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 struct proc_dir_entry
*proc_net_sctp
;
68 DEFINE_SNMP_STAT(struct sctp_mib
, sctp_statistics
) __read_mostly
;
70 struct idr sctp_assocs_id
;
71 DEFINE_SPINLOCK(sctp_assocs_id_lock
);
73 /* This is the global socket data structure used for responding to
74 * the Out-of-the-blue (OOTB) packets. A control sock will be created
75 * for this socket at the initialization time.
77 static struct sock
*sctp_ctl_sock
;
79 static struct sctp_pf
*sctp_pf_inet6_specific
;
80 static struct sctp_pf
*sctp_pf_inet_specific
;
81 static struct sctp_af
*sctp_af_v4_specific
;
82 static struct sctp_af
*sctp_af_v6_specific
;
84 struct kmem_cache
*sctp_chunk_cachep __read_mostly
;
85 struct kmem_cache
*sctp_bucket_cachep __read_mostly
;
87 int sysctl_sctp_mem
[3];
88 int sysctl_sctp_rmem
[3];
89 int sysctl_sctp_wmem
[3];
91 /* Return the address of the control sock. */
92 struct sock
*sctp_get_ctl_sock(void)
97 /* Set up the proc fs entry for the SCTP protocol. */
98 static __init
int sctp_proc_init(void)
100 if (!proc_net_sctp
) {
101 struct proc_dir_entry
*ent
;
102 ent
= proc_mkdir("sctp", init_net
.proc_net
);
104 ent
->owner
= THIS_MODULE
;
110 if (sctp_snmp_proc_init())
112 if (sctp_eps_proc_init())
114 if (sctp_assocs_proc_init())
116 if (sctp_remaddr_proc_init())
125 /* Clean up the proc fs entry for the SCTP protocol.
126 * Note: Do not make this __exit as it is used in the init error
129 static void sctp_proc_exit(void)
131 sctp_snmp_proc_exit();
132 sctp_eps_proc_exit();
133 sctp_assocs_proc_exit();
134 sctp_remaddr_proc_exit();
137 proc_net_sctp
= NULL
;
138 remove_proc_entry("sctp", init_net
.proc_net
);
142 /* Private helper to extract ipv4 address and stash them in
143 * the protocol structure.
145 static void sctp_v4_copy_addrlist(struct list_head
*addrlist
,
146 struct net_device
*dev
)
148 struct in_device
*in_dev
;
149 struct in_ifaddr
*ifa
;
150 struct sctp_sockaddr_entry
*addr
;
153 if ((in_dev
= __in_dev_get_rcu(dev
)) == NULL
) {
158 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
159 /* Add the address to the local list. */
160 addr
= t_new(struct sctp_sockaddr_entry
, GFP_ATOMIC
);
162 addr
->a
.v4
.sin_family
= AF_INET
;
163 addr
->a
.v4
.sin_port
= 0;
164 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
166 INIT_LIST_HEAD(&addr
->list
);
167 INIT_RCU_HEAD(&addr
->rcu
);
168 list_add_tail(&addr
->list
, addrlist
);
175 /* Extract our IP addresses from the system and stash them in the
176 * protocol structure.
178 static void sctp_get_local_addr_list(void)
180 struct net_device
*dev
;
181 struct list_head
*pos
;
184 read_lock(&dev_base_lock
);
185 for_each_netdev(&init_net
, dev
) {
186 __list_for_each(pos
, &sctp_address_families
) {
187 af
= list_entry(pos
, struct sctp_af
, list
);
188 af
->copy_addrlist(&sctp_local_addr_list
, dev
);
191 read_unlock(&dev_base_lock
);
194 /* Free the existing local addresses. */
195 static void sctp_free_local_addr_list(void)
197 struct sctp_sockaddr_entry
*addr
;
198 struct list_head
*pos
, *temp
;
200 list_for_each_safe(pos
, temp
, &sctp_local_addr_list
) {
201 addr
= list_entry(pos
, struct sctp_sockaddr_entry
, list
);
207 void sctp_local_addr_free(struct rcu_head
*head
)
209 struct sctp_sockaddr_entry
*e
= container_of(head
,
210 struct sctp_sockaddr_entry
, rcu
);
214 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
215 int sctp_copy_local_addr_list(struct sctp_bind_addr
*bp
, sctp_scope_t scope
,
216 gfp_t gfp
, int copy_flags
)
218 struct sctp_sockaddr_entry
*addr
;
222 list_for_each_entry_rcu(addr
, &sctp_local_addr_list
, list
) {
225 if (sctp_in_scope(&addr
->a
, scope
)) {
226 /* Now that the address is in scope, check to see if
227 * the address type is really supported by the local
228 * sock as well as the remote peer.
