1 /* SCTP kernel reference 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 reference Implementation
11 * Initialization/cleanup for SCTP protocol support.
13 * The SCTP reference 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 * The SCTP reference 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 socket
*sctp_ctl_socket
;
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)
94 return sctp_ctl_socket
->sk
;
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())
123 /* Clean up the proc fs entry for the SCTP protocol.
124 * Note: Do not make this __exit as it is used in the init error
127 static void sctp_proc_exit(void)
129 sctp_snmp_proc_exit();
130 sctp_eps_proc_exit();
131 sctp_assocs_proc_exit();
134 proc_net_sctp
= NULL
;
135 remove_proc_entry("sctp", init_net
.proc_net
);
139 /* Private helper to extract ipv4 address and stash them in
140 * the protocol structure.
142 static void sctp_v4_copy_addrlist(struct list_head
*addrlist
,
143 struct net_device
*dev
)
145 struct in_device
*in_dev
;
146 struct in_ifaddr
*ifa
;
147 struct sctp_sockaddr_entry
*addr
;
150 if ((in_dev
= __in_dev_get_rcu(dev
)) == NULL
) {
155 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
156 /* Add the address to the local list. */
157 addr
= t_new(struct sctp_sockaddr_entry
, GFP_ATOMIC
);
159 addr
->a
.v4
.sin_family
= AF_INET
;
160 addr
->a
.v4
.sin_port
= 0;
161 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
163 INIT_LIST_HEAD(&addr
->list
);
164 INIT_RCU_HEAD(&addr
->rcu
);
165 list_add_tail(&addr
->list
, addrlist
);
172 /* Extract our IP addresses from the system and stash them in the
173 * protocol structure.
175 static void sctp_get_local_addr_list(void)
177 struct net_device
*dev
;
178 struct list_head
*pos
;
181 read_lock(&dev_base_lock
);
182 for_each_netdev(&init_net
, dev
) {
183 __list_for_each(pos
, &sctp_address_families
) {
184 af
= list_entry(pos
, struct sctp_af
, list
);
185 af
->copy_addrlist(&sctp_local_addr_list
, dev
);
188 read_unlock(&dev_base_lock
);
191 /* Free the existing local addresses. */
192 static void sctp_free_local_addr_list(void)
194 struct sctp_sockaddr_entry
*addr
;
195 struct list_head
*pos
, *temp
;
197 list_for_each_safe(pos
, temp
, &sctp_local_addr_list
) {
198 addr
= list_entry(pos
, struct sctp_sockaddr_entry
, list
);
204 void sctp_local_addr_free(struct rcu_head
*head
)
206 struct sctp_sockaddr_entry
*e
= container_of(head
,
207 struct sctp_sockaddr_entry
, rcu
);
211 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
212 int sctp_copy_local_addr_list(struct sctp_bind_addr
*bp
, sctp_scope_t scope
,
213 gfp_t gfp
, int copy_flags
)
215 struct sctp_sockaddr_entry
*addr
;
219 list_for_each_entry_rcu(addr
, &sctp_local_addr_list
, list
) {
222 if (sctp_in_scope(&addr
->a
, scope
)) {
223 /* Now that the address is in scope, check to see if
224 * the address type is really supported by the local
225 * sock as well as the remote peer.
227 if ((((AF_INET
== addr
->a
.sa
.sa_family
) &&
228 (copy_flags
& SCTP_ADDR4_PEERSUPP
))) ||
229 (((AF_INET6
== addr
->a
.sa
.sa_family
) &&
230 (copy_flags
& SCTP_ADDR6_ALLOWED
) &&
231 (copy_flags
& SCTP_ADDR6_PEERSUPP
)))) {
232 error
= sctp_add_bind_addr(bp
, &addr
->a
,
233 SCTP_ADDR_SRC
, GFP_ATOMIC
);
245 /* Initialize a sctp_addr from in incoming skb. */
246 static void sctp_v4_from_skb(union sctp_addr
*addr
, struct sk_buff
*skb
,
253 port
= &addr
->v4
.sin_port
;
254 addr
->v4
.sin_family
= AF_INET
;
259 from
= &ip_hdr(skb
)->saddr
;
262 from
= &ip_hdr(skb
)->daddr
;
264 memcpy(&addr
->v4
.sin_addr
.s_addr
, from
, sizeof(struct in_addr
));
267 /* Initialize an sctp_addr from a socket. */
268 static void sctp_v4_from_sk(union sctp_addr
*addr
, struct sock
*sk
)
270 addr
->v4
.sin_family
= AF_INET
;
271 addr
->v4
.sin_port
= 0;
272 addr
->v4
.sin_addr
.s_addr
= inet_sk(sk
)->rcv_saddr
;
275 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
276 static void sctp_v4_to_sk_saddr(union sctp_addr
*addr
, struct sock
*sk
)
278 inet_sk(sk
)->rcv_saddr
= addr
->v4
.sin_addr
.s_addr
;
281 /* Initialize sk->sk_daddr from sctp_addr. */
282 static void sctp_v4_to_sk_daddr(union sctp_addr
*addr
, struct sock
*sk
)
284 inet_sk(sk
)->daddr
= addr
->v4
.sin_addr
.s_addr
;
287 /* Initialize a sctp_addr from an address parameter. */
288 static void sctp_v4_from_addr_param(union sctp_addr
*addr
,
289 union sctp_addr_param
*param
,
290 __be16 port
, int iif
)
292 addr
->v4
.sin_family
= AF_INET
;
293 addr
->v4
.sin_port
= port
;
294 addr
->v4
.sin_addr
.s_addr
= param
->v4
.addr
.s_addr
;
297 /* Initialize an address parameter from a sctp_addr and return the length
298 * of the address parameter.
