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 <linux/highmem.h>
56 #include <linux/swap.h>
57 #include <net/net_namespace.h>
58 #include <net/protocol.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>
67 /* Global data structures. */
68 struct sctp_globals sctp_globals __read_mostly;
69 DEFINE_SNMP_STAT(struct sctp_mib, sctp_statistics) __read_mostly;
72 struct proc_dir_entry *proc_net_sctp;
75 struct idr sctp_assocs_id;
76 DEFINE_SPINLOCK(sctp_assocs_id_lock);
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.
82 static struct sock *sctp_ctl_sock;
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;
89 struct kmem_cache *sctp_chunk_cachep __read_mostly;
90 struct kmem_cache *sctp_bucket_cachep __read_mostly;
92 int sysctl_sctp_mem[3];
93 int sysctl_sctp_rmem[3];
94 int sysctl_sctp_wmem[3];
96 /* Return the address of the control sock. */
97 struct sock *sctp_get_ctl_sock(void)
102 /* Set up the proc fs entry for the SCTP protocol. */
103 static __init int sctp_proc_init(void)
105 #ifdef CONFIG_PROC_FS
106 if (!proc_net_sctp) {
107 struct proc_dir_entry *ent;
108 ent = proc_mkdir("sctp", init_net.proc_net);
110 ent->owner = THIS_MODULE;
116 if (sctp_snmp_proc_init())
117 goto out_snmp_proc_init;
118 if (sctp_eps_proc_init())
119 goto out_eps_proc_init;
120 if (sctp_assocs_proc_init())
121 goto out_assocs_proc_init;
122 if (sctp_remaddr_proc_init())
123 goto out_remaddr_proc_init;
127 out_remaddr_proc_init:
128 sctp_assocs_proc_exit();
129 out_assocs_proc_init:
130 sctp_eps_proc_exit();
132 sctp_snmp_proc_exit();
135 proc_net_sctp = NULL;
136 remove_proc_entry("sctp", init_net.proc_net);
142 #endif /* CONFIG_PROC_FS */
145 /* Clean up the proc fs entry for the SCTP protocol.
146 * Note: Do not make this __exit as it is used in the init error
149 static void sctp_proc_exit(void)
151 #ifdef CONFIG_PROC_FS
152 sctp_snmp_proc_exit();
153 sctp_eps_proc_exit();
154 sctp_assocs_proc_exit();
155 sctp_remaddr_proc_exit();
158 proc_net_sctp = NULL;
159 remove_proc_entry("sctp", init_net.proc_net);
164 /* Private helper to extract ipv4 address and stash them in
165 * the protocol structure.
167 static void sctp_v4_copy_addrlist(struct list_head *addrlist,
168 struct net_device *dev)
170 struct in_device *in_dev;
171 struct in_ifaddr *ifa;
172 struct sctp_sockaddr_entry *addr;
175 if ((in_dev = __in_dev_get_rcu(dev)) == NULL) {
180 for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
181 /* Add the address to the local list. */
182 addr = t_new(struct sctp_sockaddr_entry, GFP_ATOMIC);
184 addr->a.v4.sin_family = AF_INET;
185 addr->a.v4.sin_port = 0;
186 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
188 INIT_LIST_HEAD(&addr->list);
189 INIT_RCU_HEAD(&addr->rcu);
190 list_add_tail(&addr->list, addrlist);
197 /* Extract our IP addresses from the system and stash them in the
198 * protocol structure.
200 static void sctp_get_local_addr_list(void)
202 struct net_device *dev;
203 struct list_head *pos;
206 read_lock(&dev_base_lock);
207 for_each_netdev(&init_net, dev) {
208 __list_for_each(pos, &sctp_address_families) {
209 af = list_entry(pos, struct sctp_af, list);
210 af->copy_addrlist(&sctp_local_addr_list, dev);
213 read_unlock(&dev_base_lock);
216 /* Free the existing local addresses. */
217 static void sctp_free_local_addr_list(void)
219 struct sctp_sockaddr_entry *addr;
220 struct list_head *pos, *temp;
222 list_for_each_safe(pos, temp, &sctp_local_addr_list) {
223 addr = list_entry(pos, struct sctp_sockaddr_entry, list);
229 void sctp_local_addr_free(struct rcu_head *head)
231 struct sctp_sockaddr_entry *e = container_of(head,
232 struct sctp_sockaddr_entry, rcu);
236 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
237 int sctp_copy_local_addr_list(struct sctp_bind_addr *bp, sctp_scope_t scope,
238 gfp_t gfp, int copy_flags)
240 struct sctp_sockaddr_entry *addr;
244 list_for_each_entry_rcu(addr, &sctp_local_addr_list, list) {
247 if (sctp_in_scope(&addr->a, scope)) {
248 /* Now that the address is in scope, check to see if
249 * the address type is really supported by the local
250 * sock as well as the remote peer.
