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
9 * This file is part of the SCTP kernel implementation
11 * Initialization/cleanup for SCTP protocol support.
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)
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.
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.
30 * Please send any bug reports or fixes you make to the
32 * lksctp developers <lksctp-developers@lists.sourceforge.net>
34 * Or submit a bug report through the following website:
35 * http://www.sf.net/projects/lksctp
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>
45 * Any bugs reported given to us we will try to fix... any fixes shared will
46 * be incorporated into the next SCTP release.
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 <net/net_namespace.h>
56 #include <net/protocol.h>
59 #include <net/route.h>
60 #include <net/sctp/sctp.h>
61 #include <net/addrconf.h>
62 #include <net/inet_common.h>
63 #include <net/inet_ecn.h>
65 /* Global data structures. */
66 struct sctp_globals sctp_globals __read_mostly;
67 DEFINE_SNMP_STAT(struct sctp_mib, sctp_statistics) __read_mostly;
70 struct proc_dir_entry *proc_net_sctp;
73 struct idr sctp_assocs_id;
74 DEFINE_SPINLOCK(sctp_assocs_id_lock);
76 /* This is the global socket data structure used for responding to
77 * the Out-of-the-blue (OOTB) packets. A control sock will be created
78 * for this socket at the initialization time.
80 static struct sock *sctp_ctl_sock;
82 static struct sctp_pf *sctp_pf_inet6_specific;
83 static struct sctp_pf *sctp_pf_inet_specific;
84 static struct sctp_af *sctp_af_v4_specific;
85 static struct sctp_af *sctp_af_v6_specific;
87 struct kmem_cache *sctp_chunk_cachep __read_mostly;
88 struct kmem_cache *sctp_bucket_cachep __read_mostly;
90 int sysctl_sctp_mem[3];
91 int sysctl_sctp_rmem[3];
92 int sysctl_sctp_wmem[3];
94 /* Return the address of the control sock. */
95 struct sock *sctp_get_ctl_sock(void)
100 /* Set up the proc fs entry for the SCTP protocol. */
101 static __init int sctp_proc_init(void)
103 #ifdef CONFIG_PROC_FS
104 if (!proc_net_sctp) {
105 struct proc_dir_entry *ent;
106 ent = proc_mkdir("sctp", init_net.proc_net);
108 ent->owner = THIS_MODULE;
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;
125 out_remaddr_proc_init:
126 sctp_assocs_proc_exit();
127 out_assocs_proc_init:
128 sctp_eps_proc_exit();
130 sctp_snmp_proc_exit();
133 proc_net_sctp = NULL;
134 remove_proc_entry("sctp", init_net.proc_net);
140 #endif /* CONFIG_PROC_FS */
143 /* Clean up the proc fs entry for the SCTP protocol.
144 * Note: Do not make this __exit as it is used in the init error
147 static void sctp_proc_exit(void)
149 #ifdef CONFIG_PROC_FS
150 sctp_snmp_proc_exit();
151 sctp_eps_proc_exit();
152 sctp_assocs_proc_exit();
153 sctp_remaddr_proc_exit();
156 proc_net_sctp = NULL;
157 remove_proc_entry("sctp", init_net.proc_net);
162 /* Private helper to extract ipv4 address and stash them in
163 * the protocol structure.
165 static void sctp_v4_copy_addrlist(struct list_head *addrlist,
166 struct net_device *dev)
168 struct in_device *in_dev;
169 struct in_ifaddr *ifa;
170 struct sctp_sockaddr_entry *addr;
173 if ((in_dev = __in_dev_get_rcu(dev)) == NULL) {
178 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
179 /* Add the address to the local list. */
180 addr = t_new(struct sctp_sockaddr_entry, GFP_ATOMIC);
182 addr->a.v4.sin_family = AF_INET;
183 addr->a.v4.sin_port = 0;
184 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
186 INIT_LIST_HEAD(&addr->list);
187 INIT_RCU_HEAD(&addr->rcu);
188 list_add_tail(&addr->list, addrlist);
195 /* Extract our IP addresses from the system and stash them in the
196 * protocol structure.
198 static void sctp_get_local_addr_list(void)
200 struct net_device *dev;
201 struct list_head *pos;
204 read_lock(&dev_base_lock);
205 for_each_netdev(&init_net, dev) {
206 __list_for_each(pos, &sctp_address_families) {
207 af = list_entry(pos, struct sctp_af, list);
208 af->copy_addrlist(&sctp_local_addr_list, dev);
211 read_unlock(&dev_base_lock);
214 /* Free the existing local addresses. */
215 static void sctp_free_local_addr_list(void)
217 struct sctp_sockaddr_entry *addr;
218 struct list_head *pos, *temp;
220 list_for_each_safe(pos, temp, &sctp_local_addr_list) {
221 addr = list_entry(pos, struct sctp_sockaddr_entry, list);
227 void sctp_local_addr_free(struct rcu_head *head)
229 struct sctp_sockaddr_entry *e = container_of(head,
230 struct sctp_sockaddr_entry, rcu);
234 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
235 int sctp_copy_local_addr_list(struct sctp_bind_addr *bp, sctp_scope_t scope,
236 gfp_t gfp, int copy_flags)
238 struct sctp_sockaddr_entry *addr;
242 list_for_each_entry_rcu(addr, &sctp_local_addr_list, list) {
245 if (sctp_in_scope(&addr->a, scope)) {
246 /* Now that the address is in scope, check to see if
247 * the address type is really supported by the local
248 * sock as well as the remote peer.