300 static int sctp_v4_to_addr_param(const union sctp_addr
*addr
,
301 union sctp_addr_param
*param
)
303 int length
= sizeof(sctp_ipv4addr_param_t
);
305 param
->v4
.param_hdr
.type
= SCTP_PARAM_IPV4_ADDRESS
;
306 param
->v4
.param_hdr
.length
= htons(length
);
307 param
->v4
.addr
.s_addr
= addr
->v4
.sin_addr
.s_addr
;
312 /* Initialize a sctp_addr from a dst_entry. */
313 static void sctp_v4_dst_saddr(union sctp_addr
*saddr
, struct dst_entry
*dst
,
316 struct rtable
*rt
= (struct rtable
*)dst
;
317 saddr
->v4
.sin_family
= AF_INET
;
318 saddr
->v4
.sin_port
= port
;
319 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
322 /* Compare two addresses exactly. */
323 static int sctp_v4_cmp_addr(const union sctp_addr
*addr1
,
324 const union sctp_addr
*addr2
)
326 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
328 if (addr1
->v4
.sin_port
!= addr2
->v4
.sin_port
)
330 if (addr1
->v4
.sin_addr
.s_addr
!= addr2
->v4
.sin_addr
.s_addr
)
336 /* Initialize addr struct to INADDR_ANY. */
337 static void sctp_v4_inaddr_any(union sctp_addr
*addr
, __be16 port
)
339 addr
->v4
.sin_family
= AF_INET
;
340 addr
->v4
.sin_addr
.s_addr
= INADDR_ANY
;
341 addr
->v4
.sin_port
= port
;
344 /* Is this a wildcard address? */
345 static int sctp_v4_is_any(const union sctp_addr
*addr
)
347 return INADDR_ANY
== addr
->v4
.sin_addr
.s_addr
;
350 /* This function checks if the address is a valid address to be used for
354 * Return 0 - If the address is a non-unicast or an illegal address.
355 * Return 1 - If the address is a unicast.
357 static int sctp_v4_addr_valid(union sctp_addr
*addr
,
358 struct sctp_sock
*sp
,
359 const struct sk_buff
*skb
)
361 /* Is this a non-unicast address or a unusable SCTP address? */
362 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
))
365 /* Is this a broadcast address? */
366 if (skb
&& ((struct rtable
*)skb
->dst
)->rt_flags
& RTCF_BROADCAST
)
372 /* Should this be available for binding? */
373 static int sctp_v4_available(union sctp_addr
*addr
, struct sctp_sock
*sp
)
375 int ret
= inet_addr_type(addr
->v4
.sin_addr
.s_addr
);
378 if (addr
->v4
.sin_addr
.s_addr
!= INADDR_ANY
&&
380 !sp
->inet
.freebind
&&
381 !sysctl_ip_nonlocal_bind
)
387 /* Checking the loopback, private and other address scopes as defined in
388 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
389 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
391 * Level 0 - unusable SCTP addresses
392 * Level 1 - loopback address
393 * Level 2 - link-local addresses
394 * Level 3 - private addresses.
395 * Level 4 - global addresses
396 * For INIT and INIT-ACK address list, let L be the level of
397 * of requested destination address, sender and receiver
398 * SHOULD include all of its addresses with level greater
399 * than or equal to L.
401 static sctp_scope_t
sctp_v4_scope(union sctp_addr
*addr
)
405 /* Should IPv4 scoping be a sysctl configurable option
406 * so users can turn it off (default on) for certain
407 * unconventional networking environments?
410 /* Check for unusable SCTP addresses. */
411 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
)) {
412 retval
= SCTP_SCOPE_UNUSABLE
;
413 } else if (ipv4_is_loopback(addr
->v4
.sin_addr
.s_addr
)) {
414 retval
= SCTP_SCOPE_LOOPBACK
;
415 } else if (ipv4_is_linklocal_169(addr
->v4
.sin_addr
.s_addr
)) {
416 retval
= SCTP_SCOPE_LINK
;
417 } else if (ipv4_is_private_10(addr
->v4
.sin_addr
.s_addr
) ||
418 ipv4_is_private_172(addr
->v4
.sin_addr
.s_addr
) ||
419 ipv4_is_private_192(addr
->v4
.sin_addr
.s_addr
)) {
420 retval
= SCTP_SCOPE_PRIVATE
;
422 retval
= SCTP_SCOPE_GLOBAL
;
428 /* Returns a valid dst cache entry for the given source and destination ip
429 * addresses. If an association is passed, trys to get a dst entry with a
430 * source address that matches an address in the bind address list.
