Root/
1 | /* SCTP kernel implementation |
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 | * |
9 | * This file is part of the SCTP kernel implementation |
10 | * |
11 | * Initialization/cleanup for SCTP protocol support. |
12 | * |
13 | * This SCTP implementation is free software; |
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 | * |
19 | * This SCTP implementation is distributed in the hope that it |
20 | * will be useful, but WITHOUT ANY WARRANTY; without even the implied |
21 | * ************************ |
22 | * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. |
23 | * See the GNU General Public License for more details. |
24 | * |
25 | * You should have received a copy of the GNU General Public License |
26 | * along with GNU CC; see the file COPYING. If not, write to |
27 | * the Free Software Foundation, 59 Temple Place - Suite 330, |
28 | * Boston, MA 02111-1307, USA. |
29 | * |
30 | * Please send any bug reports or fixes you make to the |
31 | * email address(es): |
32 | * lksctp developers <lksctp-developers@lists.sourceforge.net> |
33 | * |
34 | * Or submit a bug report through the following website: |
35 | * http://www.sf.net/projects/lksctp |
36 | * |
37 | * Written or modified by: |
38 | * La Monte H.P. Yarroll <piggy@acm.org> |
39 | * Karl Knutson <karl@athena.chicago.il.us> |
40 | * Jon Grimm <jgrimm@us.ibm.com> |
41 | * Sridhar Samudrala <sri@us.ibm.com> |
42 | * Daisy Chang <daisyc@us.ibm.com> |
43 | * Ardelle Fan <ardelle.fan@intel.com> |
44 | * |
45 | * Any bugs reported given to us we will try to fix... any fixes shared will |
46 | * be incorporated into the next SCTP release. |
47 | */ |
48 | |
49 | #include <linux/module.h> |
50 | #include <linux/init.h> |
51 | #include <linux/netdevice.h> |
52 | #include <linux/inetdevice.h> |
53 | #include <linux/seq_file.h> |
54 | #include <linux/bootmem.h> |
55 | #include <linux/highmem.h> |
56 | #include <linux/swap.h> |
57 | #include <net/net_namespace.h> |
58 | #include <net/protocol.h> |
59 | #include <net/ip.h> |
60 | #include <net/ipv6.h> |
61 | #include <net/route.h> |
62 | #include <net/sctp/sctp.h> |
63 | #include <net/addrconf.h> |
64 | #include <net/inet_common.h> |
65 | #include <net/inet_ecn.h> |
66 | |
67 | /* Global data structures. */ |
68 | struct sctp_globals sctp_globals __read_mostly; |
69 | DEFINE_SNMP_STAT(struct sctp_mib, sctp_statistics) __read_mostly; |
70 | |
71 | #ifdef CONFIG_PROC_FS |
72 | struct proc_dir_entry *proc_net_sctp; |
73 | #endif |
74 | |
75 | struct idr sctp_assocs_id; |
76 | DEFINE_SPINLOCK(sctp_assocs_id_lock); |
77 | |
78 | /* This is the global socket data structure used for responding to |
79 | * the Out-of-the-blue (OOTB) packets. A control sock will be created |
80 | * for this socket at the initialization time. |
81 | */ |
82 | static struct sock *sctp_ctl_sock; |
83 | |
84 | static struct sctp_pf *sctp_pf_inet6_specific; |
85 | static struct sctp_pf *sctp_pf_inet_specific; |
86 | static struct sctp_af *sctp_af_v4_specific; |
87 | static struct sctp_af *sctp_af_v6_specific; |
88 | |
89 | struct kmem_cache *sctp_chunk_cachep __read_mostly; |
90 | struct kmem_cache *sctp_bucket_cachep __read_mostly; |
91 | |
92 | int sysctl_sctp_mem[3]; |
93 | int sysctl_sctp_rmem[3]; |
94 | int sysctl_sctp_wmem[3]; |
95 | |
96 | /* Return the address of the control sock. */ |
97 | struct sock *sctp_get_ctl_sock(void) |
98 | { |
99 | return sctp_ctl_sock; |
100 | } |
101 | |
102 | /* Set up the proc fs entry for the SCTP protocol. */ |
103 | static __init int sctp_proc_init(void) |
104 | { |
105 | if (percpu_counter_init(&sctp_sockets_allocated, 0)) |
106 | goto out_nomem; |
107 | #ifdef CONFIG_PROC_FS |
108 | if (!proc_net_sctp) { |
109 | proc_net_sctp = proc_mkdir("sctp", init_net.proc_net); |
110 | if (!proc_net_sctp) |
111 | goto out_free_percpu; |
112 | } |
113 | |
114 | if (sctp_snmp_proc_init()) |
115 | goto out_snmp_proc_init; |
116 | if (sctp_eps_proc_init()) |
117 | goto out_eps_proc_init; |
118 | if (sctp_assocs_proc_init()) |
119 | goto out_assocs_proc_init; |
120 | if (sctp_remaddr_proc_init()) |
121 | goto out_remaddr_proc_init; |
122 | |
123 | return 0; |
124 | |
125 | out_remaddr_proc_init: |
126 | sctp_assocs_proc_exit(); |
127 | out_assocs_proc_init: |
128 | sctp_eps_proc_exit(); |
129 | out_eps_proc_init: |
130 | sctp_snmp_proc_exit(); |
131 | out_snmp_proc_init: |
132 | if (proc_net_sctp) { |
133 | proc_net_sctp = NULL; |
134 | remove_proc_entry("sctp", init_net.proc_net); |
135 | } |
136 | out_free_percpu: |
137 | percpu_counter_destroy(&sctp_sockets_allocated); |
138 | #else |
139 | return 0; |
140 | #endif /* CONFIG_PROC_FS */ |
141 | |
142 | out_nomem: |
143 | return -ENOMEM; |
144 | } |
145 | |
146 | /* Clean up the proc fs entry for the SCTP protocol. |
147 | * Note: Do not make this __exit as it is used in the init error |
148 | * path. |
149 | */ |
150 | static void sctp_proc_exit(void) |
151 | { |
152 | #ifdef CONFIG_PROC_FS |
153 | sctp_snmp_proc_exit(); |
154 | sctp_eps_proc_exit(); |
155 | sctp_assocs_proc_exit(); |
156 | sctp_remaddr_proc_exit(); |
157 | |
158 | if (proc_net_sctp) { |
159 | proc_net_sctp = NULL; |
160 | remove_proc_entry("sctp", init_net.proc_net); |
161 | } |
162 | #endif |
163 | percpu_counter_destroy(&sctp_sockets_allocated); |
164 | } |
165 | |
166 | /* Private helper to extract ipv4 address and stash them in |
167 | * the protocol structure. |
168 | */ |
169 | static void sctp_v4_copy_addrlist(struct list_head *addrlist, |
170 | struct net_device *dev) |
171 | { |
172 | struct in_device *in_dev; |
173 | struct in_ifaddr *ifa; |
174 | struct sctp_sockaddr_entry *addr; |
175 | |
176 | rcu_read_lock(); |
177 | if ((in_dev = __in_dev_get_rcu(dev)) == NULL) { |
178 | rcu_read_unlock(); |
179 | return; |
180 | } |
181 | |
182 | for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) { |