250 if ((((AF_INET == addr->a.sa.sa_family) &&
251 (copy_flags & SCTP_ADDR4_PEERSUPP))) ||
252 (((AF_INET6 == addr->a.sa.sa_family) &&
253 (copy_flags & SCTP_ADDR6_ALLOWED) &&
254 (copy_flags & SCTP_ADDR6_PEERSUPP)))) {
255 error = sctp_add_bind_addr(bp, &addr->a,
256 SCTP_ADDR_SRC, GFP_ATOMIC);
268 /* Initialize a sctp_addr from in incoming skb. */
269 static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
276 port = &addr->v4.sin_port;
277 addr->v4.sin_family = AF_INET;
282 from = &ip_hdr(skb)->saddr;
285 from = &ip_hdr(skb)->daddr;
287 memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr));
290 /* Initialize an sctp_addr from a socket. */
291 static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk)
293 addr->v4.sin_family = AF_INET;
294 addr->v4.sin_port = 0;
295 addr->v4.sin_addr.s_addr = inet_sk(sk)->rcv_saddr;
298 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
299 static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk)
301 inet_sk(sk)->rcv_saddr = addr->v4.sin_addr.s_addr;
304 /* Initialize sk->sk_daddr from sctp_addr. */
305 static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk)
307 inet_sk(sk)->daddr = addr->v4.sin_addr.s_addr;
310 /* Initialize a sctp_addr from an address parameter. */
311 static void sctp_v4_from_addr_param(union sctp_addr *addr,
312 union sctp_addr_param *param,
313 __be16 port, int iif)
315 addr->v4.sin_family = AF_INET;
316 addr->v4.sin_port = port;
317 addr->v4.sin_addr.s_addr = param->v4.addr.s_addr;
320 /* Initialize an address parameter from a sctp_addr and return the length
321 * of the address parameter.
323 static int sctp_v4_to_addr_param(const union sctp_addr *addr,
324 union sctp_addr_param *param)
326 int length = sizeof(sctp_ipv4addr_param_t);
328 param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS;
329 param->v4.param_hdr.length = htons(length);
330 param->v4.addr.s_addr = addr->v4.sin_addr.s_addr;
335 /* Initialize a sctp_addr from a dst_entry. */
336 static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct dst_entry *dst,
339 struct rtable *rt = (struct rtable *)dst;
340 saddr->v4.sin_family = AF_INET;
341 saddr->v4.sin_port = port;
342 saddr->v4.sin_addr.s_addr = rt->rt_src;
345 /* Compare two addresses exactly. */
346 static int sctp_v4_cmp_addr(const union sctp_addr *addr1,
347 const union sctp_addr *addr2)
349 if (addr1->sa.sa_family != addr2->sa.sa_family)
351 if (addr1->v4.sin_port != addr2->v4.sin_port)
353 if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr)
359 /* Initialize addr struct to INADDR_ANY. */
360 static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port)
362 addr->v4.sin_family = AF_INET;
363 addr->v4.sin_addr.s_addr = htonl(INADDR_ANY);
364 addr->v4.sin_port = port;
367 /* Is this a wildcard address? */
368 static int sctp_v4_is_any(const union sctp_addr *addr)
370 return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr;
373 /* This function checks if the address is a valid address to be used for
377 * Return 0 - If the address is a non-unicast or an illegal address.
378 * Return 1 - If the address is a unicast.
380 static int sctp_v4_addr_valid(union sctp_addr *addr,
381 struct sctp_sock *sp,
382 const struct sk_buff *skb)
384 /* IPv4 addresses not allowed */
385 if (sp && ipv6_only_sock(sctp_opt2sk(sp)))
388 /* Is this a non-unicast address or a unusable SCTP address? */
389 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr))
392 /* Is this a broadcast address? */
393 if (skb && skb->rtable->rt_flags & RTCF_BROADCAST)
399 /* Should this be available for binding? */
400 static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp)
402 int ret = inet_addr_type(&init_net, addr->v4.sin_addr.s_addr);
405 if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) &&
407 !sp->inet.freebind &&
408 !sysctl_ip_nonlocal_bind)
411 if (ipv6_only_sock(sctp_opt2sk(sp)))
417 /* Checking the loopback, private and other address scopes as defined in
418 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
419 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
421 * Level 0 - unusable SCTP addresses
422 * Level 1 - loopback address
423 * Level 2 - link-local addresses
424 * Level 3 - private addresses.