432 static struct dst_entry
*sctp_v4_get_dst(struct sctp_association
*asoc
,
433 union sctp_addr
*daddr
,
434 union sctp_addr
*saddr
)
438 struct sctp_bind_addr
*bp
;
439 struct sctp_sockaddr_entry
*laddr
;
440 struct dst_entry
*dst
= NULL
;
441 union sctp_addr dst_saddr
;
443 memset(&fl
, 0x0, sizeof(struct flowi
));
444 fl
.fl4_dst
= daddr
->v4
.sin_addr
.s_addr
;
445 fl
.proto
= IPPROTO_SCTP
;
447 fl
.fl4_tos
= RT_CONN_FLAGS(asoc
->base
.sk
);
448 fl
.oif
= asoc
->base
.sk
->sk_bound_dev_if
;
451 fl
.fl4_src
= saddr
->v4
.sin_addr
.s_addr
;
453 SCTP_DEBUG_PRINTK("%s: DST:%u.%u.%u.%u, SRC:%u.%u.%u.%u - ",
454 __FUNCTION__
, NIPQUAD(fl
.fl4_dst
),
455 NIPQUAD(fl
.fl4_src
));
457 if (!ip_route_output_key(&rt
, &fl
)) {
461 /* If there is no association or if a source address is passed, no
462 * more validation is required.
467 bp
= &asoc
->base
.bind_addr
;
470 /* Walk through the bind address list and look for a bind
471 * address that matches the source address of the returned dst.
474 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
475 if (!laddr
->valid
|| (laddr
->state
!= SCTP_ADDR_SRC
))
477 sctp_v4_dst_saddr(&dst_saddr
, dst
, htons(bp
->port
));
478 if (sctp_v4_cmp_addr(&dst_saddr
, &laddr
->a
))
483 /* None of the bound addresses match the source address of the
484 * dst. So release it.
490 /* Walk through the bind address list and try to get a dst that
491 * matches a bind address as the source address.
494 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
497 if ((laddr
->state
== SCTP_ADDR_SRC
) &&
498 (AF_INET
== laddr
->a
.sa
.sa_family
)) {
499 fl
.fl4_src
= laddr
->a
.v4
.sin_addr
.s_addr
;
500 if (!ip_route_output_key(&rt
, &fl
)) {
511 SCTP_DEBUG_PRINTK("rt_dst:%u.%u.%u.%u, rt_src:%u.%u.%u.%u\n",
512 NIPQUAD(rt
->rt_dst
), NIPQUAD(rt
->rt_src
));
514 SCTP_DEBUG_PRINTK("NO ROUTE\n");
519 /* For v4, the source address is cached in the route entry(dst). So no need
520 * to cache it separately and hence this is an empty routine.
522 static void sctp_v4_get_saddr(struct sctp_association
*asoc
,
523 struct dst_entry
*dst
,
524 union sctp_addr
*daddr
,
525 union sctp_addr
*saddr
)
527 struct rtable
*rt
= (struct rtable
*)dst
;
533 saddr
->v4
.sin_family
= AF_INET
;
534 saddr
->v4
.sin_port
= htons(asoc
->base
.bind_addr
.port
);
535 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
539 /* What interface did this skb arrive on? */
540 static int sctp_v4_skb_iif(const struct sk_buff
*skb
)
542 return ((struct rtable
*)skb
->dst
)->rt_iif
;
545 /* Was this packet marked by Explicit Congestion Notification? */
546 static int sctp_v4_is_ce(const struct sk_buff
*skb
)
548 return INET_ECN_is_ce(ip_hdr(skb
)->tos
);
551 /* Create and initialize a new sk for the socket returned by accept(). */
552 static struct sock
*sctp_v4_create_accept_sk(struct sock
*sk
,
553 struct sctp_association
*asoc
)
555 struct inet_sock
*inet
= inet_sk(sk
);
556 struct inet_sock
*newinet
;
557 struct sock
*newsk
= sk_alloc(sk
->sk_net
, PF_INET
, GFP_KERNEL
,
563 sock_init_data(NULL
, newsk
);
565 newsk
->sk_type
= SOCK_STREAM
;
567 newsk
->sk_no_check
= sk
->sk_no_check
;
568 newsk
->sk_reuse
= sk
->sk_reuse
;
569 newsk
->sk_shutdown
= sk
->sk_shutdown
;
571 newsk
->sk_destruct
= inet_sock_destruct
;
572 newsk
->sk_family
= PF_INET
;
573 newsk
->sk_protocol
= IPPROTO_SCTP
;
574 newsk
->sk_backlog_rcv
= sk
->sk_prot
->backlog_rcv
;
575 sock_reset_flag(newsk
, SOCK_ZAPPED
);
577 newinet
= inet_sk(newsk
);
579 /* Initialize sk's sport, dport, rcv_saddr and daddr for
580 * getsockname() and getpeername()
582 newinet
->sport
= inet
->sport
;
583 newinet
->saddr
= inet
->saddr
;
584 newinet
->rcv_saddr
= inet
->rcv_saddr
;
585 newinet
->dport
= htons(asoc
->peer
.port
);
586 newinet
->daddr
= asoc
->peer
.primary_addr
.v4
.sin_addr
.s_addr
;
587 newinet
->pmtudisc
= inet
->pmtudisc
;
588 newinet
->id
= asoc
->next_tsn
^ jiffies
;
590 newinet
->uc_ttl
= -1;
591 newinet
->mc_loop
= 1;
593 newinet
->mc_index
= 0;
594 newinet
->mc_list
= NULL
;
596 sk_refcnt_debug_inc(newsk
);
598 if (newsk
->sk_prot
->init(newsk
)) {
599 sk_common_release(newsk
);
607 /* Map address, empty for v4 family */
608 static void sctp_v4_addr_v4map(struct sctp_sock
*sp
, union sctp_addr
*addr
)
613 /* Dump the v4 addr to the seq file. */
614 static void sctp_v4_seq_dump_addr(struct seq_file
*seq
, union sctp_addr
*addr
)
616 seq_printf(seq
, "%d.%d.%d.%d ", NIPQUAD(addr
->v4
.sin_addr
));
619 /* Event handler for inet address addition/deletion events.
