sctp_defaults_init(struct net *net) |
4db67e80 | 1202 | { |
2ce95503 EB |
1203 | int status; |
1204 | ||
e1fc3b14 EB |
1205 | /* |
1206 | * 14. Suggested SCTP Protocol Parameter Values | |
1207 | */ | |
1208 | /* The following protocol parameters are RECOMMENDED: */ | |
1209 | /* RTO.Initial - 3 seconds */ | |
1210 | net->sctp.rto_initial = SCTP_RTO_INITIAL; | |
1211 | /* RTO.Min - 1 second */ | |
1212 | net->sctp.rto_min = SCTP_RTO_MIN; | |
1213 | /* RTO.Max - 60 seconds */ | |
1214 | net->sctp.rto_max = SCTP_RTO_MAX; | |
1215 | /* RTO.Alpha - 1/8 */ | |
1216 | net->sctp.rto_alpha = SCTP_RTO_ALPHA; | |
1217 | /* RTO.Beta - 1/4 */ | |
1218 | net->sctp.rto_beta = SCTP_RTO_BETA; | |
1219 | ||
1220 | /* Valid.Cookie.Life - 60 seconds */ | |
1221 | net->sctp.valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE; | |
1222 | ||
1223 | /* Whether Cookie Preservative is enabled(1) or not(0) */ | |
1224 | net->sctp.cookie_preserve_enable = 1; | |
1225 | ||
3c68198e | 1226 | /* Default sctp sockets to use md5 as their hmac alg */ |
0d0863b0 | 1227 | #if defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_MD5) |
3c68198e | 1228 | net->sctp.sctp_hmac_alg = "md5"; |
0d0863b0 | 1229 | #elif defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_SHA1) |
3c68198e NH |
1230 | net->sctp.sctp_hmac_alg = "sha1"; |
1231 | #else | |
1232 | net->sctp.sctp_hmac_alg = NULL; | |
1233 | #endif | |
1234 | ||
e1fc3b14 EB |
1235 | /* Max.Burst - 4 */ |
1236 | net->sctp.max_burst = SCTP_DEFAULT_MAX_BURST; | |
1237 | ||
566178f8 ZY |
1238 | /* Enable pf state by default */ |
1239 | net->sctp.pf_enable = 1; | |
1240 | ||
e1fc3b14 EB |
1241 | /* Association.Max.Retrans - 10 attempts |
1242 | * Path.Max.Retrans - 5 attempts (per destination address) | |
1243 | * Max.Init.Retransmits - 8 attempts | |
1244 | */ | |
1245 | net->sctp.max_retrans_association = 10; | |
1246 | net->sctp.max_retrans_path = 5; | |
1247 | net->sctp.max_retrans_init = 8; | |
1248 | ||
1249 | /* Sendbuffer growth - do per-socket accounting */ | |
1250 | net->sctp.sndbuf_policy = 0; | |
1251 | ||
1252 | /* Rcvbuffer growth - do per-socket accounting */ | |
1253 | net->sctp.rcvbuf_policy = 0; | |
1254 | ||
1255 | /* HB.interval - 30 seconds */ | |
1256 | net->sctp.hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT; | |
1257 | ||
1258 | /* delayed SACK timeout */ | |
1259 | net->sctp.sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK; | |
1260 | ||
1261 | /* Disable ADDIP by default. */ | |
1262 | net->sctp.addip_enable = 0; | |
1263 | net->sctp.addip_noauth = 0; | |
1264 | net->sctp.default_auto_asconf = 0; | |
1265 | ||
1266 | /* Enable PR-SCTP by default. */ | |
1267 | net->sctp.prsctp_enable = 1; | |
1268 | ||
c28445c3 XL |
1269 | /* Disable RECONF by default. */ |
1270 | net->sctp.reconf_enable = 0; | |
1271 | ||
e1fc3b14 EB |
1272 | /* Disable AUTH by default. */ |
1273 | net->sctp.auth_enable = 0; | |
1274 | ||
1275 | /* Set SCOPE policy to enabled */ | |
1276 | net->sctp.scope_policy = SCTP_SCOPE_POLICY_ENABLE; | |
1277 | ||
1278 | /* Set the default rwnd update threshold */ | |
1279 | net->sctp.rwnd_upd_shift = SCTP_DEFAULT_RWND_SHIFT; | |
1280 | ||
1281 | /* Initialize maximum autoclose timeout. */ | |
1282 | net->sctp.max_autoclose = INT_MAX / HZ; | |
1283 | ||
ebb7e95d EB |
1284 | status = sctp_sysctl_net_register(net); |
1285 | if (status) | |
1286 | goto err_sysctl_register; | |
1287 | ||
b01a2407 EB |
