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 if (percpu_counter_init(&sctp_sockets_allocated
, 0))
107 #ifdef CONFIG_PROC_FS
108 if (!proc_net_sctp
) {
109 struct proc_dir_entry
*ent
;
110 ent
= proc_mkdir("sctp", init_net
.proc_net
);
112 ent
->owner
= THIS_MODULE
;
115 goto out_free_percpu
;
118 if (sctp_snmp_proc_init())
119 goto out_snmp_proc_init
;
120 if (sctp_eps_proc_init())
121 goto out_eps_proc_init
;
122 if (sctp_assocs_proc_init())
123 goto out_assocs_proc_init
;
124 if (sctp_remaddr_proc_init())
125 goto out_remaddr_proc_init
;
129 out_remaddr_proc_init
:
130 sctp_assocs_proc_exit();
131 out_assocs_proc_init
:
132 sctp_eps_proc_exit();
134 sctp_snmp_proc_exit();
137 proc_net_sctp
= NULL
;
138 remove_proc_entry("sctp", init_net
.proc_net
);
141 percpu_counter_destroy(&sctp_sockets_allocated
);
144 #endif /* CONFIG_PROC_FS */
150 /* Clean up the proc fs entry for the SCTP protocol.
151 * Note: Do not make this __exit as it is used in the init error
154 static void sctp_proc_exit(void)
156 #ifdef CONFIG_PROC_FS
157 sctp_snmp_proc_exit();
158 sctp_eps_proc_exit();
159 sctp_assocs_proc_exit();
160 sctp_remaddr_proc_exit();
163 proc_net_sctp
= NULL
;
164 remove_proc_entry("sctp", init_net
.proc_net
);
169 /* Private helper to extract ipv4 address and stash them in
170 * the protocol structure.
172 static void sctp_v4_copy_addrlist(struct list_head
*addrlist
,
173 struct net_device
*dev
)
175 struct in_device
*in_dev
;
176 struct in_ifaddr
*ifa
;
177 struct sctp_sockaddr_entry
*addr
;
180 if ((in_dev
= __in_dev_get_rcu(dev
)) == NULL
) {
185 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
186 /* Add the address to the local list. */
187 addr
= t_new(struct sctp_sockaddr_entry
, GFP_ATOMIC
);
189 addr
->a
.v4
.sin_family
= AF_INET
;
190 addr
->a
.v4
.sin_port
= 0;
191 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
193 INIT_LIST_HEAD(&addr
->list
);
194 INIT_RCU_HEAD(&addr
->rcu
);
195 list_add_tail(&addr
->list
, addrlist
);
202 /* Extract our IP addresses from the system and stash them in the
203 * protocol structure.
205 static void sctp_get_local_addr_list(void)
207 struct net_device
*dev
;
208 struct list_head
*pos
;
211 read_lock(&dev_base_lock
);
212 for_each_netdev(&init_net
, dev
) {
213 __list_for_each(pos
, &sctp_address_families
) {
214 af
= list_entry(pos
, struct sctp_af
, list
);
215 af
->copy_addrlist(&sctp_local_addr_list
, dev
);
218 read_unlock(&dev_base_lock
);
221 /* Free the existing local addresses. */
222 static void sctp_free_local_addr_list(void)
224 struct sctp_sockaddr_entry
*addr
;
225 struct list_head
*pos
, *temp
;
227 list_for_each_safe(pos
, temp
, &sctp_local_addr_list
) {
228 addr
= list_entry(pos
, struct sctp_sockaddr_entry
, list
);
234 void sctp_local_addr_free(struct rcu_head
*head
)
236 struct sctp_sockaddr_entry
*e
= container_of(head
,
237 struct sctp_sockaddr_entry
, rcu
);
241 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
242 int sctp_copy_local_addr_list(struct sctp_bind_addr
*bp
, sctp_scope_t scope
,
243 gfp_t gfp
, int copy_flags
)
245 struct sctp_sockaddr_entry
*addr
;
249 list_for_each_entry_rcu(addr
, &sctp_local_addr_list
, list
) {
252 if (sctp_in_scope(&addr
->a
, scope
)) {
253 /* Now that the address is in scope, check to see if
254 * the address type is really supported by the local
255 * sock as well as the remote peer.