620 * The sctp_local_addr_list needs to be protocted by a spin lock since
621 * multiple notifiers (say IPv4 and IPv6) may be running at the same
622 * time and thus corrupt the list.
623 * The reader side is protected with RCU.
625 static int sctp_inetaddr_event(struct notifier_block
*this, unsigned long ev
,
628 struct in_ifaddr
*ifa
= (struct in_ifaddr
*)ptr
;
629 struct sctp_sockaddr_entry
*addr
= NULL
;
630 struct sctp_sockaddr_entry
*temp
;
634 addr
= kmalloc(sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
636 addr
->a
.v4
.sin_family
= AF_INET
;
637 addr
->a
.v4
.sin_port
= 0;
638 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
640 spin_lock_bh(&sctp_local_addr_lock
);
641 list_add_tail_rcu(&addr
->list
, &sctp_local_addr_list
);
642 spin_unlock_bh(&sctp_local_addr_lock
);
646 spin_lock_bh(&sctp_local_addr_lock
);
647 list_for_each_entry_safe(addr
, temp
,
648 &sctp_local_addr_list
, list
) {
649 if (addr
->a
.v4
.sin_addr
.s_addr
== ifa
->ifa_local
) {
651 list_del_rcu(&addr
->list
);
655 spin_unlock_bh(&sctp_local_addr_lock
);
656 if (addr
&& !addr
->valid
)
657 call_rcu(&addr
->rcu
, sctp_local_addr_free
);
665 * Initialize the control inode/socket with a control endpoint data
666 * structure. This endpoint is reserved exclusively for the OOTB processing.
668 static int sctp_ctl_sock_init(void)
673 if (sctp_get_pf_specific(PF_INET6
))
678 err
= sock_create_kern(family
, SOCK_SEQPACKET
, IPPROTO_SCTP
,
682 "SCTP: Failed to create the SCTP control socket.\n");
685 sctp_ctl_socket
->sk
->sk_allocation
= GFP_ATOMIC
;
686 inet_sk(sctp_ctl_socket
->sk
)->uc_ttl
= -1;
691 /* Register address family specific functions. */
692 int sctp_register_af(struct sctp_af
*af
)
694 switch (af
->sa_family
) {
696 if (sctp_af_v4_specific
)
698 sctp_af_v4_specific
= af
;
701 if (sctp_af_v6_specific
)
703 sctp_af_v6_specific
= af
;
709 INIT_LIST_HEAD(&af
->list
);
710 list_add_tail(&af
->list
, &sctp_address_families
);
714 /* Get the table of functions for manipulating a particular address
717 struct sctp_af
*sctp_get_af_specific(sa_family_t family
)
721 return sctp_af_v4_specific
;
723 return sctp_af_v6_specific
;
729 /* Common code to initialize a AF_INET msg_name. */
730 static void sctp_inet_msgname(char *msgname
, int *addr_len
)
732 struct sockaddr_in
*sin
;
734 sin
= (struct sockaddr_in
*)msgname
;
735 *addr_len
= sizeof(struct sockaddr_in
);
736 sin
->sin_family
= AF_INET
;
737 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
740 /* Copy the primary address of the peer primary address as the msg_name. */
741 static void sctp_inet_event_msgname(struct sctp_ulpevent
*event
, char *msgname
,
744 struct sockaddr_in
*sin
, *sinfrom
;
747 struct sctp_association
*asoc
;
750 sctp_inet_msgname(msgname
, addr_len
);
751 sin
= (struct sockaddr_in
*)msgname
;
752 sinfrom
= &asoc
->peer
.primary_addr
.v4
;
753 sin
->sin_port
= htons(asoc
->peer
.port
);
754 sin
->sin_addr
.s_addr
= sinfrom
->sin_addr
.s_addr
;
758 /* Initialize and copy out a msgname from an inbound skb. */
759 static void sctp_inet_skb_msgname(struct sk_buff
*skb
, char *msgname
, int *len
)
762 struct sctphdr
*sh
= sctp_hdr(skb
);
763 struct sockaddr_in
*sin
= (struct sockaddr_in
*)msgname
;
765 sctp_inet_msgname(msgname
, len
);
766 sin
->sin_port
= sh
->source
;
767 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
771 /* Do we support this AF? */
772 static int sctp_inet_af_supported(sa_family_t family
, struct sctp_sock
*sp
)
774 /* PF_INET only supports AF_INET addresses. */
775 return (AF_INET
== family
);
778 /* Address matching with wildcards allowed. */
779 static int sctp_inet_cmp_addr(const union sctp_addr
*addr1
,
780 const union sctp_addr
*addr2
,
781 struct sctp_sock
*opt
)
783 /* PF_INET only supports AF_INET addresses. */
784 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
786 if (INADDR_ANY
== addr1
->v4
.sin_addr
.s_addr
||
787 INADDR_ANY
== addr2
->v4
.sin_addr
.s_addr
)
789 if (addr1
->v4
.sin_addr
.s_addr
== addr2
->v4
.sin_addr
.s_addr
)
795 /* Verify that provided sockaddr looks bindable. Common verification has
796 * already been taken care of.