183 | /* Add the address to the local list. */ |
184 | addr = t_new(struct sctp_sockaddr_entry, GFP_ATOMIC); |
185 | if (addr) { |
186 | addr->a.v4.sin_family = AF_INET; |
187 | addr->a.v4.sin_port = 0; |
188 | addr->a.v4.sin_addr.s_addr = ifa->ifa_local; |
189 | addr->valid = 1; |
190 | INIT_LIST_HEAD(&addr->list); |
191 | INIT_RCU_HEAD(&addr->rcu); |
192 | list_add_tail(&addr->list, addrlist); |
193 | } |
194 | } |
195 | |
196 | rcu_read_unlock(); |
197 | } |
198 | |
199 | /* Extract our IP addresses from the system and stash them in the |
200 | * protocol structure. |
201 | */ |
202 | static void sctp_get_local_addr_list(void) |
203 | { |
204 | struct net_device *dev; |
205 | struct list_head *pos; |
206 | struct sctp_af *af; |
207 | |
208 | read_lock(&dev_base_lock); |
209 | for_each_netdev(&init_net, dev) { |
210 | __list_for_each(pos, &sctp_address_families) { |
211 | af = list_entry(pos, struct sctp_af, list); |
212 | af->copy_addrlist(&sctp_local_addr_list, dev); |
213 | } |
214 | } |
215 | read_unlock(&dev_base_lock); |
216 | } |
217 | |
218 | /* Free the existing local addresses. */ |
219 | static void sctp_free_local_addr_list(void) |
220 | { |
221 | struct sctp_sockaddr_entry *addr; |
222 | struct list_head *pos, *temp; |
223 | |
224 | list_for_each_safe(pos, temp, &sctp_local_addr_list) { |
225 | addr = list_entry(pos, struct sctp_sockaddr_entry, list); |
226 | list_del(pos); |
227 | kfree(addr); |
228 | } |
229 | } |
230 | |
231 | void sctp_local_addr_free(struct rcu_head *head) |
232 | { |
233 | struct sctp_sockaddr_entry *e = container_of(head, |
234 | struct sctp_sockaddr_entry, rcu); |
235 | kfree(e); |
236 | } |
237 | |
238 | /* Copy the local addresses which are valid for 'scope' into 'bp'. */ |
239 | int sctp_copy_local_addr_list(struct sctp_bind_addr *bp, sctp_scope_t scope, |
240 | gfp_t gfp, int copy_flags) |
241 | { |
242 | struct sctp_sockaddr_entry *addr; |
243 | int error = 0; |
244 | |
245 | rcu_read_lock(); |
246 | list_for_each_entry_rcu(addr, &sctp_local_addr_list, list) { |
247 | if (!addr->valid) |
248 | continue; |
249 | if (sctp_in_scope(&addr->a, scope)) { |
250 | /* Now that the address is in scope, check to see if |
251 | * the address type is really supported by the local |
252 | * sock as well as the remote peer. |
253 | */ |
254 | if ((((AF_INET == addr->a.sa.sa_family) && |
255 | (copy_flags & SCTP_ADDR4_PEERSUPP))) || |
256 | (((AF_INET6 == addr->a.sa.sa_family) && |
257 | (copy_flags & SCTP_ADDR6_ALLOWED) && |
258 | (copy_flags & SCTP_ADDR6_PEERSUPP)))) { |
259 | error = sctp_add_bind_addr(bp, &addr->a, |
260 | SCTP_ADDR_SRC, GFP_ATOMIC); |
261 | if (error) |
262 | goto end_copy; |
263 | } |
264 | } |
265 | } |
266 | |
267 | end_copy: |
268 | rcu_read_unlock(); |
269 | return error; |
270 | } |
271 | |
272 | /* Initialize a sctp_addr from in incoming skb. */ |
273 | static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb, |
274 | int is_saddr) |
275 | { |
276 | void *from; |
277 | __be16 *port; |
278 | struct sctphdr *sh; |
279 | |
280 | port = &addr->v4.sin_port; |
281 | addr->v4.sin_family = AF_INET; |
282 | |
283 | sh = sctp_hdr(skb); |
284 | if (is_saddr) { |
285 | *port = sh->source; |
286 | from = &ip_hdr(skb)->saddr; |
287 | } else { |
288 | *port = sh->dest; |
289 | from = &ip_hdr(skb)->daddr; |
290 | } |
291 | memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr)); |
292 | } |
293 | |
294 | /* Initialize an sctp_addr from a socket. */ |
295 | static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk) |
296 | { |
297 | addr->v4.sin_family = AF_INET; |
298 | addr->v4.sin_port = 0; |
299 | addr->v4.sin_addr.s_addr = inet_sk(sk)->rcv_saddr; |
300 | } |
301 | |
302 | /* Initialize sk->sk_rcv_saddr from sctp_addr. */ |
303 | static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk) |
304 | { |
305 | inet_sk(sk)->rcv_saddr = addr->v4.sin_addr.s_addr; |
306 | } |
307 | |
308 | /* Initialize sk->sk_daddr from sctp_addr. */ |
309 | static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk) |
310 | { |
311 | inet_sk(sk)->daddr = addr->v4.sin_addr.s_addr; |
312 | } |
313 | |
314 | /* Initialize a sctp_addr from an address parameter. */ |
315 | static void sctp_v4_from_addr_param(union sctp_addr *addr, |
316 | union sctp_addr_param *param, |
317 | __be16 port, int iif) |
318 | { |
319 | addr->v4.sin_family = AF_INET; |
320 | addr->v4.sin_port = port; |
321 | addr->v4.sin_addr.s_addr = param->v4.addr.s_addr; |
322 | } |
323 | |
324 | /* Initialize an address parameter from a sctp_addr and return the length |
325 | * of the address parameter. |
326 | */ |
327 | static int sctp_v4_to_addr_param(const union sctp_addr *addr, |
328 | union sctp_addr_param *param) |
329 | { |
330 | int length = sizeof(sctp_ipv4addr_param_t); |
331 | |
332 | param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS; |
333 | param->v4.param_hdr.length = htons(length); |
334 | param->v4.addr.s_addr = addr->v4.sin_addr.s_addr; |
335 | |
336 | return length; |
337 | } |
338 | |
339 | /* Initialize a sctp_addr from a dst_entry. */ |
340 | static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct dst_entry *dst, |
341 | __be16 port) |
342 | { |
343 | struct rtable *rt = (struct rtable *)dst; |
344 | saddr->v4.sin_family = AF_INET; |
345 | saddr->v4.sin_port = port; |
346 | saddr->v4.sin_addr.s_addr = rt->rt_src; |
347 | } |
348 | |
349 | /* Compare two addresses exactly. */ |
350 | static int sctp_v4_cmp_addr(const union sctp_addr *addr1, |
351 | const union sctp_addr *addr2) |
352 | { |
353 | if (addr1->sa.sa_family != addr2->sa.sa_family) |
354 | return 0; |
355 | if (addr1->v4.sin_port != addr2->v4.sin_port) |
356 | return 0; |
357 | if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr) |
358 | return 0; |
359 | |
360 | return 1; |
361 | } |
362 | |
