1 /* SCTP kernel implementation
2 * (C) Copyright IBM Corp. 2001, 2004
3 * Copyright (c) 1999-2000 Cisco, Inc.
4 * Copyright (c) 1999-2001 Motorola, Inc.
5 * Copyright (c) 2001 Intel Corp.
6 * Copyright (c) 2001 Nokia, Inc.
7 * Copyright (c) 2001 La Monte H.P. Yarroll
9 * This file is part of the SCTP kernel implementation
11 * Initialization/cleanup for SCTP protocol support.
13 * This SCTP implementation is free software;
14 * you can redistribute it and/or modify it under the terms of
15 * the GNU General Public License as published by
16 * the Free Software Foundation; either version 2, or (at your option)
19 * This SCTP implementation is distributed in the hope that it
20 * will be useful, but WITHOUT ANY WARRANTY; without even the implied
21 * ************************
22 * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
23 * See the GNU General Public License for more details.
25 * You should have received a copy of the GNU General Public License
26 * along with GNU CC; see the file COPYING. If not, write to
27 * the Free Software Foundation, 59 Temple Place - Suite 330,
28 * Boston, MA 02111-1307, USA.
30 * Please send any bug reports or fixes you make to the
32 * lksctp developers <lksctp-developers@lists.sourceforge.net>
34 * Or submit a bug report through the following website:
35 * http://www.sf.net/projects/lksctp
37 * Written or modified by:
38 * La Monte H.P. Yarroll <piggy@acm.org>
39 * Karl Knutson <karl@athena.chicago.il.us>
40 * Jon Grimm <jgrimm@us.ibm.com>
41 * Sridhar Samudrala <sri@us.ibm.com>
42 * Daisy Chang <daisyc@us.ibm.com>
43 * Ardelle Fan <ardelle.fan@intel.com>
45 * Any bugs reported given to us we will try to fix... any fixes shared will
46 * be incorporated into the next SCTP release.
49 #include <linux/module.h>
50 #include <linux/init.h>
51 #include <linux/netdevice.h>
52 #include <linux/inetdevice.h>
53 #include <linux/seq_file.h>
54 #include <linux/bootmem.h>
55 #include <linux/highmem.h>
56 #include <linux/swap.h>
57 #include <net/net_namespace.h>
58 #include <net/protocol.h>
61 #include <net/route.h>
62 #include <net/sctp/sctp.h>
63 #include <net/addrconf.h>
64 #include <net/inet_common.h>
65 #include <net/inet_ecn.h>
67 /* Global data structures. */
68 struct sctp_globals sctp_globals __read_mostly
;
69 DEFINE_SNMP_STAT(struct sctp_mib
, sctp_statistics
) __read_mostly
;
72 struct proc_dir_entry
*proc_net_sctp
;
75 struct idr sctp_assocs_id
;
76 DEFINE_SPINLOCK(sctp_assocs_id_lock
);
78 /* This is the global socket data structure used for responding to
79 * the Out-of-the-blue (OOTB) packets. A control sock will be created
80 * for this socket at the initialization time.
82 static struct sock
*sctp_ctl_sock
;
84 static struct sctp_pf
*sctp_pf_inet6_specific
;
85 static struct sctp_pf
*sctp_pf_inet_specific
;
86 static struct sctp_af
*sctp_af_v4_specific
;
87 static struct sctp_af
*sctp_af_v6_specific
;
89 struct kmem_cache
*sctp_chunk_cachep __read_mostly
;
90 struct kmem_cache
*sctp_bucket_cachep __read_mostly
;
92 int sysctl_sctp_mem
[3];
93 int sysctl_sctp_rmem
[3];
94 int sysctl_sctp_wmem
[3];
96 /* Return the address of the control sock. */
97 struct sock
*sctp_get_ctl_sock(void)
102 /* Set up the proc fs entry for the SCTP protocol. */
103 static __init
int sctp_proc_init(void)
105 if (percpu_counter_init(&sctp_sockets_allocated
, 0))
107 #ifdef CONFIG_PROC_FS
108 if (!proc_net_sctp
) {
109 struct proc_dir_entry
*ent
;
110 ent
= proc_mkdir("sctp", init_net
.proc_net
);
112 ent
->owner
= THIS_MODULE
;
115 goto out_free_percpu
;
118 if (sctp_snmp_proc_init())
119 goto out_snmp_proc_init
;
120 if (sctp_eps_proc_init())
121 goto out_eps_proc_init
;
122 if (sctp_assocs_proc_init())
123 goto out_assocs_proc_init
;
124 if (sctp_remaddr_proc_init())
125 goto out_remaddr_proc_init
;
129 out_remaddr_proc_init
:
130 sctp_assocs_proc_exit();
131 out_assocs_proc_init
:
132 sctp_eps_proc_exit();
134 sctp_snmp_proc_exit();
137 proc_net_sctp
= NULL
;
138 remove_proc_entry("sctp", init_net
.proc_net
);
141 percpu_counter_destroy(&sctp_sockets_allocated
);
146 #endif /* CONFIG_PROC_FS */
149 /* Clean up the proc fs entry for the SCTP protocol.