230 if ((((AF_INET
== addr
->a
.sa
.sa_family
) &&
231 (copy_flags
& SCTP_ADDR4_PEERSUPP
))) ||
232 (((AF_INET6
== addr
->a
.sa
.sa_family
) &&
233 (copy_flags
& SCTP_ADDR6_ALLOWED
) &&
234 (copy_flags
& SCTP_ADDR6_PEERSUPP
)))) {
235 error
= sctp_add_bind_addr(bp
, &addr
->a
,
236 SCTP_ADDR_SRC
, GFP_ATOMIC
);
248 /* Initialize a sctp_addr from in incoming skb. */
249 static void sctp_v4_from_skb(union sctp_addr
*addr
, struct sk_buff
*skb
,
256 port
= &addr
->v4
.sin_port
;
257 addr
->v4
.sin_family
= AF_INET
;
262 from
= &ip_hdr(skb
)->saddr
;
265 from
= &ip_hdr(skb
)->daddr
;
267 memcpy(&addr
->v4
.sin_addr
.s_addr
, from
, sizeof(struct in_addr
));
270 /* Initialize an sctp_addr from a socket. */
271 static void sctp_v4_from_sk(union sctp_addr
*addr
, struct sock
*sk
)
273 addr
->v4
.sin_family
= AF_INET
;
274 addr
->v4
.sin_port
= 0;
275 addr
->v4
.sin_addr
.s_addr
= inet_sk(sk
)->rcv_saddr
;
278 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
279 static void sctp_v4_to_sk_saddr(union sctp_addr
*addr
, struct sock
*sk
)
281 inet_sk(sk
)->rcv_saddr
= addr
->v4
.sin_addr
.s_addr
;
284 /* Initialize sk->sk_daddr from sctp_addr. */
285 static void sctp_v4_to_sk_daddr(union sctp_addr
*addr
, struct sock
*sk
)
287 inet_sk(sk
)->daddr
= addr
->v4
.sin_addr
.s_addr
;
290 /* Initialize a sctp_addr from an address parameter. */
291 static void sctp_v4_from_addr_param(union sctp_addr
*addr
,
292 union sctp_addr_param
*param
,
293 __be16 port
, int iif
)
295 addr
->v4
.sin_family
= AF_INET
;
296 addr
->v4
.sin_port
= port
;
297 addr
->v4
.sin_addr
.s_addr
= param
->v4
.addr
.s_addr
;
300 /* Initialize an address parameter from a sctp_addr and return the length
301 * of the address parameter.
303 static int sctp_v4_to_addr_param(const union sctp_addr
*addr
,
304 union sctp_addr_param
*param
)
306 int length
= sizeof(sctp_ipv4addr_param_t
);
308 param
->v4
.param_hdr
.type
= SCTP_PARAM_IPV4_ADDRESS
;
309 param
->v4
.param_hdr
.length
= htons(length
);
310 param
->v4
.addr
.s_addr
= addr
->v4
.sin_addr
.s_addr
;
315 /* Initialize a sctp_addr from a dst_entry. */
316 static void sctp_v4_dst_saddr(union sctp_addr
*saddr
, struct dst_entry
*dst
,
319 struct rtable
*rt
= (struct rtable
*)dst
;
320 saddr
->v4
.sin_family
= AF_INET
;
321 saddr
->v4
.sin_port
= port
;
322 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
325 /* Compare two addresses exactly. */
326 static int sctp_v4_cmp_addr(const union sctp_addr
*addr1
,
327 const union sctp_addr
*addr2
)
329 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
331 if (addr1
->v4
.sin_port
!= addr2
->v4
.sin_port
)
333 if (addr1
->v4
.sin_addr
.s_addr
!= addr2
->v4
.sin_addr
.s_addr
)
339 /* Initialize addr struct to INADDR_ANY. */
340 static void sctp_v4_inaddr_any(union sctp_addr
*addr
, __be16 port
)
342 addr
->v4
.sin_family
= AF_INET
;
343 addr
->v4
.sin_addr
.s_addr
= htonl(INADDR_ANY
);
344 addr
->v4
.sin_port
= port
;
347 /* Is this a wildcard address? */
348 static int sctp_v4_is_any(const union sctp_addr
*addr
)
350 return htonl(INADDR_ANY
) == addr
->v4
.sin_addr
.s_addr
;
353 /* This function checks if the address is a valid address to be used for
357 * Return 0 - If the address is a non-unicast or an illegal address.
358 * Return 1 - If the address is a unicast.
360 static int sctp_v4_addr_valid(union sctp_addr
*addr
,
361 struct sctp_sock
*sp
,
362 const struct sk_buff
*skb
)
364 /* Is this a non-unicast address or a unusable SCTP address? */
365 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
))
368 /* Is this a broadcast address? */
369 if (skb
&& skb
->rtable
->rt_flags
& RTCF_BROADCAST
)
375 /* Should this be available for binding? */
376 static int sctp_v4_available(union sctp_addr
*addr
, struct sctp_sock
*sp
)
378 int ret
= inet_addr_type(&init_net
, addr
->v4
.sin_addr
.s_addr
);
381 if (addr
->v4
.sin_addr
.s_addr
!= htonl(INADDR_ANY
) &&
383 !sp
->inet
.freebind
&&
384 !sysctl_ip_nonlocal_bind
)
390 /* Checking the loopback, private and other address scopes as defined in
391 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
392 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
394 * Level 0 - unusable SCTP addresses
395 * Level 1 - loopback address
396 * Level 2 - link-local addresses
397 * Level 3 - private addresses.
398 * Level 4 - global addresses
399 * For INIT and INIT-ACK address list, let L be the level of
400 * of requested destination address, sender and receiver
401 * SHOULD include all of its addresses with level greater
402 * than or equal to L.
404 static sctp_scope_t
sctp_v4_scope(union sctp_addr
*addr
)
408 /* Should IPv4 scoping be a sysctl configurable option
409 * so users can turn it off (default on) for certain
410 * unconventional networking environments?