425 * Level 4 - global addresses
426 * For INIT and INIT-ACK address list, let L be the level of
427 * of requested destination address, sender and receiver
428 * SHOULD include all of its addresses with level greater
429 * than or equal to L.
431 static sctp_scope_t sctp_v4_scope(union sctp_addr *addr)
435 /* Should IPv4 scoping be a sysctl configurable option
436 * so users can turn it off (default on) for certain
437 * unconventional networking environments?
440 /* Check for unusable SCTP addresses. */
441 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) {
442 retval = SCTP_SCOPE_UNUSABLE;
443 } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) {
444 retval = SCTP_SCOPE_LOOPBACK;
445 } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) {
446 retval = SCTP_SCOPE_LINK;
447 } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) ||
448 ipv4_is_private_172(addr->v4.sin_addr.s_addr) ||
449 ipv4_is_private_192(addr->v4.sin_addr.s_addr)) {
450 retval = SCTP_SCOPE_PRIVATE;
452 retval = SCTP_SCOPE_GLOBAL;
458 /* Returns a valid dst cache entry for the given source and destination ip
459 * addresses. If an association is passed, trys to get a dst entry with a
460 * source address that matches an address in the bind address list.
462 static struct dst_entry *sctp_v4_get_dst(struct sctp_association *asoc,
463 union sctp_addr *daddr,
464 union sctp_addr *saddr)
468 struct sctp_bind_addr *bp;
469 struct sctp_sockaddr_entry *laddr;
470 struct dst_entry *dst = NULL;
471 union sctp_addr dst_saddr;
473 memset(&fl, 0x0, sizeof(struct flowi));
474 fl.fl4_dst = daddr->v4.sin_addr.s_addr;
475 fl.proto = IPPROTO_SCTP;
477 fl.fl4_tos = RT_CONN_FLAGS(asoc->base.sk);
478 fl.oif = asoc->base.sk->sk_bound_dev_if;
481 fl.fl4_src = saddr->v4.sin_addr.s_addr;
483 SCTP_DEBUG_PRINTK("%s: DST:%u.%u.%u.%u, SRC:%u.%u.%u.%u - ",
484 __func__, NIPQUAD(fl.fl4_dst),
485 NIPQUAD(fl.fl4_src));
487 if (!ip_route_output_key(&init_net, &rt, &fl)) {
491 /* If there is no association or if a source address is passed, no
492 * more validation is required.
497 bp = &asoc->base.bind_addr;
500 /* Walk through the bind address list and look for a bind
501 * address that matches the source address of the returned dst.
503 sctp_v4_dst_saddr(&dst_saddr, dst, htons(bp->port));
505 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
506 if (!laddr->valid || (laddr->state != SCTP_ADDR_SRC))
508 if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a))
513 /* None of the bound addresses match the source address of the
514 * dst. So release it.
520 /* Walk through the bind address list and try to get a dst that
521 * matches a bind address as the source address.
524 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
527 if ((laddr->state == SCTP_ADDR_SRC) &&
528 (AF_INET == laddr->a.sa.sa_family)) {
529 fl.fl4_src = laddr->a.v4.sin_addr.s_addr;
530 if (!ip_route_output_key(&init_net, &rt, &fl)) {
541 SCTP_DEBUG_PRINTK("rt_dst:%u.%u.%u.%u, rt_src:%u.%u.%u.%u\n",
542 NIPQUAD(rt->rt_dst), NIPQUAD(rt->rt_src));
544 SCTP_DEBUG_PRINTK("NO ROUTE\n");
549 /* For v4, the source address is cached in the route entry(dst). So no need
550 * to cache it separately and hence this is an empty routine.
