1 /* SCTP kernel implementation
2 * (C) Copyright IBM Corp. 2001, 2004
3 * Copyright (c) 1999-2000 Cisco, Inc.
4 * Copyright (c) 1999-2001 Motorola, Inc.
5 * Copyright (c) 2001 Intel Corp.
6 * Copyright (c) 2001 Nokia, Inc.
7 * Copyright (c) 2001 La Monte H.P. Yarroll
9 * This file is part of the SCTP kernel implementation
11 * Initialization/cleanup for SCTP protocol support.
13 * This SCTP implementation is free software;
14 * you can redistribute it and/or modify it under the terms of
15 * the GNU General Public License as published by
16 * the Free Software Foundation; either version 2, or (at your option)
19 * This SCTP implementation is distributed in the hope that it
20 * will be useful, but WITHOUT ANY WARRANTY; without even the implied
21 * ************************
22 * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
23 * See the GNU General Public License for more details.
25 * You should have received a copy of the GNU General Public License
26 * along with GNU CC; see the file COPYING. If not, write to
27 * the Free Software Foundation, 59 Temple Place - Suite 330,
28 * Boston, MA 02111-1307, USA.
30 * Please send any bug reports or fixes you make to the
32 * lksctp developers <lksctp-developers@lists.sourceforge.net>
34 * Or submit a bug report through the following website:
35 * http://www.sf.net/projects/lksctp
37 * Written or modified by:
38 * La Monte H.P. Yarroll <piggy@acm.org>
39 * Karl Knutson <karl@athena.chicago.il.us>
40 * Jon Grimm <jgrimm@us.ibm.com>
41 * Sridhar Samudrala <sri@us.ibm.com>
42 * Daisy Chang <daisyc@us.ibm.com>
43 * Ardelle Fan <ardelle.fan@intel.com>
45 * Any bugs reported given to us we will try to fix... any fixes shared will
46 * be incorporated into the next SCTP release.
49 #include <linux/module.h>
50 #include <linux/init.h>
51 #include <linux/netdevice.h>
52 #include <linux/inetdevice.h>
53 #include <linux/seq_file.h>
54 #include <linux/bootmem.h>
55 #include <linux/highmem.h>
56 #include <linux/swap.h>
57 #include <net/net_namespace.h>
58 #include <net/protocol.h>
61 #include <net/route.h>
62 #include <net/sctp/sctp.h>
63 #include <net/addrconf.h>
64 #include <net/inet_common.h>
65 #include <net/inet_ecn.h>
67 /* Global data structures. */
68 struct sctp_globals sctp_globals __read_mostly
;
69 DEFINE_SNMP_STAT(struct sctp_mib
, sctp_statistics
) __read_mostly
;
72 struct proc_dir_entry
*proc_net_sctp
;
75 struct idr sctp_assocs_id
;
76 DEFINE_SPINLOCK(sctp_assocs_id_lock
);
78 /* This is the global socket data structure used for responding to
79 * the Out-of-the-blue (OOTB) packets. A control sock will be created
80 * for this socket at the initialization time.
82 static struct sock
*sctp_ctl_sock
;
84 static struct sctp_pf
*sctp_pf_inet6_specific
;
85 static struct sctp_pf
*sctp_pf_inet_specific
;
86 static struct sctp_af
*sctp_af_v4_specific
;
87 static struct sctp_af
*sctp_af_v6_specific
;
89 struct kmem_cache
*sctp_chunk_cachep __read_mostly
;
90 struct kmem_cache
*sctp_bucket_cachep __read_mostly
;
92 int sysctl_sctp_mem
[3];
93 int sysctl_sctp_rmem
[3];
94 int sysctl_sctp_wmem
[3];
96 /* Return the address of the control sock. */
97 struct sock
*sctp_get_ctl_sock(void)
102 /* Set up the proc fs entry for the SCTP protocol. */
103 static __init
int sctp_proc_init(void)
105 if (percpu_counter_init(&sctp_sockets_allocated
, 0))
107 #ifdef CONFIG_PROC_FS
108 if (!proc_net_sctp
) {
109 proc_net_sctp
= proc_mkdir("sctp", init_net
.proc_net
);
111 goto out_free_percpu
;
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
);
137 percpu_counter_destroy(&sctp_sockets_allocated
);
140 #endif /* CONFIG_PROC_FS */
146 /* Clean up the proc fs entry for the SCTP protocol.
147 * Note: Do not make this __exit as it is used in the init error
150 static void sctp_proc_exit(void)
152 #ifdef CONFIG_PROC_FS
153 sctp_snmp_proc_exit();
154 sctp_eps_proc_exit();
155 sctp_assocs_proc_exit();
156 sctp_remaddr_proc_exit();
159 proc_net_sctp
= NULL
;
160 remove_proc_entry("sctp", init_net
.proc_net
);
163 percpu_counter_destroy(&sctp_sockets_allocated
);
166 /* Private helper to extract ipv4 address and stash them in
167 * the protocol structure.
169 static void sctp_v4_copy_addrlist(struct list_head
*addrlist
,
170 struct net_device
*dev
)
172 struct in_device
*in_dev
;
173 struct in_ifaddr
*ifa
;
174 struct sctp_sockaddr_entry
*addr
;
177 if ((in_dev
= __in_dev_get_rcu(dev
)) == NULL
) {
182 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
183 /* Add the address to the local list. */
184 addr
= t_new(struct sctp_sockaddr_entry
, GFP_ATOMIC
);
186 addr
->a
.v4
.sin_family
= AF_INET
;
187 addr
->a
.v4
.sin_port
= 0;
188 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
190 INIT_LIST_HEAD(&addr
->list
);
191 INIT_RCU_HEAD(&addr
->rcu
);
192 list_add_tail(&addr
->list
, addrlist
);
199 /* Extract our IP addresses from the system and stash them in the
200 * protocol structure.