413 /* Check for unusable SCTP addresses. */
414 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
)) {
415 retval
= SCTP_SCOPE_UNUSABLE
;
416 } else if (ipv4_is_loopback(addr
->v4
.sin_addr
.s_addr
)) {
417 retval
= SCTP_SCOPE_LOOPBACK
;
418 } else if (ipv4_is_linklocal_169(addr
->v4
.sin_addr
.s_addr
)) {
419 retval
= SCTP_SCOPE_LINK
;
420 } else if (ipv4_is_private_10(addr
->v4
.sin_addr
.s_addr
) ||
421 ipv4_is_private_172(addr
->v4
.sin_addr
.s_addr
) ||
422 ipv4_is_private_192(addr
->v4
.sin_addr
.s_addr
)) {
423 retval
= SCTP_SCOPE_PRIVATE
;
425 retval
= SCTP_SCOPE_GLOBAL
;
431 /* Returns a valid dst cache entry for the given source and destination ip
432 * addresses. If an association is passed, trys to get a dst entry with a
433 * source address that matches an address in the bind address list.
435 static struct dst_entry
*sctp_v4_get_dst(struct sctp_association
*asoc
,
436 union sctp_addr
*daddr
,
437 union sctp_addr
*saddr
)
441 struct sctp_bind_addr
*bp
;
442 struct sctp_sockaddr_entry
*laddr
;
443 struct dst_entry
*dst
= NULL
;
444 union sctp_addr dst_saddr
;
446 memset(&fl
, 0x0, sizeof(struct flowi
));
447 fl
.fl4_dst
= daddr
->v4
.sin_addr
.s_addr
;
448 fl
.proto
= IPPROTO_SCTP
;
450 fl
.fl4_tos
= RT_CONN_FLAGS(asoc
->base
.sk
);
451 fl
.oif
= asoc
->base
.sk
->sk_bound_dev_if
;
454 fl
.fl4_src
= saddr
->v4
.sin_addr
.s_addr
;
456 SCTP_DEBUG_PRINTK("%s: DST:%u.%u.%u.%u, SRC:%u.%u.%u.%u - ",
457 __func__
, NIPQUAD(fl
.fl4_dst
),
458 NIPQUAD(fl
.fl4_src
));
460 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
464 /* If there is no association or if a source address is passed, no
465 * more validation is required.
470 bp
= &asoc
->base
.bind_addr
;
473 /* Walk through the bind address list and look for a bind
474 * address that matches the source address of the returned dst.
477 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
478 if (!laddr
->valid
|| (laddr
->state
!= SCTP_ADDR_SRC
))
480 sctp_v4_dst_saddr(&dst_saddr
, dst
, htons(bp
->port
));
481 if (sctp_v4_cmp_addr(&dst_saddr
, &laddr
->a
))
486 /* None of the bound addresses match the source address of the
487 * dst. So release it.
493 /* Walk through the bind address list and try to get a dst that
494 * matches a bind address as the source address.
497 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
500 if ((laddr
->state
== SCTP_ADDR_SRC
) &&
501 (AF_INET
== laddr
->a
.sa
.sa_family
)) {
502 fl
.fl4_src
= laddr
->a
.v4
.sin_addr
.s_addr
;
503 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
514 SCTP_DEBUG_PRINTK("rt_dst:%u.%u.%u.%u, rt_src:%u.%u.%u.%u\n",
515 NIPQUAD(rt
->rt_dst
), NIPQUAD(rt
->rt_src
));
517 SCTP_DEBUG_PRINTK("NO ROUTE\n");
522 /* For v4, the source address is cached in the route entry(dst). So no need
523 * to cache it separately and hence this is an empty routine.
525 static void sctp_v4_get_saddr(struct sctp_association
*asoc
,
526 struct dst_entry
*dst
,
527 union sctp_addr
*daddr
,
528 union sctp_addr
*saddr
)
530 struct rtable
*rt
= (struct rtable
*)dst
;
536 saddr
->v4
.sin_family
= AF_INET
;
537 saddr
->v4
.sin_port
= htons(asoc
->base
.bind_addr
.port
);
538 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
542 /* What interface did this skb arrive on? */
543 static int sctp_v4_skb_iif(const struct sk_buff
*skb
)
545 return skb
->rtable
->rt_iif
;
548 /* Was this packet marked by Explicit Congestion Notification? */
549 static int sctp_v4_is_ce(const struct sk_buff
*skb
)
551 return INET_ECN_is_ce(ip_hdr(skb
)->tos
);
554 /* Create and initialize a new sk for the socket returned by accept(). */
555 static struct sock
*sctp_v4_create_accept_sk(struct sock
*sk
,
556 struct sctp_association
*asoc
)
558 struct inet_sock
*inet
= inet_sk(sk
);
559 struct inet_sock
*newinet
;
560 struct sock
*newsk
= sk_alloc(sock_net(sk
), PF_INET
, GFP_KERNEL
,
566 sock_init_data(NULL
, newsk
);
568 newsk
->sk_type
= SOCK_STREAM
;
570 newsk
->sk_no_check
= sk
->sk_no_check
;
571 newsk
->sk_reuse
= sk
->sk_reuse
;
572 newsk
->sk_shutdown
= sk
->sk_shutdown
;
574 newsk
->sk_destruct
= inet_sock_destruct
;
575 newsk
->sk_family
= PF_INET
;
576 newsk
->sk_protocol
= IPPROTO_SCTP
;
577 newsk
->sk_backlog_rcv
= sk
->sk_prot
->backlog_rcv
;
578 sock_reset_flag(newsk
, SOCK_ZAPPED
);
580 newinet
= inet_sk(newsk
);
582 /* Initialize sk's sport, dport, rcv_saddr and daddr for
583 * getsockname() and getpeername()
585 newinet
->sport
= inet
->sport
;
586 newinet
->saddr
= inet
->saddr
;
587 newinet
->rcv_saddr
= inet
->rcv_saddr
;
588 newinet
->dport
= htons(asoc
->peer
.port
);
589 newinet
->daddr
= asoc
->peer
.primary_addr
.v4
.sin_addr
.s_addr
;
590 newinet
->pmtudisc
= inet
->pmtudisc
;
591 newinet
->id
= asoc
->next_tsn
^ jiffies
;