252 if ((((AF_INET == addr->a.sa.sa_family) &&
253 (copy_flags & SCTP_ADDR4_PEERSUPP))) ||
254 (((AF_INET6 == addr->a.sa.sa_family) &&
255 (copy_flags & SCTP_ADDR6_ALLOWED) &&
256 (copy_flags & SCTP_ADDR6_PEERSUPP)))) {
257 error = sctp_add_bind_addr(bp, &addr->a,
258 SCTP_ADDR_SRC, GFP_ATOMIC);
270 /* Initialize a sctp_addr from in incoming skb. */
271 static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
278 port = &addr->v4.sin_port;
279 addr->v4.sin_family = AF_INET;
284 from = &ip_hdr(skb)->saddr;
287 from = &ip_hdr(skb)->daddr;
289 memcpy(&addr->v4.sin_addr.s_addr, from, sizeof(struct in_addr));
292 /* Initialize an sctp_addr from a socket. */
293 static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk)
295 addr->v4.sin_family = AF_INET;
296 addr->v4.sin_port = 0;
297 addr->v4.sin_addr.s_addr = inet_sk(sk)->rcv_saddr;
300 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
301 static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk)
303 inet_sk(sk)->rcv_saddr = addr->v4.sin_addr.s_addr;
306 /* Initialize sk->sk_daddr from sctp_addr. */
307 static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk)
309 inet_sk(sk)->daddr = addr->v4.sin_addr.s_addr;
312 /* Initialize a sctp_addr from an address parameter. */
313 static void sctp_v4_from_addr_param(union sctp_addr *addr,
314 union sctp_addr_param *param,
315 __be16 port, int iif)
317 addr->v4.sin_family = AF_INET;
318 addr->v4.sin_port = port;
319 addr->v4.sin_addr.s_addr = param->v4.addr.s_addr;
322 /* Initialize an address parameter from a sctp_addr and return the length
323 * of the address parameter.
325 static int sctp_v4_to_addr_param(const union sctp_addr *addr,
326 union sctp_addr_param *param)
328 int length = sizeof(sctp_ipv4addr_param_t);
330 param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS;
331 param->v4.param_hdr.length = htons(length);
332 param->v4.addr.s_addr = addr->v4.sin_addr.s_addr;
337 /* Initialize a sctp_addr from a dst_entry. */
338 static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct dst_entry *dst,
341 struct rtable *rt = (struct rtable *)dst;
342 saddr->v4.sin_family = AF_INET;
343 saddr->v4.sin_port = port;
344 saddr->v4.sin_addr.s_addr = rt->rt_src;
347 /* Compare two addresses exactly. */
348 static int sctp_v4_cmp_addr(const union sctp_addr *addr1,
349 const union sctp_addr *addr2)
351 if (addr1->sa.sa_family != addr2->sa.sa_family)
353 if (addr1->v4.sin_port != addr2->v4.sin_port)
355 if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr)
361 /* Initialize addr struct to INADDR_ANY. */
362 static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port)
364 addr->v4.sin_family = AF_INET;
365 addr->v4.sin_addr.s_addr = htonl(INADDR_ANY);
366 addr->v4.sin_port = port;
369 /* Is this a wildcard address? */
370 static int sctp_v4_is_any(const union sctp_addr *addr)
372 return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr;
375 /* This function checks if the address is a valid address to be used for
379 * Return 0 - If the address is a non-unicast or an illegal address.
380 * Return 1 - If the address is a unicast.
382 static int sctp_v4_addr_valid(union sctp_addr *addr,
383 struct sctp_sock *sp,
384 const struct sk_buff *skb)
386 /* IPv4 addresses not allowed */
387 if (sp && ipv6_only_sock(sctp_opt2sk(sp)))
390 /* Is this a non-unicast address or a unusable SCTP address? */
391 if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr))
394 /* Is this a broadcast address? */
395 if (skb && skb->rtable->rt_flags & RTCF_BROADCAST)
401 /* Should this be available for binding? */
402 static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp)
404 int ret = inet_addr_type(&init_net, addr->v4.sin_addr.s_addr);
407 if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) &&
409 !sp->inet.freebind &&
410 !sysctl_ip_nonlocal_bind)
413 if (ipv6_only_sock(sctp_opt2sk(sp)))
419 /* Checking the loopback, private and other address scopes as defined in
420 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
421 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
423 * Level 0 - unusable SCTP addresses
424 * Level 1 - loopback address
425 * Level 2 - link-local addresses
426 * Level 3 - private addresses.