798 static int sctp_inet_bind_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
800 return sctp_v4_available(addr
, opt
);
803 /* Verify that sockaddr looks sendable. Common verification has already
804 * been taken care of.
806 static int sctp_inet_send_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
811 /* Fill in Supported Address Type information for INIT and INIT-ACK
812 * chunks. Returns number of addresses supported.
814 static int sctp_inet_supported_addrs(const struct sctp_sock
*opt
,
817 types
[0] = SCTP_PARAM_IPV4_ADDRESS
;
821 /* Wrapper routine that calls the ip transmit routine. */
822 static inline int sctp_v4_xmit(struct sk_buff
*skb
,
823 struct sctp_transport
*transport
, int ipfragok
)
825 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, "
826 "src:%u.%u.%u.%u, dst:%u.%u.%u.%u\n",
827 __FUNCTION__
, skb
, skb
->len
,
828 NIPQUAD(((struct rtable
*)skb
->dst
)->rt_src
),
829 NIPQUAD(((struct rtable
*)skb
->dst
)->rt_dst
));
831 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS
);
832 return ip_queue_xmit(skb
, ipfragok
);
835 static struct sctp_af sctp_ipv4_specific
;
837 static struct sctp_pf sctp_pf_inet
= {
838 .event_msgname
= sctp_inet_event_msgname
,
839 .skb_msgname
= sctp_inet_skb_msgname
,
840 .af_supported
= sctp_inet_af_supported
,
841 .cmp_addr
= sctp_inet_cmp_addr
,
842 .bind_verify
= sctp_inet_bind_verify
,
843 .send_verify
= sctp_inet_send_verify
,
844 .supported_addrs
= sctp_inet_supported_addrs
,
845 .create_accept_sk
= sctp_v4_create_accept_sk
,
846 .addr_v4map
= sctp_v4_addr_v4map
,
847 .af
= &sctp_ipv4_specific
,
850 /* Notifier for inetaddr addition/deletion events. */
851 static struct notifier_block sctp_inetaddr_notifier
= {
852 .notifier_call
= sctp_inetaddr_event
,
855 /* Socket operations. */
856 static const struct proto_ops inet_seqpacket_ops
= {
858 .owner
= THIS_MODULE
,
859 .release
= inet_release
, /* Needs to be wrapped... */
861 .connect
= inet_dgram_connect
,
862 .socketpair
= sock_no_socketpair
,
863 .accept
= inet_accept
,
864 .getname
= inet_getname
, /* Semantics are different. */
867 .listen
= sctp_inet_listen
,
868 .shutdown
= inet_shutdown
, /* Looks harmless. */
869 .setsockopt
= sock_common_setsockopt
, /* IP_SOL IP_OPTION is a problem */
870 .getsockopt
= sock_common_getsockopt
,
871 .sendmsg
= inet_sendmsg
,
872 .recvmsg
= sock_common_recvmsg
,
873 .mmap
= sock_no_mmap
,
874 .sendpage
= sock_no_sendpage
,
876 .compat_setsockopt
= compat_sock_common_setsockopt
,
877 .compat_getsockopt
= compat_sock_common_getsockopt
,
881 /* Registration with AF_INET family. */
882 static struct inet_protosw sctp_seqpacket_protosw
= {
883 .type
= SOCK_SEQPACKET
,
884 .protocol
= IPPROTO_SCTP
,
886 .ops
= &inet_seqpacket_ops
,
889 .flags
= SCTP_PROTOSW_FLAG
891 static struct inet_protosw sctp_stream_protosw
= {
893 .protocol
= IPPROTO_SCTP
,
895 .ops
= &inet_seqpacket_ops
,
898 .flags
= SCTP_PROTOSW_FLAG
901 /* Register with IP layer. */
902 static struct net_protocol sctp_protocol
= {
904 .err_handler
= sctp_v4_err
,
908 /* IPv4 address related functions. */
909 static struct sctp_af sctp_ipv4_specific
= {
910 .sa_family
= AF_INET
,
911 .sctp_xmit
= sctp_v4_xmit
,
912 .setsockopt
= ip_setsockopt
,
913 .getsockopt
= ip_getsockopt
,
914 .get_dst
= sctp_v4_get_dst
,
915 .get_saddr
= sctp_v4_get_saddr
,
916 .copy_addrlist
= sctp_v4_copy_addrlist
,
917 .from_skb
= sctp_v4_from_skb
,
918 .from_sk
= sctp_v4_from_sk
,
919 .to_sk_saddr
= sctp_v4_to_sk_saddr
,
920 .to_sk_daddr
= sctp_v4_to_sk_daddr
,
921 .from_addr_param
= sctp_v4_from_addr_param
,
922 .to_addr_param
= sctp_v4_to_addr_param
,
923 .dst_saddr
= sctp_v4_dst_saddr
,
924 .cmp_addr
= sctp_v4_cmp_addr
,
925 .addr_valid
= sctp_v4_addr_valid
,
926 .inaddr_any
= sctp_v4_inaddr_any
,
927 .is_any
= sctp_v4_is_any
,
928 .available