363 | /* Initialize addr struct to INADDR_ANY. */ |
364 | static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port) |
365 | { |
366 | addr->v4.sin_family = AF_INET; |
367 | addr->v4.sin_addr.s_addr = htonl(INADDR_ANY); |
368 | addr->v4.sin_port = port; |
369 | } |
370 | |
371 | /* Is this a wildcard address? */ |
372 | static int sctp_v4_is_any(const union sctp_addr *addr) |
373 | { |
374 | return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr; |
375 | } |
376 | |
377 | /* This function checks if the address is a valid address to be used for |
378 | * SCTP binding. |
379 | * |
380 | * Output: |
381 | * Return 0 - If the address is a non-unicast or an illegal address. |
382 | * Return 1 - If the address is a unicast. |
383 | */ |
384 | static int sctp_v4_addr_valid(union sctp_addr *addr, |
385 | struct sctp_sock *sp, |
386 | const struct sk_buff *skb) |
387 | { |
388 | /* IPv4 addresses not allowed */ |
389 | if (sp && ipv6_only_sock(sctp_opt2sk(sp))) |
390 | return 0; |
391 | |
392 | /* Is this a non-unicast address or a unusable SCTP address? */ |
393 | if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) |
394 | return 0; |
395 | |
396 | /* Is this a broadcast address? */ |
397 | if (skb && skb_rtable(skb)->rt_flags & RTCF_BROADCAST) |
398 | return 0; |
399 | |
400 | return 1; |
401 | } |
402 | |
403 | /* Should this be available for binding? */ |
404 | static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp) |
405 | { |
406 | int ret = inet_addr_type(&init_net, addr->v4.sin_addr.s_addr); |
407 | |
408 | |
409 | if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) && |
410 | ret != RTN_LOCAL && |
411 | !sp->inet.freebind && |
412 | !sysctl_ip_nonlocal_bind) |
413 | return 0; |
414 | |
415 | if (ipv6_only_sock(sctp_opt2sk(sp))) |
416 | return 0; |
417 | |
418 | return 1; |
419 | } |
420 | |
421 | /* Checking the loopback, private and other address scopes as defined in |
422 | * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4 |
423 | * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>. |
424 | * |
425 | * Level 0 - unusable SCTP addresses |
426 | * Level 1 - loopback address |
427 | * Level 2 - link-local addresses |
428 | * Level 3 - private addresses. |
429 | * Level 4 - global addresses |
430 | * For INIT and INIT-ACK address list, let L be the level of |
431 | * of requested destination address, sender and receiver |
432 | * SHOULD include all of its addresses with level greater |
433 | * than or equal to L. |
434 | * |
435 | * IPv4 scoping can be controlled through sysctl option |
436 | * net.sctp.addr_scope_policy |
437 | */ |
438 | static sctp_scope_t sctp_v4_scope(union sctp_addr *addr) |
439 | { |
440 | sctp_scope_t retval; |
441 | |
442 | /* Check for unusable SCTP addresses. */ |
443 | if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) { |
444 | retval = SCTP_SCOPE_UNUSABLE; |
445 | } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) { |
446 | retval = SCTP_SCOPE_LOOPBACK; |
447 | } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) { |
448 | retval = SCTP_SCOPE_LINK; |
449 | } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) || |
450 | ipv4_is_private_172(addr->v4.sin_addr.s_addr) || |
451 | ipv4_is_private_192(addr->v4.sin_addr.s_addr)) { |
452 | retval = SCTP_SCOPE_PRIVATE; |
453 | } else { |
454 | retval = SCTP_SCOPE_GLOBAL; |
455 | } |
456 | |
457 | return retval; |
458 | } |
459 | |
460 | /* Returns a valid dst cache entry for the given source and destination ip |
461 | * addresses. If an association is passed, trys to get a dst entry with a |
462 | * source address that matches an address in the bind address list. |
463 | */ |
464 | static struct dst_entry *sctp_v4_get_dst(struct sctp_association *asoc, |
465 | union sctp_addr *daddr, |
466 | union sctp_addr *saddr) |
467 | { |
468 | struct rtable *rt; |
469 | struct flowi fl; |
470 | struct sctp_bind_addr *bp; |
471 | struct sctp_sockaddr_entry *laddr; |
472 | struct dst_entry *dst = NULL; |
473 | union sctp_addr dst_saddr; |
474 | |
475 | memset(&fl, 0x0, sizeof(struct flowi)); |
476 | fl.fl4_dst = daddr->v4.sin_addr.s_addr; |
477 | fl.proto = IPPROTO_SCTP; |
478 | if (asoc) { |
479 | fl.fl4_tos = RT_CONN_FLAGS(asoc->base.sk); |
480 | fl.oif = asoc->base.sk->sk_bound_dev_if; |
481 | } |
482 | if (saddr) |
483 | fl.fl4_src = saddr->v4.sin_addr.s_addr; |
484 | |
485 | SCTP_DEBUG_PRINTK("%s: DST:%pI4, SRC:%pI4 - ", |
486 | __func__, &fl.fl4_dst, &fl.fl4_src); |
487 | |
488 | if (!ip_route_output_key(&init_net, &rt, &fl)) { |
489 | dst = &rt->u.dst; |
490 | } |
491 | |
492 | /* If there is no association or if a source address is passed, no |
493 | * more validation is required. |
494 | */ |
495 | if (!asoc || saddr) |
496 | goto out; |
497 | |
498 | bp = &asoc->base.bind_addr; |
499 | |
500 | if (dst) { |
501 | /* Walk through the bind address list and look for a bind |
502 | * address that matches the source address of the returned dst. |
503 | */ |
504 | sctp_v4_dst_saddr(&dst_saddr, dst, htons(bp->port)); |
505 | rcu_read_lock(); |
506 | list_for_each_entry_rcu(laddr, &bp->address_list, list) { |
507 | if (!laddr->valid || (laddr->state != SCTP_ADDR_SRC)) |
508 | continue; |
509 | if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a)) |
510 | goto out_unlock; |
511 | } |
512 | rcu_read_unlock(); |
513 | |
514 | /* None of the bound addresses match the source address of the |
515 | * dst. So release it. |
516 | */ |
517 | dst_release(dst); |
518 | dst = NULL; |
519 | } |
520 | |
521 | /* Walk through the bind address list and try to get a dst that |
522 | * matches a bind address as the source address. |
523 | */ |
524 | rcu_read_lock(); |
525 | list_for_each_entry_rcu(laddr, &bp->address_list, list) { |
526 | if (!laddr->valid) |
527 | continue; |
528 | if ((laddr->state == SCTP_ADDR_SRC) && |
529 | (AF_INET == laddr->a.sa.sa_family)) { |