202 static void sctp_get_local_addr_list(void)
204 struct net_device
*dev
;
205 struct list_head
*pos
;
208 read_lock(&dev_base_lock
);
209 for_each_netdev(&init_net
, dev
) {
210 __list_for_each(pos
, &sctp_address_families
) {
211 af
= list_entry(pos
, struct sctp_af
, list
);
212 af
->copy_addrlist(&sctp_local_addr_list
, dev
);
215 read_unlock(&dev_base_lock
);
218 /* Free the existing local addresses. */
219 static void sctp_free_local_addr_list(void)
221 struct sctp_sockaddr_entry
*addr
;
222 struct list_head
*pos
, *temp
;
224 list_for_each_safe(pos
, temp
, &sctp_local_addr_list
) {
225 addr
= list_entry(pos
, struct sctp_sockaddr_entry
, list
);
231 void sctp_local_addr_free(struct rcu_head
*head
)
233 struct sctp_sockaddr_entry
*e
= container_of(head
,
234 struct sctp_sockaddr_entry
, rcu
);
238 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
239 int sctp_copy_local_addr_list(struct sctp_bind_addr
*bp
, sctp_scope_t scope
,
240 gfp_t gfp
, int copy_flags
)
242 struct sctp_sockaddr_entry
*addr
;
246 list_for_each_entry_rcu(addr
, &sctp_local_addr_list
, list
) {
249 if (sctp_in_scope(&addr
->a
, scope
)) {
250 /* Now that the address is in scope, check to see if
251 * the address type is really supported by the local
252 * sock as well as the remote peer.
254 if ((((AF_INET
== addr
->a
.sa
.sa_family
) &&
255 (copy_flags
& SCTP_ADDR4_PEERSUPP
))) ||
256 (((AF_INET6
== addr
->a
.sa
.sa_family
) &&
257 (copy_flags
& SCTP_ADDR6_ALLOWED
) &&
258 (copy_flags
& SCTP_ADDR6_PEERSUPP
)))) {
259 error
= sctp_add_bind_addr(bp
, &addr
->a
,
260 SCTP_ADDR_SRC
, GFP_ATOMIC
);
272 /* Initialize a sctp_addr from in incoming skb. */
273 static void sctp_v4_from_skb(union sctp_addr
*addr
, struct sk_buff
*skb
,
280 port
= &addr
->v4
.sin_port
;
281 addr
->v4
.sin_family
= AF_INET
;
286 from
= &ip_hdr(skb
)->saddr
;
289 from
= &ip_hdr(skb
)->daddr
;
291 memcpy(&addr
->v4
.sin_addr
.s_addr
, from
, sizeof(struct in_addr
));
294 /* Initialize an sctp_addr from a socket. */
295 static void sctp_v4_from_sk(union sctp_addr
*addr
, struct sock
*sk
)
297 addr
->v4
.sin_family
= AF_INET
;
298 addr
->v4
.sin_port
= 0;
299 addr
->v4
.sin_addr
.s_addr
= inet_sk(sk
)->rcv_saddr
;
302 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
303 static void sctp_v4_to_sk_saddr(union sctp_addr
*addr
, struct sock
*sk
)
305 inet_sk(sk
)->rcv_saddr
= addr
->v4
.sin_addr
.s_addr
;
308 /* Initialize sk->sk_daddr from sctp_addr. */
309 static void sctp_v4_to_sk_daddr(union sctp_addr
*addr
, struct sock
*sk
)
311 inet_sk(sk
)->daddr
= addr
->v4
.sin_addr
.s_addr
;
314 /* Initialize a sctp_addr from an address parameter. */
315 static void sctp_v4_from_addr_param(union sctp_addr
*addr
,
316 union sctp_addr_param
*param
,
317 __be16 port
, int iif
)
319 addr
->v4
.sin_family
= AF_INET
;
320 addr
->v4
.sin_port
= port
;
321 addr
->v4
.sin_addr
.s_addr
= param
->v4
.addr
.s_addr
;
324 /* Initialize an address parameter from a sctp_addr and return the length
325 * of the address parameter.
327 static int sctp_v4_to_addr_param(const union sctp_addr
*addr
,
328 union sctp_addr_param
*param
)
330 int length
= sizeof(sctp_ipv4addr_param_t
);
332 param
->v4
.param_hdr
.type
= SCTP_PARAM_IPV4_ADDRESS
;
333 param
->v4
.param_hdr
.length
= htons(length
);
334 param
->v4
.addr
.s_addr
= addr
->v4
.sin_addr
.s_addr
;
339 /* Initialize a sctp_addr from a dst_entry. */
340 static void sctp_v4_dst_saddr(union sctp_addr
*saddr
, struct dst_entry
*dst
,
343 struct rtable
*rt
= (struct rtable
*)dst
;
344 saddr
->v4
.sin_family
= AF_INET
;
345 saddr
->v4
.sin_port
= port
;
346 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
349 /* Compare two addresses exactly. */
350 static int sctp_v4_cmp_addr(const union sctp_addr
*addr1
,
351 const union sctp_addr
*addr2
)
353 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
355 if (addr1
->v4
.sin_port
!= addr2
->v4
.sin_port
)
357 if (addr1
->v4
.sin_addr
.s_addr
!= addr2
->v4
.sin_addr
.s_addr
)
363 /* Initialize addr struct to INADDR_ANY. */
364 static void sctp_v4_inaddr_any(union sctp_addr
*addr
, __be16 port
)
366 addr
->v4
.sin_family
= AF_INET
;
367 addr
->v4
.sin_addr
.s_addr
= htonl(INADDR_ANY
);
368 addr
->v4
.sin_port
= port
;
371 /* Is this a wildcard address? */
372 static int sctp_v4_is_any(const union sctp_addr
*addr
)
374 return htonl(INADDR_ANY
) == addr
->v4
.sin_addr
.s_addr
;
377 /* This function checks if the address is a valid address to be used for
381 * Return 0 - If the address is a non-unicast or an illegal address.