1288 | /* Allocate and initialise sctp mibs. */ |
1289 | status = init_sctp_mibs(net); | |
1290 | if (status) | |
1291 | goto err_init_mibs; | |
1292 | ||
13d782f6 EB |
1293 | /* Initialize proc fs directory. */ |
1294 | status = sctp_proc_init(net); | |
1295 | if (status) | |
1296 | goto err_init_proc; | |
1297 | ||
1298 | sctp_dbg_objcnt_init(net); | |
1299 | ||
4db67e80 EB |
1300 | /* Initialize the local address list. */ |
1301 | INIT_LIST_HEAD(&net->sctp.local_addr_list); | |
1302 | spin_lock_init(&net->sctp.local_addr_lock); | |
1303 | sctp_get_local_addr_list(net); | |
1304 | ||
1305 | /* Initialize the address event list */ | |
1306 | INIT_LIST_HEAD(&net->sctp.addr_waitq); | |
1307 | INIT_LIST_HEAD(&net->sctp.auto_asconf_splist); | |
1308 | spin_lock_init(&net->sctp.addr_wq_lock); | |
1309 | net->sctp.addr_wq_timer.expires = 0; | |
9c3b5751 | 1310 | timer_setup(&net->sctp.addr_wq_timer, sctp_addr_wq_timeout_handler, 0); |
4db67e80 EB |
1311 | |
1312 | return 0; | |
2ce95503 | 1313 | |
13d782f6 | 1314 | err_init_proc: |
b01a2407 EB |
1315 | cleanup_sctp_mibs(net); |
1316 | err_init_mibs: | |
ebb7e95d EB |
1317 | sctp_sysctl_net_unregister(net); |
1318 | err_sysctl_register: | |
2ce95503 | 1319 | return status; |
4db67e80 EB |
1320 | } |
1321 | ||
8e2d61e0 | 1322 | static void __net_exit sctp_defaults_exit(struct net *net) |
4db67e80 EB |
1323 | { |
1324 | /* Free the local address list */ | |
1325 | sctp_free_addr_wq(net); | |
1326 | sctp_free_local_addr_list(net); | |
2ce95503 | 1327 | |
13d782f6 EB |
1328 | sctp_dbg_objcnt_exit(net); |
1329 | ||
1330 | sctp_proc_exit(net); | |
b01a2407 | 1331 | cleanup_sctp_mibs(net); |
ebb7e95d | 1332 | sctp_sysctl_net_unregister(net); |
4db67e80 EB |
1333 | } |
1334 | ||
8e2d61e0 MRL |
1335 | static struct pernet_operations sctp_defaults_ops = { |
1336 | .init = sctp_defaults_init, | |
1337 | .exit = sctp_defaults_exit, | |
1338 | }; | |
1339 | ||
1340 | static int __net_init sctp_ctrlsock_init(struct net *net) | |
1341 | { | |
1342 | int status; | |
1343 | ||
1344 | /* Initialize the control inode/socket for handling OOTB packets. */ | |
1345 | status = sctp_ctl_sock_init(net); | |
1346 | if (status) | |
1347 | pr_err("Failed to initialize the SCTP control sock\n"); | |
1348 | ||
1349 | return status; | |
1350 | } | |
1351 | ||
d95c7c7e | 1352 | static void __net_exit sctp_ctrlsock_exit(struct net *net) |
8e2d61e0 MRL |
1353 | { |
1354 | /* Free the control endpoint. */ | |
1355 | inet_ctl_sock_destroy(net->sctp.ctl_sock); | |
1356 | } | |
1357 | ||
1358 | static struct pernet_operations sctp_ctrlsock_ops = { | |
1359 | .init = sctp_ctrlsock_init, | |
1360 | .exit = sctp_ctrlsock_exit, | |
4db67e80 EB |
1361 | }; |
1362 | ||
1da177e4 | 1363 | /* Initialize the universe into something sensible. */ |
dda91928 | 1364 | static __init int sctp_init(void) |
1da177e4 LT |
1365 | { |
1366 | int i; | |
1367 | int status = -EINVAL; | |
1368 | unsigned long goal; | |
4d93df0a NH |
1369 | unsigned long limit; |
1370 | int max_share; | |
1da177e4 | 1371 | int order; |
d9749fb5 NH |
1372 | int num_entries; |
1373 | int max_entry_order; | |
1da177e4 | 1374 | |
b4772ef8 | 1375 | sock_skb_cb_check_size(sizeof(struct sctp_ulpevent)); |
1da177e4 | 1376 | |
827bf122 | 1377 | /* Allocate bind_bucket and chunk caches. */ |
1da177e4 LT |
1378 | status = -ENOBUFS; |
1379 | sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket", | |