257 if ((((AF_INET
== addr
->a
.sa
.sa_family
) &&
258 (copy_flags
& SCTP_ADDR4_PEERSUPP
))) ||
259 (((AF_INET6
== addr
->a
.sa
.sa_family
) &&
260 (copy_flags
& SCTP_ADDR6_ALLOWED
) &&
261 (copy_flags
& SCTP_ADDR6_PEERSUPP
)))) {
262 error
= sctp_add_bind_addr(bp
, &addr
->a
,
263 SCTP_ADDR_SRC
, GFP_ATOMIC
);
275 /* Initialize a sctp_addr from in incoming skb. */
276 static void sctp_v4_from_skb(union sctp_addr
*addr
, struct sk_buff
*skb
,
283 port
= &addr
->v4
.sin_port
;
284 addr
->v4
.sin_family
= AF_INET
;
289 from
= &ip_hdr(skb
)->saddr
;
292 from
= &ip_hdr(skb
)->daddr
;
294 memcpy(&addr
->v4
.sin_addr
.s_addr
, from
, sizeof(struct in_addr
));
297 /* Initialize an sctp_addr from a socket. */
298 static void sctp_v4_from_sk(union sctp_addr
*addr
, struct sock
*sk
)
300 addr
->v4
.sin_family
= AF_INET
;
301 addr
->v4
.sin_port
= 0;
302 addr
->v4
.sin_addr
.s_addr
= inet_sk(sk
)->rcv_saddr
;
305 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
306 static void sctp_v4_to_sk_saddr(union sctp_addr
*addr
, struct sock
*sk
)
308 inet_sk(sk
)->rcv_saddr
= addr
->v4
.sin_addr
.s_addr
;
311 /* Initialize sk->sk_daddr from sctp_addr. */
312 static void sctp_v4_to_sk_daddr(union sctp_addr
*addr
, struct sock
*sk
)
314 inet_sk(sk
)->daddr
= addr
->v4
.sin_addr
.s_addr
;
317 /* Initialize a sctp_addr from an address parameter. */
318 static void sctp_v4_from_addr_param(union sctp_addr
*addr
,
319 union sctp_addr_param
*param
,
320 __be16 port
, int iif
)
322 addr
->v4
.sin_family
= AF_INET
;
323 addr
->v4
.sin_port
= port
;
324 addr
->v4
.sin_addr
.s_addr
= param
->v4
.addr
.s_addr
;
327 /* Initialize an address parameter from a sctp_addr and return the length
328 * of the address parameter.
330 static int sctp_v4_to_addr_param(const union sctp_addr
*addr
,
331 union sctp_addr_param
*param
)
333 int length
= sizeof(sctp_ipv4addr_param_t
);
335 param
->v4
.param_hdr
.type
= SCTP_PARAM_IPV4_ADDRESS
;
336 param
->v4
.param_hdr
.length
= htons(length
);
337 param
->v4
.addr
.s_addr
= addr
->v4
.sin_addr
.s_addr
;
342 /* Initialize a sctp_addr from a dst_entry. */
343 static void sctp_v4_dst_saddr(union sctp_addr
*saddr
, struct dst_entry
*dst
,
346 struct rtable
*rt
= (struct rtable
*)dst
;
347 saddr
->v4
.sin_family
= AF_INET
;
348 saddr
->v4
.sin_port
= port
;
349 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
352 /* Compare two addresses exactly. */
353 static int sctp_v4_cmp_addr(const union sctp_addr
*addr1
,
354 const union sctp_addr
*addr2
)
356 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
358 if (addr1
->v4
.sin_port
!= addr2
->v4
.sin_port
)
360 if (addr1
->v4
.sin_addr
.s_addr
!= addr2
->v4
.sin_addr
.s_addr
)
366 /* Initialize addr struct to INADDR_ANY. */
367 static void sctp_v4_inaddr_any(union sctp_addr
*addr
, __be16 port
)
369 addr
->v4
.sin_family
= AF_INET
;
370 addr
->v4
.sin_addr
.s_addr
= htonl(INADDR_ANY
);
371 addr
->v4
.sin_port
= port
;
374 /* Is this a wildcard address? */
375 static int sctp_v4_is_any(const union sctp_addr
*addr
)
377 return htonl(INADDR_ANY
) == addr
->v4
.sin_addr
.s_addr
;
380 /* This function checks if the address is a valid address to be used for
384 * Return 0 - If the address is a non-unicast or an illegal address.
385 * Return 1 - If the address is a unicast.
387 static int sctp_v4_addr_valid(union sctp_addr
*addr
,
388 struct sctp_sock
*sp
,
389 const struct sk_buff
*skb
)
391 /* IPv4 addresses not allowed */
392 if (sp
&& ipv6_only_sock(sctp_opt2sk(sp
)))
395 /* Is this a non-unicast address or a unusable SCTP address? */
396 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
))
399 /* Is this a broadcast address? */
400 if (skb
&& skb
->rtable
->rt_flags
& RTCF_BROADCAST
)
406 /* Should this be available for binding? */
407 static int sctp_v4_available(union sctp_addr
*addr
, struct sctp_sock
*sp
)
409 int ret
= inet_addr_type(&init_net
, addr
->v4
.sin_addr
.s_addr
);
412 if (addr
->v4
.sin_addr
.s_addr
!= htonl(INADDR_ANY
) &&
414 !sp
->inet
.freebind
&&
415 !sysctl_ip_nonlocal_bind
)
418 if (ipv6_only_sock(sctp_opt2sk(sp
)))
424 /* Checking the loopback, private and other address scopes as defined in
425 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
426 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
428 * Level 0 - unusable SCTP addresses
429 * Level 1 - loopback address
430 * Level 2 - link-local addresses
431 * Level 3 - private addresses.
432 * Level 4 - global addresses
433 * For INIT and INIT-ACK address list, let L be the level of
434 * of requested destination address, sender and receiver
435 * SHOULD include all of its addresses with level greater
436 * than or equal to L.
438 static sctp_scope_t
sctp_v4_scope(union sctp_addr
*addr
)
442 /* Should IPv4 scoping be a sysctl configurable option
443 * so users can turn it off (default on) for certain
444 * unconventional networking environments?