382 * Return 1 - If the address is a unicast.
384 static int sctp_v4_addr_valid(union sctp_addr
*addr
,
385 struct sctp_sock
*sp
,
386 const struct sk_buff
*skb
)
388 /* IPv4 addresses not allowed */
389 if (sp
&& ipv6_only_sock(sctp_opt2sk(sp
)))
392 /* Is this a non-unicast address or a unusable SCTP address? */
393 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
))
396 /* Is this a broadcast address? */
397 if (skb
&& skb_rtable(skb
)->rt_flags
& RTCF_BROADCAST
)
403 /* Should this be available for binding? */
404 static int sctp_v4_available(union sctp_addr
*addr
, struct sctp_sock
*sp
)
406 int ret
= inet_addr_type(&init_net
, addr
->v4
.sin_addr
.s_addr
);
409 if (addr
->v4
.sin_addr
.s_addr
!= htonl(INADDR_ANY
) &&
411 !sp
->inet
.freebind
&&
412 !sysctl_ip_nonlocal_bind
)
415 if (ipv6_only_sock(sctp_opt2sk(sp
)))
421 /* Checking the loopback, private and other address scopes as defined in
422 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
423 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
425 * Level 0 - unusable SCTP addresses
426 * Level 1 - loopback address
427 * Level 2 - link-local addresses
428 * Level 3 - private addresses.
429 * Level 4 - global addresses
430 * For INIT and INIT-ACK address list, let L be the level of
431 * of requested destination address, sender and receiver
432 * SHOULD include all of its addresses with level greater
433 * than or equal to L.
435 static sctp_scope_t
sctp_v4_scope(union sctp_addr
*addr
)
439 /* Should IPv4 scoping be a sysctl configurable option
440 * so users can turn it off (default on) for certain
441 * unconventional networking environments?
444 /* Check for unusable SCTP addresses. */
445 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
)) {
446 retval
= SCTP_SCOPE_UNUSABLE
;
447 } else if (ipv4_is_loopback(addr
->v4
.sin_addr
.s_addr
)) {
448 retval
= SCTP_SCOPE_LOOPBACK
;
449 } else if (ipv4_is_linklocal_169(addr
->v4
.sin_addr
.s_addr
)) {
450 retval
= SCTP_SCOPE_LINK
;
451 } else if (ipv4_is_private_10(addr
->v4
.sin_addr
.s_addr
) ||
452 ipv4_is_private_172(addr
->v4
.sin_addr
.s_addr
) ||
453 ipv4_is_private_192(addr
->v4
.sin_addr
.s_addr
)) {
454 retval
= SCTP_SCOPE_PRIVATE
;
456 retval
= SCTP_SCOPE_GLOBAL
;
462 /* Returns a valid dst cache entry for the given source and destination ip
463 * addresses. If an association is passed, trys to get a dst entry with a
464 * source address that matches an address in the bind address list.
466 static struct dst_entry
*sctp_v4_get_dst(struct sctp_association
*asoc
,
467 union sctp_addr
*daddr
,
468 union sctp_addr
*saddr
)
472 struct sctp_bind_addr
*bp
;
473 struct sctp_sockaddr_entry
*laddr
;
474 struct dst_entry
*dst
= NULL
;
475 union sctp_addr dst_saddr
;
477 memset(&fl
, 0x0, sizeof(struct flowi
));
478 fl
.fl4_dst
= daddr
->v4
.sin_addr
.s_addr
;
479 fl
.proto
= IPPROTO_SCTP
;
481 fl
.fl4_tos
= RT_CONN_FLAGS(asoc
->base
.sk
);
482 fl
.oif
= asoc
->base
.sk
->sk_bound_dev_if
;
485 fl
.fl4_src
= saddr
->v4
.sin_addr
.s_addr
;
487 SCTP_DEBUG_PRINTK("%s: DST:%pI4, SRC:%pI4 - ",
488 __func__
, &fl
.fl4_dst
, &fl
.fl4_src
);
490 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
494 /* If there is no association or if a source address is passed, no
495 * more validation is required.
500 bp
= &asoc
->base
.bind_addr
;
503 /* Walk through the bind address list and look for a bind
504 * address that matches the source address of the returned dst.
506 sctp_v4_dst_saddr(&dst_saddr
, dst
, htons(bp
->port
));
508 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
509 if (!laddr
->valid
|| (laddr
->state
!= SCTP_ADDR_SRC
))
511 if (sctp_v4_cmp_addr(&dst_saddr
, &laddr
->a
))
516 /* None of the bound addresses match the source address of the
517 * dst. So release it.
523 /* Walk through the bind address list and try to get a dst that
524 * matches a bind address as the source address.
527 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
530 if ((laddr
->state
== SCTP_ADDR_SRC
) &&
531 (AF_INET
== laddr
->a
.sa
.sa_family
)) {
532 fl
.fl4_src
= laddr
->a
.v4
.sin_addr
.s_addr
;
533 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
544 SCTP_DEBUG_PRINTK("rt_dst:%pI4, rt_src:%pI4\n",
545 &rt
->rt_dst
, &rt
->rt_src
);
547 SCTP_DEBUG_PRINTK("NO ROUTE\n");
552 /* For v4, the source address is cached in the route entry(dst). So no need
553 * to cache it separately and hence this is an empty routine.
