1 /* SCTP kernel implementation
2 * (C) Copyright IBM Corp. 2001, 2004
3 * Copyright (c) 1999-2000 Cisco, Inc.
4 * Copyright (c) 1999-2001 Motorola, Inc.
5 * Copyright (c) 2001 Intel Corp.
6 * Copyright (c) 2001 Nokia, Inc.
7 * Copyright (c) 2001 La Monte H.P. Yarroll
9 * This file is part of the SCTP kernel implementation
11 * Initialization/cleanup for SCTP protocol support.
13 * This SCTP implementation is free software;
14 * you can redistribute it and/or modify it under the terms of
15 * the GNU General Public License as published by
16 * the Free Software Foundation; either version 2, or (at your option)
19 * This SCTP implementation is distributed in the hope that it
20 * will be useful, but WITHOUT ANY WARRANTY; without even the implied
21 * ************************
22 * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
23 * See the GNU General Public License for more details.
25 * You should have received a copy of the GNU General Public License
26 * along with GNU CC; see the file COPYING. If not, write to
27 * the Free Software Foundation, 59 Temple Place - Suite 330,
28 * Boston, MA 02111-1307, USA.
30 * Please send any bug reports or fixes you make to the
32 * lksctp developers <lksctp-developers@lists.sourceforge.net>
34 * Or submit a bug report through the following website:
35 * http://www.sf.net/projects/lksctp
37 * Written or modified by:
38 * La Monte H.P. Yarroll <piggy@acm.org>
39 * Karl Knutson <karl@athena.chicago.il.us>
40 * Jon Grimm <jgrimm@us.ibm.com>
41 * Sridhar Samudrala <sri@us.ibm.com>
42 * Daisy Chang <daisyc@us.ibm.com>
43 * Ardelle Fan <ardelle.fan@intel.com>
45 * Any bugs reported given to us we will try to fix... any fixes shared will
46 * be incorporated into the next SCTP release.
49 #include <linux/module.h>
50 #include <linux/init.h>
51 #include <linux/netdevice.h>
52 #include <linux/inetdevice.h>
53 #include <linux/seq_file.h>
54 #include <linux/bootmem.h>
55 #include <net/net_namespace.h>
56 #include <net/protocol.h>
59 #include <net/route.h>
60 #include <net/sctp/sctp.h>
61 #include <net/addrconf.h>
62 #include <net/inet_common.h>
63 #include <net/inet_ecn.h>
65 /* Global data structures. */
66 struct sctp_globals sctp_globals __read_mostly
;
67 DEFINE_SNMP_STAT(struct sctp_mib
, sctp_statistics
) __read_mostly
;
70 struct proc_dir_entry
*proc_net_sctp
;
73 struct idr sctp_assocs_id
;
74 DEFINE_SPINLOCK(sctp_assocs_id_lock
);
76 /* This is the global socket data structure used for responding to
77 * the Out-of-the-blue (OOTB) packets. A control sock will be created
78 * for this socket at the initialization time.
80 static struct sock
*sctp_ctl_sock
;
82 static struct sctp_pf
*sctp_pf_inet6_specific
;
83 static struct sctp_pf
*sctp_pf_inet_specific
;
84 static struct sctp_af
*sctp_af_v4_specific
;
85 static struct sctp_af
*sctp_af_v6_specific
;
87 struct kmem_cache
*sctp_chunk_cachep __read_mostly
;
88 struct kmem_cache
*sctp_bucket_cachep __read_mostly
;
90 int sysctl_sctp_mem
[3];
91 int sysctl_sctp_rmem
[3];
92 int sysctl_sctp_wmem
[3];
94 /* Return the address of the control sock. */
95 struct sock
*sctp_get_ctl_sock(void)
100 /* Set up the proc fs entry for the SCTP protocol. */
101 static __init
int sctp_proc_init(void)
103 #ifdef CONFIG_PROC_FS
104 if (!proc_net_sctp
) {
105 struct proc_dir_entry
*ent
;
106 ent
= proc_mkdir("sctp", init_net
.proc_net
);
108 ent
->owner
= THIS_MODULE
;
114 if (sctp_snmp_proc_init())
115 goto out_snmp_proc_init
;
116 if (sctp_eps_proc_init())
117 goto out_eps_proc_init
;
118 if (sctp_assocs_proc_init())
119 goto out_assocs_proc_init
;
120 if (sctp_remaddr_proc_init())
121 goto out_remaddr_proc_init
;
125 out_remaddr_proc_init
:
126 sctp_assocs_proc_exit();
127 out_assocs_proc_init
:
128 sctp_eps_proc_exit();
130 sctp_snmp_proc_exit();
133 proc_net_sctp
= NULL
;
134 remove_proc_entry("sctp", init_net
.proc_net
);
140 #endif /* CONFIG_PROC_FS */
143 /* Clean up the proc fs entry for the SCTP protocol.
144 * Note: Do not make this __exit as it is used in the init error
147 static void sctp_proc_exit(void)
149 #ifdef CONFIG_PROC_FS
150 sctp_snmp_proc_exit();
151 sctp_eps_proc_exit();
152 sctp_assocs_proc_exit();
153 sctp_remaddr_proc_exit();
156 proc_net_sctp
= NULL
;
157 remove_proc_entry("sctp", init_net
.proc_net
);
162 /* Private helper to extract ipv4 address and stash them in
163 * the protocol structure.