552 static void sctp_v4_get_saddr(struct sctp_sock *sk,
553 struct sctp_association *asoc,
554 struct dst_entry *dst,
555 union sctp_addr *daddr,
556 union sctp_addr *saddr)
558 struct rtable *rt = (struct rtable *)dst;
564 saddr->v4.sin_family = AF_INET;
565 saddr->v4.sin_port = htons(asoc->base.bind_addr.port);
566 saddr->v4.sin_addr.s_addr = rt->rt_src;
570 /* What interface did this skb arrive on? */
571 static int sctp_v4_skb_iif(const struct sk_buff *skb)
573 return skb->rtable->rt_iif;
576 /* Was this packet marked by Explicit Congestion Notification? */
577 static int sctp_v4_is_ce(const struct sk_buff *skb)
579 return INET_ECN_is_ce(ip_hdr(skb)->tos);
582 /* Create and initialize a new sk for the socket returned by accept(). */
583 static struct sock *sctp_v4_create_accept_sk(struct sock *sk,
584 struct sctp_association *asoc)
586 struct inet_sock *inet = inet_sk(sk);
587 struct inet_sock *newinet;
588 struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL,
594 sock_init_data(NULL, newsk);
596 newsk->sk_type = SOCK_STREAM;
598 newsk->sk_no_check = sk->sk_no_check;
599 newsk->sk_reuse = sk->sk_reuse;
600 newsk->sk_shutdown = sk->sk_shutdown;
602 newsk->sk_destruct = inet_sock_destruct;
603 newsk->sk_family = PF_INET;
604 newsk->sk_protocol = IPPROTO_SCTP;
605 newsk->sk_backlog_rcv = sk->sk_prot->backlog_rcv;
606 sock_reset_flag(newsk, SOCK_ZAPPED);
608 newinet = inet_sk(newsk);
610 /* Initialize sk's sport, dport, rcv_saddr and daddr for
611 * getsockname() and getpeername()
613 newinet->sport = inet->sport;
614 newinet->saddr = inet->saddr;
615 newinet->rcv_saddr = inet->rcv_saddr;
616 newinet->dport = htons(asoc->peer.port);
617 newinet->daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr;
618 newinet->pmtudisc = inet->pmtudisc;
619 newinet->id = asoc->next_tsn ^ jiffies;
621 newinet->uc_ttl = -1;
622 newinet->mc_loop = 1;
624 newinet->mc_index = 0;
625 newinet->mc_list = NULL;
627 sk_refcnt_debug_inc(newsk);
629 if (newsk->sk_prot->init(newsk)) {
630 sk_common_release(newsk);
638 /* Map address, empty for v4 family */
639 static void sctp_v4_addr_v4map(struct sctp_sock *sp, union sctp_addr *addr)
644 /* Dump the v4 addr to the seq file. */
645 static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr)
647 seq_printf(seq, "%d.%d.%d.%d ", NIPQUAD(addr->v4.sin_addr));
650 static void sctp_v4_ecn_capable(struct sock *sk)
655 /* Event handler for inet address addition/deletion events.
656 * The sctp_local_addr_list needs to be protocted by a spin lock since
657 * multiple notifiers (say IPv4 and IPv6) may be running at the same
658 * time and thus corrupt the list.
659 * The reader side is protected with RCU.
661 static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev,
664 struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
665 struct sctp_sockaddr_entry *addr = NULL;
666 struct sctp_sockaddr_entry *temp;
669 if (dev_net(ifa->ifa_dev->dev) != &init_net)
674 addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
676 addr->a.v4.sin_family = AF_INET;
677 addr->a.v4.sin_port = 0;
678 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
680 spin_lock_bh(&sctp_local_addr_lock);
681 list_add_tail_rcu(&addr->list, &sctp_local_addr_list);
682 spin_unlock_bh(&sctp_local_addr_lock);
686 spin_lock_bh(&sctp_local_addr_lock);
687 list_for_each_entry_safe(addr, temp,
688 &sctp_local_addr_list, list) {
689 if (addr->a.sa.sa_family == AF_INET &&
690 addr->a.v4.sin_addr.s_addr ==
694 list_del_rcu(&addr->list);
698 spin_unlock_bh(&sctp_local_addr_lock);
700 call_rcu(&addr->rcu, sctp_local_addr_free);
708 * Initialize the control inode/socket with a control endpoint data
709 * structure. This endpoint is reserved exclusively for the OOTB processing.
711 static int sctp_ctl_sock_init(void)
716 if (sctp_get_pf_specific(PF_INET6))
721 err = inet_ctl_sock_create(&sctp_ctl_sock, family,
722 SOCK_SEQPACKET, IPPROTO_SCTP, &init_net);
725 "SCTP: Failed to create the SCTP control socket.\n");
731 /* Register address family specific functions. */
732 int sctp_register_af(struct sctp_af *af)
734 switch (af->sa_family) {
736 if (sctp_af_v4_specific)
738 sctp_af_v4_specific = af;
741 if (sctp_af_v6_specific)
743 sctp_af_v6_specific = af;
749 INIT_LIST_HEAD(&af->list);
750 list_add_tail(&af->list, &sctp_address_families);
754 /* Get the table of functions for manipulating a particular address
757 struct sctp_af *sctp_get_af_specific(sa_family_t family)
761 return sctp_af_v4_specific;
763 return sctp_af_v6_specific;
769 /* Common code to initialize a AF_INET msg_name. */
770 static void sctp_inet_msgname(char *msgname, int *addr_len)
772 struct sockaddr_in *sin;
774 sin = (struct sockaddr_in *)msgname;
775 *addr_len = sizeof(struct sockaddr_in);
776 sin->sin_family = AF_INET;
777 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
780 /* Copy the primary address of the peer primary address as the msg_name. */
781 static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname,
784 struct sockaddr_in *sin, *sinfrom;
787 struct sctp_association *asoc;
790 sctp_inet_msgname(msgname, addr_len);
791 sin = (struct sockaddr_in *)msgname;
792 sinfrom = &asoc->peer.primary_addr.v4;
793 sin->sin_port = htons(asoc->peer.port);
794 sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr;
798 /* Initialize and copy out a msgname from an inbound skb. */
799 static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len)
802 struct sctphdr *sh = sctp_hdr(skb);
803 struct sockaddr_in *sin = (struct sockaddr_in *)msgname;
805 sctp_inet_msgname(msgname, len);
806 sin->sin_port = sh->source;
807 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
811 /* Do we support this AF? */
812 static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp)
814 /* PF_INET only supports AF_INET addresses. */
815 return (AF_INET == family);
818 /* Address matching with wildcards allowed. */
819 static int sctp_inet_cmp_addr(const union sctp_addr *addr1,
820 const union sctp_addr *addr2,
821 struct sctp_sock *opt)
823 /* PF_INET only supports AF_INET addresses. */
824 if (addr1->sa.sa_family != addr2->sa.sa_family)
826 if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr ||
827 htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr)
829 if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr)
835 /* Verify that provided sockaddr looks bindable. Common verification has
836 * already been taken care of.