150 * Note: Do not make this __exit as it is used in the init error
153 static void sctp_proc_exit(void)
155 #ifdef CONFIG_PROC_FS
156 sctp_snmp_proc_exit();
157 sctp_eps_proc_exit();
158 sctp_assocs_proc_exit();
159 sctp_remaddr_proc_exit();
162 proc_net_sctp
= NULL
;
163 remove_proc_entry("sctp", init_net
.proc_net
);
168 /* Private helper to extract ipv4 address and stash them in
169 * the protocol structure.
171 static void sctp_v4_copy_addrlist(struct list_head
*addrlist
,
172 struct net_device
*dev
)
174 struct in_device
*in_dev
;
175 struct in_ifaddr
*ifa
;
176 struct sctp_sockaddr_entry
*addr
;
179 if ((in_dev
= __in_dev_get_rcu(dev
)) == NULL
) {
184 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
185 /* Add the address to the local list. */
186 addr
= t_new(struct sctp_sockaddr_entry
, GFP_ATOMIC
);
188 addr
->a
.v4
.sin_family
= AF_INET
;
189 addr
->a
.v4
.sin_port
= 0;
190 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
192 INIT_LIST_HEAD(&addr
->list
);
193 INIT_RCU_HEAD(&addr
->rcu
);
194 list_add_tail(&addr
->list
, addrlist
);
201 /* Extract our IP addresses from the system and stash them in the
202 * protocol structure.
204 static void sctp_get_local_addr_list(void)
206 struct net_device
*dev
;
207 struct list_head
*pos
;
210 read_lock(&dev_base_lock
);
211 for_each_netdev(&init_net
, dev
) {
212 __list_for_each(pos
, &sctp_address_families
) {
213 af
= list_entry(pos
, struct sctp_af
, list
);
214 af
->copy_addrlist(&sctp_local_addr_list
, dev
);
217 read_unlock(&dev_base_lock
);
220 /* Free the existing local addresses. */
221 static void sctp_free_local_addr_list(void)
223 struct sctp_sockaddr_entry
*addr
;
224 struct list_head
*pos
, *temp
;
226 list_for_each_safe(pos
, temp
, &sctp_local_addr_list
) {
227 addr
= list_entry(pos
, struct sctp_sockaddr_entry
, list
);
233 void sctp_local_addr_free(struct rcu_head
*head
)
235 struct sctp_sockaddr_entry
*e
= container_of(head
,
236 struct sctp_sockaddr_entry
, rcu
);
240 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
241 int sctp_copy_local_addr_list(struct sctp_bind_addr
*bp
, sctp_scope_t scope
,
242 gfp_t gfp
, int copy_flags
)
244 struct sctp_sockaddr_entry
*addr
;
248 list_for_each_entry_rcu(addr
, &sctp_local_addr_list
, list
) {
251 if (sctp_in_scope(&addr
->a
, scope
)) {
252 /* Now that the address is in scope, check to see if
253 * the address type is really supported by the local
254 * sock as well as the remote peer.