447 /* Check for unusable SCTP addresses. */
448 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
)) {
449 retval
= SCTP_SCOPE_UNUSABLE
;
450 } else if (ipv4_is_loopback(addr
->v4
.sin_addr
.s_addr
)) {
451 retval
= SCTP_SCOPE_LOOPBACK
;
452 } else if (ipv4_is_linklocal_169(addr
->v4
.sin_addr
.s_addr
)) {
453 retval
= SCTP_SCOPE_LINK
;
454 } else if (ipv4_is_private_10(addr
->v4
.sin_addr
.s_addr
) ||
455 ipv4_is_private_172(addr
->v4
.sin_addr
.s_addr
) ||
456 ipv4_is_private_192(addr
->v4
.sin_addr
.s_addr
)) {
457 retval
= SCTP_SCOPE_PRIVATE
;
459 retval
= SCTP_SCOPE_GLOBAL
;
465 /* Returns a valid dst cache entry for the given source and destination ip
466 * addresses. If an association is passed, trys to get a dst entry with a
467 * source address that matches an address in the bind address list.
469 static struct dst_entry
*sctp_v4_get_dst(struct sctp_association
*asoc
,
470 union sctp_addr
*daddr
,
471 union sctp_addr
*saddr
)
475 struct sctp_bind_addr
*bp
;
476 struct sctp_sockaddr_entry
*laddr
;
477 struct dst_entry
*dst
= NULL
;
478 union sctp_addr dst_saddr
;
480 memset(&fl
, 0x0, sizeof(struct flowi
));
481 fl
.fl4_dst
= daddr
->v4
.sin_addr
.s_addr
;
482 fl
.proto
= IPPROTO_SCTP
;
484 fl
.fl4_tos
= RT_CONN_FLAGS(asoc
->base
.sk
);
485 fl
.oif
= asoc
->base
.sk
->sk_bound_dev_if
;
488 fl
.fl4_src
= saddr
->v4
.sin_addr
.s_addr
;
490 SCTP_DEBUG_PRINTK("%s: DST:%pI4, SRC:%pI4 - ",
491 __func__
, &fl
.fl4_dst
, &fl
.fl4_src
);
493 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
497 /* If there is no association or if a source address is passed, no
498 * more validation is required.
503 bp
= &asoc
->base
.bind_addr
;
506 /* Walk through the bind address list and look for a bind
507 * address that matches the source address of the returned dst.
509 sctp_v4_dst_saddr(&dst_saddr
, dst
, htons(bp
->port
));
511 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
512 if (!laddr
->valid
|| (laddr
->state
!= SCTP_ADDR_SRC
))
514 if (sctp_v4_cmp_addr(&dst_saddr
, &laddr
->a
))
519 /* None of the bound addresses match the source address of the
520 * dst. So release it.
526 /* Walk through the bind address list and try to get a dst that
527 * matches a bind address as the source address.
530 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
533 if ((laddr
->state
== SCTP_ADDR_SRC
) &&
534 (AF_INET
== laddr
->a
.sa
.sa_family
)) {
535 fl
.fl4_src
= laddr
->a
.v4
.sin_addr
.s_addr
;
536 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
547 SCTP_DEBUG_PRINTK("rt_dst:%pI4, rt_src:%pI4\n",
548 &rt
->rt_dst
, &rt
->rt_src
);
550 SCTP_DEBUG_PRINTK("NO ROUTE\n");
555 /* For v4, the source address is cached in the route entry(dst). So no need
556 * to cache it separately and hence this is an empty routine.
558 static void sctp_v4_get_saddr(struct sctp_sock
*sk
,
559 struct sctp_association
*asoc
,
560 struct dst_entry
*dst
,
561 union sctp_addr
*daddr
,
562 union sctp_addr
*saddr
)
564 struct rtable
*rt
= (struct rtable
*)dst
;
570 saddr
->v4
.sin_family
= AF_INET
;
571 saddr
->v4
.sin_port
= htons(asoc
->base
.bind_addr
.port
);
572 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
576 /* What interface did this skb arrive on? */
577 static int sctp_v4_skb_iif(const struct sk_buff
*skb
)
579 return skb
->rtable
->rt_iif
;
582 /* Was this packet marked by Explicit Congestion Notification? */
583 static int sctp_v4_is_ce(const struct sk_buff
*skb
)
585 return INET_ECN_is_ce(ip_hdr(skb
)->tos
);
588 /* Create and initialize a new sk for the socket returned by accept(). */
589 static struct sock
*sctp_v4_create_accept_sk(struct sock
*sk
,
590 struct sctp_association
*asoc
)
592 struct sock
*newsk
= sk_alloc(sock_net(sk
), PF_INET
, GFP_KERNEL
,
594 struct inet_sock
*newinet
;
599 sock_init_data(NULL
, newsk
);
601 sctp_copy_sock(newsk
, sk
, asoc
);
602 sock_reset_flag(newsk
, SOCK_ZAPPED
);
604 newinet
= inet_sk(newsk
);
606 newinet
->daddr
= asoc
->peer
.primary_addr
.v4
.sin_addr
.s_addr
;
608 sk_refcnt_debug_inc(newsk
);
610 if (newsk
->sk_prot
->init(newsk
)) {
611 sk_common_release(newsk
);
619 /* Map address, empty for v4 family */
620 static void sctp_v4_addr_v4map(struct sctp_sock
*sp
, union sctp_addr
*addr
)
625 /* Dump the v4 addr to the seq file. */
626 static void sctp_v4_seq_dump_addr(struct seq_file
*seq
, union sctp_addr
*addr
)
628 seq_printf(seq
, "%pI4 ", &addr
->v4
.sin_addr
);
631 static void sctp_v4_ecn_capable(struct sock
*sk
)
636 /* Event handler for inet address addition/deletion events.