165 static void sctp_v4_copy_addrlist(struct list_head
*addrlist
,
166 struct net_device
*dev
)
168 struct in_device
*in_dev
;
169 struct in_ifaddr
*ifa
;
170 struct sctp_sockaddr_entry
*addr
;
173 if ((in_dev
= __in_dev_get_rcu(dev
)) == NULL
) {
178 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
179 /* Add the address to the local list. */
180 addr
= t_new(struct sctp_sockaddr_entry
, GFP_ATOMIC
);
182 addr
->a
.v4
.sin_family
= AF_INET
;
183 addr
->a
.v4
.sin_port
= 0;
184 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
186 INIT_LIST_HEAD(&addr
->list
);
187 INIT_RCU_HEAD(&addr
->rcu
);
188 list_add_tail(&addr
->list
, addrlist
);
195 /* Extract our IP addresses from the system and stash them in the
196 * protocol structure.
198 static void sctp_get_local_addr_list(void)
200 struct net_device
*dev
;
201 struct list_head
*pos
;
204 read_lock(&dev_base_lock
);
205 for_each_netdev(&init_net
, dev
) {
206 __list_for_each(pos
, &sctp_address_families
) {
207 af
= list_entry(pos
, struct sctp_af
, list
);
208 af
->copy_addrlist(&sctp_local_addr_list
, dev
);
211 read_unlock(&dev_base_lock
);
214 /* Free the existing local addresses. */
215 static void sctp_free_local_addr_list(void)
217 struct sctp_sockaddr_entry
*addr
;
218 struct list_head
*pos
, *temp
;
220 list_for_each_safe(pos
, temp
, &sctp_local_addr_list
) {
221 addr
= list_entry(pos
, struct sctp_sockaddr_entry
, list
);
227 void sctp_local_addr_free(struct rcu_head
*head
)
229 struct sctp_sockaddr_entry
*e
= container_of(head
,
230 struct sctp_sockaddr_entry
, rcu
);
234 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
235 int sctp_copy_local_addr_list(struct sctp_bind_addr
*bp
, sctp_scope_t scope
,
236 gfp_t gfp
, int copy_flags
)
238 struct sctp_sockaddr_entry
*addr
;
242 list_for_each_entry_rcu(addr
, &sctp_local_addr_list
, list
) {
245 if (sctp_in_scope(&addr
->a
, scope
)) {
246 /* Now that the address is in scope, check to see if
247 * the address type is really supported by the local
248 * sock as well as the remote peer.
250 if ((((AF_INET
== addr
->a
.sa
.sa_family
) &&
251 (copy_flags
& SCTP_ADDR4_PEERSUPP
))) ||
252 (((AF_INET6
== addr
->a
.sa
.sa_family
) &&
253 (copy_flags
& SCTP_ADDR6_ALLOWED
) &&
254 (copy_flags
& SCTP_ADDR6_PEERSUPP
)))) {
255 error
= sctp_add_bind_addr(bp
, &addr
->a
,
256 SCTP_ADDR_SRC
, GFP_ATOMIC
);
268 /* Initialize a sctp_addr from in incoming skb. */
269 static void sctp_v4_from_skb(union sctp_addr
*addr
, struct sk_buff
*skb
,
276 port
= &addr
->v4
.sin_port
;
277 addr
->v4
.sin_family
= AF_INET
;
282 from
= &ip_hdr(skb
)->saddr
;
285 from
= &ip_hdr(skb
)->daddr
;
287 memcpy(&addr
->v4
.sin_addr
.s_addr
, from
, sizeof(struct in_addr
));
290 /* Initialize an sctp_addr from a socket. */
291 static void sctp_v4_from_sk(union sctp_addr
*addr
, struct sock
*sk
)
293 addr
->v4
.sin_family
= AF_INET
;
294 addr
->v4
.sin_port
= 0;
295 addr
->v4
.sin_addr
.s_addr
= inet_sk(sk
)->rcv_saddr
;
298 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
299 static void sctp_v4_to_sk_saddr(union sctp_addr
*addr
, struct sock
*sk
)
301 inet_sk(sk
)->rcv_saddr
= addr
->v4
.sin_addr
.s_addr
;
304 /* Initialize sk->sk_daddr from sctp_addr. */
305 static void sctp_v4_to_sk_daddr(union sctp_addr
*addr
, struct sock
*sk
)
307 inet_sk(sk
)->daddr
= addr
->v4
.sin_addr
.s_addr
;
310 /* Initialize a sctp_addr from an address parameter. */
311 static void sctp_v4_from_addr_param(union sctp_addr
*addr
,
312 union sctp_addr_param
*param
,
313 __be16 port
, int iif
)
315 addr
->v4
.sin_family
= AF_INET
;
316 addr
->v4
.sin_port
= port
;
317 addr
->v4
.sin_addr
.s_addr
= param
->v4
.addr
.s_addr
;
320 /* Initialize an address parameter from a sctp_addr and return the length
321 * of the address parameter.
323 static int sctp_v4_to_addr_param(const union sctp_addr
*addr
,
324 union sctp_addr_param
*param
)
326 int length
= sizeof(sctp_ipv4addr_param_t
);
328 param
->v4
.param_hdr
.type
= SCTP_PARAM_IPV4_ADDRESS
;
329 param
->v4
.param_hdr
.length
= htons(length
);
330 param
->v4
.addr
.s_addr
= addr
->v4
.sin_addr
.s_addr
;
335 /* Initialize a sctp_addr from a dst_entry. */
336 static void sctp_v4_dst_saddr(union sctp_addr
*saddr
, struct dst_entry
*dst
,
339 struct rtable
*rt
= (struct rtable
*)dst
;
340 saddr
->v4
.sin_family
= AF_INET
;
341 saddr
->v4
.sin_port
= port
;
342 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
345 /* Compare two addresses exactly. */
346 static int sctp_v4_cmp_addr(const union sctp_addr
*addr1
,
347 const union sctp_addr
*addr2
)
349 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
351 if (addr1
->v4
.sin_port
!= addr2
->v4
.sin_port
)
353 if (addr1
->v4
.sin_addr
.s_addr
!= addr2
->v4
.sin_addr
.s_addr
)
359 /* Initialize addr struct to INADDR_ANY. */
360 static void sctp_v4_inaddr_any(union sctp_addr
*addr
, __be16 port
)
362 addr
->v4
.sin_family
= AF_INET
;
363 addr
->v4
.sin_addr
.s_addr
= htonl(INADDR_ANY
);
364 addr
->v4
.sin_port
= port
;
367 /* Is this a wildcard address? */
368 static int sctp_v4_is_any(const union sctp_addr
*addr
)
370 return htonl(INADDR_ANY
) == addr
->v4
.sin_addr
.s_addr
;
373 /* This function checks if the address is a valid address to be used for
377 * Return 0 - If the address is a non-unicast or an illegal address.