= sctp_v4_available
,
929 .scope
= sctp_v4_scope
,
930 .skb_iif
= sctp_v4_skb_iif
,
931 .is_ce
= sctp_v4_is_ce
,
932 .seq_dump_addr
= sctp_v4_seq_dump_addr
,
933 .net_header_len
= sizeof(struct iphdr
),
934 .sockaddr_len
= sizeof(struct sockaddr_in
),
936 .compat_setsockopt
= compat_ip_setsockopt
,
937 .compat_getsockopt
= compat_ip_getsockopt
,
941 struct sctp_pf
*sctp_get_pf_specific(sa_family_t family
) {
945 return sctp_pf_inet_specific
;
947 return sctp_pf_inet6_specific
;
953 /* Register the PF specific function table. */
954 int sctp_register_pf(struct sctp_pf
*pf
, sa_family_t family
)
958 if (sctp_pf_inet_specific
)
960 sctp_pf_inet_specific
= pf
;
963 if (sctp_pf_inet6_specific
)
965 sctp_pf_inet6_specific
= pf
;
973 static int __init
init_sctp_mibs(void)
975 sctp_statistics
[0] = alloc_percpu(struct sctp_mib
);
976 if (!sctp_statistics
[0])
978 sctp_statistics
[1] = alloc_percpu(struct sctp_mib
);
979 if (!sctp_statistics
[1]) {
980 free_percpu(sctp_statistics
[0]);
987 static void cleanup_sctp_mibs(void)
989 free_percpu(sctp_statistics
[0]);
990 free_percpu(sctp_statistics
[1]);
993 /* Initialize the universe into something sensible. */
994 SCTP_STATIC __init
int sctp_init(void)
997 int status
= -EINVAL
;
1003 /* SCTP_DEBUG sanity check. */
1004 if (!sctp_sanity_check())
1007 /* Allocate bind_bucket and chunk caches. */
1009 sctp_bucket_cachep
= kmem_cache_create("sctp_bind_bucket",
1010 sizeof(struct sctp_bind_bucket
),
1011 0, SLAB_HWCACHE_ALIGN
,
1013 if (!sctp_bucket_cachep
)
1016 sctp_chunk_cachep
= kmem_cache_create("sctp_chunk",
1017 sizeof(struct sctp_chunk
),
1018 0, SLAB_HWCACHE_ALIGN
,
1020 if (!sctp_chunk_cachep
)
1021 goto err_chunk_cachep
;
1023 /* Allocate and initialise sctp mibs. */
1024 status
= init_sctp_mibs();
1028 /* Initialize proc fs directory. */
1029 status
= sctp_proc_init();
1033 /* Initialize object count debugging. */
1034 sctp_dbg_objcnt_init();
1036 /* Initialize the SCTP specific PF functions. */
1037 sctp_register_pf(&sctp_pf_inet
, PF_INET
);
1039 * 14. Suggested SCTP Protocol Parameter Values
1041 /* The following protocol parameters are RECOMMENDED: */
1042 /* RTO.Initial - 3 seconds */
1043 sctp_rto_initial
= SCTP_RTO_INITIAL
;
1044 /* RTO.Min - 1 second */
1045 sctp_rto_min
= SCTP_RTO_MIN
;
1046 /* RTO.Max - 60 seconds */
1047 sctp_rto_max
= SCTP_RTO_MAX
;
1048 /* RTO.Alpha - 1/8 */
1049 sctp_rto_alpha
= SCTP_RTO_ALPHA
;
1050 /* RTO.Beta - 1/4 */
1051 sctp_rto_beta
= SCTP_RTO_BETA
;
1053 /* Valid.Cookie.Life - 60 seconds */
1054 sctp_valid_cookie_life
= SCTP_DEFAULT_COOKIE_LIFE
;
1056 /* Whether Cookie Preservative is enabled(1) or not(0) */
1057 sctp_cookie_preserve_enable
= 1;
1060 sctp_max_burst
= SCTP_DEFAULT_MAX_BURST
;
1062 /* Association.Max.Retrans - 10 attempts
1063 * Path.Max.Retrans - 5 attempts (per destination address)
1064 * Max.Init.Retransmits - 8 attempts
1066 sctp_max_retrans_association
= 10;
1067 sctp_max_retrans_path
= 5;
1068 sctp_max_retrans_init
= 8;
1070 /* Sendbuffer growth - do per-socket accounting */
1071 sctp_sndbuf_policy
= 0;
1073 /* Rcvbuffer growth - do per-socket accounting */
1074 sctp_rcvbuf_policy
= 0;
1076 /* HB.interval - 30 seconds */
1077 sctp_hb_interval
= SCTP_DEFAULT_TIMEOUT_HEARTBEAT
;
1079 /* delayed SACK timeout */
1080 sctp_sack_timeout
= SCTP_DEFAULT_TIMEOUT_SACK
;
1082 /* Implementation specific variables. */
1084 /* Initialize default stream count setup information. */
1085 sctp_max_instreams
= SCTP_DEFAULT_INSTREAMS
;
1086 sctp_max_outstreams
= SCTP_DEFAULT_OUTSTREAMS
;
1088 /* Initialize handle used for association ids. */
1089 idr_init(&sctp_assocs_id
);
1091 /* Set the pressure threshold to be a fraction of global memory that
1092 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1093 * memory, with a floor of 128 pages.