530 | fl.fl4_src = laddr->a.v4.sin_addr.s_addr; |
531 | if (!ip_route_output_key(&init_net, &rt, &fl)) { |
532 | dst = &rt->u.dst; |
533 | goto out_unlock; |
534 | } |
535 | } |
536 | } |
537 | |
538 | out_unlock: |
539 | rcu_read_unlock(); |
540 | out: |
541 | if (dst) |
542 | SCTP_DEBUG_PRINTK("rt_dst:%pI4, rt_src:%pI4\n", |
543 | &rt->rt_dst, &rt->rt_src); |
544 | else |
545 | SCTP_DEBUG_PRINTK("NO ROUTE\n"); |
546 | |
547 | return dst; |
548 | } |
549 | |
550 | /* For v4, the source address is cached in the route entry(dst). So no need |
551 | * to cache it separately and hence this is an empty routine. |
552 | */ |
553 | static void sctp_v4_get_saddr(struct sctp_sock *sk, |
554 | struct sctp_association *asoc, |
555 | struct dst_entry *dst, |
556 | union sctp_addr *daddr, |
557 | union sctp_addr *saddr) |
558 | { |
559 | struct rtable *rt = (struct rtable *)dst; |
560 | |
561 | if (!asoc) |
562 | return; |
563 | |
564 | if (rt) { |
565 | saddr->v4.sin_family = AF_INET; |
566 | saddr->v4.sin_port = htons(asoc->base.bind_addr.port); |
567 | saddr->v4.sin_addr.s_addr = rt->rt_src; |
568 | } |
569 | } |
570 | |
571 | /* What interface did this skb arrive on? */ |
572 | static int sctp_v4_skb_iif(const struct sk_buff *skb) |
573 | { |
574 | return skb_rtable(skb)->rt_iif; |
575 | } |
576 | |
577 | /* Was this packet marked by Explicit Congestion Notification? */ |
578 | static int sctp_v4_is_ce(const struct sk_buff *skb) |
579 | { |
580 | return INET_ECN_is_ce(ip_hdr(skb)->tos); |
581 | } |
582 | |
583 | /* Create and initialize a new sk for the socket returned by accept(). */ |
584 | static struct sock *sctp_v4_create_accept_sk(struct sock *sk, |
585 | struct sctp_association *asoc) |
586 | { |
587 | struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL, |
588 | sk->sk_prot); |
589 | struct inet_sock *newinet; |
590 | |
591 | if (!newsk) |
592 | goto out; |
593 | |
594 | sock_init_data(NULL, newsk); |
595 | |
596 | sctp_copy_sock(newsk, sk, asoc); |
597 | sock_reset_flag(newsk, SOCK_ZAPPED); |
598 | |
599 | newinet = inet_sk(newsk); |
600 | |
601 | newinet->daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr; |
602 | |
603 | sk_refcnt_debug_inc(newsk); |
604 | |
605 | if (newsk->sk_prot->init(newsk)) { |
606 | sk_common_release(newsk); |
607 | newsk = NULL; |
608 | } |
609 | |
610 | out: |
611 | return newsk; |
612 | } |
613 | |
614 | /* Map address, empty for v4 family */ |
615 | static void sctp_v4_addr_v4map(struct sctp_sock *sp, union sctp_addr *addr) |
616 | { |
617 | /* Empty */ |
618 | } |
619 | |
620 | /* Dump the v4 addr to the seq file. */ |
621 | static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr) |
622 | { |
623 | seq_printf(seq, "%pI4 ", &addr->v4.sin_addr); |
624 | } |
625 | |
626 | static void sctp_v4_ecn_capable(struct sock *sk) |
627 | { |
628 | INET_ECN_xmit(sk); |
629 | } |
630 | |
631 | /* Event handler for inet address addition/deletion events. |
632 | * The sctp_local_addr_list needs to be protocted by a spin lock since |
633 | * multiple notifiers (say IPv4 and IPv6) may be running at the same |
634 | * time and thus corrupt the list. |
635 | * The reader side is protected with RCU. |
636 | */ |
637 | static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev, |
638 | void *ptr) |
639 | { |
640 | struct in_ifaddr *ifa = (struct in_ifaddr *)ptr; |
641 | struct sctp_sockaddr_entry *addr = NULL; |
642 | struct sctp_sockaddr_entry *temp; |
643 | int found = 0; |
644 | |
645 | if (!net_eq(dev_net(ifa->ifa_dev->dev), &init_net)) |
646 | return NOTIFY_DONE; |
647 | |
648 | switch (ev) { |
649 | case NETDEV_UP: |
650 | addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC); |
651 | if (addr) { |
652 | addr->a.v4.sin_family = AF_INET; |
653 | addr->a.v4.sin_port = 0; |
654 | addr->a.v4.sin_addr.s_addr = ifa->ifa_local; |
655 | addr->valid = 1; |
656 | spin_lock_bh(&sctp_local_addr_lock); |
657 | list_add_tail_rcu(&addr->list, &sctp_local_addr_list); |
658 | spin_unlock_bh(&sctp_local_addr_lock); |
659 | } |
660 | break; |
661 | case NETDEV_DOWN: |
662 | spin_lock_bh(&sctp_local_addr_lock); |
663 | list_for_each_entry_safe(addr, temp, |
664 | &sctp_local_addr_list, list) { |
665 | if (addr->a.sa.sa_family == AF_INET && |
666 | addr->a.v4.sin_addr.s_addr == |
667 | ifa->ifa_local) { |
668 | found = 1; |
669 | addr->valid = 0; |
670 | list_del_rcu(&addr->list); |
671 | break; |
672 | } |
673 | } |
674 | spin_unlock_bh(&sctp_local_addr_lock); |
675 | if (found) |
676 | call_rcu(&addr->rcu, sctp_local_addr_free); |
677 | break; |
678 | } |
679 | |
680 | return NOTIFY_DONE; |
681 | } |
682 | |
683 | /* |
684 | * Initialize the control inode/socket with a control endpoint data |
685 | * structure. This endpoint is reserved exclusively for the OOTB processing. |
686 | */ |
687 | static int sctp_ctl_sock_init(void) |
688 | { |
689 | int err; |
690 | sa_family_t family = PF_INET; |
691 | |
692 | if (sctp_get_pf_specific(PF_INET6)) |
693 | family = PF_INET6; |
694 | |
695 | err = inet_ctl_sock_create(&sctp_ctl_sock, family, |
696 | SOCK_SEQPACKET, IPPROTO_SCTP, &init_net); |
697 | |
698 | /* If IPv6 socket could not be created, try the IPv4 socket */ |
699 | if (err < 0 && family == PF_INET6) |
700 | err = inet_ctl_sock_create(&sctp_ctl_sock, AF_INET, |
701 | SOCK_SEQPACKET, IPPROTO_SCTP, |
702 | &init_net); |
703 | |
704 | if (err < 0) { |
705 | printk(KERN_ERR |
706 | "SCTP: Failed to create the SCTP control socket.\n"); |
707 | return err; |
708 | } |
709 | return 0; |
710 | } |
711 | |
712 | /* Register address family specific functions. */ |
713 | int sctp_register_af(struct sctp_af *af) |
714 | { |
715 | switch (af->sa_family) { |
716 | case AF_INET: |
717 | if (sctp_af_v4_specific) |
718 | return 0; |
719 | sctp_af_v4_specific = af; |
720 | break; |
721 | case AF_INET6: |