378 * Return 1 - If the address is a unicast.
380 static int sctp_v4_addr_valid(union sctp_addr
*addr
,
381 struct sctp_sock
*sp
,
382 const struct sk_buff
*skb
)
384 /* Is this a non-unicast address or a unusable SCTP address? */
385 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
))
388 /* Is this a broadcast address? */
389 if (skb
&& skb
->rtable
->rt_flags
& RTCF_BROADCAST
)
395 /* Should this be available for binding? */
396 static int sctp_v4_available(union sctp_addr
*addr
, struct sctp_sock
*sp
)
398 int ret
= inet_addr_type(&init_net
, addr
->v4
.sin_addr
.s_addr
);
401 if (addr
->v4
.sin_addr
.s_addr
!= htonl(INADDR_ANY
) &&
403 !sp
->inet
.freebind
&&
404 !sysctl_ip_nonlocal_bind
)
410 /* Checking the loopback, private and other address scopes as defined in
411 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
412 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
414 * Level 0 - unusable SCTP addresses
415 * Level 1 - loopback address
416 * Level 2 - link-local addresses
417 * Level 3 - private addresses.
418 * Level 4 - global addresses
419 * For INIT and INIT-ACK address list, let L be the level of
420 * of requested destination address, sender and receiver
421 * SHOULD include all of its addresses with level greater
422 * than or equal to L.
424 static sctp_scope_t
sctp_v4_scope(union sctp_addr
*addr
)
428 /* Should IPv4 scoping be a sysctl configurable option
429 * so users can turn it off (default on) for certain
430 * unconventional networking environments?
433 /* Check for unusable SCTP addresses. */
434 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
)) {
435 retval
= SCTP_SCOPE_UNUSABLE
;
436 } else if (ipv4_is_loopback(addr
->v4
.sin_addr
.s_addr
)) {
437 retval
= SCTP_SCOPE_LOOPBACK
;
438 } else if (ipv4_is_linklocal_169(addr
->v4
.sin_addr
.s_addr
)) {
439 retval
= SCTP_SCOPE_LINK
;
440 } else if (ipv4_is_private_10(addr
->v4
.sin_addr
.s_addr
) ||
441 ipv4_is_private_172(addr
->v4
.sin_addr
.s_addr
) ||
442 ipv4_is_private_192(addr
->v4
.sin_addr
.s_addr
)) {
443 retval
= SCTP_SCOPE_PRIVATE
;
445 retval
= SCTP_SCOPE_GLOBAL
;
451 /* Returns a valid dst cache entry for the given source and destination ip
452 * addresses. If an association is passed, trys to get a dst entry with a
453 * source address that matches an address in the bind address list.
455 static struct dst_entry
*sctp_v4_get_dst(struct sctp_association
*asoc
,
456 union sctp_addr
*daddr
,
457 union sctp_addr
*saddr
)
461 struct sctp_bind_addr
*bp
;
462 struct sctp_sockaddr_entry
*laddr
;
463 struct dst_entry
*dst
= NULL
;
464 union sctp_addr dst_saddr
;
466 memset(&fl
, 0x0, sizeof(struct flowi
));
467 fl
.fl4_dst
= daddr
->v4
.sin_addr
.s_addr
;
468 fl
.proto
= IPPROTO_SCTP
;
470 fl
.fl4_tos
= RT_CONN_FLAGS(asoc
->base
.sk
);
471 fl
.oif
= asoc
->base
.sk
->sk_bound_dev_if
;
474 fl
.fl4_src
= saddr
->v4
.sin_addr
.s_addr
;
476 SCTP_DEBUG_PRINTK("%s: DST:%u.%u.%u.%u, SRC:%u.%u.%u.%u - ",
477 __func__
, NIPQUAD(fl
.fl4_dst
),
478 NIPQUAD(fl
.fl4_src
));
480 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
484 /* If there is no association or if a source address is passed, no
485 * more validation is required.
490 bp
= &asoc
->base
.bind_addr
;
493 /* Walk through the bind address list and look for a bind
494 * address that matches the source address of the returned dst.
496 sctp_v4_dst_saddr(&dst_saddr
, dst
, htons(bp
->port
));
498 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
499 if (!laddr
->valid
|| (laddr
->state
!= SCTP_ADDR_SRC
))
501 if (sctp_v4_cmp_addr(&dst_saddr
, &laddr
->a
))
506 /* None of the bound addresses match the source address of the
507 * dst. So release it.
513 /* Walk through the bind address list and try to get a dst that
514 * matches a bind address as the source address.
517 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
520 if ((laddr
->state
== SCTP_ADDR_SRC
) &&
521 (AF_INET
== laddr
->a
.sa
.sa_family
)) {
522 fl
.fl4_src
= laddr
->a
.v4
.sin_addr
.s_addr
;
523 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
534 SCTP_DEBUG_PRINTK("rt_dst:%u.%u.%u.%u, rt_src:%u.%u.%u.%u\n",
535 NIPQUAD(rt
->rt_dst
), NIPQUAD(rt
->rt_src
));
537 SCTP_DEBUG_PRINTK("NO ROUTE\n");
542 /* For v4, the source address is cached in the route entry(dst). So no need
543 * to cache it separately and hence this is an empty routine.