1380 | sizeof(struct sctp_bind_bucket), | |
1381 | 0, SLAB_HWCACHE_ALIGN, | |
20c2df83 | 1382 | NULL); |
1da177e4 | 1383 | if (!sctp_bucket_cachep) |
827bf122 | 1384 | goto out; |
1da177e4 LT |
1385 | |
1386 | sctp_chunk_cachep = kmem_cache_create("sctp_chunk", | |
1387 | sizeof(struct sctp_chunk), | |
1388 | 0, SLAB_HWCACHE_ALIGN, | |
20c2df83 | 1389 | NULL); |
1da177e4 LT |
1390 | if (!sctp_chunk_cachep) |
1391 | goto err_chunk_cachep; | |
1392 | ||
908c7f19 | 1393 | status = percpu_counter_init(&sctp_sockets_allocated, 0, GFP_KERNEL); |
632c928a EB |
1394 | if (status) |
1395 | goto err_percpu_counter_init; | |
1396 | ||
1da177e4 LT |
1397 | /* Implementation specific variables. */ |
1398 | ||
1399 | /* Initialize default stream count setup information. */ | |
1400 | sctp_max_instreams = SCTP_DEFAULT_INSTREAMS; | |
1401 | sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS; | |
1402 | ||
1403 | /* Initialize handle used for association ids. */ | |
1404 | idr_init(&sctp_assocs_id); | |
1405 | ||
f03d78db | 1406 | limit = nr_free_buffer_pages() / 8; |
4d93df0a NH |
1407 | limit = max(limit, 128UL); |
1408 | sysctl_sctp_mem[0] = limit / 4 * 3; | |
1409 | sysctl_sctp_mem[1] = limit; | |
1410 | sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2; | |
1411 | ||
1412 | /* Set per-socket limits to no more than 1/128 the pressure threshold*/ | |
1413 | limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7); | |
1414 | max_share = min(4UL*1024*1024, limit); | |
1415 | ||
845525a6 | 1416 | sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */ |
87fb4b7b | 1417 | sysctl_sctp_rmem[1] = 1500 * SKB_TRUESIZE(1); |
4d93df0a NH |
1418 | sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share); |
1419 | ||
3ab224be | 1420 | sysctl_sctp_wmem[0] = SK_MEM_QUANTUM; |
4d93df0a NH |
1421 | sysctl_sctp_wmem[1] = 16*1024; |
1422 | sysctl_sctp_wmem[2] = max(64*1024, max_share); | |
1423 | ||
1da177e4 LT |
1424 | /* Size and allocate the association hash table. |
1425 | * The methodology is similar to that of the tcp hash tables. | |
d9749fb5 | 1426 | * Though not identical. Start by getting a goal size |
1da177e4 | 1427 | */ |
4481374c JB |
1428 | if (totalram_pages >= (128 * 1024)) |
1429 | goal = totalram_pages >> (22 - PAGE_SHIFT); | |
1da177e4 | 1430 | else |
4481374c | 1431 | goal = totalram_pages >> (24 - PAGE_SHIFT); |
1da177e4 | 1432 | |
d9749fb5 NH |
1433 | /* Then compute the page order for said goal */ |
1434 | order = get_order(goal); | |
1435 | ||
1436 | /* Now compute the required page order for the maximum sized table we | |
1437 | * want to create | |
1438 | */ | |
1439 | max_entry_order = get_order(MAX_SCTP_PORT_HASH_ENTRIES * | |
1440 | sizeof(struct sctp_bind_hashbucket)); | |
1441 | ||
1442 | /* Limit the page order by that maximum hash table size */ | |
1443 | order = min(order, max_entry_order); | |
1da177e4 | 1444 | |
1da177e4 LT |
1445 | /* Allocate and initialize the endpoint hash table. */ |
1446 | sctp_ep_hashsize = 64; | |
3b77d661 | 1447 | sctp_ep_hashtable = |
1da177e4 LT |
1448 | kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL); |
1449 | if (!sctp_ep_hashtable) { | |
145ce502 | 1450 | pr_err("Failed endpoint_hash alloc\n"); |
1da177e4 LT |
1451 | status = -ENOMEM; |
1452 | goto err_ehash_alloc; | |
1453 | } | |
1454 | for (i = 0; i < sctp_ep_hashsize; i++) { | |
1455 | rwlock_init(&sctp_ep_hashtable[i].lock); | |