256 if ((((AF_INET
== addr
->a
.sa
.sa_family
) &&
257 (copy_flags
& SCTP_ADDR4_PEERSUPP
))) ||
258 (((AF_INET6
== addr
->a
.sa
.sa_family
) &&
259 (copy_flags
& SCTP_ADDR6_ALLOWED
) &&
260 (copy_flags
& SCTP_ADDR6_PEERSUPP
)))) {
261 error
= sctp_add_bind_addr(bp
, &addr
->a
,
262 SCTP_ADDR_SRC
, GFP_ATOMIC
);
274 /* Initialize a sctp_addr from in incoming skb. */
275 static void sctp_v4_from_skb(union sctp_addr
*addr
, struct sk_buff
*skb
,
282 port
= &addr
->v4
.sin_port
;
283 addr
->v4
.sin_family
= AF_INET
;
288 from
= &ip_hdr(skb
)->saddr
;
291 from
= &ip_hdr(skb
)->daddr
;
293 memcpy(&addr
->v4
.sin_addr
.s_addr
, from
, sizeof(struct in_addr
));
296 /* Initialize an sctp_addr from a socket. */
297 static void sctp_v4_from_sk(union sctp_addr
*addr
, struct sock
*sk
)
299 addr
->v4
.sin_family
= AF_INET
;
300 addr
->v4
.sin_port
= 0;
301 addr
->v4
.sin_addr
.s_addr
= inet_sk(sk
)->rcv_saddr
;
304 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
305 static void sctp_v4_to_sk_saddr(union sctp_addr
*addr
, struct sock
*sk
)
307 inet_sk(sk
)->rcv_saddr
= addr
->v4
.sin_addr
.s_addr
;
310 /* Initialize sk->sk_daddr from sctp_addr. */
311 static void sctp_v4_to_sk_daddr(union sctp_addr
*addr
, struct sock
*sk
)
313 inet_sk(sk
)->daddr
= addr
->v4
.sin_addr
.s_addr
;
316 /* Initialize a sctp_addr from an address parameter. */
317 static void sctp_v4_from_addr_param(union sctp_addr
*addr
,
318 union sctp_addr_param
*param
,
319 __be16 port
, int iif
)
321 addr
->v4
.sin_family
= AF_INET
;
322 addr
->v4
.sin_port
= port
;
323 addr
->v4
.sin_addr
.s_addr
= param
->v4
.addr
.s_addr
;
326 /* Initialize an address parameter from a sctp_addr and return the length
327 * of the address parameter.
329 static int sctp_v4_to_addr_param(const union sctp_addr
*addr
,
330 union sctp_addr_param
*param
)
332 int length
= sizeof(sctp_ipv4addr_param_t
);
334 param
->v4
.param_hdr
.type
= SCTP_PARAM_IPV4_ADDRESS
;
335 param
->v4
.param_hdr
.length
= htons(length
);
336 param
->v4
.addr
.s_addr
= addr
->v4
.sin_addr
.s_addr
;
341 /* Initialize a sctp_addr from a dst_entry. */
342 static void sctp_v4_dst_saddr(union sctp_addr
*saddr
, struct dst_entry
*dst
,
345 struct rtable
*rt
= (struct rtable
*)dst
;
346 saddr
->v4
.sin_family
= AF_INET
;
347 saddr
->v4
.sin_port
= port
;
348 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
351 /* Compare two addresses exactly. */
352 static int sctp_v4_cmp_addr(const union sctp_addr
*addr1
,
353 const union sctp_addr
*addr2
)
355 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
357 if (addr1
->v4
.sin_port
!= addr2
->v4
.sin_port
)
359 if (addr1
->v4
.sin_addr
.s_addr
!= addr2
->v4
.sin_addr
.s_addr
)
365 /* Initialize addr struct to INADDR_ANY. */
366 static void sctp_v4_inaddr_any(union sctp_addr
*addr
, __be16 port
)
368 addr
->v4
.sin_family
= AF_INET
;
369 addr
->v4
.sin_addr
.s_addr
= htonl(INADDR_ANY
);
370 addr
->v4
.sin_port
= port
;
373 /* Is this a wildcard address? */
374 static int sctp_v4_is_any(const union sctp_addr
*addr
)
376 return htonl(INADDR_ANY
) == addr
->v4
.sin_addr
.s_addr
;
379 /* This function checks if the address is a valid address to be used for
383 * Return 0 - If the address is a non-unicast or an illegal address.