555 static void sctp_v4_get_saddr(struct sctp_sock
*sk
,
556 struct sctp_association
*asoc
,
557 struct dst_entry
*dst
,
558 union sctp_addr
*daddr
,
559 union sctp_addr
*saddr
)
561 struct rtable
*rt
= (struct rtable
*)dst
;
567 saddr
->v4
.sin_family
= AF_INET
;
568 saddr
->v4
.sin_port
= htons(asoc
->base
.bind_addr
.port
);
569 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
573 /* What interface did this skb arrive on? */
574 static int sctp_v4_skb_iif(const struct sk_buff
*skb
)
576 return skb_rtable(skb
)->rt_iif
;
579 /* Was this packet marked by Explicit Congestion Notification? */
580 static int sctp_v4_is_ce(const struct sk_buff
*skb
)
582 return INET_ECN_is_ce(ip_hdr(skb
)->tos
);
585 /* Create and initialize a new sk for the socket returned by accept(). */
586 static struct sock
*sctp_v4_create_accept_sk(struct sock
*sk
,
587 struct sctp_association
*asoc
)
589 struct sock
*newsk
= sk_alloc(sock_net(sk
), PF_INET
, GFP_KERNEL
,
591 struct inet_sock
*newinet
;
596 sock_init_data(NULL
, newsk
);
598 sctp_copy_sock(newsk
, sk
, asoc
);
599 sock_reset_flag(newsk
, SOCK_ZAPPED
);
601 newinet
= inet_sk(newsk
);
603 newinet
->daddr
= asoc
->peer
.primary_addr
.v4
.sin_addr
.s_addr
;
605 sk_refcnt_debug_inc(newsk
);
607 if (newsk
->sk_prot
->init(newsk
)) {
608 sk_common_release(newsk
);
616 /* Map address, empty for v4 family */
617 static void sctp_v4_addr_v4map(struct sctp_sock
*sp
, union sctp_addr
*addr
)
622 /* Dump the v4 addr to the seq file. */
623 static void sctp_v4_seq_dump_addr(struct seq_file
*seq
, union sctp_addr
*addr
)
625 seq_printf(seq
, "%pI4 ", &addr
->v4
.sin_addr
);
628 static void sctp_v4_ecn_capable(struct sock
*sk
)
633 /* Event handler for inet address addition/deletion events.
634 * The sctp_local_addr_list needs to be protocted by a spin lock since
635 * multiple notifiers (say IPv4 and IPv6) may be running at the same
636 * time and thus corrupt the list.
637 * The reader side is protected with RCU.
639 static int sctp_inetaddr_event(struct notifier_block
*this, unsigned long ev
,
642 struct in_ifaddr
*ifa
= (struct in_ifaddr
*)ptr
;
643 struct sctp_sockaddr_entry
*addr
= NULL
;
644 struct sctp_sockaddr_entry
*temp
;
647 if (!net_eq(dev_net(ifa
->ifa_dev
->dev
), &init_net
))
652 addr
= kmalloc(sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
654 addr
->a
.v4
.sin_family
= AF_INET
;
655 addr
->a
.v4
.sin_port
= 0;
656 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
658 spin_lock_bh(&sctp_local_addr_lock
);
659 list_add_tail_rcu(&addr
->list
, &sctp_local_addr_list
);
660 spin_unlock_bh(&sctp_local_addr_lock
);
664 spin_lock_bh(&sctp_local_addr_lock
);
665 list_for_each_entry_safe(addr
, temp
,
666 &sctp_local_addr_list
, list
) {
667 if (addr
->a
.sa
.sa_family
== AF_INET
&&
668 addr
->a
.v4
.sin_addr
.s_addr
==
672 list_del_rcu(&addr
->list
);
676 spin_unlock_bh(&sctp_local_addr_lock
);
678 call_rcu(&addr
->rcu
, sctp_local_addr_free
);
686 * Initialize the control inode/socket with a control endpoint data
687 * structure. This endpoint is reserved exclusively for the OOTB processing.