427 * Level 4 - global addresses
428 * For INIT and INIT-ACK address list, let L be the level of
429 * of requested destination address, sender and receiver
430 * SHOULD include all of its addresses with level greater
431 * than or equal to L.
433 static sctp_scope_t sctp_v4_scope(union sctp_addr *addr)
437 /* Should IPv4 scoping be a sysctl configurable option
438 * so users can turn it off (default on) for certain
439 * unconventional networking environments?
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;
454 retval = SCTP_SCOPE_GLOBAL;
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.
464 static struct dst_entry *sctp_v4_get_dst(struct sctp_association *asoc,
465 union sctp_addr *daddr,
466 union sctp_addr *saddr)
470 struct sctp_bind_addr *bp;
471 struct sctp_sockaddr_entry *laddr;
472 struct dst_entry *dst = NULL;
473 union sctp_addr dst_saddr;
475 memset(&fl, 0x0, sizeof(struct flowi));
476 fl.fl4_dst = daddr->v4.sin_addr.s_addr;
477 fl.proto = IPPROTO_SCTP;
479 fl.fl4_tos = RT_CONN_FLAGS(asoc->base.sk);
480 fl.oif = asoc->base.sk->sk_bound_dev_if;
483 fl.fl4_src = saddr->v4.sin_addr.s_addr;
485 SCTP_DEBUG_PRINTK("%s: DST:%pI4, SRC:%pI4 - ",
486 __func__, &fl.fl4_dst, &fl.fl4_src);
488 if (!ip_route_output_key(&init_net, &rt, &fl)) {
492 /* If there is no association or if a source address is passed, no
493 * more validation is required.
498 bp = &asoc->base.bind_addr;
501 /* Walk through the bind address list and look for a bind
502 * address that matches the source address of the returned dst.
504 sctp_v4_dst_saddr(&dst_saddr, dst, htons(bp->port));
506 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
507 if (!laddr->valid || (laddr->state != SCTP_ADDR_SRC))
509 if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a))
514 /* None of the bound addresses match the source address of the
515 * dst. So release it.
521 /* Walk through the bind address list and try to get a dst that
522 * matches a bind address as the source address.
525 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
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)) {
542 SCTP_DEBUG_PRINTK("rt_dst:%pI4, rt_src:%pI4\n",
543 &rt->rt_dst, &rt->rt_src);
545 SCTP_DEBUG_PRINTK("NO ROUTE\n");
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.
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)
559 struct rtable *rt = (struct rtable *)dst;
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;
571 /* What interface did this skb arrive on? */
572 static int sctp_v4_skb_iif(const struct sk_buff *skb)
574 return skb->rtable->rt_iif;
577 /* Was this packet marked by Explicit Congestion Notification? */
578 static int sctp_v4_is_ce(const struct sk_buff *skb)
580 return INET_ECN_is_ce(ip_hdr(skb)->tos);
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)
587 struct inet_sock *inet = inet_sk(sk);
588 struct inet_sock *newinet;
589 struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL,
595 sock_init_data(NULL, newsk);
597 newsk->sk_type = SOCK_STREAM;
599 newsk->sk_no_check = sk->sk_no_check;
600 newsk->sk_reuse = sk->sk_reuse;
601 newsk->sk_shutdown = sk->sk_shutdown;
603 newsk->sk_destruct = inet_sock_destruct;
604 newsk->sk_family = PF_INET;
605 newsk->sk_protocol = IPPROTO_SCTP;
606 newsk->sk_backlog_rcv = sk->sk_prot->backlog_rcv;
607 sock_reset_flag(newsk, SOCK_ZAPPED);
609 newinet = inet_sk(newsk);
611 /* Initialize sk's sport, dport, rcv_saddr and daddr for
612 * getsockname() and getpeername()
614 newinet->sport = inet->sport;
615 newinet->saddr = inet->saddr;
616 newinet->rcv_saddr = inet->rcv_saddr;
617 newinet->dport = htons(asoc->peer.port);
618 newinet->daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr;
619 newinet->pmtudisc = inet->pmtudisc;
620 newinet->id = asoc->next_tsn ^ jiffies;
622 newinet->uc_ttl = -1;
623 newinet->mc_loop = 1;
625 newinet->mc_index = 0;
626 newinet->mc_list = NULL;
628 sk_refcnt_debug_inc(newsk);
630 if (newsk->sk_prot->init(newsk)) {
631 sk_common_release(newsk);
639 /* Map address, empty for v4 family */
640 static void sctp_v4_addr_v4map(struct sctp_sock *sp, union sctp_addr *addr)
645 /* Dump the v4 addr to the seq file. */
646 static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr)
648 seq_printf(seq, "%pI4 ", &addr->v4.sin_addr);
651 static void sctp_v4_ecn_capable(struct sock *sk)
656 /* Event handler for inet address addition/deletion events.