384 * Return 1 - If the address is a unicast.
386 static int sctp_v4_addr_valid(union sctp_addr
*addr
,
387 struct sctp_sock
*sp
,
388 const struct sk_buff
*skb
)
390 /* IPv4 addresses not allowed */
391 if (sp
&& ipv6_only_sock(sctp_opt2sk(sp
)))
394 /* Is this a non-unicast address or a unusable SCTP address? */
395 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
))
398 /* Is this a broadcast address? */
399 if (skb
&& skb
->rtable
->rt_flags
& RTCF_BROADCAST
)
405 /* Should this be available for binding? */
406 static int sctp_v4_available(union sctp_addr
*addr
, struct sctp_sock
*sp
)
408 int ret
= inet_addr_type(&init_net
, addr
->v4
.sin_addr
.s_addr
);
411 if (addr
->v4
.sin_addr
.s_addr
!= htonl(INADDR_ANY
) &&
413 !sp
->inet
.freebind
&&
414 !sysctl_ip_nonlocal_bind
)
417 if (ipv6_only_sock(sctp_opt2sk(sp
)))
423 /* Checking the loopback, private and other address scopes as defined in
424 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
425 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
427 * Level 0 - unusable SCTP addresses
428 * Level 1 - loopback address
429 * Level 2 - link-local addresses
430 * Level 3 - private addresses.
431 * Level 4 - global addresses
432 * For INIT and INIT-ACK address list, let L be the level of
433 * of requested destination address, sender and receiver
434 * SHOULD include all of its addresses with level greater
435 * than or equal to L.
437 static sctp_scope_t
sctp_v4_scope(union sctp_addr
*addr
)
441 /* Should IPv4 scoping be a sysctl configurable option
442 * so users can turn it off (default on) for certain
443 * unconventional networking environments?
446 /* Check for unusable SCTP addresses. */
447 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
)) {
448 retval
= SCTP_SCOPE_UNUSABLE
;
449 } else if (ipv4_is_loopback(addr
->v4
.sin_addr
.s_addr
)) {
450 retval
= SCTP_SCOPE_LOOPBACK
;
451 } else if (ipv4_is_linklocal_169(addr
->v4
.sin_addr
.s_addr
)) {
452 retval
= SCTP_SCOPE_LINK
;
453 } else if (ipv4_is_private_10(addr
->v4
.sin_addr
.s_addr
) ||
454 ipv4_is_private_172(addr
->v4
.sin_addr
.s_addr
) ||
455 ipv4_is_private_192(addr
->v4
.sin_addr
.s_addr
)) {
456 retval
= SCTP_SCOPE_PRIVATE
;
458 retval
= SCTP_SCOPE_GLOBAL
;
464 /* Returns a valid dst cache entry for the given source and destination ip
465 * addresses. If an association is passed, trys to get a dst entry with a
466 * source address that matches an address in the bind address list.
468 static struct dst_entry
*sctp_v4_get_dst(struct sctp_association
*asoc
,
469 union sctp_addr
*daddr
,
470 union sctp_addr
*saddr
)
474 struct sctp_bind_addr
*bp
;
475 struct sctp_sockaddr_entry
*laddr
;
476 struct dst_entry
*dst
= NULL
;
477 union sctp_addr dst_saddr
;
479 memset(&fl
, 0x0, sizeof(struct flowi
));
480 fl
.fl4_dst
= daddr
->v4
.sin_addr
.s_addr
;
481 fl
.proto
= IPPROTO_SCTP
;
483 fl
.fl4_tos
= RT_CONN_FLAGS(asoc
->base
.sk
);
484 fl
.oif
= asoc
->base
.sk
->sk_bound_dev_if
;
487 fl
.fl4_src
= saddr
->v4
.sin_addr
.s_addr
;
489 SCTP_DEBUG_PRINTK("%s: DST:%pI4, SRC:%pI4 - ",
490 __func__
, &fl
.fl4_dst
, &fl
.fl4_src
);
492 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
496 /* If there is no association or if a source address is passed, no
497 * more validation is required.
502 bp
= &asoc
->base
.bind_addr
;
505 /* Walk through the bind address list and look for a bind
506 * address that matches the source address of the returned dst.