689 static int sctp_ctl_sock_init(void)
692 sa_family_t family
= PF_INET
;
694 if (sctp_get_pf_specific(PF_INET6
))
697 err
= inet_ctl_sock_create(&sctp_ctl_sock
, family
,
698 SOCK_SEQPACKET
, IPPROTO_SCTP
, &init_net
);
700 /* If IPv6 socket could not be created, try the IPv4 socket */
701 if (err
< 0 && family
== PF_INET6
)
702 err
= inet_ctl_sock_create(&sctp_ctl_sock
, AF_INET
,
703 SOCK_SEQPACKET
, IPPROTO_SCTP
,
708 "SCTP: Failed to create the SCTP control socket.\n");
714 /* Register address family specific functions. */
715 int sctp_register_af(struct sctp_af
*af
)
717 switch (af
->sa_family
) {
719 if (sctp_af_v4_specific
)
721 sctp_af_v4_specific
= af
;
724 if (sctp_af_v6_specific
)
726 sctp_af_v6_specific
= af
;
732 INIT_LIST_HEAD(&af
->list
);
733 list_add_tail(&af
->list
, &sctp_address_families
);
737 /* Get the table of functions for manipulating a particular address
740 struct sctp_af
*sctp_get_af_specific(sa_family_t family
)
744 return sctp_af_v4_specific
;
746 return sctp_af_v6_specific
;
752 /* Common code to initialize a AF_INET msg_name. */
753 static void sctp_inet_msgname(char *msgname
, int *addr_len
)
755 struct sockaddr_in
*sin
;
757 sin
= (struct sockaddr_in
*)msgname
;
758 *addr_len
= sizeof(struct sockaddr_in
);
759 sin
->sin_family
= AF_INET
;
760 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
763 /* Copy the primary address of the peer primary address as the msg_name. */
764 static void sctp_inet_event_msgname(struct sctp_ulpevent
*event
, char *msgname
,
767 struct sockaddr_in
*sin
, *sinfrom
;
770 struct sctp_association
*asoc
;
773 sctp_inet_msgname(msgname
, addr_len
);
774 sin
= (struct sockaddr_in
*)msgname
;
775 sinfrom
= &asoc
->peer
.primary_addr
.v4
;
776 sin
->sin_port
= htons(asoc
->peer
.port
);
777 sin
->sin_addr
.s_addr
= sinfrom
->sin_addr
.s_addr
;
781 /* Initialize and copy out a msgname from an inbound skb. */
782 static void sctp_inet_skb_msgname(struct sk_buff
*skb
, char *msgname
, int *len
)
785 struct sctphdr
*sh
= sctp_hdr(skb
);
786 struct sockaddr_in
*sin
= (struct sockaddr_in
*)msgname
;
788 sctp_inet_msgname(msgname
, len
);
789 sin
->sin_port
= sh
->source
;
790 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
794 /* Do we support this AF? */
795 static int sctp_inet_af_supported(sa_family_t family
, struct sctp_sock
*sp
)
797 /* PF_INET only supports AF_INET addresses. */
798 return (AF_INET
== family
);
801 /* Address matching with wildcards allowed. */
802 static int sctp_inet_cmp_addr(const union sctp_addr
*addr1
,
803 const union sctp_addr
*addr2
,
804 struct sctp_sock
*opt
)
806 /* PF_INET only supports AF_INET addresses. */
807 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
809 if (htonl(INADDR_ANY
) == addr1
->v4
.sin_addr
.s_addr
||
810 htonl(INADDR_ANY
) == addr2
->v4
.sin_addr
.s_addr
)
812 if (addr1
->v4
.sin_addr
.s_addr
== addr2
->v4
.sin_addr
.s_addr
)
818 /* Verify that provided sockaddr looks bindable. Common verification has
819 * already been taken care of.
821 static int sctp_inet_bind_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
823 return sctp_v4_available(addr
, opt
);
826 /* Verify that sockaddr looks sendable. Common verification has already
827 * been taken care of.
829 static int sctp_inet_send_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
834 /* Fill in Supported Address Type information for INIT and INIT-ACK
835 * chunks. Returns number of addresses supported.
837 static int sctp_inet_supported_addrs(const struct sctp_sock
*opt
,
840 types
[0] = SCTP_PARAM_IPV4_ADDRESS
;
844 /* Wrapper routine that calls the ip transmit routine. */
845 static inline int sctp_v4_xmit(struct sk_buff
*skb
,
846 struct sctp_transport
*transport
)
848 struct inet_sock
*inet
= inet_sk(skb
->sk
);
850 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n",
851 __func__
, skb
, skb
->len
,
852 &skb_rtable(skb
)->rt_src
,
853 &skb_rtable(skb
)->rt_dst
);
855 inet
->pmtudisc
= transport
->param_flags
& SPP_PMTUD_ENABLE
?
856 IP_PMTUDISC_DO
: IP_PMTUDISC_DONT
;
858 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS
);
859 return ip_queue_xmit(skb
, 0);
862 static struct sctp_af sctp_af_inet
;
864 static struct sctp_pf sctp_pf_inet
= {
865 .event_msgname
= sctp_inet_event_msgname
,
866 .skb_msgname
= sctp_inet_skb_msgname
,
867 .af_supported
= sctp_inet_af_supported
,
868 .cmp_addr
= sctp_inet_cmp_addr
,
869 .bind_verify
= sctp_inet_bind_verify
,
870 .send_verify
= sctp_inet_send_verify
,
871 .supported_addrs
= sctp_inet_supported_addrs
,