637 * The sctp_local_addr_list needs to be protocted by a spin lock since
638 * multiple notifiers (say IPv4 and IPv6) may be running at the same
639 * time and thus corrupt the list.
640 * The reader side is protected with RCU.
642 static int sctp_inetaddr_event(struct notifier_block
*this, unsigned long ev
,
645 struct in_ifaddr
*ifa
= (struct in_ifaddr
*)ptr
;
646 struct sctp_sockaddr_entry
*addr
= NULL
;
647 struct sctp_sockaddr_entry
*temp
;
650 if (!net_eq(dev_net(ifa
->ifa_dev
->dev
), &init_net
))
655 addr
= kmalloc(sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
657 addr
->a
.v4
.sin_family
= AF_INET
;
658 addr
->a
.v4
.sin_port
= 0;
659 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
661 spin_lock_bh(&sctp_local_addr_lock
);
662 list_add_tail_rcu(&addr
->list
, &sctp_local_addr_list
);
663 spin_unlock_bh(&sctp_local_addr_lock
);
667 spin_lock_bh(&sctp_local_addr_lock
);
668 list_for_each_entry_safe(addr
, temp
,
669 &sctp_local_addr_list
, list
) {
670 if (addr
->a
.sa
.sa_family
== AF_INET
&&
671 addr
->a
.v4
.sin_addr
.s_addr
==
675 list_del_rcu(&addr
->list
);
679 spin_unlock_bh(&sctp_local_addr_lock
);
681 call_rcu(&addr
->rcu
, sctp_local_addr_free
);
689 * Initialize the control inode/socket with a control endpoint data
690 * structure. This endpoint is reserved exclusively for the OOTB processing.
692 static int sctp_ctl_sock_init(void)
695 sa_family_t family
= PF_INET
;
697 if (sctp_get_pf_specific(PF_INET6
))
700 err
= inet_ctl_sock_create(&sctp_ctl_sock
, family
,
701 SOCK_SEQPACKET
, IPPROTO_SCTP
, &init_net
);
703 /* If IPv6 socket could not be created, try the IPv4 socket */
704 if (err
< 0 && family
== PF_INET6
)
705 err
= inet_ctl_sock_create(&sctp_ctl_sock
, AF_INET
,
706 SOCK_SEQPACKET
, IPPROTO_SCTP
,
711 "SCTP: Failed to create the SCTP control socket.\n");
717 /* Register address family specific functions. */
718 int sctp_register_af(struct sctp_af
*af
)
720 switch (af
->sa_family
) {
722 if (sctp_af_v4_specific
)
724 sctp_af_v4_specific
= af
;
727 if (sctp_af_v6_specific
)
729 sctp_af_v6_specific
= af
;
735 INIT_LIST_HEAD(&af
->list
);
736 list_add_tail(&af
->list
, &sctp_address_families
);
740 /* Get the table of functions for manipulating a particular address
743 struct sctp_af
*sctp_get_af_specific(sa_family_t family
)
747 return sctp_af_v4_specific
;
749 return sctp_af_v6_specific
;
755 /* Common code to initialize a AF_INET msg_name. */
756 static void sctp_inet_msgname(char *msgname
, int *addr_len
)
758 struct sockaddr_in
*sin
;
760 sin
= (struct sockaddr_in
*)msgname
;
761 *addr_len
= sizeof(struct sockaddr_in
);
762 sin
->sin_family
= AF_INET
;
763 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
766 /* Copy the primary address of the peer primary address as the msg_name. */
767 static void sctp_inet_event_msgname(struct sctp_ulpevent
*event
, char *msgname
,
770 struct sockaddr_in
*sin
, *sinfrom
;
773 struct sctp_association
*asoc
;
776 sctp_inet_msgname(msgname
, addr_len
);
777 sin
= (struct sockaddr_in
*)msgname
;
778 sinfrom
= &asoc
->peer
.primary_addr
.v4
;
779 sin
->sin_port
= htons(asoc
->peer
.port
);
780 sin
->sin_addr
.s_addr
= sinfrom
->sin_addr
.s_addr
;
784 /* Initialize and copy out a msgname from an inbound skb. */
785 static void sctp_inet_skb_msgname(struct sk_buff
*skb
, char *msgname
, int *len
)
788 struct sctphdr
*sh
= sctp_hdr(skb
);
789 struct sockaddr_in
*sin
= (struct sockaddr_in
*)msgname
;
791 sctp_inet_msgname(msgname
, len
);
792 sin
->sin_port
= sh
->source
;
793 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
797 /* Do we support this AF? */
798 static int sctp_inet_af_supported(sa_family_t family
, struct sctp_sock
*sp
)
800 /* PF_INET only supports AF_INET addresses. */
801 return (AF_INET
== family
);
804 /* Address matching with wildcards allowed. */
805 static int sctp_inet_cmp_addr(const union sctp_addr
*addr1
,
806 const union sctp_addr
*addr2
,
807 struct sctp_sock
*opt
)
809 /* PF_INET only supports AF_INET addresses. */
810 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
812 if (htonl(INADDR_ANY
) == addr1
->v4
.sin_addr
.s_addr
||
813 htonl(INADDR_ANY
) == addr2
->v4
.sin_addr
.s_addr
)
815 if (addr1
->v4
.sin_addr
.s_addr
== addr2
->v4
.sin_addr
.s_addr
)
821 /* Verify that provided sockaddr looks bindable. Common verification has
822 * already been taken care of.