d970dbf8 | 1456 | INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain); |
1da177e4 LT |
1457 | } |
1458 | ||
d9749fb5 NH |
1459 | /* Allocate and initialize the SCTP port hash table. |
1460 | * Note that order is initalized to start at the max sized | |
1461 | * table we want to support. If we can't get that many pages | |
1462 | * reduce the order and try again | |
1463 | */ | |
1da177e4 | 1464 | do { |
1da177e4 | 1465 | sctp_port_hashtable = (struct sctp_bind_hashbucket *) |
6857a02a | 1466 | __get_free_pages(GFP_KERNEL | __GFP_NOWARN, order); |
1da177e4 | 1467 | } while (!sctp_port_hashtable && --order > 0); |
d9749fb5 | 1468 | |
1da177e4 | 1469 | if (!sctp_port_hashtable) { |
145ce502 | 1470 | pr_err("Failed bind hash alloc\n"); |
1da177e4 LT |
1471 | status = -ENOMEM; |
1472 | goto err_bhash_alloc; | |
1473 | } | |
d9749fb5 NH |
1474 | |
1475 | /* Now compute the number of entries that will fit in the | |
1476 | * port hash space we allocated | |
1477 | */ | |
1478 | num_entries = (1UL << order) * PAGE_SIZE / | |
1479 | sizeof(struct sctp_bind_hashbucket); | |
1480 | ||
1481 | /* And finish by rounding it down to the nearest power of two | |
1482 | * this wastes some memory of course, but its needed because | |
1483 | * the hash function operates based on the assumption that | |
1484 | * that the number of entries is a power of two | |
1485 | */ | |
1486 | sctp_port_hashsize = rounddown_pow_of_two(num_entries); | |
1487 | ||
1da177e4 LT |
1488 | for (i = 0; i < sctp_port_hashsize; i++) { |
1489 | spin_lock_init(&sctp_port_hashtable[i].lock); | |
d970dbf8 | 1490 | INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain); |
1da177e4 LT |
1491 | } |
1492 | ||
a5e27d18 WY |
1493 | status = sctp_transport_hashtable_init(); |
1494 | if (status) | |
4f008781 XL |
1495 | goto err_thash_alloc; |
1496 | ||
d9749fb5 NH |
1497 | pr_info("Hash tables configured (bind %d/%d)\n", sctp_port_hashsize, |
1498 | num_entries); | |
1da177e4 | 1499 | |
1da177e4 LT |
1500 | sctp_sysctl_register(); |
1501 | ||
1502 | INIT_LIST_HEAD(&sctp_address_families); | |
270637ab VY |
1503 | sctp_v4_pf_init(); |
1504 | sctp_v6_pf_init(); | |
1ba896f6 | 1505 | sctp_sched_ops_init(); |
1da177e4 | 1506 | |
8e2d61e0 MRL |
1507 | status = register_pernet_subsys(&sctp_defaults_ops); |
1508 | if (status) | |
1509 | goto err_register_defaults; | |
827bf122 | 1510 | |
8e2d61e0 | 1511 | status = sctp_v4_protosw_init(); |
1da177e4 | 1512 | if (status) |
270637ab VY |
1513 | goto err_protosw_init; |
1514 | ||
1515 | status = sctp_v6_protosw_init(); | |
1516 | if (status) | |
1517 | goto err_v6_protosw_init; | |
1da177e4 | 1518 | |
8e2d61e0 | 1519 | status = register_pernet_subsys(&sctp_ctrlsock_ops); |
4db67e80 | 1520 | if (status) |
8e2d61e0 | 1521 | goto err_register_ctrlsock; |
4db67e80 | 1522 | |
270637ab VY |
1523 | status = sctp_v4_add_protocol(); |
1524 | if (status) | |
827bf122 | 1525 | goto err_add_protocol; |
827bf122 SS |
1526 | |
1527 | /* Register SCTP with inet6 layer. */ | |
1528 | status = sctp_v6_add_protocol(); | |
1529 | if (status) | |
1530 | goto err_v6_add_protocol; | |
1531 | ||
90017acc MRL |
1532 | if (sctp_offload_init() < 0) |
1533 | pr_crit("%s: Cannot add SCTP protocol offload\n", __func__); | |
1534 | ||
1da177e4 LT |
1535 | out: |
1536 | return status; | |
827bf122 | 1537 | err_v6_add_protocol: |
270637ab | 1538 | sctp_v4_del_protocol(); |
d1dd5247 | 1539 | err_add_protocol: |
8e2d61e0 MRL |