545 static void sctp_v4_get_saddr(struct sctp_sock
*sk
,
546 struct sctp_association
*asoc
,
547 struct dst_entry
*dst
,
548 union sctp_addr
*daddr
,
549 union sctp_addr
*saddr
)
551 struct rtable
*rt
= (struct rtable
*)dst
;
557 saddr
->v4
.sin_family
= AF_INET
;
558 saddr
->v4
.sin_port
= htons(asoc
->base
.bind_addr
.port
);
559 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
563 /* What interface did this skb arrive on? */
564 static int sctp_v4_skb_iif(const struct sk_buff
*skb
)
566 return skb
->rtable
->rt_iif
;
569 /* Was this packet marked by Explicit Congestion Notification? */
570 static int sctp_v4_is_ce(const struct sk_buff
*skb
)
572 return INET_ECN_is_ce(ip_hdr(skb
)->tos
);
575 /* Create and initialize a new sk for the socket returned by accept(). */
576 static struct sock
*sctp_v4_create_accept_sk(struct sock
*sk
,
577 struct sctp_association
*asoc
)
579 struct inet_sock
*inet
= inet_sk(sk
);
580 struct inet_sock
*newinet
;
581 struct sock
*newsk
= sk_alloc(sock_net(sk
), PF_INET
, GFP_KERNEL
,
587 sock_init_data(NULL
, newsk
);
589 newsk
->sk_type
= SOCK_STREAM
;
591 newsk
->sk_no_check
= sk
->sk_no_check
;
592 newsk
->sk_reuse
= sk
->sk_reuse
;
593 newsk
->sk_shutdown
= sk
->sk_shutdown
;
595 newsk
->sk_destruct
= inet_sock_destruct
;
596 newsk
->sk_family
= PF_INET
;
597 newsk
->sk_protocol
= IPPROTO_SCTP
;
598 newsk
->sk_backlog_rcv
= sk
->sk_prot
->backlog_rcv
;
599 sock_reset_flag(newsk
, SOCK_ZAPPED
);
601 newinet
= inet_sk(newsk
);
603 /* Initialize sk's sport, dport, rcv_saddr and daddr for
604 * getsockname() and getpeername()
606 newinet
->sport
= inet
->sport
;
607 newinet
->saddr
= inet
->saddr
;
608 newinet
->rcv_saddr
= inet
->rcv_saddr
;
609 newinet
->dport
= htons(asoc
->peer
.port
);
610 newinet
->daddr
= asoc
->peer
.primary_addr
.v4
.sin_addr
.s_addr
;
611 newinet
->pmtudisc
= inet
->pmtudisc
;
612 newinet
->id
= asoc
->next_tsn
^ jiffies
;
614 newinet
->uc_ttl
= -1;
615 newinet
->mc_loop
= 1;
617 newinet
->mc_index
= 0;
618 newinet
->mc_list
= NULL
;
620 sk_refcnt_debug_inc(newsk
);
622 if (newsk
->sk_prot
->init(newsk
)) {
623 sk_common_release(newsk
);
631 /* Map address, empty for v4 family */
632 static void sctp_v4_addr_v4map(struct sctp_sock
*sp
, union sctp_addr
*addr
)
637 /* Dump the v4 addr to the seq file. */
638 static void sctp_v4_seq_dump_addr(struct seq_file
*seq
, union sctp_addr
*addr
)
640 seq_printf(seq
, "%d.%d.%d.%d ", NIPQUAD(addr
->v4
.sin_addr
));
643 static void sctp_v4_ecn_capable(struct sock
*sk
)
648 /* Event handler for inet address addition/deletion events.
649 * The sctp_local_addr_list needs to be protocted by a spin lock since
650 * multiple notifiers (say IPv4 and IPv6) may be running at the same
651 * time and thus corrupt the list.
652 * The reader side is protected with RCU.
654 static int sctp_inetaddr_event(struct notifier_block
*this, unsigned long ev
,
657 struct in_ifaddr
*ifa
= (struct in_ifaddr
*)ptr
;
658 struct sctp_sockaddr_entry
*addr
= NULL
;
659 struct sctp_sockaddr_entry
*temp
;
662 if (dev_net(ifa
->ifa_dev
->dev
) != &init_net
)
667 addr
= kmalloc(sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
669 addr
->a
.v4
.sin_family
= AF_INET
;
670 addr
->a
.v4
.sin_port
= 0;
671 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
673 spin_lock_bh(&sctp_local_addr_lock
);
674 list_add_tail_rcu(&addr
->list
, &sctp_local_addr_list
);
675 spin_unlock_bh(&sctp_local_addr_lock
);
679 spin_lock_bh(&sctp_local_addr_lock
);
680 list_for_each_entry_safe(addr
, temp
,
681 &sctp_local_addr_list
, list
) {
682 if (addr
->a
.sa
.sa_family
== AF_INET
&&
683 addr
->a
.v4
.sin_addr
.s_addr
==
687 list_del_rcu(&addr
->list
);
691 spin_unlock_bh(&sctp_local_addr_lock
);
693 call_rcu(&addr
->rcu
, sctp_local_addr_free
);
701 * Initialize the control inode/socket with a control endpoint data
702 * structure. This endpoint is reserved exclusively for the OOTB processing.