1094 * Note this initalizes the data in sctpv6_prot too
1095 * Unabashedly stolen from tcp_init
1097 limit
= min(num_physpages
, 1UL<<(28-PAGE_SHIFT
)) >> (20-PAGE_SHIFT
);
1098 limit
= (limit
* (num_physpages
>> (20-PAGE_SHIFT
))) >> (PAGE_SHIFT
-11);
1099 limit
= max(limit
, 128UL);
1100 sysctl_sctp_mem
[0] = limit
/ 4 * 3;
1101 sysctl_sctp_mem
[1] = limit
;
1102 sysctl_sctp_mem
[2] = sysctl_sctp_mem
[0] * 2;
1104 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1105 limit
= (sysctl_sctp_mem
[1]) << (PAGE_SHIFT
- 7);
1106 max_share
= min(4UL*1024*1024, limit
);
1108 sysctl_sctp_rmem
[0] = PAGE_SIZE
; /* give each asoc 1 page min */
1109 sysctl_sctp_rmem
[1] = (1500 *(sizeof(struct sk_buff
) + 1));
1110 sysctl_sctp_rmem
[2] = max(sysctl_sctp_rmem
[1], max_share
);
1112 sysctl_sctp_wmem
[0] = SK_MEM_QUANTUM
;
1113 sysctl_sctp_wmem
[1] = 16*1024;
1114 sysctl_sctp_wmem
[2] = max(64*1024, max_share
);
1116 /* Size and allocate the association hash table.
1117 * The methodology is similar to that of the tcp hash tables.
1119 if (num_physpages
>= (128 * 1024))
1120 goal
= num_physpages
>> (22 - PAGE_SHIFT
);
1122 goal
= num_physpages
>> (24 - PAGE_SHIFT
);
1124 for (order
= 0; (1UL << order
) < goal
; order
++)
1128 sctp_assoc_hashsize
= (1UL << order
) * PAGE_SIZE
/
1129 sizeof(struct sctp_hashbucket
);
1130 if ((sctp_assoc_hashsize
> (64 * 1024)) && order
> 0)
1132 sctp_assoc_hashtable
= (struct sctp_hashbucket
*)
1133 __get_free_pages(GFP_ATOMIC
, order
);
1134 } while (!sctp_assoc_hashtable
&& --order
> 0);
1135 if (!sctp_assoc_hashtable
) {
1136 printk(KERN_ERR
"SCTP: Failed association hash alloc.\n");
1138 goto err_ahash_alloc
;
1140 for (i
= 0; i
< sctp_assoc_hashsize
; i
++) {
1141 rwlock_init(&sctp_assoc_hashtable
[i
].lock
);
1142 INIT_HLIST_HEAD(&sctp_assoc_hashtable
[i
].chain
);
1145 /* Allocate and initialize the endpoint hash table. */
1146 sctp_ep_hashsize
= 64;
1147 sctp_ep_hashtable
= (struct sctp_hashbucket
*)
1148 kmalloc(64 * sizeof(struct sctp_hashbucket
), GFP_KERNEL
);
1149 if (!sctp_ep_hashtable
) {
1150 printk(KERN_ERR
"SCTP: Failed endpoint_hash alloc.\n");
1152 goto err_ehash_alloc
;
1154 for (i
= 0; i
< sctp_ep_hashsize
; i
++) {
1155 rwlock_init(&sctp_ep_hashtable
[i
].lock
);
1156 INIT_HLIST_HEAD(&sctp_ep_hashtable
[i
].chain
);
1159 /* Allocate and initialize the SCTP port hash table. */
1161 sctp_port_hashsize
= (1UL << order
) * PAGE_SIZE
/
1162 sizeof(struct sctp_bind_hashbucket
);
1163 if ((sctp_port_hashsize
> (64 * 1024)) && order
> 0)
1165 sctp_port_hashtable
= (struct sctp_bind_hashbucket
*)
1166 __get_free_pages(GFP_ATOMIC
, order
);
1167 } while (!sctp_port_hashtable
&& --order
> 0);
1168 if (!sctp_port_hashtable
) {
1169 printk(KERN_ERR
"SCTP: Failed bind hash alloc.");
1171 goto err_bhash_alloc
;
1173 for (i
= 0; i
< sctp_port_hashsize
; i
++) {
1174 spin_lock_init(&sctp_port_hashtable
[i
].lock
);
1175 INIT_HLIST_HEAD(&sctp_port_hashtable
[i
].chain
);
1178 printk(KERN_INFO
"SCTP: Hash tables configured "
1179 "(established %d bind %d)\n",
1180 sctp_assoc_hashsize
, sctp_port_hashsize
);
1182 /* Disable ADDIP by default. */
1183 sctp_addip_enable
= 0;
1184 sctp_addip_noauth
= 0;
1186 /* Enable PR-SCTP by default. */
1187 sctp_prsctp_enable
= 1;
1189 /* Disable AUTH by default. */
1190 sctp_auth_enable
= 0;
1192 sctp_sysctl_register();
1194 INIT_LIST_HEAD(&sctp_address_families
);
1195 sctp_register_af(&sctp_ipv4_specific
);
1197 status
= proto_register(&sctp_prot
, 1);
1199 goto err_proto_register
;
1201 /* Register SCTP(UDP and TCP style) with socket layer. */