1540 | unregister_pernet_subsys(&sctp_ctrlsock_ops); |
1541 | err_register_ctrlsock: | |
270637ab VY |
1542 | sctp_v6_protosw_exit(); |
1543 | err_v6_protosw_init: | |
1544 | sctp_v4_protosw_exit(); | |
1545 | err_protosw_init: | |
8e2d61e0 MRL |
1546 | unregister_pernet_subsys(&sctp_defaults_ops); |
1547 | err_register_defaults: | |
270637ab VY |
1548 | sctp_v4_pf_exit(); |
1549 | sctp_v6_pf_exit(); | |
1da177e4 | 1550 | sctp_sysctl_unregister(); |
1da177e4 LT |
1551 | free_pages((unsigned long)sctp_port_hashtable, |
1552 | get_order(sctp_port_hashsize * | |
1553 | sizeof(struct sctp_bind_hashbucket))); | |
1554 | err_bhash_alloc: | |
4f008781 XL |
1555 | sctp_transport_hashtable_destroy(); |
1556 | err_thash_alloc: | |
1da177e4 LT |
1557 | kfree(sctp_ep_hashtable); |
1558 | err_ehash_alloc: | |
632c928a EB |
1559 | percpu_counter_destroy(&sctp_sockets_allocated); |
1560 | err_percpu_counter_init: | |
1da177e4 LT |
1561 | kmem_cache_destroy(sctp_chunk_cachep); |
1562 | err_chunk_cachep: | |
1563 | kmem_cache_destroy(sctp_bucket_cachep); | |
1da177e4 LT |
1564 | goto out; |
1565 | } | |
1566 | ||
1567 | /* Exit handler for the SCTP protocol. */ | |
dda91928 | 1568 | static __exit void sctp_exit(void) |
1da177e4 LT |
1569 | { |
1570 | /* BUG. This should probably do something useful like clean | |
1571 | * up all the remaining associations and all that memory. | |
1572 | */ | |
1573 | ||
827bf122 SS |
1574 | /* Unregister with inet6/inet layers. */ |
1575 | sctp_v6_del_protocol(); | |
270637ab | 1576 | sctp_v4_del_protocol(); |
1da177e4 | 1577 | |
8e2d61e0 | 1578 | unregister_pernet_subsys(&sctp_ctrlsock_ops); |
4db67e80 | 1579 | |
270637ab VY |
1580 | /* Free protosw registrations */ |
1581 | sctp_v6_protosw_exit(); | |
1582 | sctp_v4_protosw_exit(); | |
1583 | ||
8e2d61e0 MRL |
1584 | unregister_pernet_subsys(&sctp_defaults_ops); |
1585 | ||
827bf122 | 1586 | /* Unregister with socket layer. */ |
270637ab VY |
1587 | sctp_v6_pf_exit(); |
1588 | sctp_v4_pf_exit(); | |
827bf122 | 1589 | |
1da177e4 | 1590 | sctp_sysctl_unregister(); |
1da177e4 | 1591 | |
1da177e4 LT |
1592 | free_pages((unsigned long)sctp_port_hashtable, |
1593 | get_order(sctp_port_hashsize * | |
1594 | sizeof(struct sctp_bind_hashbucket))); | |
b5eff712 | 1595 | kfree(sctp_ep_hashtable); |
4f008781 | 1596 | sctp_transport_hashtable_destroy(); |
1da177e4 | 1597 | |
632c928a | 1598 | percpu_counter_destroy(&sctp_sockets_allocated); |
1da177e4 | 1599 | |
eaa184a1 JDB |
1600 | rcu_barrier(); /* Wait for completion of call_rcu()'s */ |
1601 | ||
827bf122 SS |
1602 | kmem_cache_destroy(sctp_chunk_cachep); |
1603 | kmem_cache_destroy(sctp_bucket_cachep); | |
1da177e4 LT |
1604 | } |
1605 | ||
1606 | module_init(sctp_init); | |
1607 | module_exit(sctp_exit); | |
1608 | ||
bb97d31f ACM |
1609 | /* |
1610 | * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly. | |
1611 | */ | |
1612 | MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132"); | |
882a382c | 1613 | MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132"); |
91705c61 | 1614 | MODULE_AUTHOR("Linux Kernel SCTP developers <linux-sctp@vger.kernel.org>"); |
1da177e4 | 1615 | MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)"); |
71acc0dd DM |
1616 | module_param_named(no_checksums, sctp_checksum_disable, bool, 0644); |
1617 | MODULE_PARM_DESC(no_checksums, "Disable checksums computing and verification"); | |
1da177e4 | 1618 | MODULE_LICENSE("GPL"); |