657 * The sctp_local_addr_list needs to be protocted by a spin lock since
658 * multiple notifiers (say IPv4 and IPv6) may be running at the same
659 * time and thus corrupt the list.
660 * The reader side is protected with RCU.
662 static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev,
665 struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
666 struct sctp_sockaddr_entry *addr = NULL;
667 struct sctp_sockaddr_entry *temp;
670 if (!net_eq(dev_net(ifa->ifa_dev->dev), &init_net))
675 addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
677 addr->a.v4.sin_family = AF_INET;
678 addr->a.v4.sin_port = 0;
679 addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
681 spin_lock_bh(&sctp_local_addr_lock);
682 list_add_tail_rcu(&addr->list, &sctp_local_addr_list);
683 spin_unlock_bh(&sctp_local_addr_lock);
687 spin_lock_bh(&sctp_local_addr_lock);
688 list_for_each_entry_safe(addr, temp,
689 &sctp_local_addr_list, list) {
690 if (addr->a.sa.sa_family == AF_INET &&
691 addr->a.v4.sin_addr.s_addr ==
695 list_del_rcu(&addr->list);
699 spin_unlock_bh(&sctp_local_addr_lock);
701 call_rcu(&addr->rcu, sctp_local_addr_free);
709 * Initialize the control inode/socket with a control endpoint data
710 * structure. This endpoint is reserved exclusively for the OOTB processing.
712 static int sctp_ctl_sock_init(void)
717 if (sctp_get_pf_specific(PF_INET6))
722 err = inet_ctl_sock_create(&sctp_ctl_sock, family,
723 SOCK_SEQPACKET, IPPROTO_SCTP, &init_net);
726 "SCTP: Failed to create the SCTP control socket.\n");
732 /* Register address family specific functions. */
733 int sctp_register_af(struct sctp_af *af)
735 switch (af->sa_family) {
737 if (sctp_af_v4_specific)
739 sctp_af_v4_specific = af;
742 if (sctp_af_v6_specific)
744 sctp_af_v6_specific = af;
750 INIT_LIST_HEAD(&af->list);
751 list_add_tail(&af->list, &sctp_address_families);
755 /* Get the table of functions for manipulating a particular address
758 struct sctp_af *sctp_get_af_specific(sa_family_t family)
762 return sctp_af_v4_specific;
764 return sctp_af_v6_specific;
770 /* Common code to initialize a AF_INET msg_name. */
771 static void sctp_inet_msgname(char *msgname, int *addr_len)
773 struct sockaddr_in *sin;
775 sin = (struct sockaddr_in *)msgname;
776 *addr_len = sizeof(struct sockaddr_in);
777 sin->sin_family = AF_INET;
778 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
781 /* Copy the primary address of the peer primary address as the msg_name. */
782 static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname,
785 struct sockaddr_in *sin, *sinfrom;
788 struct sctp_association *asoc;
791 sctp_inet_msgname(msgname, addr_len);
792 sin = (struct sockaddr_in *)msgname;
793 sinfrom = &asoc->peer.primary_addr.v4;
794 sin->sin_port = htons(asoc->peer.port);
795 sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr;
799 /* Initialize and copy out a msgname from an inbound skb. */
800 static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len)
803 struct sctphdr *sh = sctp_hdr(skb);
804 struct sockaddr_in *sin = (struct sockaddr_in *)msgname;
806 sctp_inet_msgname(msgname, len);
807 sin->sin_port = sh->source;
808 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
812 /* Do we support this AF? */
813 static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp)
815 /* PF_INET only supports AF_INET addresses. */
816 return (AF_INET == family);
819 /* Address matching with wildcards allowed. */
820 static int sctp_inet_cmp_addr(const union sctp_addr *addr1,
821 const union sctp_addr *addr2,
822 struct sctp_sock *opt)
824 /* PF_INET only supports AF_INET addresses. */
825 if (addr1->sa.sa_family != addr2->sa.sa_family)
827 if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr ||
828 htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr)
830 if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr)
836 /* Verify that provided sockaddr looks bindable. Common verification has
837 * already been taken care of.
839 static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr)
841 return sctp_v4_available(addr, opt);
844 /* Verify that sockaddr looks sendable. Common verification has already
845 * been taken care of.