872 .create_accept_sk
= sctp_v4_create_accept_sk
,
873 .addr_v4map
= sctp_v4_addr_v4map
,
877 /* Notifier for inetaddr addition/deletion events. */
878 static struct notifier_block sctp_inetaddr_notifier
= {
879 .notifier_call
= sctp_inetaddr_event
,
882 /* Socket operations. */
883 static const struct proto_ops inet_seqpacket_ops
= {
885 .owner
= THIS_MODULE
,
886 .release
= inet_release
, /* Needs to be wrapped... */
888 .connect
= inet_dgram_connect
,
889 .socketpair
= sock_no_socketpair
,
890 .accept
= inet_accept
,
891 .getname
= inet_getname
, /* Semantics are different. */
894 .listen
= sctp_inet_listen
,
895 .shutdown
= inet_shutdown
, /* Looks harmless. */
896 .setsockopt
= sock_common_setsockopt
, /* IP_SOL IP_OPTION is a problem */
897 .getsockopt
= sock_common_getsockopt
,
898 .sendmsg
= inet_sendmsg
,
899 .recvmsg
= sock_common_recvmsg
,
900 .mmap
= sock_no_mmap
,
901 .sendpage
= sock_no_sendpage
,
903 .compat_setsockopt
= compat_sock_common_setsockopt
,
904 .compat_getsockopt
= compat_sock_common_getsockopt
,
908 /* Registration with AF_INET family. */
909 static struct inet_protosw sctp_seqpacket_protosw
= {
910 .type
= SOCK_SEQPACKET
,
911 .protocol
= IPPROTO_SCTP
,
913 .ops
= &inet_seqpacket_ops
,
916 .flags
= SCTP_PROTOSW_FLAG
918 static struct inet_protosw sctp_stream_protosw
= {
920 .protocol
= IPPROTO_SCTP
,
922 .ops
= &inet_seqpacket_ops
,
925 .flags
= SCTP_PROTOSW_FLAG
928 /* Register with IP layer. */
929 static struct net_protocol sctp_protocol
= {
931 .err_handler
= sctp_v4_err
,
935 /* IPv4 address related functions. */
936 static struct sctp_af sctp_af_inet
= {
937 .sa_family
= AF_INET
,
938 .sctp_xmit
= sctp_v4_xmit
,
939 .setsockopt
= ip_setsockopt
,
940 .getsockopt
= ip_getsockopt
,
941 .get_dst
= sctp_v4_get_dst
,
942 .get_saddr
= sctp_v4_get_saddr
,
943 .copy_addrlist
= sctp_v4_copy_addrlist
,
944 .from_skb
= sctp_v4_from_skb
,
945 .from_sk
= sctp_v4_from_sk
,
946 .to_sk_saddr
= sctp_v4_to_sk_saddr
,
947 .to_sk_daddr
= sctp_v4_to_sk_daddr
,
948 .from_addr_param
= sctp_v4_from_addr_param
,
949 .to_addr_param
= sctp_v4_to_addr_param
,
950 .dst_saddr
= sctp_v4_dst_saddr
,
951 .cmp_addr
= sctp_v4_cmp_addr
,
952 .addr_valid
= sctp_v4_addr_valid
,
953 .inaddr_any
= sctp_v4_inaddr_any
,
954 .is_any
= sctp_v4_is_any
,
955 .available
= sctp_v4_available
,
956 .scope
= sctp_v4_scope
,
957 .skb_iif
= sctp_v4_skb_iif
,
958 .is_ce
= sctp_v4_is_ce
,
959 .seq_dump_addr
= sctp_v4_seq_dump_addr
,
960 .ecn_capable
= sctp_v4_ecn_capable
,
961 .net_header_len
= sizeof(struct iphdr
),
962 .sockaddr_len
= sizeof(struct sockaddr_in
),
964 .compat_setsockopt
= compat_ip_setsockopt
,
965 .compat_getsockopt
= compat_ip_getsockopt
,
969 struct sctp_pf
*sctp_get_pf_specific(sa_family_t family
) {
973 return sctp_pf_inet_specific
;
975 return sctp_pf_inet6_specific
;
981 /* Register the PF specific function table. */
982 int sctp_register_pf(struct sctp_pf
*pf
, sa_family_t family
)
986 if (sctp_pf_inet_specific
)
988 sctp_pf_inet_specific
= pf
;
991 if (sctp_pf_inet6_specific
)
993 sctp_pf_inet6_specific
= pf
;
1001 static inline int init_sctp_mibs(void)
1003 return snmp_mib_init((void**)sctp_statistics
, sizeof(struct sctp_mib
));
1006 static inline void cleanup_sctp_mibs(void)
1008 snmp_mib_free((void**)sctp_statistics
);
1011 static void sctp_v4_pf_init(void)
1013 /* Initialize the SCTP specific PF functions. */
1014 sctp_register_pf(&sctp_pf_inet
, PF_INET
);
1015 sctp_register_af(&sctp_af_inet
);
1018 static void sctp_v4_pf_exit(void)
1020 list_del(&sctp_af_inet
.list
);
1023 static int sctp_v4_protosw_init(void)
1027 rc
= proto_register(&sctp_prot
, 1);
1031 /* Register SCTP(UDP and TCP style) with socket layer. */
1032 inet_register_protosw(&sctp_seqpacket_protosw
);
1033 inet_register_protosw(&sctp_stream_protosw
);
1038 static void sctp_v4_protosw_exit(void)
1040 inet_unregister_protosw(&sctp_stream_protosw
);
1041 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1042 proto_unregister(&sctp_prot
);
1045 static int sctp_v4_add_protocol(void)
1047 /* Register notifier for inet address additions/deletions. */
1048 register_inetaddr_notifier(&sctp_inetaddr_notifier
);
1050 /* Register SCTP with inet layer. */
1051 if (inet_add_protocol(&sctp_protocol
, IPPROTO_SCTP
) < 0)
1057 static void sctp_v4_del_protocol(void)
1059 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1060 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1063 /* Initialize the universe into something sensible. */
1064 SCTP_STATIC __init
int sctp_init(void)