824 static int sctp_inet_bind_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
826 return sctp_v4_available(addr
, opt
);
829 /* Verify that sockaddr looks sendable. Common verification has already
830 * been taken care of.
832 static int sctp_inet_send_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
837 /* Fill in Supported Address Type information for INIT and INIT-ACK
838 * chunks. Returns number of addresses supported.
840 static int sctp_inet_supported_addrs(const struct sctp_sock
*opt
,
843 types
[0] = SCTP_PARAM_IPV4_ADDRESS
;
847 /* Wrapper routine that calls the ip transmit routine. */
848 static inline int sctp_v4_xmit(struct sk_buff
*skb
,
849 struct sctp_transport
*transport
)
851 struct inet_sock
*inet
= inet_sk(skb
->sk
);
853 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n",
854 __func__
, skb
, skb
->len
,
855 &skb
->rtable
->rt_src
,
856 &skb
->rtable
->rt_dst
);
858 inet
->pmtudisc
= transport
->param_flags
& SPP_PMTUD_ENABLE
?
859 IP_PMTUDISC_DO
: IP_PMTUDISC_DONT
;
861 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS
);
862 return ip_queue_xmit(skb
, 0);
865 static struct sctp_af sctp_af_inet
;
867 static struct sctp_pf sctp_pf_inet
= {
868 .event_msgname
= sctp_inet_event_msgname
,
869 .skb_msgname
= sctp_inet_skb_msgname
,
870 .af_supported
= sctp_inet_af_supported
,
871 .cmp_addr
= sctp_inet_cmp_addr
,
872 .bind_verify
= sctp_inet_bind_verify
,
873 .send_verify
= sctp_inet_send_verify
,
874 .supported_addrs
= sctp_inet_supported_addrs
,
875 .create_accept_sk
= sctp_v4_create_accept_sk
,
876 .addr_v4map
= sctp_v4_addr_v4map
,
880 /* Notifier for inetaddr addition/deletion events. */
881 static struct notifier_block sctp_inetaddr_notifier
= {
882 .notifier_call
= sctp_inetaddr_event
,
885 /* Socket operations. */
886 static const struct proto_ops inet_seqpacket_ops
= {
888 .owner
= THIS_MODULE
,
889 .release
= inet_release
, /* Needs to be wrapped... */
891 .connect
= inet_dgram_connect
,
892 .socketpair
= sock_no_socketpair
,
893 .accept
= inet_accept
,
894 .getname
= inet_getname
, /* Semantics are different. */
897 .listen
= sctp_inet_listen
,
898 .shutdown
= inet_shutdown
, /* Looks harmless. */
899 .setsockopt
= sock_common_setsockopt
, /* IP_SOL IP_OPTION is a problem */
900 .getsockopt
= sock_common_getsockopt
,
901 .sendmsg
= inet_sendmsg
,
902 .recvmsg
= sock_common_recvmsg
,
903 .mmap
= sock_no_mmap
,
904 .sendpage
= sock_no_sendpage
,
906 .compat_setsockopt
= compat_sock_common_setsockopt
,
907 .compat_getsockopt
= compat_sock_common_getsockopt
,
911 /* Registration with AF_INET family. */
912 static struct inet_protosw sctp_seqpacket_protosw
= {
913 .type
= SOCK_SEQPACKET
,
914 .protocol
= IPPROTO_SCTP
,
916 .ops
= &inet_seqpacket_ops
,
919 .flags
= SCTP_PROTOSW_FLAG
921 static struct inet_protosw sctp_stream_protosw
= {
923 .protocol
= IPPROTO_SCTP
,
925 .ops
= &inet_seqpacket_ops
,
928 .flags
= SCTP_PROTOSW_FLAG
931 /* Register with IP layer. */
932 static struct net_protocol sctp_protocol
= {
934 .err_handler
= sctp_v4_err
,
938 /* IPv4 address related functions. */
939 static struct sctp_af sctp_af_inet
= {
940 .sa_family
= AF_INET
,
941 .sctp_xmit
= sctp_v4_xmit
,
942 .setsockopt
= ip_setsockopt
,
943 .getsockopt
= ip_getsockopt
,
944 .get_dst
= sctp_v4_get_dst
,
945 .get_saddr
= sctp_v4_get_saddr
,
946 .copy_addrlist
= sctp_v4_copy_addrlist
,
947 .from_skb
= sctp_v4_from_skb
,
948 .from_sk
= sctp_v4_from_sk
,
949 .to_sk_saddr
= sctp_v4_to_sk_saddr
,
950 .to_sk_daddr
= sctp_v4_to_sk_daddr
,
951 .from_addr_param
= sctp_v4_from_addr_param
,
952 .to_addr_param
= sctp_v4_to_addr_param
,
953 .dst_saddr
= sctp_v4_dst_saddr
,
954 .cmp_addr