593 newinet
->uc_ttl
= -1;
594 newinet
->mc_loop
= 1;
596 newinet
->mc_index
= 0;
597 newinet
->mc_list
= NULL
;
599 sk_refcnt_debug_inc(newsk
);
601 if (newsk
->sk_prot
->init(newsk
)) {
602 sk_common_release(newsk
);
610 /* Map address, empty for v4 family */
611 static void sctp_v4_addr_v4map(struct sctp_sock
*sp
, union sctp_addr
*addr
)
616 /* Dump the v4 addr to the seq file. */
617 static void sctp_v4_seq_dump_addr(struct seq_file
*seq
, union sctp_addr
*addr
)
619 seq_printf(seq
, "%d.%d.%d.%d ", NIPQUAD(addr
->v4
.sin_addr
));
622 /* Event handler for inet address addition/deletion events.
623 * The sctp_local_addr_list needs to be protocted by a spin lock since
624 * multiple notifiers (say IPv4 and IPv6) may be running at the same
625 * time and thus corrupt the list.
626 * The reader side is protected with RCU.
628 static int sctp_inetaddr_event(struct notifier_block
*this, unsigned long ev
,
631 struct in_ifaddr
*ifa
= (struct in_ifaddr
*)ptr
;
632 struct sctp_sockaddr_entry
*addr
= NULL
;
633 struct sctp_sockaddr_entry
*temp
;
636 if (dev_net(ifa
->ifa_dev
->dev
) != &init_net
)
641 addr
= kmalloc(sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
643 addr
->a
.v4
.sin_family
= AF_INET
;
644 addr
->a
.v4
.sin_port
= 0;
645 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
647 spin_lock_bh(&sctp_local_addr_lock
);
648 list_add_tail_rcu(&addr
->list
, &sctp_local_addr_list
);
649 spin_unlock_bh(&sctp_local_addr_lock
);
653 spin_lock_bh(&sctp_local_addr_lock
);
654 list_for_each_entry_safe(addr
, temp
,
655 &sctp_local_addr_list
, list
) {
656 if (addr
->a
.sa
.sa_family
== AF_INET
&&
657 addr
->a
.v4
.sin_addr
.s_addr
==
661 list_del_rcu(&addr
->list
);
665 spin_unlock_bh(&sctp_local_addr_lock
);
667 call_rcu(&addr
->rcu
, sctp_local_addr_free
);
675 * Initialize the control inode/socket with a control endpoint data
676 * structure. This endpoint is reserved exclusively for the OOTB processing.
678 static int sctp_ctl_sock_init(void)
683 if (sctp_get_pf_specific(PF_INET6
))
688 err
= inet_ctl_sock_create(&sctp_ctl_sock
, family
,
689 SOCK_SEQPACKET
, IPPROTO_SCTP
, &init_net
);
692 "SCTP: Failed to create the SCTP control socket.\n");
698 /* Register address family specific functions. */
699 int sctp_register_af(struct sctp_af
*af
)
701 switch (af
->sa_family
) {
703 if (sctp_af_v4_specific
)
705 sctp_af_v4_specific
= af
;
708 if (sctp_af_v6_specific
)
710 sctp_af_v6_specific
= af
;
716 INIT_LIST_HEAD(&af
->list
);
717 list_add_tail(&af
->list
, &sctp_address_families
);
721 /* Get the table of functions for manipulating a particular address
724 struct sctp_af
*sctp_get_af_specific(sa_family_t family
)
728 return sctp_af_v4_specific
;
730 return sctp_af_v6_specific
;
736 /* Common code to initialize a AF_INET msg_name. */
737 static void sctp_inet_msgname(char *msgname
, int *addr_len
)
739 struct sockaddr_in
*sin
;
741 sin
= (struct sockaddr_in
*)msgname
;
742 *addr_len
= sizeof(struct sockaddr_in
);
743 sin
->sin_family
= AF_INET
;
744 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
747 /* Copy the primary address of the peer primary address as the msg_name. */
748 static void sctp_inet_event_msgname(struct sctp_ulpevent
*event
, char *msgname
,
751 struct sockaddr_in
*sin
, *sinfrom
;
754 struct sctp_association
*asoc
;
757 sctp_inet_msgname(msgname
, addr_len
);
758 sin
= (struct sockaddr_in
*)msgname
;
759 sinfrom
= &asoc
->peer
.primary_addr
.v4
;
760 sin
->sin_port
= htons(asoc
->peer
.port
);
761 sin
->sin_addr
.s_addr
= sinfrom
->sin_addr
.s_addr
;
765 /* Initialize and copy out a msgname from an inbound skb. */
766 static void sctp_inet_skb_msgname(struct sk_buff
*skb
, char *msgname
, int *len
)
769 struct sctphdr
*sh
= sctp_hdr(skb
);
770 struct sockaddr_in
*sin
= (struct sockaddr_in
*)msgname
;
772 sctp_inet_msgname(msgname
, len
);
773 sin
->sin_port
= sh
->source
;
774 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
778 /* Do we support this AF? */
779 static int sctp_inet_af_supported(sa_family_t family
, struct sctp_sock
*sp
)
781 /* PF_INET only supports AF_INET addresses. */
782 return (AF_INET
== family
);
785 /* Address matching with wildcards allowed. */
786 static int sctp_inet_cmp_addr(const union sctp_addr
*addr1
,
787 const union sctp_addr
*addr2
,
788 struct sctp_sock
*opt
)
790 /* PF_INET only supports AF_INET addresses. */
791 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
793 if (htonl(INADDR_ANY
) == addr1
->v4
.sin_addr
.s_addr
||
794 htonl(INADDR_ANY
) == addr2
->v4
.sin_addr
.s_addr
)
796 if (addr1
->v4
.sin_addr
.s_addr
== addr2
->v4
.sin_addr
.s_addr
)
802 /* Verify that provided sockaddr looks bindable. Common verification has
803 * already been taken care of.