1202 inet_register_protosw(&sctp_seqpacket_protosw
);
1203 inet_register_protosw(&sctp_stream_protosw
);
1205 status
= sctp_v6_init();
1209 /* Initialize the control inode/socket for handling OOTB packets. */
1210 if ((status
= sctp_ctl_sock_init())) {
1212 "SCTP: Failed to initialize the SCTP control sock.\n");
1213 goto err_ctl_sock_init
;
1216 /* Initialize the local address list. */
1217 INIT_LIST_HEAD(&sctp_local_addr_list
);
1218 spin_lock_init(&sctp_local_addr_lock
);
1219 sctp_get_local_addr_list();
1221 /* Register notifier for inet address additions/deletions. */
1222 register_inetaddr_notifier(&sctp_inetaddr_notifier
);
1224 /* Register SCTP with inet layer. */
1225 if (inet_add_protocol(&sctp_protocol
, IPPROTO_SCTP
) < 0) {
1227 goto err_add_protocol
;
1230 /* Register SCTP with inet6 layer. */
1231 status
= sctp_v6_add_protocol();
1233 goto err_v6_add_protocol
;
1238 err_v6_add_protocol
:
1239 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1240 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1242 sctp_free_local_addr_list();
1243 sock_release(sctp_ctl_socket
);
1247 inet_unregister_protosw(&sctp_stream_protosw
);
1248 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1249 proto_unregister(&sctp_prot
);
1251 sctp_sysctl_unregister();
1252 list_del(&sctp_ipv4_specific
.list
);
1253 free_pages((unsigned long)sctp_port_hashtable
,
1254 get_order(sctp_port_hashsize
*
1255 sizeof(struct sctp_bind_hashbucket
)));
1257 kfree(sctp_ep_hashtable
);
1259 free_pages((unsigned long)sctp_assoc_hashtable
,
1260 get_order(sctp_assoc_hashsize
*
1261 sizeof(struct sctp_hashbucket
)));
1263 sctp_dbg_objcnt_exit();
1266 cleanup_sctp_mibs();
1268 kmem_cache_destroy(sctp_chunk_cachep
);
1270 kmem_cache_destroy(sctp_bucket_cachep
);
1274 /* Exit handler for the SCTP protocol. */
1275 SCTP_STATIC __exit
void sctp_exit(void)
1277 /* BUG. This should probably do something useful like clean
1278 * up all the remaining associations and all that memory.
1281 /* Unregister with inet6/inet layers. */
1282 sctp_v6_del_protocol();
1283 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1285 /* Unregister notifier for inet address additions/deletions. */
1286 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1288 /* Free the local address list. */
1289 sctp_free_local_addr_list();
1291 /* Free the control endpoint. */
1292 sock_release(sctp_ctl_socket
);
1294 /* Cleanup v6 initializations. */
1297 /* Unregister with socket layer. */
1298 inet_unregister_protosw(&sctp_stream_protosw
);
1299 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1301 sctp_sysctl_unregister();
1302 list_del(&sctp_ipv4_specific
.list
);
1304 free_pages((unsigned long)sctp_assoc_hashtable
,
1305 get_order(sctp_assoc_hashsize
*
1306 sizeof(struct sctp_hashbucket
)));
1307 kfree(sctp_ep_hashtable
);
1308 free_pages((unsigned long)sctp_port_hashtable
,
1309 get_order(sctp_port_hashsize
*
1310 sizeof(struct sctp_bind_hashbucket
)));
1312 sctp_dbg_objcnt_exit();
1314 cleanup_sctp_mibs();
1316 kmem_cache_destroy(sctp_chunk_cachep
);
1317 kmem_cache_destroy(sctp_bucket_cachep
);
1319 proto_unregister(&sctp_prot
);
1322 module_init(sctp_init
);
1323 module_exit(sctp_exit
);
1326 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1328 MODULE_ALIAS("net-pf-" __stringify(PF_INET
) "-proto-132");
1329 MODULE_ALIAS("net-pf-" __stringify(PF_INET6
) "-proto-132");
1330 MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1331 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1332 MODULE_LICENSE("GPL");