847 static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr)
852 /* Fill in Supported Address Type information for INIT and INIT-ACK
853 * chunks. Returns number of addresses supported.
855 static int sctp_inet_supported_addrs(const struct sctp_sock *opt,
858 types[0] = SCTP_PARAM_IPV4_ADDRESS;
862 /* Wrapper routine that calls the ip transmit routine. */
863 static inline int sctp_v4_xmit(struct sk_buff *skb,
864 struct sctp_transport *transport)
866 struct inet_sock *inet = inet_sk(skb->sk);
868 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n",
869 __func__, skb, skb->len,
870 &skb->rtable->rt_src,
871 &skb->rtable->rt_dst);
873 inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ?
874 IP_PMTUDISC_DO : IP_PMTUDISC_DONT;
876 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS);
877 return ip_queue_xmit(skb, 0);
880 static struct sctp_af sctp_af_inet;
882 static struct sctp_pf sctp_pf_inet = {
883 .event_msgname = sctp_inet_event_msgname,
884 .skb_msgname = sctp_inet_skb_msgname,
885 .af_supported = sctp_inet_af_supported,
886 .cmp_addr = sctp_inet_cmp_addr,
887 .bind_verify = sctp_inet_bind_verify,
888 .send_verify = sctp_inet_send_verify,
889 .supported_addrs = sctp_inet_supported_addrs,
890 .create_accept_sk = sctp_v4_create_accept_sk,
891 .addr_v4map = sctp_v4_addr_v4map,
895 /* Notifier for inetaddr addition/deletion events. */
896 static struct notifier_block sctp_inetaddr_notifier = {
897 .notifier_call = sctp_inetaddr_event,
900 /* Socket operations. */
901 static const struct proto_ops inet_seqpacket_ops = {
903 .owner = THIS_MODULE,
904 .release = inet_release, /* Needs to be wrapped... */
906 .connect = inet_dgram_connect,
907 .socketpair = sock_no_socketpair,
908 .accept = inet_accept,
909 .getname = inet_getname, /* Semantics are different. */
912 .listen = sctp_inet_listen,
913 .shutdown = inet_shutdown, /* Looks harmless. */
914 .setsockopt = sock_common_setsockopt, /* IP_SOL IP_OPTION is a problem */
915 .getsockopt = sock_common_getsockopt,
916 .sendmsg = inet_sendmsg,
917 .recvmsg = sock_common_recvmsg,
918 .mmap = sock_no_mmap,
919 .sendpage = sock_no_sendpage,
921 .compat_setsockopt = compat_sock_common_setsockopt,
922 .compat_getsockopt = compat_sock_common_getsockopt,
926 /* Registration with AF_INET family. */
927 static struct inet_protosw sctp_seqpacket_protosw = {
928 .type = SOCK_SEQPACKET,
929 .protocol = IPPROTO_SCTP,
931 .ops = &inet_seqpacket_ops,
934 .flags = SCTP_PROTOSW_FLAG
936 static struct inet_protosw sctp_stream_protosw = {
938 .protocol = IPPROTO_SCTP,
940 .ops = &inet_seqpacket_ops,
943 .flags = SCTP_PROTOSW_FLAG
946 /* Register with IP layer. */
947 static struct net_protocol sctp_protocol = {
949 .err_handler = sctp_v4_err,
953 /* IPv4 address related functions. */
954 static struct sctp_af sctp_af_inet = {
955 .sa_family = AF_INET,
956 .sctp_xmit = sctp_v4_xmit,
957 .setsockopt = ip_setsockopt,
958 .getsockopt = ip_getsockopt,
959 .get_dst = sctp_v4_get_dst,
960 .get_saddr = sctp_v4_get_saddr,
961 .copy_addrlist = sctp_v4_copy_addrlist,
962 .from_skb = sctp_v4_from_skb,
963 .from_sk = sctp_v4_from_sk,
964 .to_sk_saddr = sctp_v4_to_sk_saddr,
965 .to_sk_daddr = sctp_v4_to_sk_daddr,
966 .from_addr_param = sctp_v4_from_addr_param,
967 .to_addr_param = sctp_v4_to_addr_param,
968 .dst_saddr = sctp_v4_dst_saddr,
969 .cmp_addr = sctp_v4_cmp_addr,
970 .addr_valid = sctp_v4_addr_valid,
971 .inaddr_any = sctp_v4_inaddr_any,
972 .is_any = sctp_v4_is_any,
973 .available = sctp_v4_available,
974 .scope = sctp_v4_scope,
975 .skb_iif = sctp_v4_skb_iif,
976 .is_ce = sctp_v4_is_ce,
977 .seq_dump_addr = sctp_v4_seq_dump_addr,
978 .ecn_capable = sctp_v4_ecn_capable,
979 .net_header_len = sizeof(struct iphdr),
980 .sockaddr_len = sizeof(struct sockaddr_in),
982 .compat_setsockopt = compat_ip_setsockopt,
983 .compat_getsockopt = compat_ip_getsockopt,
987 struct sctp_pf *sctp_get_pf_specific(sa_family_t family) {
991 return sctp_pf_inet_specific;
993 return sctp_pf_inet6_specific;
999 /* Register the PF specific function table. */
1000 int sctp_register_pf(struct sctp_pf *pf, sa_family_t family)