805 static int sctp_inet_bind_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
807 return sctp_v4_available(addr
, opt
);
810 /* Verify that sockaddr looks sendable. Common verification has already
811 * been taken care of.
813 static int sctp_inet_send_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
818 /* Fill in Supported Address Type information for INIT and INIT-ACK
819 * chunks. Returns number of addresses supported.
821 static int sctp_inet_supported_addrs(const struct sctp_sock
*opt
,
824 types
[0] = SCTP_PARAM_IPV4_ADDRESS
;
828 /* Wrapper routine that calls the ip transmit routine. */
829 static inline int sctp_v4_xmit(struct sk_buff
*skb
,
830 struct sctp_transport
*transport
, int ipfragok
)
832 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, "
833 "src:%u.%u.%u.%u, dst:%u.%u.%u.%u\n",
834 __func__
, skb
, skb
->len
,
835 NIPQUAD(skb
->rtable
->rt_src
),
836 NIPQUAD(skb
->rtable
->rt_dst
));
838 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS
);
839 return ip_queue_xmit(skb
, ipfragok
);
842 static struct sctp_af sctp_af_inet
;
844 static struct sctp_pf sctp_pf_inet
= {
845 .event_msgname
= sctp_inet_event_msgname
,
846 .skb_msgname
= sctp_inet_skb_msgname
,
847 .af_supported
= sctp_inet_af_supported
,
848 .cmp_addr
= sctp_inet_cmp_addr
,
849 .bind_verify
= sctp_inet_bind_verify
,
850 .send_verify
= sctp_inet_send_verify
,
851 .supported_addrs
= sctp_inet_supported_addrs
,
852 .create_accept_sk
= sctp_v4_create_accept_sk
,
853 .addr_v4map
= sctp_v4_addr_v4map
,
857 /* Notifier for inetaddr addition/deletion events. */
858 static struct notifier_block sctp_inetaddr_notifier
= {
859 .notifier_call
= sctp_inetaddr_event
,
862 /* Socket operations. */
863 static const struct proto_ops inet_seqpacket_ops
= {
865 .owner
= THIS_MODULE
,
866 .release
= inet_release
, /* Needs to be wrapped... */
868 .connect
= inet_dgram_connect
,
869 .socketpair
= sock_no_socketpair
,
870 .accept
= inet_accept
,
871 .getname
= inet_getname
, /* Semantics are different. */
874 .listen
= sctp_inet_listen
,
875 .shutdown
= inet_shutdown
, /* Looks harmless. */
876 .setsockopt
= sock_common_setsockopt
, /* IP_SOL IP_OPTION is a problem */
877 .getsockopt
= sock_common_getsockopt
,
878 .sendmsg
= inet_sendmsg
,
879 .recvmsg
= sock_common_recvmsg
,
880 .mmap
= sock_no_mmap
,
881 .sendpage
= sock_no_sendpage
,
883 .compat_setsockopt
= compat_sock_common_setsockopt
,
884 .compat_getsockopt
= compat_sock_common_getsockopt
,
888 /* Registration with AF_INET family. */
889 static struct inet_protosw sctp_seqpacket_protosw
= {
890 .type
= SOCK_SEQPACKET
,
891 .protocol
= IPPROTO_SCTP
,
893 .ops
= &inet_seqpacket_ops
,
896 .flags
= SCTP_PROTOSW_FLAG
898 static struct inet_protosw sctp_stream_protosw
= {
900 .protocol
= IPPROTO_SCTP
,
902 .ops
= &inet_seqpacket_ops
,
905 .flags
= SCTP_PROTOSW_FLAG
908 /* Register with IP layer. */
909 static struct net_protocol sctp_protocol
= {
911 .err_handler
= sctp_v4_err
,
915 /* IPv4 address related functions. */
916 static struct sctp_af sctp_af_inet
= {
917 .sa_family
= AF_INET
,
918 .sctp_xmit
= sctp_v4_xmit
,
919 .setsockopt
= ip_setsockopt
,
920 .getsockopt
= ip_getsockopt
,
921 .get_dst
= sctp_v4_get_dst
,
922 .get_saddr
= sctp_v4_get_saddr
,
923 .copy_addrlist
= sctp_v4_copy_addrlist
,
924 .from_skb
= sctp_v4_from_skb
,
925 .from_sk
= sctp_v4_from_sk
,
926 .to_sk_saddr
= sctp_v4_to_sk_saddr
,
927 .to_sk_daddr
= sctp_v4_to_sk_daddr
,
928 .from_addr_param
= sctp_v4_from_addr_param
,
929 .to_addr_param
= sctp_v4_to_addr_param
,
930 .dst_saddr
= sctp_v4_dst_saddr
,
931 .cmp_addr
= sctp_v4_cmp_addr
,
932 .addr_valid
= sctp_v4_addr_valid
,
933 .inaddr_any
= sctp_v4_inaddr_any
,
934 .is_any