838 static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr)
840 return sctp_v4_available(addr, opt);
843 /* Verify that sockaddr looks sendable. Common verification has already
844 * been taken care of.
846 static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr)
851 /* Fill in Supported Address Type information for INIT and INIT-ACK
852 * chunks. Returns number of addresses supported.
854 static int sctp_inet_supported_addrs(const struct sctp_sock *opt,
857 types[0] = SCTP_PARAM_IPV4_ADDRESS;
861 /* Wrapper routine that calls the ip transmit routine. */
862 static inline int sctp_v4_xmit(struct sk_buff *skb,
863 struct sctp_transport *transport, int ipfragok)
865 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, "
866 "src:%u.%u.%u.%u, dst:%u.%u.%u.%u\n",
867 __func__, skb, skb->len,
868 NIPQUAD(skb->rtable->rt_src),
869 NIPQUAD(skb->rtable->rt_dst));
871 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS);
872 return ip_queue_xmit(skb, ipfragok);
875 static struct sctp_af sctp_af_inet;
877 static struct sctp_pf sctp_pf_inet = {
878 .event_msgname = sctp_inet_event_msgname,
879 .skb_msgname = sctp_inet_skb_msgname,
880 .af_supported = sctp_inet_af_supported,
881 .cmp_addr = sctp_inet_cmp_addr,
882 .bind_verify = sctp_inet_bind_verify,
883 .send_verify = sctp_inet_send_verify,
884 .supported_addrs = sctp_inet_supported_addrs,
885 .create_accept_sk = sctp_v4_create_accept_sk,
886 .addr_v4map = sctp_v4_addr_v4map,
890 /* Notifier for inetaddr addition/deletion events. */
891 static struct notifier_block sctp_inetaddr_notifier = {
892 .notifier_call = sctp_inetaddr_event,
895 /* Socket operations. */
896 static const struct proto_ops inet_seqpacket_ops = {
898 .owner = THIS_MODULE,
899 .release = inet_release, /* Needs to be wrapped... */
901 .connect = inet_dgram_connect,
902 .socketpair = sock_no_socketpair,
903 .accept = inet_accept,
904 .getname = inet_getname, /* Semantics are different. */
907 .listen = sctp_inet_listen,
908 .shutdown = inet_shutdown, /* Looks harmless. */
909 .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */
910 .getsockopt = sock_common_getsockopt,
911 .sendmsg = inet_sendmsg,
912 .recvmsg = sock_common_recvmsg,
913 .mmap = sock_no_mmap,
914 .sendpage = sock_no_sendpage,
916 .compat_setsockopt = compat_sock_common_setsockopt,
917 .compat_getsockopt = compat_sock_common_getsockopt,
921 /* Registration with AF_INET family. */
922 static struct inet_protosw sctp_seqpacket_protosw = {
923 .type = SOCK_SEQPACKET,
924 .protocol = IPPROTO_SCTP,
926 .ops = &inet_seqpacket_ops,
929 .flags = SCTP_PROTOSW_FLAG
931 static struct inet_protosw sctp_stream_protosw = {
933 .protocol = IPPROTO_SCTP,
935 .ops = &inet_seqpacket_ops,
938 .flags = SCTP_PROTOSW_FLAG
941 /* Register with IP layer. */
942 static struct net_protocol sctp_protocol = {
944 .err_handler = sctp_v4_err,
948 /* IPv4 address related functions. */
949 static struct sctp_af sctp_af_inet = {
950 .sa_family = AF_INET,
951 .sctp_xmit = sctp_v4_xmit,
952 .setsockopt = ip_setsockopt,
953 .getsockopt = ip_getsockopt,
954 .get_dst = sctp_v4_get_dst,
955 .get_saddr = sctp_v4_get_saddr,
956 .copy_addrlist = sctp_v4_copy_addrlist,
957 .from_skb = sctp_v4_from_skb,
958 .from_sk = sctp_v4_from_sk,
959 .to_sk_saddr = sctp_v4_to_sk_saddr,
960 .to_sk_daddr = sctp_v4_to_sk_daddr,
961 .from_addr_param = sctp_v4_from_addr_param,
962 .to_addr_param = sctp_v4_to_addr_param,
963 .dst_saddr = sctp_v4_dst_saddr,
964 .cmp_addr = sctp_v4_cmp_addr,
965 .addr_valid = sctp_v4_addr_valid,
966 .inaddr_any = sctp_v4_inaddr_any,
967 .is_any = sctp_v4_is_any,
968 .available = sctp_v4_available,
969 .scope = sctp_v4_scope,