1004 if (sctp_pf_inet_specific)
1006 sctp_pf_inet_specific = pf;
1009 if (sctp_pf_inet6_specific)
1011 sctp_pf_inet6_specific = pf;
1019 static inline int init_sctp_mibs(void)
1021 return snmp_mib_init((void**)sctp_statistics, sizeof(struct sctp_mib));
1024 static inline void cleanup_sctp_mibs(void)
1026 snmp_mib_free((void**)sctp_statistics);
1029 static void sctp_v4_pf_init(void)
1031 /* Initialize the SCTP specific PF functions. */
1032 sctp_register_pf(&sctp_pf_inet, PF_INET);
1033 sctp_register_af(&sctp_af_inet);
1036 static void sctp_v4_pf_exit(void)
1038 list_del(&sctp_af_inet.list);
1041 static int sctp_v4_protosw_init(void)
1045 rc = proto_register(&sctp_prot, 1);
1049 /* Register SCTP(UDP and TCP style) with socket layer. */
1050 inet_register_protosw(&sctp_seqpacket_protosw);
1051 inet_register_protosw(&sctp_stream_protosw);
1056 static void sctp_v4_protosw_exit(void)
1058 inet_unregister_protosw(&sctp_stream_protosw);
1059 inet_unregister_protosw(&sctp_seqpacket_protosw);
1060 proto_unregister(&sctp_prot);
1063 static int sctp_v4_add_protocol(void)
1065 /* Register notifier for inet address additions/deletions. */
1066 register_inetaddr_notifier(&sctp_inetaddr_notifier);
1068 /* Register SCTP with inet layer. */
1069 if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0)
1075 static void sctp_v4_del_protocol(void)
1077 inet_del_protocol(&sctp_protocol, IPPROTO_SCTP);
1078 unregister_inetaddr_notifier(&sctp_inetaddr_notifier);
1081 /* Initialize the universe into something sensible. */
1082 SCTP_STATIC __init int sctp_init(void)
1085 int status = -EINVAL;
1087 unsigned long limit;
1088 unsigned long nr_pages;
1092 /* SCTP_DEBUG sanity check. */
1093 if (!sctp_sanity_check())
1096 /* Allocate bind_bucket and chunk caches. */
1098 sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket",
1099 sizeof(struct sctp_bind_bucket),
1100 0, SLAB_HWCACHE_ALIGN,
1102 if (!sctp_bucket_cachep)
1105 sctp_chunk_cachep = kmem_cache_create("sctp_chunk",
1106 sizeof(struct sctp_chunk),
1107 0, SLAB_HWCACHE_ALIGN,
1109 if (!sctp_chunk_cachep)
1110 goto err_chunk_cachep;
1112 /* Allocate and initialise sctp mibs. */
1113 status = init_sctp_mibs();
1117 /* Initialize proc fs directory. */
1118 status = sctp_proc_init();
1122 /* Initialize object count debugging. */
1123 sctp_dbg_objcnt_init();
1126 * 14. Suggested SCTP Protocol Parameter Values
1128 /* The following protocol parameters are RECOMMENDED: */
1129 /* RTO.Initial - 3 seconds */
1130 sctp_rto_initial = SCTP_RTO_INITIAL;
1131 /* RTO.Min - 1 second */
1132 sctp_rto_min = SCTP_RTO_MIN;
1133 /* RTO.Max - 60 seconds */
1134 sctp_rto_max = SCTP_RTO_MAX;
1135 /* RTO.Alpha - 1/8 */
1136 sctp_rto_alpha = SCTP_RTO_ALPHA;
1137 /* RTO.Beta - 1/4 */
1138 sctp_rto_beta = SCTP_RTO_BETA;
1140 /* Valid.Cookie.Life - 60 seconds */
1141 sctp_valid_cookie_life = SCTP_DEFAULT_COOKIE_LIFE;
1143 /* Whether Cookie Preservative is enabled(1) or not(0) */
1144 sctp_cookie_preserve_enable = 1;
1147 sctp_max_burst = SCTP_DEFAULT_MAX_BURST;
1149 /* Association.Max.Retrans - 10 attempts
1150 * Path.Max.Retrans - 5 attempts (per destination address)
1151 * Max.Init.Retransmits - 8 attempts
1153 sctp_max_retrans_association = 10;
1154 sctp_max_retrans_path = 5;
1155 sctp_max_retrans_init = 8;
1157 /* Sendbuffer growth - do per-socket accounting */
1158 sctp_sndbuf_policy = 0;
1160 /* Rcvbuffer growth - do per-socket accounting */
1161 sctp_rcvbuf_policy = 0;
1163 /* HB.interval - 30 seconds */
1164 sctp_hb_interval = SCTP_DEFAULT_TIMEOUT_HEARTBEAT;
1166 /* delayed SACK timeout */
1167 sctp_sack_timeout = SCTP_DEFAULT_TIMEOUT_SACK;
1169 /* Implementation specific variables. */
1171 /* Initialize default stream count setup information. */
1172 sctp_max_instreams = SCTP_DEFAULT_INSTREAMS;
1173 sctp_max_outstreams = SCTP_DEFAULT_OUTSTREAMS;
1175 /* Initialize handle used for association ids. */
1176 idr_init(&sctp_assocs_id);
1178 /* Set the pressure threshold to be a fraction of global memory that
1179 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1180 * memory, with a floor of 128 pages.