722 | if (sctp_af_v6_specific) |
723 | return 0; |
724 | sctp_af_v6_specific = af; |
725 | break; |
726 | default: |
727 | return 0; |
728 | } |
729 | |
730 | INIT_LIST_HEAD(&af->list); |
731 | list_add_tail(&af->list, &sctp_address_families); |
732 | return 1; |
733 | } |
734 | |
735 | /* Get the table of functions for manipulating a particular address |
736 | * family. |
737 | */ |
738 | struct sctp_af *sctp_get_af_specific(sa_family_t family) |
739 | { |
740 | switch (family) { |
741 | case AF_INET: |
742 | return sctp_af_v4_specific; |
743 | case AF_INET6: |
744 | return sctp_af_v6_specific; |
745 | default: |
746 | return NULL; |
747 | } |
748 | } |
749 | |
750 | /* Common code to initialize a AF_INET msg_name. */ |
751 | static void sctp_inet_msgname(char *msgname, int *addr_len) |
752 | { |
753 | struct sockaddr_in *sin; |
754 | |
755 | sin = (struct sockaddr_in *)msgname; |
756 | *addr_len = sizeof(struct sockaddr_in); |
757 | sin->sin_family = AF_INET; |
758 | memset(sin->sin_zero, 0, sizeof(sin->sin_zero)); |
759 | } |
760 | |
761 | /* Copy the primary address of the peer primary address as the msg_name. */ |
762 | static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname, |
763 | int *addr_len) |
764 | { |
765 | struct sockaddr_in *sin, *sinfrom; |
766 | |
767 | if (msgname) { |
768 | struct sctp_association *asoc; |
769 | |
770 | asoc = event->asoc; |
771 | sctp_inet_msgname(msgname, addr_len); |
772 | sin = (struct sockaddr_in *)msgname; |
773 | sinfrom = &asoc->peer.primary_addr.v4; |
774 | sin->sin_port = htons(asoc->peer.port); |
775 | sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr; |
776 | } |
777 | } |
778 | |
779 | /* Initialize and copy out a msgname from an inbound skb. */ |
780 | static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len) |
781 | { |
782 | if (msgname) { |
783 | struct sctphdr *sh = sctp_hdr(skb); |
784 | struct sockaddr_in *sin = (struct sockaddr_in *)msgname; |
785 | |
786 | sctp_inet_msgname(msgname, len); |
787 | sin->sin_port = sh->source; |
788 | sin->sin_addr.s_addr = ip_hdr(skb)->saddr; |
789 | } |
790 | } |
791 | |
792 | /* Do we support this AF? */ |
793 | static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp) |
794 | { |
795 | /* PF_INET only supports AF_INET addresses. */ |
796 | return (AF_INET == family); |
797 | } |
798 | |
799 | /* Address matching with wildcards allowed. */ |
800 | static int sctp_inet_cmp_addr(const union sctp_addr *addr1, |
801 | const union sctp_addr *addr2, |
802 | struct sctp_sock *opt) |
803 | { |
804 | /* PF_INET only supports AF_INET addresses. */ |
805 | if (addr1->sa.sa_family != addr2->sa.sa_family) |
806 | return 0; |
807 | if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr || |
808 | htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr) |
809 | return 1; |
810 | if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr) |
811 | return 1; |
812 | |
813 | return 0; |
814 | } |
815 | |
816 | /* Verify that provided sockaddr looks bindable. Common verification has |
817 | * already been taken care of. |
818 | */ |
819 | static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr) |
820 | { |
821 | return sctp_v4_available(addr, opt); |
822 | } |
823 | |
824 | /* Verify that sockaddr looks sendable. Common verification has already |
825 | * been taken care of. |
826 | */ |
827 | static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr) |
828 | { |
829 | return 1; |
830 | } |
831 | |
832 | /* Fill in Supported Address Type information for INIT and INIT-ACK |
833 | * chunks. Returns number of addresses supported. |
834 | */ |
835 | static int sctp_inet_supported_addrs(const struct sctp_sock *opt, |
836 | __be16 *types) |
837 | { |
838 | types[0] = SCTP_PARAM_IPV4_ADDRESS; |
839 | return 1; |
840 | } |
841 | |
842 | /* Wrapper routine that calls the ip transmit routine. */ |
843 | static inline int sctp_v4_xmit(struct sk_buff *skb, |
844 | struct sctp_transport *transport) |
845 | { |
846 | struct inet_sock *inet = inet_sk(skb->sk); |
847 | |
848 | SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n", |
849 | __func__, skb, skb->len, |
850 | &skb_rtable(skb)->rt_src, |
851 | &skb_rtable(skb)->rt_dst); |
852 | |
853 | inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ? |
854 | IP_PMTUDISC_DO : IP_PMTUDISC_DONT; |
855 | |
856 | SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS); |
857 | return ip_queue_xmit(skb, 0); |
858 | } |
859 | |
860 | static struct sctp_af sctp_af_inet; |
861 | |
862 | static struct sctp_pf sctp_pf_inet = { |
863 | .event_msgname = sctp_inet_event_msgname, |
864 | .skb_msgname = sctp_inet_skb_msgname, |
865 | .af_supported = sctp_inet_af_supported, |
866 | .cmp_addr = sctp_inet_cmp_addr, |
867 | .bind_verify = sctp_inet_bind_verify, |
868 | .send_verify = sctp_inet_send_verify, |
869 | .supported_addrs = sctp_inet_supported_addrs, |
870 | .create_accept_sk = sctp_v4_create_accept_sk, |
871 | .addr_v4map = sctp_v4_addr_v4map, |
872 | .af = &sctp_af_inet |
873 | }; |
874 | |
875 | /* Notifier for inetaddr addition/deletion events. */ |
876 | static struct notifier_block sctp_inetaddr_notifier = { |
877 | .notifier_call = sctp_inetaddr_event, |
878 | }; |
879 | |
880 | /* Socket operations. */ |
881 | static const struct proto_ops inet_seqpacket_ops = { |
882 | .family = PF_INET, |
883 | .owner = THIS_MODULE, |
884 | .release = inet_release, /* Needs to be wrapped... */ |
885 | .bind = inet_bind, |
886 | .connect = inet_dgram_connect, |
887 | .socketpair = sock_no_socketpair, |
888 | .accept = inet_accept, |
889 | .getname = inet_getname, /* Semantics are different. */ |
890 | .poll = sctp_poll, |
891 | .ioctl = inet_ioctl, |
892 | .listen = sctp_inet_listen, |
893 | .shutdown = inet_shutdown, /* Looks harmless. */ |
894 | .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */ |
895 | .getsockopt = sock_common_getsockopt, |
896 | .sendmsg = inet_sendmsg, |
897 | .recvmsg = sock_common_recvmsg, |
898 | .mmap = sock_no_mmap, |
899 | .sendpage = sock_no_sendpage, |
900 | #ifdef CONFIG_COMPAT |
901 | .compat_setsockopt = compat_sock_common_setsockopt, |
902 | .compat_getsockopt = compat_sock_common_getsockopt, |
903 | #endif |
904 | }; |
905 | |
906 | /* Registration with AF_INET family. */ |
907 | static struct inet_protosw sctp_seqpacket_protosw = { |
908 | .type = SOCK_SEQPACKET, |
909 | .protocol = IPPROTO_SCTP, |
910 | .prot = &sctp_prot, |
911 | .ops = &inet_seqpacket_ops, |
912 | .capability = -1, |
913 | .no_check = 0, |
914 | .flags = SCTP_PROTOSW_FLAG |
915 | }; |
916 | static struct inet_protosw sctp_stream_protosw = { |
917 | .type = SOCK_STREAM, |
918 | .protocol = IPPROTO_SCTP, |
919 | .prot = &sctp_prot, |
920 | .ops = &inet_seqpacket_ops, |
921 | .capability = -1, |
922 | .no_check = 0, |
923 | .flags = SCTP_PROTOSW_FLAG |
924 | }; |
925 | |
926 | /* Register with IP layer. */ |
927 | static const struct net_protocol sctp_protocol = { |
928 | .handler = sctp_rcv, |
929 | .err_handler = sctp_v4_err, |
930 | .no_policy = 1, |
931 | }; |
932 | |
933 | /* IPv4 address related functions. */ |
934 | static struct sctp_af sctp_af_inet = { |
935 | .sa_family = AF_INET, |
936 | .sctp_xmit = sctp_v4_xmit, |
937 | .setsockopt = ip_setsockopt, |
938 | .getsockopt = ip_getsockopt, |
939 | .get_dst = sctp_v4_get_dst, |
940 | .get_saddr = sctp_v4_get_saddr, |
941 | .copy_addrlist = sctp_v4_copy_addrlist, |
942 | .from_skb = sctp_v4_from_skb, |
943 | .from_sk = sctp_v4_from_sk, |
944 | .to_sk_saddr = sctp_v4_to_sk_saddr, |
945 | .to_sk_daddr = sctp_v4_to_sk_daddr, |
946 | .from_addr_param = sctp_v4_from_addr_param, |
947 | .to_addr_param = sctp_v4_to_addr_param, |
948 | .dst_saddr = sctp_v4_dst_saddr, |
949 | .cmp_addr = sctp_v4_cmp_addr, |
950 | .addr_valid = sctp_v4_addr_valid, |
951 | .inaddr_any = sctp_v4_inaddr_any, |
952 | .is_any = sctp_v4_is_any, |
953 | .available = sctp_v4_available, |
954 | .scope = sctp_v4_scope, |
955 | .skb_iif = sctp_v4_skb_iif, |
956 | .is_ce = sctp_v4_is_ce, |
957 | .seq_dump_addr = sctp_v4_seq_dump_addr, |
958 | .ecn_capable = sctp_v4_ecn_capable, |
959 | .net_header_len = sizeof(struct iphdr), |
960 | .sockaddr_len = sizeof(struct sockaddr_in), |
961 | #ifdef CONFIG_COMPAT |
962 | .compat_setsockopt = compat_ip_setsockopt, |
963 | .compat_getsockopt = compat_ip_getsockopt, |
964 | #endif |
965 | }; |
966 | |
967 | struct sctp_pf *sctp_get_pf_specific(sa_family_t family) { |
968 | |
969 | switch (family) { |
970 | case PF_INET: |
971 | return sctp_pf_inet_specific; |
972 | case PF_INET6: |
973 | return sctp_pf_inet6_specific; |
974 | default: |
975 | return NULL; |
976 | } |
977 | } |
978 | |
979 | /* Register the PF specific function table. */ |
980 | int sctp_register_pf(struct sctp_pf *pf, sa_family_t family) |
981 | { |
982 | switch (family) { |
983 | case PF_INET: |
984 | if (sctp_pf_inet_specific) |
985 | return 0; |
986 | sctp_pf_inet_specific = pf; |
987 | break; |
988 | case PF_INET6: |
989 | if (sctp_pf_inet6_specific) |
990 | return 0; |
991 | sctp_pf_inet6_specific = pf; |
992 | break; |
993 | default: |
994 | return 0; |
995 | } |
996 | return 1; |
997 | } |
998 | |
999 | static inline int init_sctp_mibs(void) |
1000 | { |
1001 | return snmp_mib_init((void**)sctp_statistics, sizeof(struct sctp_mib)); |
1002 | } |
1003 | |
1004 | static inline void cleanup_sctp_mibs(void) |
1005 | { |
1006 | snmp_mib_free((void**)sctp_statistics); |
1007 | } |
1008 | |
1009 | static void sctp_v4_pf_init(void) |
1010 | { |
1011 | /* Initialize the SCTP specific PF functions. */ |
1012 | sctp_register_pf(&sctp_pf_inet, PF_INET); |
1013 | sctp_register_af(&sctp_af_inet); |
1014 | } |
1015 | |
1016 | static void sctp_v4_pf_exit(void) |
1017 | { |
1018 | list_del(&sctp_af_inet.list); |
1019 | } |
1020 | |
1021 | static int sctp_v4_protosw_init(void) |
1022 | { |
1023 | int rc; |
1024 | |
1025 | rc = proto_register(&sctp_prot, 1); |
1026 | if (rc) |
1027 | return rc; |
1028 | |
1029 | /* Register SCTP(UDP and TCP style) with socket layer. */ |
1030 | inet_register_protosw(&sctp_seqpacket_protosw); |
1031 | inet_register_protosw(&sctp_stream_protosw); |
1032 | |
1033 | return 0; |
1034 | } |
1035 | |
1036 | static void sctp_v4_protosw_exit(void) |
1037 | { |
1038 | inet_unregister_protosw(&sctp_stream_protosw); |
1039 | inet_unregister_protosw(&sctp_seqpacket_protosw); |
1040 | proto_unregister(&sctp_prot); |
1041 | } |
1042 | |
1043 | static int sctp_v4_add_protocol(void) |
1044 | { |
1045 | /* Register notifier for inet address additions/deletions. */ |
1046 | register_inetaddr_notifier(&sctp_inetaddr_notifier); |
1047 | |
1048 | /* Register SCTP with inet layer. */ |
1049 | if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0) |
1050 | return -EAGAIN; |
1051 | |
1052 | return 0; |
1053 | } |
1054 | |
1055 | static void sctp_v4_del_protocol(void) |
1056 | { |
1057 | inet_del_protocol(&sctp_protocol, IPPROTO_SCTP); |
1058 | unregister_inetaddr_notifier(&sctp_inetaddr_notifier); |
1059 | } |
1060 | |
1061 | /* Initialize the universe into something sensible. */ |
1062 | SCTP_STATIC __init int sctp_init(void) |
1063 | { |
1064 | int i; |
1065 | int status = -EINVAL; |
1066 | unsigned long goal; |
1067 | unsigned long limit; |
1068 | unsigned long nr_pages; |
1069 | int max_share; |
1070 | int order; |
1071 | |
1072 | /* SCTP_DEBUG sanity check. */ |
1073 | if (!sctp_sanity_check()) |
1074 | goto out; |
1075 | |
1076 | /* Allocate bind_bucket and chunk caches. */ |
1077 | status = -ENOBUFS; |
1078 | sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket", |
1079 | sizeof(struct sctp_bind_bucket), |
1080 | 0, SLAB_HWCACHE_ALIGN, |
1081 | NULL); |
1082 | if (!sctp_bucket_cachep) |