= sctp_v4_cmp_addr
,
955 .addr_valid
= sctp_v4_addr_valid
,
956 .inaddr_any
= sctp_v4_inaddr_any
,
957 .is_any
= sctp_v4_is_any
,
958 .available
= sctp_v4_available
,
959 .scope
= sctp_v4_scope
,
960 .skb_iif
= sctp_v4_skb_iif
,
961 .is_ce
= sctp_v4_is_ce
,
962 .seq_dump_addr
= sctp_v4_seq_dump_addr
,
963 .ecn_capable
= sctp_v4_ecn_capable
,
964 .net_header_len
= sizeof(struct iphdr
),
965 .sockaddr_len
= sizeof(struct sockaddr_in
),
967 .compat_setsockopt
= compat_ip_setsockopt
,
968 .compat_getsockopt
= compat_ip_getsockopt
,
972 struct sctp_pf
*sctp_get_pf_specific(sa_family_t family
) {
976 return sctp_pf_inet_specific
;
978 return sctp_pf_inet6_specific
;
984 /* Register the PF specific function table. */
985 int sctp_register_pf(struct sctp_pf
*pf
, sa_family_t family
)
989 if (sctp_pf_inet_specific
)
991 sctp_pf_inet_specific
= pf
;
994 if (sctp_pf_inet6_specific
)
996 sctp_pf_inet6_specific
= pf
;
1004 static inline int init_sctp_mibs(void)
1006 return snmp_mib_init((void**)sctp_statistics
, sizeof(struct sctp_mib
));
1009 static inline void cleanup_sctp_mibs(void)
1011 snmp_mib_free((void**)sctp_statistics
);
1014 static void sctp_v4_pf_init(void)
1016 /* Initialize the SCTP specific PF functions. */
1017 sctp_register_pf(&sctp_pf_inet
, PF_INET
);
1018 sctp_register_af(&sctp_af_inet
);
1021 static void sctp_v4_pf_exit(void)
1023 list_del(&sctp_af_inet
.list
);
1026 static int sctp_v4_protosw_init(void)
1030 rc
= proto_register(&sctp_prot
, 1);
1034 /* Register SCTP(UDP and TCP style) with socket layer. */
1035 inet_register_protosw(&sctp_seqpacket_protosw
);
1036 inet_register_protosw(&sctp_stream_protosw
);
1041 static void sctp_v4_protosw_exit(void)
1043 inet_unregister_protosw(&sctp_stream_protosw
);
1044 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1045 proto_unregister(&sctp_prot
);
1048 static int sctp_v4_add_protocol(void)
1050 /* Register notifier for inet address additions/deletions. */
1051 register_inetaddr_notifier(&sctp_inetaddr_notifier
);
1053 /* Register SCTP with inet layer. */
1054 if (inet_add_protocol(&sctp_protocol
, IPPROTO_SCTP
) < 0)
1060 static void sctp_v4_del_protocol(void)
1062 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1063 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1066 /* Initialize the universe into something sensible. */
1067 SCTP_STATIC __init
int sctp_init(void)
1070 int status
= -EINVAL
;
1072 unsigned long limit
;
1073 unsigned long nr_pages
;
1077 /* SCTP_DEBUG sanity check. */
1078 if (!sctp_sanity_check())
1081 /* Allocate bind_bucket and chunk caches. */
1083 sctp_bucket_cachep
= kmem_cache_create("sctp_bind_bucket",
1084 sizeof(struct sctp_bind_bucket
),
1085 0, SLAB_HWCACHE_ALIGN
,
1087 if (!sctp_bucket_cachep
)
1090 sctp_chunk_cachep
= kmem_cache_create("sctp_chunk",
1091 sizeof(struct sctp_chunk
),
1092 0, SLAB_HWCACHE_ALIGN
,
1094 if (!sctp_chunk_cachep
)
1095 goto err_chunk_cachep
;
1097 /* Allocate and initialise sctp mibs. */
1098 status
= init_sctp_mibs();
1102 /* Initialize proc fs directory. */
1103 status
= sctp_proc_init();
1107 /* Initialize object count debugging. */
1108 sctp_dbg_objcnt_init();
1111 * 14. Suggested SCTP Protocol Parameter Values
1113 /* The following protocol parameters are RECOMMENDED: */
1114 /* RTO.Initial - 3 seconds */
1115 sctp_rto_initial
= SCTP_RTO_INITIAL
;
1116 /* RTO.Min - 1 second */
1117 sctp_rto_min
= SCTP_RTO_MIN
;
1118 /* RTO.Max - 60 seconds */
1119 sctp_rto_max
= SCTP_RTO_MAX
;
1120 /* RTO.Alpha - 1/8 */
1121 sctp_rto_alpha
= SCTP_RTO_ALPHA
;
1122 /* RTO.Beta - 1/4 */
1123 sctp_rto_beta
= SCTP_RTO_BETA
;
1125 /* Valid.Cookie.Life - 60 seconds */
1126 sctp_valid_cookie_life