= sctp_v4_is_any
,
935 .available
= sctp_v4_available
,
936 .scope
= sctp_v4_scope
,
937 .skb_iif
= sctp_v4_skb_iif
,
938 .is_ce
= sctp_v4_is_ce
,
939 .seq_dump_addr
= sctp_v4_seq_dump_addr
,
940 .net_header_len
= sizeof(struct iphdr
),
941 .sockaddr_len
= sizeof(struct sockaddr_in
),
943 .compat_setsockopt
= compat_ip_setsockopt
,
944 .compat_getsockopt
= compat_ip_getsockopt
,
948 struct sctp_pf
*sctp_get_pf_specific(sa_family_t family
) {
952 return sctp_pf_inet_specific
;
954 return sctp_pf_inet6_specific
;
960 /* Register the PF specific function table. */
961 int sctp_register_pf(struct sctp_pf
*pf
, sa_family_t family
)
965 if (sctp_pf_inet_specific
)
967 sctp_pf_inet_specific
= pf
;
970 if (sctp_pf_inet6_specific
)
972 sctp_pf_inet6_specific
= pf
;
980 static inline int init_sctp_mibs(void)
982 return snmp_mib_init((void**)sctp_statistics
, sizeof(struct sctp_mib
));
985 static inline void cleanup_sctp_mibs(void)
987 snmp_mib_free((void**)sctp_statistics
);
990 static void sctp_v4_pf_init(void)
992 /* Initialize the SCTP specific PF functions. */
993 sctp_register_pf(&sctp_pf_inet
, PF_INET
);
994 sctp_register_af(&sctp_af_inet
);
997 static void sctp_v4_pf_exit(void)
999 list_del(&sctp_af_inet
.list
);
1002 static int sctp_v4_protosw_init(void)
1006 rc
= proto_register(&sctp_prot
, 1);
1010 /* Register SCTP(UDP and TCP style) with socket layer. */
1011 inet_register_protosw(&sctp_seqpacket_protosw
);
1012 inet_register_protosw(&sctp_stream_protosw
);
1017 static void sctp_v4_protosw_exit(void)
1019 inet_unregister_protosw(&sctp_stream_protosw
);
1020 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1021 proto_unregister(&sctp_prot
);
1024 static int sctp_v4_add_protocol(void)
1026 /* Register notifier for inet address additions/deletions. */
1027 register_inetaddr_notifier(&sctp_inetaddr_notifier
);
1029 /* Register SCTP with inet layer. */
1030 if (inet_add_protocol(&sctp_protocol
, IPPROTO_SCTP
) < 0)
1036 static void sctp_v4_del_protocol(void)
1038 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1039 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1042 /* Initialize the universe into something sensible. */
1043 SCTP_STATIC __init
int sctp_init(void)
1046 int status
= -EINVAL
;
1048 unsigned long limit
;
1052 /* SCTP_DEBUG sanity check. */
1053 if (!sctp_sanity_check())
1056 /* Allocate bind_bucket and chunk caches. */
1058 sctp_bucket_cachep
= kmem_cache_create("sctp_bind_bucket",
1059 sizeof(struct sctp_bind_bucket
),
1060 0, SLAB_HWCACHE_ALIGN
,
1062 if (!sctp_bucket_cachep
)
1065 sctp_chunk_cachep
= kmem_cache_create("sctp_chunk",
1066 sizeof(struct sctp_chunk
),
1067 0, SLAB_HWCACHE_ALIGN
,
1069 if (!sctp_chunk_cachep
)
1070 goto err_chunk_cachep
;
1072 /* Allocate and initialise sctp mibs. */
1073 status
= init_sctp_mibs();
1077 /* Initialize proc fs directory. */
1078 status
= sctp_proc_init();
1082 /* Initialize object count debugging. */
1083 sctp_dbg_objcnt_init();
1086 * 14. Suggested SCTP Protocol Parameter Values
1088 /* The following protocol parameters are RECOMMENDED: */
1089 /* RTO.Initial - 3 seconds */
1090 sctp_rto_initial
= SCTP_RTO_INITIAL
;
1091 /* RTO.Min - 1 second */
1092 sctp_rto_min
= SCTP_RTO_MIN
;
1093 /* RTO.Max - 60 seconds */
1094 sctp_rto_max
= SCTP_RTO_MAX
;
1095 /* RTO.Alpha - 1/8 */
1096 sctp_rto_alpha
= SCTP_RTO_ALPHA
;
1097 /* RTO.Beta - 1/4 */
1098 sctp_rto_beta
= SCTP_RTO_BETA
;
1100 /* Valid.Cookie.Life - 60 seconds */
1101 sctp_valid_cookie_life
= SCTP_DEFAULT_COOKIE_LIFE
;
1103 /* Whether Cookie Preservative is enabled(1) or not(0) */
1104 sctp_cookie_preserve_enable
= 1;
1107 sctp_max_burst
= SCTP_DEFAULT_MAX_BURST
;
1109 /* Association.Max.Retrans - 10 attempts