1083 | goto out; |
1084 | |
1085 | sctp_chunk_cachep = kmem_cache_create("sctp_chunk", |
1086 | sizeof(struct sctp_chunk), |
1087 | 0, SLAB_HWCACHE_ALIGN, |
1088 | NULL); |
1089 | if (!sctp_chunk_cachep) |
1090 | goto err_chunk_cachep; |
1091 | |
1092 | /* Allocate and initialise sctp mibs. */ |
1093 | status = init_sctp_mibs(); |
1094 | if (status) |
1095 | goto err_init_mibs; |
1096 | |
1097 | /* Initialize proc fs directory. */ |
1098 | status = sctp_proc_init(); |
1099 | if (status) |
1100 | goto err_init_proc; |
1101 | |
1102 | /* Initialize object count debugging. */ |
1103 | sctp_dbg_objcnt_init(); |
1104 | |
1105 | /* |
1106 | * 14. Suggested SCTP Protocol Parameter Values |
1107 | */ |
1108 | /* The following protocol parameters are RECOMMENDED: */ |
1109 | /* RTO.Initial - 3 seconds */ |
1110 | sctp_rto_initial = SCTP_RTO_INITIAL; |
1111 | /* RTO.Min - 1 second */ |
1112 | sctp_rto_min = SCTP_RTO_MIN; |
1113 | /* RTO.Max - 60 seconds */ |
1114 | sctp_rto_max = SCTP_RTO_MAX; |
1115 | /* RTO.Alpha - 1/8 */ |
1116 | sctp_rto_alpha = SCTP_RTO_ALPHA; |
1117 | /* RTO.Beta - 1/4 */ |
1118 | sctp_rto_beta = SCTP_RTO_BETA; |
1119 | |
1120 | /* Valid.Cookie.Life - 60 seconds */ |
1121 | sctp_valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE; |
1122 | |
1123 | /* Whether Cookie Preservative is enabled(1) or not(0) */ |
1124 | sctp_cookie_preserve_enable = 1; |
1125 | |
1126 | /* Max.Burst - 4 */ |
1127 | sctp_max_burst = SCTP_DEFAULT_MAX_BURST; |
1128 | |
1129 | /* Association.Max.Retrans - 10 attempts |
1130 | * Path.Max.Retrans - 5 attempts (per destination address) |
1131 | * Max.Init.Retransmits - 8 attempts |
1132 | */ |
1133 | sctp_max_retrans_association = 10; |
1134 | sctp_max_retrans_path = 5; |
1135 | sctp_max_retrans_init = 8; |
1136 | |
1137 | /* Sendbuffer growth - do per-socket accounting */ |
1138 | sctp_sndbuf_policy = 0; |
1139 | |
1140 | /* Rcvbuffer growth - do per-socket accounting */ |
1141 | sctp_rcvbuf_policy = 0; |
1142 | |
1143 | /* HB.interval - 30 seconds */ |
1144 | sctp_hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT; |
1145 | |
1146 | /* delayed SACK timeout */ |
1147 | sctp_sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK; |
1148 | |
1149 | /* Implementation specific variables. */ |
1150 | |
1151 | /* Initialize default stream count setup information. */ |
1152 | sctp_max_instreams = SCTP_DEFAULT_INSTREAMS; |
1153 | sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS; |
1154 | |
1155 | /* Initialize handle used for association ids. */ |
1156 | idr_init(&sctp_assocs_id); |
1157 | |
1158 | /* Set the pressure threshold to be a fraction of global memory that |
1159 | * is up to 1/2 at 256 MB, decreasing toward zero with the amount of |
1160 | * memory, with a floor of 128 pages. |
1161 | * Note this initalizes the data in sctpv6_prot too |
1162 | * Unabashedly stolen from tcp_init |
1163 | */ |
1164 | nr_pages = totalram_pages - totalhigh_pages; |
1165 | limit = min(nr_pages, 1UL<<(28-PAGE_SHIFT)) >> (20-PAGE_SHIFT); |
1166 | limit = (limit * (nr_pages >> (20-PAGE_SHIFT))) >> (PAGE_SHIFT-11); |
1167 | limit = max(limit, 128UL); |
1168 | sysctl_sctp_mem[0] = limit / 4 * 3; |
1169 | sysctl_sctp_mem[1] = limit; |
1170 | sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2; |
1171 | |
1172 | /* Set per-socket limits to no more than 1/128 the pressure threshold*/ |
1173 | limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7); |
1174 | max_share = min(4UL*1024*1024, limit); |
1175 | |
1176 | sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */ |
1177 | sysctl_sctp_rmem[1] = (1500 *(sizeof(struct sk_buff) + 1)); |
1178 | sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share); |
1179 | |
1180 | sysctl_sctp_wmem[0] = SK_MEM_QUANTUM; |
1181 | sysctl_sctp_wmem[1] = 16*1024; |
1182 | sysctl_sctp_wmem[2] = max(64*1024, max_share); |
1183 | |
1184 | /* Size and allocate the association hash table. |
1185 | * The methodology is similar to that of the tcp hash tables. |
1186 | */ |
1187 | if (totalram_pages >= (128 * 1024)) |
1188 | goal = totalram_pages >> (22 - PAGE_SHIFT); |
1189 | else |
1190 | goal = totalram_pages >> (24 - PAGE_SHIFT); |
1191 | |
1192 | for (order = 0; (1UL << order) < goal; order++) |
1193 | ; |
1194 | |
1195 | do { |
1196 | sctp_assoc_hashsize = (1UL << order) * PAGE_SIZE / |
1197 | sizeof(struct sctp_hashbucket); |
1198 | if ((sctp_assoc_hashsize > (64 * 1024)) && order > 0) |
1199 | continue; |
1200 | sctp_assoc_hashtable = (struct sctp_hashbucket *) |
1201 | __get_free_pages(GFP_ATOMIC, order); |
1202 | } while (!sctp_assoc_hashtable && --order > 0); |
1203 | if (!sctp_assoc_hashtable) { |
1204 | printk(KERN_ERR "SCTP: Failed association hash alloc.\n"); |
1205 | status = -ENOMEM; |
1206 | goto err_ahash_alloc; |
1207 | } |
1208 | for (i = 0; i < sctp_assoc_hashsize; i++) { |
1209 | rwlock_init(&sctp_assoc_hashtable[i].lock); |
1210 | INIT_HLIST_HEAD(&sctp_assoc_hashtable[i].chain); |
1211 | } |
1212 | |
1213 | /* Allocate and initialize the endpoint hash table. */ |
1214 | sctp_ep_hashsize = 64; |
1215 | sctp_ep_hashtable = (struct sctp_hashbucket *) |
1216 | kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL); |
1217 | if (!sctp_ep_hashtable) { |
1218 | printk(KERN_ERR "SCTP: Failed endpoint_hash alloc.\n"); |
1219 | status = -ENOMEM; |
1220 | goto err_ehash_alloc; |
1221 | } |
1222 | for (i = 0; i < sctp_ep_hashsize; i++) { |
1223 | rwlock_init(&sctp_ep_hashtable[i].lock); |
1224 | INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain); |
1225 | } |
1226 | |
1227 | /* Allocate and initialize the SCTP port hash table. */ |
1228 | do { |
1229 | sctp_port_hashsize = (1UL << order) * PAGE_SIZE / |
1230 | sizeof(struct sctp_bind_hashbucket); |
1231 | if ((sctp_port_hashsize > (64 * 1024)) && order > 0) |
1232 | continue; |