1181 * Note this initalizes the data in sctpv6_prot too
1182 * Unabashedly stolen from tcp_init
1184 nr_pages = totalram_pages - totalhigh_pages;
1185 limit = min(nr_pages, 1UL<<(28-PAGE_SHIFT)) >> (20-PAGE_SHIFT);
1186 limit = (limit * (nr_pages >> (20-PAGE_SHIFT))) >> (PAGE_SHIFT-11);
1187 limit = max(limit, 128UL);
1188 sysctl_sctp_mem[0] = limit / 4 * 3;
1189 sysctl_sctp_mem[1] = limit;
1190 sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2;
1192 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1193 limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7);
1194 max_share = min(4UL*1024*1024, limit);
1196 sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */
1197 sysctl_sctp_rmem[1] = (1500 *(sizeof(struct sk_buff) + 1));
1198 sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share);
1200 sysctl_sctp_wmem[0] = SK_MEM_QUANTUM;
1201 sysctl_sctp_wmem[1] = 16*1024;
1202 sysctl_sctp_wmem[2] = max(64*1024, max_share);
1204 /* Size and allocate the association hash table.
1205 * The methodology is similar to that of the tcp hash tables.
1207 if (num_physpages >= (128 * 1024))
1208 goal = num_physpages >> (22 - PAGE_SHIFT);
1210 goal = num_physpages >> (24 - PAGE_SHIFT);
1212 for (order = 0; (1UL << order) < goal; order++)
1216 sctp_assoc_hashsize = (1UL << order) * PAGE_SIZE /
1217 sizeof(struct sctp_hashbucket);
1218 if ((sctp_assoc_hashsize > (64 * 1024)) && order > 0)
1220 sctp_assoc_hashtable = (struct sctp_hashbucket *)
1221 __get_free_pages(GFP_ATOMIC, order);
1222 } while (!sctp_assoc_hashtable && --order > 0);
1223 if (!sctp_assoc_hashtable) {
1224 printk(KERN_ERR "SCTP: Failed association hash alloc.\n");
1226 goto err_ahash_alloc;
1228 for (i = 0; i < sctp_assoc_hashsize; i++) {
1229 rwlock_init(&sctp_assoc_hashtable[i].lock);
1230 INIT_HLIST_HEAD(&sctp_assoc_hashtable[i].chain);
1233 /* Allocate and initialize the endpoint hash table. */
1234 sctp_ep_hashsize = 64;
1235 sctp_ep_hashtable = (struct sctp_hashbucket *)
1236 kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL);
1237 if (!sctp_ep_hashtable) {
1238 printk(KERN_ERR "SCTP: Failed endpoint_hash alloc.\n");
1240 goto err_ehash_alloc;
1242 for (i = 0; i < sctp_ep_hashsize; i++) {
1243 rwlock_init(&sctp_ep_hashtable[i].lock);
1244 INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain);
1247 /* Allocate and initialize the SCTP port hash table. */
1249 sctp_port_hashsize = (1UL << order) * PAGE_SIZE /
1250 sizeof(struct sctp_bind_hashbucket);
1251 if ((sctp_port_hashsize > (64 * 1024)) && order > 0)
1253 sctp_port_hashtable = (struct sctp_bind_hashbucket *)
1254 __get_free_pages(GFP_ATOMIC, order);
1255 } while (!sctp_port_hashtable && --order > 0);
1256 if (!sctp_port_hashtable) {
1257 printk(KERN_ERR "SCTP: Failed bind hash alloc.");
1259 goto err_bhash_alloc;
1261 for (i = 0; i < sctp_port_hashsize; i++) {
1262 spin_lock_init(&sctp_port_hashtable[i].lock);
1263 INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain);
1266 printk(KERN_INFO "SCTP: Hash tables configured "
1267 "(established %d bind %d)\n",
1268 sctp_assoc_hashsize, sctp_port_hashsize);
1270 /* Disable ADDIP by default. */
1271 sctp_addip_enable = 0;