970 .skb_iif = sctp_v4_skb_iif,
971 .is_ce = sctp_v4_is_ce,
972 .seq_dump_addr = sctp_v4_seq_dump_addr,
973 .ecn_capable = sctp_v4_ecn_capable,
974 .net_header_len = sizeof(struct iphdr),
975 .sockaddr_len = sizeof(struct sockaddr_in),
977 .compat_setsockopt = compat_ip_setsockopt,
978 .compat_getsockopt = compat_ip_getsockopt,
982 struct sctp_pf *sctp_get_pf_specific(sa_family_t family) {
986 return sctp_pf_inet_specific;
988 return sctp_pf_inet6_specific;
994 /* Register the PF specific function table. */
995 int sctp_register_pf(struct sctp_pf *pf, sa_family_t family)
999 if (sctp_pf_inet_specific)
1001 sctp_pf_inet_specific = pf;
1004 if (sctp_pf_inet6_specific)
1006 sctp_pf_inet6_specific = pf;
1014 static inline int init_sctp_mibs(void)
1016 return snmp_mib_init((void**)sctp_statistics, sizeof(struct sctp_mib));
1019 static inline void cleanup_sctp_mibs(void)
1021 snmp_mib_free((void**)sctp_statistics);
1024 static void sctp_v4_pf_init(void)
1026 /* Initialize the SCTP specific PF functions. */
1027 sctp_register_pf(&sctp_pf_inet, PF_INET);
1028 sctp_register_af(&sctp_af_inet);
1031 static void sctp_v4_pf_exit(void)
1033 list_del(&sctp_af_inet.list);
1036 static int sctp_v4_protosw_init(void)
1040 rc = proto_register(&sctp_prot, 1);
1044 /* Register SCTP(UDP and TCP style) with socket layer. */
1045 inet_register_protosw(&sctp_seqpacket_protosw);
1046 inet_register_protosw(&sctp_stream_protosw);
1051 static void sctp_v4_protosw_exit(void)
1053 inet_unregister_protosw(&sctp_stream_protosw);
1054 inet_unregister_protosw(&sctp_seqpacket_protosw);
1055 proto_unregister(&sctp_prot);
1058 static int sctp_v4_add_protocol(void)
1060 /* Register notifier for inet address additions/deletions. */
1061 register_inetaddr_notifier(&sctp_inetaddr_notifier);
1063 /* Register SCTP with inet layer. */
1064 if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0)
1070 static void sctp_v4_del_protocol(void)
1072 inet_del_protocol(&sctp_protocol, IPPROTO_SCTP);
1073 unregister_inetaddr_notifier(&sctp_inetaddr_notifier);
1076 /* Initialize the universe into something sensible. */
1077 SCTP_STATIC __init int sctp_init(void)
1080 int status = -EINVAL;
1082 unsigned long limit;
1086 /* SCTP_DEBUG sanity check. */
1087 if (!sctp_sanity_check())
1090 /* Allocate bind_bucket and chunk caches. */
1092 sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket",
1093 sizeof(struct sctp_bind_bucket),
1094 0, SLAB_HWCACHE_ALIGN,
1096 if (!sctp_bucket_cachep)
1099 sctp_chunk_cachep = kmem_cache_create("sctp_chunk",
1100 sizeof(struct sctp_chunk),
1101 0, SLAB_HWCACHE_ALIGN,
1103 if (!sctp_chunk_cachep)
1104 goto err_chunk_cachep;
1106 /* Allocate and initialise sctp mibs. */
1107 status = init_sctp_mibs();
1111 /* Initialize proc fs directory. */
1112 status = sctp_proc_init();
1116 /* Initialize object count debugging. */
1117 sctp_dbg_objcnt_init();
1120 * 14. Suggested SCTP Protocol Parameter Values
1122 /* The following protocol parameters are RECOMMENDED: */
1123 /* RTO.Initial - 3 seconds */
1124 sctp_rto_initial = SCTP_RTO_INITIAL;
1125 /* RTO.Min - 1 second */
1126 sctp_rto_min = SCTP_RTO_MIN;
1127 /* RTO.Max - 60 seconds */
1128 sctp_rto_max = SCTP_RTO_MAX;
1129 /* RTO.Alpha - 1/8 */
1130 sctp_rto_alpha = SCTP_RTO_ALPHA;
1131 /* RTO.Beta - 1/4 */
1132 sctp_rto_beta = SCTP_RTO_BETA;
1134 /* Valid.Cookie.Life - 60 seconds */
1135 sctp_valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE;
1137 /* Whether Cookie Preservative is enabled(1) or not(0) */
1138 sctp_cookie_preserve_enable = 1;
1141 sctp_max_burst = SCTP_DEFAULT_MAX_BURST;
1143 /* Association.Max.Retrans - 10 attempts
1144 * Path.Max.Retrans - 5 attempts (per destination address)