704 static int sctp_ctl_sock_init(void)
709 if (sctp_get_pf_specific(PF_INET6
))
714 err
= inet_ctl_sock_create(&sctp_ctl_sock
, family
,
715 SOCK_SEQPACKET
, IPPROTO_SCTP
, &init_net
);
718 "SCTP: Failed to create the SCTP control socket.\n");
724 /* Register address family specific functions. */
725 int sctp_register_af(struct sctp_af
*af
)
727 switch (af
->sa_family
) {
729 if (sctp_af_v4_specific
)
731 sctp_af_v4_specific
= af
;
734 if (sctp_af_v6_specific
)
736 sctp_af_v6_specific
= af
;
742 INIT_LIST_HEAD(&af
->list
);
743 list_add_tail(&af
->list
, &sctp_address_families
);
747 /* Get the table of functions for manipulating a particular address
750 struct sctp_af
*sctp_get_af_specific(sa_family_t family
)
754 return sctp_af_v4_specific
;
756 return sctp_af_v6_specific
;
762 /* Common code to initialize a AF_INET msg_name. */
763 static void sctp_inet_msgname(char *msgname
, int *addr_len
)
765 struct sockaddr_in
*sin
;
767 sin
= (struct sockaddr_in
*)msgname
;
768 *addr_len
= sizeof(struct sockaddr_in
);
769 sin
->sin_family
= AF_INET
;
770 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
773 /* Copy the primary address of the peer primary address as the msg_name. */
774 static void sctp_inet_event_msgname(struct sctp_ulpevent
*event
, char *msgname
,
777 struct sockaddr_in
*sin
, *sinfrom
;
780 struct sctp_association
*asoc
;
783 sctp_inet_msgname(msgname
, addr_len
);
784 sin
= (struct sockaddr_in
*)msgname
;
785 sinfrom
= &asoc
->peer
.primary_addr
.v4
;
786 sin
->sin_port
= htons(asoc
->peer
.port
);
787 sin
->sin_addr
.s_addr
= sinfrom
->sin_addr
.s_addr
;
791 /* Initialize and copy out a msgname from an inbound skb. */
792 static void sctp_inet_skb_msgname(struct sk_buff
*skb
, char *msgname
, int *len
)
795 struct sctphdr
*sh
= sctp_hdr(skb
);
796 struct sockaddr_in
*sin
= (struct sockaddr_in
*)msgname
;
798 sctp_inet_msgname(msgname
, len
);
799 sin
->sin_port
= sh
->source
;
800 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
804 /* Do we support this AF? */
805 static int sctp_inet_af_supported(sa_family_t family
, struct sctp_sock
*sp
)
807 /* PF_INET only supports AF_INET addresses. */
808 return (AF_INET
== family
);
811 /* Address matching with wildcards allowed. */
812 static int sctp_inet_cmp_addr(const union sctp_addr
*addr1
,
813 const union sctp_addr
*addr2
,
814 struct sctp_sock
*opt
)
816 /* PF_INET only supports AF_INET addresses. */
817 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
819 if (htonl(INADDR_ANY
) == addr1
->v4
.sin_addr
.s_addr
||
820 htonl(INADDR_ANY
) == addr2
->v4
.sin_addr
.s_addr
)
822 if (addr1
->v4
.sin_addr
.s_addr
== addr2
->v4
.sin_addr
.s_addr
)
828 /* Verify that provided sockaddr looks bindable. Common verification has
829 * already been taken care of.
831 static int sctp_inet_bind_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
833 return sctp_v4_available(addr
, opt
);
836 /* Verify that sockaddr looks sendable. Common verification has already
837 * been taken care of.
839 static int sctp_inet_send_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
844 /* Fill in Supported Address Type information for INIT and INIT-ACK
845 * chunks. Returns number of addresses supported.
847 static int sctp_inet_supported_addrs(const struct sctp_sock
*opt
,
850 types
[0] = SCTP_PARAM_IPV4_ADDRESS
;
854 /* Wrapper routine that calls the ip transmit routine. */
855 static inline int sctp_v4_xmit(struct sk_buff
*skb
,
856 struct sctp_transport
*transport
, int ipfragok
)
858 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, "
859 "src:%u.%u.%u.%u, dst:%u.%u.%u.%u\n",
860 __func__
, skb
, skb
->len
,
861 NIPQUAD(skb
->rtable
->rt_src
),
862 NIPQUAD(skb
->rtable
->rt_dst
));
864 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS
);
865 return ip_queue_xmit(skb
, ipfragok
);
868 static struct sctp_af sctp_af_inet
;
870 static struct sctp_pf sctp_pf_inet
= {
871 .event_msgname
= sctp_inet_event_msgname
,
872 .skb_msgname
= sctp_inet_skb_msgname
,
873 .af_supported
= sctp_inet_af_supported
,
874 .cmp_addr
= sctp_inet_cmp_addr
,
875 .bind_verify