1067 int status
= -EINVAL
;
1069 unsigned long limit
;
1070 unsigned long nr_pages
;
1074 /* SCTP_DEBUG sanity check. */
1075 if (!sctp_sanity_check())
1078 /* Allocate bind_bucket and chunk caches. */
1080 sctp_bucket_cachep
= kmem_cache_create("sctp_bind_bucket",
1081 sizeof(struct sctp_bind_bucket
),
1082 0, SLAB_HWCACHE_ALIGN
,
1084 if (!sctp_bucket_cachep
)
1087 sctp_chunk_cachep
= kmem_cache_create("sctp_chunk",
1088 sizeof(struct sctp_chunk
),
1089 0, SLAB_HWCACHE_ALIGN
,
1091 if (!sctp_chunk_cachep
)
1092 goto err_chunk_cachep
;
1094 /* Allocate and initialise sctp mibs. */
1095 status
= init_sctp_mibs();
1099 /* Initialize proc fs directory. */
1100 status
= sctp_proc_init();
1104 /* Initialize object count debugging. */
1105 sctp_dbg_objcnt_init();
1108 * 14. Suggested SCTP Protocol Parameter Values
1110 /* The following protocol parameters are RECOMMENDED: */
1111 /* RTO.Initial - 3 seconds */
1112 sctp_rto_initial
= SCTP_RTO_INITIAL
;
1113 /* RTO.Min - 1 second */
1114 sctp_rto_min
= SCTP_RTO_MIN
;
1115 /* RTO.Max - 60 seconds */
1116 sctp_rto_max
= SCTP_RTO_MAX
;
1117 /* RTO.Alpha - 1/8 */
1118 sctp_rto_alpha
= SCTP_RTO_ALPHA
;
1119 /* RTO.Beta - 1/4 */
1120 sctp_rto_beta
= SCTP_RTO_BETA
;
1122 /* Valid.Cookie.Life - 60 seconds */
1123 sctp_valid_cookie_life
= SCTP_DEFAULT_COOKIE_LIFE
;
1125 /* Whether Cookie Preservative is enabled(1) or not(0) */
1126 sctp_cookie_preserve_enable
= 1;
1129 sctp_max_burst
= SCTP_DEFAULT_MAX_BURST
;
1131 /* Association.Max.Retrans - 10 attempts
1132 * Path.Max.Retrans - 5 attempts (per destination address)
1133 * Max.Init.Retransmits - 8 attempts
1135 sctp_max_retrans_association
= 10;
1136 sctp_max_retrans_path
= 5;
1137 sctp_max_retrans_init
= 8;
1139 /* Sendbuffer growth - do per-socket accounting */
1140 sctp_sndbuf_policy
= 0;
1142 /* Rcvbuffer growth - do per-socket accounting */
1143 sctp_rcvbuf_policy
= 0;
1145 /* HB.interval - 30 seconds */
1146 sctp_hb_interval
= SCTP_DEFAULT_TIMEOUT_HEARTBEAT
;
1148 /* delayed SACK timeout */
1149 sctp_sack_timeout
= SCTP_DEFAULT_TIMEOUT_SACK
;
1151 /* Implementation specific variables. */
1153 /* Initialize default stream count setup information. */
1154 sctp_max_instreams
= SCTP_DEFAULT_INSTREAMS
;
1155 sctp_max_outstreams
= SCTP_DEFAULT_OUTSTREAMS
;
1157 /* Initialize handle used for association ids. */
1158 idr_init(&sctp_assocs_id
);
1160 /* Set the pressure threshold to be a fraction of global memory that
1161 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1162 * memory, with a floor of 128 pages.
1163 * Note this initalizes the data in sctpv6_prot too
1164 * Unabashedly stolen from tcp_init
1166 nr_pages
= totalram_pages
- totalhigh_pages
;
1167 limit
= min(nr_pages
, 1UL<<(28-PAGE_SHIFT
)) >> (20-PAGE_SHIFT
);
1168 limit
= (limit
* (nr_pages
>> (20-PAGE_SHIFT
))) >> (PAGE_SHIFT
-11);
1169 limit
= max(limit
, 128UL);
1170 sysctl_sctp_mem
[0] = limit
/ 4 * 3;
1171 sysctl_sctp_mem
[1] = limit
;
1172 sysctl_sctp_mem
[2] = sysctl_sctp_mem
[0] * 2;
1174 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1175 limit
= (sysctl_sctp_mem
[1]) << (PAGE_SHIFT
- 7);
1176 max_share
= min(4UL*1024*1024, limit
);
1178 sysctl_sctp_rmem
[0] = SK_MEM_QUANTUM
; /* give each asoc 1 page min */
1179 sysctl_sctp_rmem
[1] = (1500 *(sizeof(struct sk_buff
) + 1));
1180 sysctl_sctp_rmem
[2] = max(sysctl_sctp_rmem
[1], max_share
);
1182 sysctl_sctp_wmem
[0] = SK_MEM_QUANTUM
;
1183 sysctl_sctp_wmem
[1] = 16*1024;
1184 sysctl_sctp_wmem
[2] = max(64*1024, max_share
);
1186 /* Size and allocate the association hash table.
1187 * The methodology is similar to that of the tcp hash tables.
1189 if (num_physpages
>= (128 * 1024))
1190 goal
= num_physpages
>> (22 - PAGE_SHIFT
);
1192 goal
= num_physpages
>> (24 - PAGE_SHIFT
);
1194 for (order
= 0; (1UL << order
) < goal
; order
++)
1198 sctp_assoc_hashsize
= (1UL << order
) * PAGE_SIZE
/
1199 sizeof(struct sctp_hashbucket
);
1200 if ((sctp_assoc_hashsize
> (64 * 1024)) && order
> 0)
1202 sctp_assoc_hashtable
= (struct sctp_hashbucket
*)
1203 __get_free_pages(GFP_ATOMIC
, order
);
1204 } while (!sctp_assoc_hashtable
&& --order
> 0);
1205 if (!sctp_assoc_hashtable
) {
1206 printk(KERN_ERR
"SCTP: Failed association hash alloc.\n");
1208 goto err_ahash_alloc
;
1210 for (i
= 0; i
< sctp_assoc_hashsize
; i
++) {
1211 rwlock_init(&sctp_assoc_hashtable
[i
].lock
);
1212 INIT_HLIST_HEAD(&sctp_assoc_hashtable
[i
].chain
);
1215 /* Allocate and initialize the endpoint hash table. */
1216 sctp_ep_hashsize
= 64;
1217 sctp_ep_hashtable