1145 * Max.Init.Retransmits - 8 attempts
1147 sctp_max_retrans_association = 10;
1148 sctp_max_retrans_path = 5;
1149 sctp_max_retrans_init = 8;
1151 /* Sendbuffer growth - do per-socket accounting */
1152 sctp_sndbuf_policy = 0;
1154 /* Rcvbuffer growth - do per-socket accounting */
1155 sctp_rcvbuf_policy = 0;
1157 /* HB.interval - 30 seconds */
1158 sctp_hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT;
1160 /* delayed SACK timeout */
1161 sctp_sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK;
1163 /* Implementation specific variables. */
1165 /* Initialize default stream count setup information. */
1166 sctp_max_instreams = SCTP_DEFAULT_INSTREAMS;
1167 sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS;
1169 /* Initialize handle used for association ids. */
1170 idr_init(&sctp_assocs_id);
1172 /* Set the pressure threshold to be a fraction of global memory that
1173 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1174 * memory, with a floor of 128 pages.
1175 * Note this initalizes the data in sctpv6_prot too
1176 * Unabashedly stolen from tcp_init
1178 limit = min(num_physpages, 1UL<<(28-PAGE_SHIFT)) >> (20-PAGE_SHIFT);
1179 limit = (limit * (num_physpages >> (20-PAGE_SHIFT))) >> (PAGE_SHIFT-11);
1180 limit = max(limit, 128UL);
1181 sysctl_sctp_mem[0] = limit / 4 * 3;
1182 sysctl_sctp_mem[1] = limit;
1183 sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2;
1185 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1186 limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7);
1187 max_share = min(4UL*1024*1024, limit);
1189 sysctl_sctp_rmem[0] = PAGE_SIZE; /* give each asoc 1 page min */
1190 sysctl_sctp_rmem[1] = (1500 *(sizeof(struct sk_buff) + 1));
1191 sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share);
1193 sysctl_sctp_wmem[0] = SK_MEM_QUANTUM;
1194 sysctl_sctp_wmem[1] = 16*1024;
1195 sysctl_sctp_wmem[2] = max(64*1024, max_share);
1197 /* Size and allocate the association hash table.
1198 * The methodology is similar to that of the tcp hash tables.
1200 if (num_physpages >= (128 * 1024))
1201 goal = num_physpages >> (22 - PAGE_SHIFT);
1203 goal = num_physpages >> (24 - PAGE_SHIFT);
1205 for (order = 0; (1UL << order) < goal; order++)
1209 sctp_assoc_hashsize = (1UL << order) * PAGE_SIZE /
1210 sizeof(struct sctp_hashbucket);
1211 if ((sctp_assoc_hashsize > (64 * 1024)) && order > 0)
1213 sctp_assoc_hashtable = (struct sctp_hashbucket *)
1214 __get_free_pages(GFP_ATOMIC, order);
1215 } while (!sctp_assoc_hashtable && --order > 0);
1216 if (!sctp_assoc_hashtable) {
1217 printk(KERN_ERR "SCTP: Failed association hash alloc.\n");
1219 goto err_ahash_alloc;
1221 for (i = 0; i < sctp_assoc_hashsize; i++) {
1222 rwlock_init(&sctp_assoc_hashtable[i].lock);
1223 INIT_HLIST_HEAD(&sctp_assoc_hashtable[i].chain);
1226 /* Allocate and initialize the endpoint hash table. */
1227 sctp_ep_hashsize = 64;
1228 sctp_ep_hashtable = (struct sctp_hashbucket *)
1229 kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL);
1230 if (!sctp_ep_hashtable) {
1231 printk(KERN_ERR "SCTP: Failed endpoint_hash alloc.\n");
1233 goto err_ehash_alloc;
1235 for (i = 0; i < sctp_ep_hashsize; i++) {
1236 rwlock_init(&sctp_ep_hashtable[i].lock);
1237 INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain);
1240 /* Allocate and initialize the SCTP port hash table. */
1242 sctp_port_hashsize = (1UL << order) * PAGE_SIZE /
1243 sizeof(struct sctp_bind_hashbucket);
1244 if ((sctp_port_hashsize > (64 * 1024)) && order > 0)
1246 sctp_port_hashtable = (struct sctp_bind_hashbucket *)
1247 __get_free_pages(GFP_ATOMIC, order);
1248 } while (!sctp_port_hashtable && --order > 0);