= SCTP_DEFAULT_COOKIE_LIFE
;
1128 /* Whether Cookie Preservative is enabled(1) or not(0) */
1129 sctp_cookie_preserve_enable
= 1;
1132 sctp_max_burst
= SCTP_DEFAULT_MAX_BURST
;
1134 /* Association.Max.Retrans - 10 attempts
1135 * Path.Max.Retrans - 5 attempts (per destination address)
1136 * Max.Init.Retransmits - 8 attempts
1138 sctp_max_retrans_association
= 10;
1139 sctp_max_retrans_path
= 5;
1140 sctp_max_retrans_init
= 8;
1142 /* Sendbuffer growth - do per-socket accounting */
1143 sctp_sndbuf_policy
= 0;
1145 /* Rcvbuffer growth - do per-socket accounting */
1146 sctp_rcvbuf_policy
= 0;
1148 /* HB.interval - 30 seconds */
1149 sctp_hb_interval
= SCTP_DEFAULT_TIMEOUT_HEARTBEAT
;
1151 /* delayed SACK timeout */
1152 sctp_sack_timeout
= SCTP_DEFAULT_TIMEOUT_SACK
;
1154 /* Implementation specific variables. */
1156 /* Initialize default stream count setup information. */
1157 sctp_max_instreams
= SCTP_DEFAULT_INSTREAMS
;
1158 sctp_max_outstreams
= SCTP_DEFAULT_OUTSTREAMS
;
1160 /* Initialize handle used for association ids. */
1161 idr_init(&sctp_assocs_id
);
1163 /* Set the pressure threshold to be a fraction of global memory that
1164 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1165 * memory, with a floor of 128 pages.
1166 * Note this initalizes the data in sctpv6_prot too
1167 * Unabashedly stolen from tcp_init
1169 nr_pages
= totalram_pages
- totalhigh_pages
;
1170 limit
= min(nr_pages
, 1UL<<(28-PAGE_SHIFT
)) >> (20-PAGE_SHIFT
);
1171 limit
= (limit
* (nr_pages
>> (20-PAGE_SHIFT
))) >> (PAGE_SHIFT
-11);
1172 limit
= max(limit
, 128UL);
1173 sysctl_sctp_mem
[0] = limit
/ 4 * 3;
1174 sysctl_sctp_mem
[1] = limit
;
1175 sysctl_sctp_mem
[2] = sysctl_sctp_mem
[0] * 2;
1177 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1178 limit
= (sysctl_sctp_mem
[1]) << (PAGE_SHIFT
- 7);
1179 max_share
= min(4UL*1024*1024, limit
);
1181 sysctl_sctp_rmem
[0] = SK_MEM_QUANTUM
; /* give each asoc 1 page min */
1182 sysctl_sctp_rmem
[1] = (1500 *(sizeof(struct sk_buff
) + 1));
1183 sysctl_sctp_rmem
[2] = max(sysctl_sctp_rmem
[1], max_share
);
1185 sysctl_sctp_wmem
[0] = SK_MEM_QUANTUM
;
1186 sysctl_sctp_wmem
[1] = 16*1024;
1187 sysctl_sctp_wmem
[2] = max(64*1024, max_share
);
1189 /* Size and allocate the association hash table.
1190 * The methodology is similar to that of the tcp hash tables.
1192 if (num_physpages
>= (128 * 1024))
1193 goal
= num_physpages
>> (22 - PAGE_SHIFT
);
1195 goal
= num_physpages
>> (24 - PAGE_SHIFT
);
1197 for (order
= 0; (1UL << order
) < goal
; order
++)
1201 sctp_assoc_hashsize
= (1UL << order
) * PAGE_SIZE
/
1202 sizeof(struct sctp_hashbucket
);
1203 if ((sctp_assoc_hashsize
> (64 * 1024)) && order
> 0)
1205 sctp_assoc_hashtable
= (struct sctp_hashbucket
*)
1206 __get_free_pages(GFP_ATOMIC
, order
);
1207 } while (!sctp_assoc_hashtable
&& --order
> 0);
1208 if (!sctp_assoc_hashtable
) {
1209 printk(KERN_ERR
"SCTP: Failed association hash alloc.\n");
1211 goto err_ahash_alloc
;
1213 for (i
= 0; i
< sctp_assoc_hashsize
; i
++) {
1214 rwlock_init(&sctp_assoc_hashtable
[i
].lock
);
1215 INIT_HLIST_HEAD(&sctp_assoc_hashtable
[i
].chain
);
1218 /* Allocate and initialize the endpoint hash table. */
1219 sctp_ep_hashsize
= 64;
1220 sctp_ep_hashtable
= (struct sctp_hashbucket
*)
1221 kmalloc(64 * sizeof(struct sctp_hashbucket
), GFP_KERNEL
);
1222 if (!sctp_ep_hashtable
) {
1223 printk(KERN_ERR
"SCTP: Failed endpoint_hash alloc.\n");
1225 goto err_ehash_alloc
;
1227 for (i
= 0; i
< sctp_ep_hashsize
; i
++) {
1228 rwlock_init(&sctp_ep_hashtable
[i
].lock
);
1229 INIT_HLIST_HEAD(&sctp_ep_hashtable
[i
].chain
);
1232 /* Allocate and initialize the SCTP port hash table. */