1233 | sctp_port_hashtable = (struct sctp_bind_hashbucket *) |
1234 | __get_free_pages(GFP_ATOMIC, order); |
1235 | } while (!sctp_port_hashtable && --order > 0); |
1236 | if (!sctp_port_hashtable) { |
1237 | printk(KERN_ERR "SCTP: Failed bind hash alloc."); |
1238 | status = -ENOMEM; |
1239 | goto err_bhash_alloc; |
1240 | } |
1241 | for (i = 0; i < sctp_port_hashsize; i++) { |
1242 | spin_lock_init(&sctp_port_hashtable[i].lock); |
1243 | INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain); |
1244 | } |
1245 | |
1246 | printk(KERN_INFO "SCTP: Hash tables configured " |
1247 | "(established %d bind %d)\n", |
1248 | sctp_assoc_hashsize, sctp_port_hashsize); |
1249 | |
1250 | /* Disable ADDIP by default. */ |
1251 | sctp_addip_enable = 0; |
1252 | sctp_addip_noauth = 0; |
1253 | |
1254 | /* Enable PR-SCTP by default. */ |
1255 | sctp_prsctp_enable = 1; |
1256 | |
1257 | /* Disable AUTH by default. */ |
1258 | sctp_auth_enable = 0; |
1259 | |
1260 | /* Set SCOPE policy to enabled */ |
1261 | sctp_scope_policy = SCTP_SCOPE_POLICY_ENABLE; |
1262 | |
1263 | sctp_sysctl_register(); |
1264 | |
1265 | INIT_LIST_HEAD(&sctp_address_families); |
1266 | sctp_v4_pf_init(); |
1267 | sctp_v6_pf_init(); |
1268 | |
1269 | /* Initialize the local address list. */ |
1270 | INIT_LIST_HEAD(&sctp_local_addr_list); |
1271 | spin_lock_init(&sctp_local_addr_lock); |
1272 | sctp_get_local_addr_list(); |
1273 | |
1274 | status = sctp_v4_protosw_init(); |
1275 | |
1276 | if (status) |
1277 | goto err_protosw_init; |
1278 | |
1279 | status = sctp_v6_protosw_init(); |
1280 | if (status) |
1281 | goto err_v6_protosw_init; |
1282 | |
1283 | /* Initialize the control inode/socket for handling OOTB packets. */ |
1284 | if ((status = sctp_ctl_sock_init())) { |
1285 | printk (KERN_ERR |
1286 | "SCTP: Failed to initialize the SCTP control sock.\n"); |
1287 | goto err_ctl_sock_init; |
1288 | } |
1289 | |
1290 | status = sctp_v4_add_protocol(); |
1291 | if (status) |
1292 | goto err_add_protocol; |
1293 | |
1294 | /* Register SCTP with inet6 layer. */ |
1295 | status = sctp_v6_add_protocol(); |
1296 | if (status) |
1297 | goto err_v6_add_protocol; |
1298 | |
1299 | status = 0; |
1300 | out: |
1301 | return status; |
1302 | err_v6_add_protocol: |
1303 | sctp_v4_del_protocol(); |
1304 | err_add_protocol: |
1305 | inet_ctl_sock_destroy(sctp_ctl_sock); |
1306 | err_ctl_sock_init: |
1307 | sctp_v6_protosw_exit(); |
1308 | err_v6_protosw_init: |
1309 | sctp_v4_protosw_exit(); |
1310 | err_protosw_init: |
1311 | sctp_free_local_addr_list(); |
1312 | sctp_v4_pf_exit(); |
1313 | sctp_v6_pf_exit(); |
1314 | sctp_sysctl_unregister(); |
1315 | free_pages((unsigned long)sctp_port_hashtable, |
1316 | get_order(sctp_port_hashsize * |
1317 | sizeof(struct sctp_bind_hashbucket))); |
1318 | err_bhash_alloc: |
1319 | kfree(sctp_ep_hashtable); |
1320 | err_ehash_alloc: |
1321 | free_pages((unsigned long)sctp_assoc_hashtable, |
1322 | get_order(sctp_assoc_hashsize * |
1323 | sizeof(struct sctp_hashbucket))); |
1324 | err_ahash_alloc: |
1325 | sctp_dbg_objcnt_exit(); |
1326 | sctp_proc_exit(); |
1327 | err_init_proc: |
1328 | cleanup_sctp_mibs(); |
1329 | err_init_mibs: |
1330 | kmem_cache_destroy(sctp_chunk_cachep); |
1331 | err_chunk_cachep: |
1332 | kmem_cache_destroy(sctp_bucket_cachep); |
1333 | goto out; |
1334 | } |
1335 | |
1336 | /* Exit handler for the SCTP protocol. */ |
1337 | SCTP_STATIC __exit void sctp_exit(void) |
1338 | { |
1339 | /* BUG. This should probably do something useful like clean |
1340 | * up all the remaining associations and all that memory. |
1341 | */ |
1342 | |
1343 | /* Unregister with inet6/inet layers. */ |
1344 | sctp_v6_del_protocol(); |
1345 | sctp_v4_del_protocol(); |
1346 | |
1347 | /* Free the control endpoint. */ |
1348 | inet_ctl_sock_destroy(sctp_ctl_sock); |
1349 | |
1350 | /* Free protosw registrations */ |
1351 | sctp_v6_protosw_exit(); |
1352 | sctp_v4_protosw_exit(); |
1353 | |
1354 | /* Free the local address list. */ |
1355 | sctp_free_local_addr_list(); |
1356 | |
1357 | /* Unregister with socket layer. */ |
1358 | sctp_v6_pf_exit(); |
1359 | sctp_v4_pf_exit(); |
1360 | |
1361 | sctp_sysctl_unregister(); |
1362 | |
1363 | free_pages((unsigned long)sctp_assoc_hashtable, |
1364 | get_order(sctp_assoc_hashsize * |
1365 | sizeof(struct sctp_hashbucket))); |
1366 | kfree(sctp_ep_hashtable); |
1367 | free_pages((unsigned long)sctp_port_hashtable, |
1368 | get_order(sctp_port_hashsize * |
1369 | sizeof(struct sctp_bind_hashbucket))); |
1370 | |
1371 | sctp_dbg_objcnt_exit(); |
1372 | sctp_proc_exit(); |
1373 | cleanup_sctp_mibs(); |
1374 | |
1375 | rcu_barrier(); /* Wait for completion of call_rcu()'s */ |
1376 | |
1377 | kmem_cache_destroy(sctp_chunk_cachep); |
1378 | kmem_cache_destroy(sctp_bucket_cachep); |
1379 | } |
1380 | |
1381 | module_init(sctp_init); |
1382 | module_exit(sctp_exit); |
1383 | |
1384 | /* |
1385 | * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly. |
1386 | */ |
1387 | MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132"); |
1388 | MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132"); |
1389 | MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>"); |
1390 | MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)"); |
1391 | module_param_named(no_checksums, sctp_checksum_disable, bool, 0644); |
1392 | MODULE_PARM_DESC(no_checksums, "Disable checksums computing and verification"); |
1393 | MODULE_LICENSE("GPL"); |
1394 |
Branches:
ben-wpan
ben-wpan-stefan
javiroman/ks7010
jz-2.6.34
jz-2.6.34-rc5
jz-2.6.34-rc6
jz-2.6.34-rc7
jz-2.6.35
jz-2.6.36
jz-2.6.37
jz-2.6.38
jz-2.6.39
jz-3.0
jz-3.1
jz-3.11
jz-3.12
jz-3.13
jz-3.15
jz-3.16
jz-3.18-dt
jz-3.2
jz-3.3
jz-3.4
jz-3.5
jz-3.6
jz-3.6-rc2-pwm
jz-3.9
jz-3.9-clk
jz-3.9-rc8
jz47xx
jz47xx-2.6.38
master
Tags:
od-2011-09-04
od-2011-09-18
v2.6.34-rc5
v2.6.34-rc6
v2.6.34-rc7
v3.9