1249 if (!sctp_port_hashtable) {
1250 printk(KERN_ERR "SCTP: Failed bind hash alloc.");
1252 goto err_bhash_alloc;
1254 for (i = 0; i < sctp_port_hashsize; i++) {
1255 spin_lock_init(&sctp_port_hashtable[i].lock);
1256 INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain);
1259 printk(KERN_INFO "SCTP: Hash tables configured "
1260 "(established %d bind %d)\n",
1261 sctp_assoc_hashsize, sctp_port_hashsize);
1263 /* Disable ADDIP by default. */
1264 sctp_addip_enable = 0;
1265 sctp_addip_noauth = 0;
1267 /* Enable PR-SCTP by default. */
1268 sctp_prsctp_enable = 1;
1270 /* Disable AUTH by default. */
1271 sctp_auth_enable = 0;
1273 sctp_sysctl_register();
1275 INIT_LIST_HEAD(&sctp_address_families);
1279 /* Initialize the local address list. */
1280 INIT_LIST_HEAD(&sctp_local_addr_list);
1281 spin_lock_init(&sctp_local_addr_lock);
1282 sctp_get_local_addr_list();
1284 status = sctp_v4_protosw_init();
1287 goto err_protosw_init;
1289 status = sctp_v6_protosw_init();
1291 goto err_v6_protosw_init;
1293 /* Initialize the control inode/socket for handling OOTB packets. */
1294 if ((status = sctp_ctl_sock_init())) {
1296 "SCTP: Failed to initialize the SCTP control sock.\n");
1297 goto err_ctl_sock_init;
1300 status = sctp_v4_add_protocol();
1302 goto err_add_protocol;
1304 /* Register SCTP with inet6 layer. */
1305 status = sctp_v6_add_protocol();
1307 goto err_v6_add_protocol;
1312 err_v6_add_protocol:
1313 sctp_v6_del_protocol();
1315 sctp_v4_del_protocol();
1316 inet_ctl_sock_destroy(sctp_ctl_sock);
1318 sctp_v6_protosw_exit();
1319 err_v6_protosw_init:
1320 sctp_v4_protosw_exit();
1322 sctp_free_local_addr_list();
1325 sctp_sysctl_unregister();
1326 list_del(&sctp_af_inet.list);
1327 free_pages((unsigned long)sctp_port_hashtable,
1328 get_order(sctp_port_hashsize *
1329 sizeof(struct sctp_bind_hashbucket)));
1331 kfree(sctp_ep_hashtable);
1333 free_pages((unsigned long)sctp_assoc_hashtable,
1334 get_order(sctp_assoc_hashsize *
1335 sizeof(struct sctp_hashbucket)));
1337 sctp_dbg_objcnt_exit();
1340 cleanup_sctp_mibs();
1342 kmem_cache_destroy(sctp_chunk_cachep);
1344 kmem_cache_destroy(sctp_bucket_cachep);
1348 /* Exit handler for the SCTP protocol. */
1349 SCTP_STATIC __exit void sctp_exit(void)
1351 /* BUG. This should probably do something useful like clean
1352 * up all the remaining associations and all that memory.
1355 /* Unregister with inet6/inet layers. */
1356 sctp_v6_del_protocol();
1357 sctp_v4_del_protocol();
1359 /* Free the control endpoint. */
1360 inet_ctl_sock_destroy(sctp_ctl_sock);
1362 /* Free protosw registrations */
1363 sctp_v6_protosw_exit();
1364 sctp_v4_protosw_exit();
1366 /* Free the local address list. */
1367 sctp_free_local_addr_list();
1369 /* Unregister with socket layer. */
1373 sctp_sysctl_unregister();
1374 list_del(&sctp_af_inet.list);
1376 free_pages((unsigned long)sctp_assoc_hashtable,
1377 get_order(sctp_assoc_hashsize *
1378 sizeof(struct sctp_hashbucket)));
1379 kfree(sctp_ep_hashtable);
1380 free_pages((unsigned long)sctp_port_hashtable,
1381 get_order(sctp_port_hashsize *
1382 sizeof(struct sctp_bind_hashbucket)));
1384 sctp_dbg_objcnt_exit();
1386 cleanup_sctp_mibs();
1388 kmem_cache_destroy(sctp_chunk_cachep);
1389 kmem_cache_destroy(sctp_bucket_cachep);
1392 module_init(sctp_init);
1393 module_exit(sctp_exit);
1396 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1398 MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132");
1399 MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132");
1400 MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1401 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1402 MODULE_LICENSE("GPL");