1272 sctp_addip_noauth = 0;
1274 /* Enable PR-SCTP by default. */
1275 sctp_prsctp_enable = 1;
1277 /* Disable AUTH by default. */
1278 sctp_auth_enable = 0;
1280 sctp_sysctl_register();
1282 INIT_LIST_HEAD(&sctp_address_families);
1286 /* Initialize the local address list. */
1287 INIT_LIST_HEAD(&sctp_local_addr_list);
1288 spin_lock_init(&sctp_local_addr_lock);
1289 sctp_get_local_addr_list();
1291 status = sctp_v4_protosw_init();
1294 goto err_protosw_init;
1296 status = sctp_v6_protosw_init();
1298 goto err_v6_protosw_init;
1300 /* Initialize the control inode/socket for handling OOTB packets. */
1301 if ((status = sctp_ctl_sock_init())) {
1303 "SCTP: Failed to initialize the SCTP control sock.\n");
1304 goto err_ctl_sock_init;
1307 status = sctp_v4_add_protocol();
1309 goto err_add_protocol;
1311 /* Register SCTP with inet6 layer. */
1312 status = sctp_v6_add_protocol();
1314 goto err_v6_add_protocol;
1319 err_v6_add_protocol:
1320 sctp_v6_del_protocol();
1322 sctp_v4_del_protocol();
1323 inet_ctl_sock_destroy(sctp_ctl_sock);
1325 sctp_v6_protosw_exit();
1326 err_v6_protosw_init:
1327 sctp_v4_protosw_exit();
1329 sctp_free_local_addr_list();
1332 sctp_sysctl_unregister();
1333 list_del(&sctp_af_inet.list);
1334 free_pages((unsigned long)sctp_port_hashtable,
1335 get_order(sctp_port_hashsize *
1336 sizeof(struct sctp_bind_hashbucket)));
1338 kfree(sctp_ep_hashtable);
1340 free_pages((unsigned long)sctp_assoc_hashtable,
1341 get_order(sctp_assoc_hashsize *
1342 sizeof(struct sctp_hashbucket)));
1344 sctp_dbg_objcnt_exit();
1347 cleanup_sctp_mibs();
1349 kmem_cache_destroy(sctp_chunk_cachep);
1351 kmem_cache_destroy(sctp_bucket_cachep);
1355 /* Exit handler for the SCTP protocol. */
1356 SCTP_STATIC __exit void sctp_exit(void)
1358 /* BUG. This should probably do something useful like clean
1359 * up all the remaining associations and all that memory.
1362 /* Unregister with inet6/inet layers. */
1363 sctp_v6_del_protocol();
1364 sctp_v4_del_protocol();
1366 /* Free the control endpoint. */
1367 inet_ctl_sock_destroy(sctp_ctl_sock);
1369 /* Free protosw registrations */
1370 sctp_v6_protosw_exit();
1371 sctp_v4_protosw_exit();
1373 /* Free the local address list. */
1374 sctp_free_local_addr_list();
1376 /* Unregister with socket layer. */
1380 sctp_sysctl_unregister();
1381 list_del(&sctp_af_inet.list);
1383 free_pages((unsigned long)sctp_assoc_hashtable,
1384 get_order(sctp_assoc_hashsize *
1385 sizeof(struct sctp_hashbucket)));
1386 kfree(sctp_ep_hashtable);
1387 free_pages((unsigned long)sctp_port_hashtable,
1388 get_order(sctp_port_hashsize *
1389 sizeof(struct sctp_bind_hashbucket)));
1391 sctp_dbg_objcnt_exit();
1393 cleanup_sctp_mibs();
1395 kmem_cache_destroy(sctp_chunk_cachep);
1396 kmem_cache_destroy(sctp_bucket_cachep);
1399 module_init(sctp_init);
1400 module_exit(sctp_exit);
1403 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1405 MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132");
1406 MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132");
1407 MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1408 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1409 MODULE_LICENSE("GPL");