= (struct sctp_hashbucket
*)
1218 kmalloc(64 * sizeof(struct sctp_hashbucket
), GFP_KERNEL
);
1219 if (!sctp_ep_hashtable
) {
1220 printk(KERN_ERR
"SCTP: Failed endpoint_hash alloc.\n");
1222 goto err_ehash_alloc
;
1224 for (i
= 0; i
< sctp_ep_hashsize
; i
++) {
1225 rwlock_init(&sctp_ep_hashtable
[i
].lock
);
1226 INIT_HLIST_HEAD(&sctp_ep_hashtable
[i
].chain
);
1229 /* Allocate and initialize the SCTP port hash table. */
1231 sctp_port_hashsize
= (1UL << order
) * PAGE_SIZE
/
1232 sizeof(struct sctp_bind_hashbucket
);
1233 if ((sctp_port_hashsize
> (64 * 1024)) && order
> 0)
1235 sctp_port_hashtable
= (struct sctp_bind_hashbucket
*)
1236 __get_free_pages(GFP_ATOMIC
, order
);
1237 } while (!sctp_port_hashtable
&& --order
> 0);
1238 if (!sctp_port_hashtable
) {
1239 printk(KERN_ERR
"SCTP: Failed bind hash alloc.");
1241 goto err_bhash_alloc
;
1243 for (i
= 0; i
< sctp_port_hashsize
; i
++) {
1244 spin_lock_init(&sctp_port_hashtable
[i
].lock
);
1245 INIT_HLIST_HEAD(&sctp_port_hashtable
[i
].chain
);
1248 printk(KERN_INFO
"SCTP: Hash tables configured "
1249 "(established %d bind %d)\n",
1250 sctp_assoc_hashsize
, sctp_port_hashsize
);
1252 /* Disable ADDIP by default. */
1253 sctp_addip_enable
= 0;
1254 sctp_addip_noauth
= 0;
1256 /* Enable PR-SCTP by default. */
1257 sctp_prsctp_enable
= 1;
1259 /* Disable AUTH by default. */
1260 sctp_auth_enable
= 0;
1262 sctp_sysctl_register();
1264 INIT_LIST_HEAD(&sctp_address_families
);
1268 /* Initialize the local address list. */
1269 INIT_LIST_HEAD(&sctp_local_addr_list
);
1270 spin_lock_init(&sctp_local_addr_lock
);
1271 sctp_get_local_addr_list();
1273 status
= sctp_v4_protosw_init();
1276 goto err_protosw_init
;
1278 status
= sctp_v6_protosw_init();
1280 goto err_v6_protosw_init
;
1282 /* Initialize the control inode/socket for handling OOTB packets. */
1283 if ((status
= sctp_ctl_sock_init())) {
1285 "SCTP: Failed to initialize the SCTP control sock.\n");
1286 goto err_ctl_sock_init
;
1289 status
= sctp_v4_add_protocol();
1291 goto err_add_protocol
;
1293 /* Register SCTP with inet6 layer. */
1294 status
= sctp_v6_add_protocol();
1296 goto err_v6_add_protocol
;
1301 err_v6_add_protocol
:
1302 sctp_v4_del_protocol();
1304 inet_ctl_sock_destroy(sctp_ctl_sock
);
1306 sctp_v6_protosw_exit();
1307 err_v6_protosw_init
:
1308 sctp_v4_protosw_exit();
1310 sctp_free_local_addr_list();
1313 sctp_sysctl_unregister();
1314 free_pages((unsigned long)sctp_port_hashtable
,
1315 get_order(sctp_port_hashsize
*
1316 sizeof(struct sctp_bind_hashbucket
)));
1318 kfree(sctp_ep_hashtable
);
1320 free_pages((unsigned long)sctp_assoc_hashtable
,
1321 get_order(sctp_assoc_hashsize
*
1322 sizeof(struct sctp_hashbucket
)));
1324 sctp_dbg_objcnt_exit();
1327 cleanup_sctp_mibs();
1329 kmem_cache_destroy(sctp_chunk_cachep
);
1331 kmem_cache_destroy(sctp_bucket_cachep
);
1335 /* Exit handler for the SCTP protocol. */
1336 SCTP_STATIC __exit
void sctp_exit(void)
1338 /* BUG. This should probably do something useful like clean
1339 * up all the remaining associations and all that memory.
1342 /* Unregister with inet6/inet layers. */
1343 sctp_v6_del_protocol();
1344 sctp_v4_del_protocol();
1346 /* Free the control endpoint. */
1347 inet_ctl_sock_destroy(sctp_ctl_sock
);
1349 /* Free protosw registrations */
1350 sctp_v6_protosw_exit();
1351 sctp_v4_protosw_exit();
1353 /* Free the local address list. */
1354 sctp_free_local_addr_list();
1356 /* Unregister with socket layer. */
1360 sctp_sysctl_unregister();
1362 free_pages((unsigned long)sctp_assoc_hashtable
,
1363 get_order(sctp_assoc_hashsize
*
1364 sizeof(struct sctp_hashbucket
)));
1365 kfree(sctp_ep_hashtable
);
1366 free_pages((unsigned long)sctp_port_hashtable
,
1367 get_order(sctp_port_hashsize
*
1368 sizeof(struct sctp_bind_hashbucket
)));
1370 sctp_dbg_objcnt_exit();
1372 cleanup_sctp_mibs();
1374 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1376 kmem_cache_destroy(sctp_chunk_cachep
);
1377 kmem_cache_destroy(sctp_bucket_cachep
);
1380 module_init(sctp_init
);
1381 module_exit(sctp_exit
);
1384 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1386 MODULE_ALIAS("net-pf-" __stringify(PF_INET
) "-proto-132");
1387 MODULE_ALIAS("net-pf-" __stringify(PF_INET6
) "-proto-132");
1388 MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1389 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1390 module_param_named(no_checksums
, sctp_checksum_disable
, bool, 0644);
1391 MODULE_PARM_DESC(no_checksums
, "Disable checksums computing and verification");
1392 MODULE_LICENSE("GPL");