1234 sctp_port_hashsize
= (1UL << order
) * PAGE_SIZE
/
1235 sizeof(struct sctp_bind_hashbucket
);
1236 if ((sctp_port_hashsize
> (64 * 1024)) && order
> 0)
1238 sctp_port_hashtable
= (struct sctp_bind_hashbucket
*)
1239 __get_free_pages(GFP_ATOMIC
, order
);
1240 } while (!sctp_port_hashtable
&& --order
> 0);
1241 if (!sctp_port_hashtable
) {
1242 printk(KERN_ERR
"SCTP: Failed bind hash alloc.");
1244 goto err_bhash_alloc
;
1246 for (i
= 0; i
< sctp_port_hashsize
; i
++) {
1247 spin_lock_init(&sctp_port_hashtable
[i
].lock
);
1248 INIT_HLIST_HEAD(&sctp_port_hashtable
[i
].chain
);
1251 printk(KERN_INFO
"SCTP: Hash tables configured "
1252 "(established %d bind %d)\n",
1253 sctp_assoc_hashsize
, sctp_port_hashsize
);
1255 /* Disable ADDIP by default. */
1256 sctp_addip_enable
= 0;
1257 sctp_addip_noauth
= 0;
1259 /* Enable PR-SCTP by default. */
1260 sctp_prsctp_enable
= 1;
1262 /* Disable AUTH by default. */
1263 sctp_auth_enable
= 0;
1265 sctp_sysctl_register();
1267 INIT_LIST_HEAD(&sctp_address_families
);
1271 /* Initialize the local address list. */
1272 INIT_LIST_HEAD(&sctp_local_addr_list
);
1273 spin_lock_init(&sctp_local_addr_lock
);
1274 sctp_get_local_addr_list();
1276 status
= sctp_v4_protosw_init();
1279 goto err_protosw_init
;
1281 status
= sctp_v6_protosw_init();
1283 goto err_v6_protosw_init
;
1285 /* Initialize the control inode/socket for handling OOTB packets. */
1286 if ((status
= sctp_ctl_sock_init())) {
1288 "SCTP: Failed to initialize the SCTP control sock.\n");
1289 goto err_ctl_sock_init
;
1292 status
= sctp_v4_add_protocol();
1294 goto err_add_protocol
;
1296 /* Register SCTP with inet6 layer. */
1297 status
= sctp_v6_add_protocol();
1299 goto err_v6_add_protocol
;
1304 err_v6_add_protocol
:
1305 sctp_v4_del_protocol();
1307 inet_ctl_sock_destroy(sctp_ctl_sock
);
1309 sctp_v6_protosw_exit();
1310 err_v6_protosw_init
:
1311 sctp_v4_protosw_exit();
1313 sctp_free_local_addr_list();
1316 sctp_sysctl_unregister();
1317 free_pages((unsigned long)sctp_port_hashtable
,
1318 get_order(sctp_port_hashsize
*
1319 sizeof(struct sctp_bind_hashbucket
)));
1321 kfree(sctp_ep_hashtable
);
1323 free_pages((unsigned long)sctp_assoc_hashtable
,
1324 get_order(sctp_assoc_hashsize
*
1325 sizeof(struct sctp_hashbucket
)));
1327 sctp_dbg_objcnt_exit();
1330 cleanup_sctp_mibs();
1332 kmem_cache_destroy(sctp_chunk_cachep
);
1334 kmem_cache_destroy(sctp_bucket_cachep
);
1338 /* Exit handler for the SCTP protocol. */
1339 SCTP_STATIC __exit
void sctp_exit(void)
1341 /* BUG. This should probably do something useful like clean
1342 * up all the remaining associations and all that memory.
1345 /* Unregister with inet6/inet layers. */
1346 sctp_v6_del_protocol();
1347 sctp_v4_del_protocol();
1349 /* Free the control endpoint. */
1350 inet_ctl_sock_destroy(sctp_ctl_sock
);
1352 /* Free protosw registrations */
1353 sctp_v6_protosw_exit();
1354 sctp_v4_protosw_exit();
1356 /* Free the local address list. */
1357 sctp_free_local_addr_list();
1359 /* Unregister with socket layer. */
1363 sctp_sysctl_unregister();
1365 free_pages((unsigned long)sctp_assoc_hashtable
,
1366 get_order(sctp_assoc_hashsize
*
1367 sizeof(struct sctp_hashbucket
)));
1368 kfree(sctp_ep_hashtable
);
1369 free_pages((unsigned long)sctp_port_hashtable
,
1370 get_order(sctp_port_hashsize
*
1371 sizeof(struct sctp_bind_hashbucket
)));
1373 sctp_dbg_objcnt_exit();
1375 cleanup_sctp_mibs();
1377 kmem_cache_destroy(sctp_chunk_cachep
);
1378 kmem_cache_destroy(sctp_bucket_cachep
);
1381 module_init(sctp_init
);
1382 module_exit(sctp_exit
);
1385 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
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");