= sctp_inet_bind_verify
,
876 .send_verify
= sctp_inet_send_verify
,
877 .supported_addrs
= sctp_inet_supported_addrs
,
878 .create_accept_sk
= sctp_v4_create_accept_sk
,
879 .addr_v4map
= sctp_v4_addr_v4map
,
883 /* Notifier for inetaddr addition/deletion events. */
884 static struct notifier_block sctp_inetaddr_notifier
= {
885 .notifier_call
= sctp_inetaddr_event
,
888 /* Socket operations. */
889 static const struct proto_ops inet_seqpacket_ops
= {
891 .owner
= THIS_MODULE
,
892 .release
= inet_release
, /* Needs to be wrapped... */
894 .connect
= inet_dgram_connect
,
895 .socketpair
= sock_no_socketpair
,
896 .accept
= inet_accept
,
897 .getname
= inet_getname
, /* Semantics are different. */
900 .listen
= sctp_inet_listen
,
901 .shutdown
= inet_shutdown
, /* Looks harmless. */
902 .setsockopt
= sock_common_setsockopt
, /* IP_SOL IP_OPTION is a problem */
903 .getsockopt
= sock_common_getsockopt
,
904 .sendmsg
= inet_sendmsg
,
905 .recvmsg
= sock_common_recvmsg
,
906 .mmap
= sock_no_mmap
,
907 .sendpage
= sock_no_sendpage
,
909 .compat_setsockopt
= compat_sock_common_setsockopt
,
910 .compat_getsockopt
= compat_sock_common_getsockopt
,
914 /* Registration with AF_INET family. */
915 static struct inet_protosw sctp_seqpacket_protosw
= {
916 .type
= SOCK_SEQPACKET
,
917 .protocol
= IPPROTO_SCTP
,
919 .ops
= &inet_seqpacket_ops
,
922 .flags
= SCTP_PROTOSW_FLAG
924 static struct inet_protosw sctp_stream_protosw
= {
926 .protocol
= IPPROTO_SCTP
,
928 .ops
= &inet_seqpacket_ops
,
931 .flags
= SCTP_PROTOSW_FLAG
934 /* Register with IP layer. */
935 static struct net_protocol sctp_protocol
= {
937 .err_handler
= sctp_v4_err
,
941 /* IPv4 address related functions. */
942 static struct sctp_af sctp_af_inet
= {
943 .sa_family
= AF_INET
,
944 .sctp_xmit
= sctp_v4_xmit
,
945 .setsockopt
= ip_setsockopt
,
946 .getsockopt
= ip_getsockopt
,
947 .get_dst
= sctp_v4_get_dst
,
948 .get_saddr
= sctp_v4_get_saddr
,
949 .copy_addrlist
= sctp_v4_copy_addrlist
,
950 .from_skb
= sctp_v4_from_skb
,
951 .from_sk
= sctp_v4_from_sk
,
952 .to_sk_saddr
= sctp_v4_to_sk_saddr
,
953 .to_sk_daddr
= sctp_v4_to_sk_daddr
,
954 .from_addr_param
= sctp_v4_from_addr_param
,
955 .to_addr_param
= sctp_v4_to_addr_param
,
956 .dst_saddr
= sctp_v4_dst_saddr
,
957 .cmp_addr
= sctp_v4_cmp_addr
,
958 .addr_valid
= sctp_v4_addr_valid
,
959 .inaddr_any
= sctp_v4_inaddr_any
,
960 .is_any
= sctp_v4_is_any
,
961 .available
= sctp_v4_available
,
962 .scope
= sctp_v4_scope
,
963 .skb_iif
= sctp_v4_skb_iif
,
964 .is_ce
= sctp_v4_is_ce
,
965 .seq_dump_addr
= sctp_v4_seq_dump_addr
,
966 .ecn_capable
= sctp_v4_ecn_capable
,
967 .net_header_len
= sizeof(struct iphdr
),
968 .sockaddr_len
= sizeof(struct sockaddr_in
),
970 .compat_setsockopt
= compat_ip_setsockopt
,
971 .compat_getsockopt
= compat_ip_getsockopt
,
975 struct sctp_pf
*sctp_get_pf_specific(sa_family_t family
) {
979 return sctp_pf_inet_specific
;
981 return sctp_pf_inet6_specific
;
987 /* Register the PF specific function table. */
988 int sctp_register_pf(struct sctp_pf
*pf
, sa_family_t family
)
992 if (sctp_pf_inet_specific
)
994 sctp_pf_inet_specific
= pf
;
997 if (sctp_pf_inet6_specific
)
999 sctp_pf_inet6_specific
= pf
;
1007 static inline int init_sctp_mibs(void)
1009 return snmp_mib_init((void**)sctp_statistics
, sizeof(struct sctp_mib
));
1012 static inline void cleanup_sctp_mibs(void)
1014 snmp_mib_free((void**)sctp_statistics
);
1017 static void sctp_v4_pf_init(void)
1019 /* Initialize the SCTP specific PF functions. */
1020 sctp_register_pf(&sctp_pf_inet
, PF_INET
);
1021 sctp_register_af(&sctp_af_inet
);
1024 static void sctp_v4_pf_exit(void)
1026 list_del(&sctp_af_inet
.list
);
1029 static int sctp_v4_protosw_init(void)
1033 rc
= proto_register(&sctp_prot
, 1);
1037 /* Register SCTP(UDP and TCP style) with socket layer. */
1038 inet_register_protosw(&sctp_seqpacket_protosw
);
1039 inet_register_protosw(&sctp_stream_protosw
);
1044 static void sctp_v4_protosw_exit(void)
1046 inet_unregister_protosw(&sctp_stream_protosw
);
1047 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1048 proto_unregister(&sctp_prot