1110 * Path.Max.Retrans - 5 attempts (per destination address)
1111 * Max.Init.Retransmits - 8 attempts
1113 sctp_max_retrans_association
= 10;
1114 sctp_max_retrans_path
= 5;
1115 sctp_max_retrans_init
= 8;
1117 /* Sendbuffer growth - do per-socket accounting */
1118 sctp_sndbuf_policy
= 0;
1120 /* Rcvbuffer growth - do per-socket accounting */
1121 sctp_rcvbuf_policy
= 0;
1123 /* HB.interval - 30 seconds */
1124 sctp_hb_interval
= SCTP_DEFAULT_TIMEOUT_HEARTBEAT
;
1126 /* delayed SACK timeout */
1127 sctp_sack_timeout
= SCTP_DEFAULT_TIMEOUT_SACK
;
1129 /* Implementation specific variables. */
1131 /* Initialize default stream count setup information. */
1132 sctp_max_instreams
= SCTP_DEFAULT_INSTREAMS
;
1133 sctp_max_outstreams
= SCTP_DEFAULT_OUTSTREAMS
;
1135 /* Initialize handle used for association ids. */
1136 idr_init(&sctp_assocs_id
);
1138 /* Set the pressure threshold to be a fraction of global memory that
1139 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1140 * memory, with a floor of 128 pages.
1141 * Note this initalizes the data in sctpv6_prot too
1142 * Unabashedly stolen from tcp_init
1144 limit
= min(num_physpages
, 1UL<<(28-PAGE_SHIFT
)) >> (20-PAGE_SHIFT
);
1145 limit
= (limit
* (num_physpages
>> (20-PAGE_SHIFT
))) >> (PAGE_SHIFT
-11);
1146 limit
= max(limit
, 128UL);
1147 sysctl_sctp_mem
[0] = limit
/ 4 * 3;
1148 sysctl_sctp_mem
[1] = limit
;
1149 sysctl_sctp_mem
[2] = sysctl_sctp_mem
[0] * 2;
1151 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1152 limit
= (sysctl_sctp_mem
[1]) << (PAGE_SHIFT
- 7);
1153 max_share
= min(4UL*1024*1024, limit
);
1155 sysctl_sctp_rmem
[0] = PAGE_SIZE
; /* give each asoc 1 page min */
1156 sysctl_sctp_rmem
[1] = (1500 *(sizeof(struct sk_buff
) + 1));
1157 sysctl_sctp_rmem
[2] = max(sysctl_sctp_rmem
[1], max_share
);
1159 sysctl_sctp_wmem
[0] = SK_MEM_QUANTUM
;
1160 sysctl_sctp_wmem
[1] = 16*1024;
1161 sysctl_sctp_wmem
[2] = max(64*1024, max_share
);
1163 /* Size and allocate the association hash table.
1164 * The methodology is similar to that of the tcp hash tables.
1166 if (num_physpages
>= (128 * 1024))
1167 goal
= num_physpages
>> (22 - PAGE_SHIFT
);
1169 goal
= num_physpages
>> (24 - PAGE_SHIFT
);
1171 for (order
= 0; (1UL << order
) < goal
; order
++)
1175 sctp_assoc_hashsize
= (1UL << order
) * PAGE_SIZE
/
1176 sizeof(struct sctp_hashbucket
);
1177 if ((sctp_assoc_hashsize
> (64 * 1024)) && order
> 0)
1179 sctp_assoc_hashtable
= (struct sctp_hashbucket
*)
1180 __get_free_pages(GFP_ATOMIC
, order
);
1181 } while (!sctp_assoc_hashtable
&& --order
> 0);
1182 if (!sctp_assoc_hashtable
) {
1183 printk(KERN_ERR
"SCTP: Failed association hash alloc.\n");
1185 goto err_ahash_alloc
;
1187 for (i
= 0; i
< sctp_assoc_hashsize
; i
++) {
1188 rwlock_init(&sctp_assoc_hashtable
[i
].lock
);
1189 INIT_HLIST_HEAD(&sctp_assoc_hashtable
[i
].chain
);
1192 /* Allocate and initialize the endpoint hash table. */
1193 sctp_ep_hashsize
= 64;
1194 sctp_ep_hashtable
= (struct sctp_hashbucket
*)
1195 kmalloc(64 * sizeof(struct sctp_hashbucket
), GFP_KERNEL
);
1196 if (!sctp_ep_hashtable
) {
1197 printk(KERN_ERR
"SCTP: Failed endpoint_hash alloc.\n");
1199 goto err_ehash_alloc
;
1201 for (i
= 0; i
< sctp_ep_hashsize
; i
++) {
1202 rwlock_init(&sctp_ep_hashtable
[i
].lock
);
1203 INIT_HLIST_HEAD(&sctp_ep_hashtable
[i
].chain
);
1206 /* Allocate and initialize the SCTP port hash table. */
1208 sctp_port_hashsize
= (1UL << order
) * PAGE_SIZE
/
1209 sizeof(struct sctp_bind_hashbucket
);
1210 if ((sctp_port_hashsize
> (64 * 1024)) && order
> 0)
1212 sctp_port_hashtable
= (struct sctp_bind_hashbucket
*)
1213 __get_free_pages(GFP_ATOMIC
, order
);
1214 } while (!sctp_port_hashtable
&& --order