);
1051 static int sctp_v4_add_protocol(void)
1053 /* Register notifier for inet address additions/deletions. */
1054 register_inetaddr_notifier(&sctp_inetaddr_notifier
);
1056 /* Register SCTP with inet layer. */
1057 if (inet_add_protocol(&sctp_protocol
, IPPROTO_SCTP
) < 0)
1063 static void sctp_v4_del_protocol(void)
1065 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1066 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1069 /* Initialize the universe into something sensible. */
1070 SCTP_STATIC __init
int sctp_init(void)
1073 int status
= -EINVAL
;
1075 unsigned long limit
;
1079 /* SCTP_DEBUG sanity check. */
1080 if (!sctp_sanity_check())
1083 /* Allocate bind_bucket and chunk caches. */
1085 sctp_bucket_cachep
= kmem_cache_create("sctp_bind_bucket",
1086 sizeof(struct sctp_bind_bucket
),
1087 0, SLAB_HWCACHE_ALIGN
,
1089 if (!sctp_bucket_cachep
)
1092 sctp_chunk_cachep
= kmem_cache_create("sctp_chunk",
1093 sizeof(struct sctp_chunk
),
1094 0, SLAB_HWCACHE_ALIGN
,
1096 if (!sctp_chunk_cachep
)
1097 goto err_chunk_cachep
;
1099 /* Allocate and initialise sctp mibs. */
1100 status
= init_sctp_mibs();
1104 /* Initialize proc fs directory. */
1105 status
= sctp_proc_init();
1109 /* Initialize object count debugging. */
1110 sctp_dbg_objcnt_init();
1113 * 14. Suggested SCTP Protocol Parameter Values
1115 /* The following protocol parameters are RECOMMENDED: */
1116 /* RTO.Initial - 3 seconds */
1117 sctp_rto_initial
= SCTP_RTO_INITIAL
;
1118 /* RTO.Min - 1 second */
1119 sctp_rto_min
= SCTP_RTO_MIN
;
1120 /* RTO.Max - 60 seconds */
1121 sctp_rto_max
= SCTP_RTO_MAX
;
1122 /* RTO.Alpha - 1/8 */
1123 sctp_rto_alpha
= SCTP_RTO_ALPHA
;
1124 /* RTO.Beta - 1/4 */
1125 sctp_rto_beta
= SCTP_RTO_BETA
;
1127 /* Valid.Cookie.Life - 60 seconds */
1128 sctp_valid_cookie_life
= SCTP_DEFAULT_COOKIE_LIFE
;
1130 /* Whether Cookie Preservative is enabled(1) or not(0) */
1131 sctp_cookie_preserve_enable
= 1;
1134 sctp_max_burst
= SCTP_DEFAULT_MAX_BURST
;
1136 /* Association.Max.Retrans - 10 attempts
1137 * Path.Max.Retrans - 5 attempts (per destination address)
1138 * Max.Init.Retransmits - 8 attempts
1140 sctp_max_retrans_association
= 10;
1141 sctp_max_retrans_path
= 5;
1142 sctp_max_retrans_init
= 8;
1144 /* Sendbuffer growth - do per-socket accounting */
1145 sctp_sndbuf_policy
= 0;
1147 /* Rcvbuffer growth - do per-socket accounting */
1148 sctp_rcvbuf_policy
= 0;
1150 /* HB.interval - 30 seconds */
1151 sctp_hb_interval
= SCTP_DEFAULT_TIMEOUT_HEARTBEAT
;
1153 /* delayed SACK timeout */
1154 sctp_sack_timeout
= SCTP_DEFAULT_TIMEOUT_SACK
;
1156 /* Implementation specific variables. */
1158 /* Initialize default stream count setup information. */
1159 sctp_max_instreams
= SCTP_DEFAULT_INSTREAMS
;
1160 sctp_max_outstreams
= SCTP_DEFAULT_OUTSTREAMS
;
1162 /* Initialize handle used for association ids. */
1163 idr_init(&sctp_assocs_id
);
1165 /* Set the pressure threshold to be a fraction of global memory that
1166 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1167 * memory, with a floor of 128 pages.
1168 * Note this initalizes the data in sctpv6_prot too
1169 * Unabashedly stolen from tcp_init
1171 limit
= min(num_physpages
, 1UL<<(28-PAGE_SHIFT
)) >> (20-PAGE_SHIFT
);
1172 limit
= (limit
* (num_physpages
>> (20-PAGE_SHIFT
))) >> (PAGE_SHIFT
-11);
1173 limit
= max(limit
, 128UL);
1174 sysctl_sctp_mem
[0] = limit
/ 4 * 3;
1175 sysctl_sctp_mem
[1] = limit
;
1176 sysctl_sctp_mem
[2] = sysctl_sctp_mem
[0] * 2;
1178 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1179 limit
= (sysctl_sctp_mem
[1]) << (PAGE_SHIFT
- 7);
1180 max_share
= min(4UL*1024*1024, limit
);
1182 sysctl_sctp_rmem
[0] = PAGE_SIZE
; /* give each asoc 1 page min */
1183 sysctl_sctp_rmem
[1] = (1500 *(sizeof(struct sk_buff
) + 1));
1184 sysctl_sctp_rmem
[2] = max(sysctl_sctp_rmem
[1], max_share
);
1186 sysctl_sctp_wmem
[0] = SK_MEM_QUANTUM
;
1187 sysctl_sctp_wmem
[1] = 16*1024;
1188 sysctl_sctp_wmem
[2] = max(64*1024, max_share
);
1190 /* Size and allocate the association hash table.
1191 * The methodology is similar to that of the tcp hash tables.
1193 if (num_physpages
>= (128 * 1024))
1194 goal
= num_physpages
>> (22 - PAGE_SHIFT
);
1196 goal
= num_physpages
>> (24 - PAGE_SHIFT
);
1198 for (order
= 0; (1UL << order
) < goal
; order
++)
1202 sctp_assoc_hashsize
= (1UL << order
) * PAGE_SIZE
/
1203 sizeof(struct sctp_hashbucket
);
1204 if ((sctp_assoc_hashsize
> (64 * 1024)) && order
> 0)
1206 sctp_assoc_hashtable
= (struct sctp_hashbucket
*)
1207 __get_free_pages(GFP_ATOMIC
, order
);
1208 } while (!sctp_assoc_hashtable
&& --order
> 0);
1209 if (!sctp_assoc_hashtable
) {
1210 printk(KERN_ERR
"SCTP: Failed association hash alloc.\n");
1212 goto err_ahash_alloc
;
1214 for (i
= 0; i
< sctp_assoc_hashsize
; i
++) {
1215 rwlock_init(&sctp_assoc_hashtable
[i
].lock
);
1216 INIT_HLIST_HEAD(&sctp_assoc_hashtable
[i
].chain
);
1219 /* Allocate and initialize the endpoint hash table. */
1220 sctp_ep_hashsize
= 64;
1221 sctp_ep_hashtable
= (struct sctp_hashbucket
*)
1222 kmalloc(64 * sizeof(struct sctp_hashbucket
), GFP_KERNEL
);
1223 if (!sctp_ep_hashtable
) {
1224 printk(KERN_ERR
"SCTP: Failed endpoint_hash alloc.\n");
1226 goto err_ehash_alloc
;
1228 for (i
= 0; i
< sctp_ep_hashsize
; i
++) {
1229 rwlock_init(&sctp_ep_hashtable
[i
].lock
);
1230 INIT_HLIST_HEAD(&sctp_ep_hashtable
[i
].chain
);
1233 /* Allocate and initialize the SCTP port hash table. */
1235 sctp_port_hashsize
= (1UL << order
) * PAGE_SIZE
/
1236 sizeof(struct sctp_bind_hashbucket
);
1237 if ((sctp_port_hashsize
> (64 * 1024)) && order
> 0)
1239 sctp_port_hashtable
= (struct sctp_bind_hashbucket
*)
1240 __get_free_pages(GFP_ATOMIC
, order
);
1241 } while (!sctp_port_hashtable
&& --order
> 0);
1242 if (!sctp_port_hashtable
) {
1243 printk(KERN_ERR
"SCTP: Failed bind hash alloc.");
1245 goto err_bhash_alloc
;
1247 for (i
= 0; i
< sctp_port_hashsize
; i
++) {
1248 spin_lock_init(&sctp_port_hashtable
[i
].lock
);
1249 INIT_HLIST_HEAD(&sctp_port_hashtable
[i
].chain
);
1252 printk(KERN_INFO
"SCTP: Hash tables configured "
1253 "(established %d bind %d)\n",
1254 sctp_assoc_hashsize
, sctp_port_hashsize
);
1256 /* Disable ADDIP by default. */
1257 sctp_addip_enable
= 0;
1258 sctp_addip_noauth
= 0;
1260 /* Enable PR-SCTP by default. */
1261 sctp_prsctp_enable
= 1;
1263 /* Disable AUTH by default. */
1264 sctp_auth_enable
= 0;
1266 sctp_sysctl_register();
1268 INIT_LIST_HEAD(&sctp_address_families
);
1272 /* Initialize the local address list. */
1273 INIT_LIST_HEAD(&sctp_local_addr_list
);
1274 spin_lock_init(&sctp_local_addr_lock
);
1275 sctp_get_local_addr_list();
1277 status
= sctp_v4_protosw_init();
1280 goto err_protosw_init
;
1282 status
= sctp_v6_protosw_init();
1284 goto err_v6_protosw_init
;
1286 /* Initialize the control inode/socket for handling OOTB packets. */
1287 if ((status
= sctp_ctl_sock_init())) {
1289 "SCTP: Failed to initialize the SCTP control sock.\n");
1290 goto err_ctl_sock_init
;
1293 status
= sctp_v4_add_protocol();
1295 goto err_add_protocol
;
1297 /* Register SCTP with inet6 layer. */
1298 status
= sctp_v6_add_protocol();
1300 goto err_v6_add_protocol
;
1305 err_v6_add_protocol
:
1306 sctp_v6_del_protocol();
1308 sctp_v4_del_protocol();
1309 inet_ctl_sock_destroy(sctp_ctl_sock
);
1311 sctp_v6_protosw_exit();
1312 err_v6_protosw_init
:
1313 sctp_v4_protosw_exit();
1315 sctp_free_local_addr_list();
1318 sctp_sysctl_unregister();
1319 list_del(&sctp_af_inet
.list
);
1320 free_pages((unsigned long)sctp_port_hashtable
,
1321 get_order(sctp_port_hashsize
*
1322 sizeof(struct sctp_bind_hashbucket
)));
1324 kfree(sctp_ep_hashtable
);
1326 free_pages((unsigned long)sctp_assoc_hashtable
,
1327 get_order(sctp_assoc_hashsize
*
1328 sizeof(struct sctp_hashbucket
)));
1330 sctp_dbg_objcnt_exit();
1333 cleanup_sctp_mibs();
1335 kmem_cache_destroy(sctp_chunk_cachep
);
1337 kmem_cache_destroy(sctp_bucket_cachep
);
1341 /* Exit handler for the SCTP protocol. */
1342 SCTP_STATIC __exit
void sctp_exit(void)
1344 /* BUG. This should probably do something useful like clean
1345 * up all the remaining associations and all that memory.
1348 /* Unregister with inet6/inet layers. */
1349 sctp_v6_del_protocol();
1350 sctp_v4_del_protocol();
1352 /* Free the control endpoint. */
1353 inet_ctl_sock_destroy(sctp_ctl_sock
);
1355 /* Free protosw registrations */
1356 sctp_v6_protosw_exit();
1357 sctp_v4_protosw_exit();
1359 /* Free the local address list. */
1360 sctp_free_local_addr_list();
1362 /* Unregister with socket layer. */
1366 sctp_sysctl_unregister();
1367 list_del(&sctp_af_inet
.list
);
1369 free_pages((unsigned long)sctp_assoc_hashtable
,
1370 get_order(sctp_assoc_hashsize
*
1371 sizeof(struct sctp_hashbucket
)));
1372 kfree(sctp_ep_hashtable
);
1373 free_pages((unsigned long)sctp_port_hashtable
,
1374 get_order(sctp_port_hashsize
*
1375 sizeof(struct sctp_bind_hashbucket
)));
1377 sctp_dbg_objcnt_exit();
1379 cleanup_sctp_mibs();
1381 kmem_cache_destroy(sctp_chunk_cachep
);
1382 kmem_cache_destroy(sctp_bucket_cachep
);
1385 module_init(sctp_init
);
1386 module_exit(sctp_exit
);
1389 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1391 MODULE_ALIAS("net-pf-" __stringify(PF_INET
) "-proto-132");
1392 MODULE_ALIAS("net-pf-" __stringify(PF_INET6
) "-proto-132");
1393 MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1394 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1395 MODULE_LICENSE("GPL");