> 0);
1215 if (!sctp_port_hashtable
) {
1216 printk(KERN_ERR
"SCTP: Failed bind hash alloc.");
1218 goto err_bhash_alloc
;
1220 for (i
= 0; i
< sctp_port_hashsize
; i
++) {
1221 spin_lock_init(&sctp_port_hashtable
[i
].lock
);
1222 INIT_HLIST_HEAD(&sctp_port_hashtable
[i
].chain
);
1225 printk(KERN_INFO
"SCTP: Hash tables configured "
1226 "(established %d bind %d)\n",
1227 sctp_assoc_hashsize
, sctp_port_hashsize
);
1229 /* Disable ADDIP by default. */
1230 sctp_addip_enable
= 0;
1231 sctp_addip_noauth
= 0;
1233 /* Enable PR-SCTP by default. */
1234 sctp_prsctp_enable
= 1;
1236 /* Disable AUTH by default. */
1237 sctp_auth_enable
= 0;
1239 sctp_sysctl_register();
1241 INIT_LIST_HEAD(&sctp_address_families
);
1245 /* Initialize the local address list. */
1246 INIT_LIST_HEAD(&sctp_local_addr_list
);
1247 spin_lock_init(&sctp_local_addr_lock
);
1248 sctp_get_local_addr_list();
1250 status
= sctp_v4_protosw_init();
1253 goto err_protosw_init
;
1255 status
= sctp_v6_protosw_init();
1257 goto err_v6_protosw_init
;
1259 /* Initialize the control inode/socket for handling OOTB packets. */
1260 if ((status
= sctp_ctl_sock_init())) {
1262 "SCTP: Failed to initialize the SCTP control sock.\n");
1263 goto err_ctl_sock_init
;
1266 status
= sctp_v4_add_protocol();
1268 goto err_add_protocol
;
1270 /* Register SCTP with inet6 layer. */
1271 status
= sctp_v6_add_protocol();
1273 goto err_v6_add_protocol
;
1278 err_v6_add_protocol
:
1279 sctp_v6_del_protocol();
1281 sctp_v4_del_protocol();
1282 inet_ctl_sock_destroy(sctp_ctl_sock
);
1284 sctp_v6_protosw_exit();
1285 err_v6_protosw_init
:
1286 sctp_v4_protosw_exit();
1288 sctp_free_local_addr_list();
1291 sctp_sysctl_unregister();
1292 list_del(&sctp_af_inet
.list
);
1293 free_pages((unsigned long)sctp_port_hashtable
,
1294 get_order(sctp_port_hashsize
*
1295 sizeof(struct sctp_bind_hashbucket
)));
1297 kfree(sctp_ep_hashtable
);
1299 free_pages((unsigned long)sctp_assoc_hashtable
,
1300 get_order(sctp_assoc_hashsize
*
1301 sizeof(struct sctp_hashbucket
)));
1303 sctp_dbg_objcnt_exit();
1306 cleanup_sctp_mibs();
1308 kmem_cache_destroy(sctp_chunk_cachep
);
1310 kmem_cache_destroy(sctp_bucket_cachep
);
1314 /* Exit handler for the SCTP protocol. */
1315 SCTP_STATIC __exit
void sctp_exit(void)
1317 /* BUG. This should probably do something useful like clean
1318 * up all the remaining associations and all that memory.
1321 /* Unregister with inet6/inet layers. */
1322 sctp_v6_del_protocol();
1323 sctp_v4_del_protocol();
1325 /* Free the control endpoint. */
1326 inet_ctl_sock_destroy(sctp_ctl_sock
);
1328 /* Free protosw registrations */
1329 sctp_v6_protosw_exit();
1330 sctp_v4_protosw_exit();
1332 /* Free the local address list. */
1333 sctp_free_local_addr_list();
1335 /* Unregister with socket layer. */
1339 sctp_sysctl_unregister();
1340 list_del(&sctp_af_inet
.list
);
1342 free_pages((unsigned long)sctp_assoc_hashtable
,
1343 get_order(sctp_assoc_hashsize
*
1344 sizeof(struct sctp_hashbucket
)));
1345 kfree(sctp_ep_hashtable
);
1346 free_pages((unsigned long)sctp_port_hashtable
,
1347 get_order(sctp_port_hashsize
*
1348 sizeof(struct sctp_bind_hashbucket
)));
1350 sctp_dbg_objcnt_exit();
1352 cleanup_sctp_mibs();
1354 kmem_cache_destroy(sctp_chunk_cachep
);
1355 kmem_cache_destroy(sctp_bucket_cachep
);
1358 module_init(sctp_init
);
1359 module_exit(sctp_exit
);
1362 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1364 MODULE_ALIAS("net-pf-" __stringify(PF_INET
) "-proto-132");
1365 MODULE_ALIAS("net-pf-" __stringify(PF_INET6
) "-proto-132");
1366 MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1367 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1368 MODULE_LICENSE("GPL");