508 sctp_v4_dst_saddr(&dst_saddr
, dst
, htons(bp
->port
));
510 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
511 if (!laddr
->valid
|| (laddr
->state
!= SCTP_ADDR_SRC
))
513 if (sctp_v4_cmp_addr(&dst_saddr
, &laddr
->a
))
518 /* None of the bound addresses match the source address of the
519 * dst. So release it.
525 /* Walk through the bind address list and try to get a dst that
526 * matches a bind address as the source address.
529 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
532 if ((laddr
->state
== SCTP_ADDR_SRC
) &&
533 (AF_INET
== laddr
->a
.sa
.sa_family
)) {
534 fl
.fl4_src
= laddr
->a
.v4
.sin_addr
.s_addr
;
535 if (!ip_route_output_key(&init_net
, &rt
, &fl
)) {
546 SCTP_DEBUG_PRINTK("rt_dst:%pI4, rt_src:%pI4\n",
547 &rt
->rt_dst
, &rt
->rt_src
);
549 SCTP_DEBUG_PRINTK("NO ROUTE\n");
554 /* For v4, the source address is cached in the route entry(dst). So no need
555 * to cache it separately and hence this is an empty routine.
557 static void sctp_v4_get_saddr(struct sctp_sock
*sk
,
558 struct sctp_association
*asoc
,
559 struct dst_entry
*dst
,
560 union sctp_addr
*daddr
,
561 union sctp_addr
*saddr
)
563 struct rtable
*rt
= (struct rtable
*)dst
;
569 saddr
->v4
.sin_family
= AF_INET
;
570 saddr
->v4
.sin_port
= htons(asoc
->base
.bind_addr
.port
);
571 saddr
->v4
.sin_addr
.s_addr
= rt
->rt_src
;
575 /* What interface did this skb arrive on? */
576 static int sctp_v4_skb_iif(const struct sk_buff
*skb
)
578 return skb
->rtable
->rt_iif
;
581 /* Was this packet marked by Explicit Congestion Notification? */
582 static int sctp_v4_is_ce(const struct sk_buff
*skb
)
584 return INET_ECN_is_ce(ip_hdr(skb
)->tos
);
587 /* Create and initialize a new sk for the socket returned by accept(). */
588 static struct sock
*sctp_v4_create_accept_sk(struct sock
*sk
,
589 struct sctp_association
*asoc
)
591 struct inet_sock
*inet
= inet_sk(sk
);
592 struct inet_sock
*newinet
;
593 struct sock
*newsk
= sk_alloc(sock_net(sk
), PF_INET
, GFP_KERNEL
,
599 sock_init_data(NULL
, newsk
);
601 newsk
->sk_type
= SOCK_STREAM
;
603 newsk
->sk_no_check
= sk
->sk_no_check
;
604 newsk
->sk_reuse
= sk
->sk_reuse
;
605 newsk
->sk_shutdown
= sk
->sk_shutdown
;
607 newsk
->sk_destruct
= inet_sock_destruct
;
608 newsk
->sk_family
= PF_INET
;
609 newsk
->sk_protocol
= IPPROTO_SCTP
;
610 newsk
->sk_backlog_rcv
= sk
->sk_prot
->backlog_rcv
;
611 sock_reset_flag(newsk
, SOCK_ZAPPED
);
613 newinet
= inet_sk(newsk
);
615 /* Initialize sk's sport, dport, rcv_saddr and daddr for
616 * getsockname() and getpeername()
618 newinet
->sport
= inet
->sport
;
619 newinet
->saddr
= inet
->saddr
;
620 newinet
->rcv_saddr
= inet
->rcv_saddr
;
621 newinet
->dport
= htons(asoc
->peer
.port
);
622 newinet
->daddr
= asoc
->peer
.primary_addr
.v4
.sin_addr
.s_addr
;
623 newinet
->pmtudisc
= inet
->pmtudisc
;
624 newinet
->id
= asoc
->next_tsn
^ jiffies
;
626 newinet
->uc_ttl
= -1;
627 newinet
->mc_loop
= 1;
629 newinet
->mc_index
= 0;
630 newinet
->mc_list
= NULL
;
632 sk_refcnt_debug_inc(newsk
);
634 if (newsk
->sk_prot
->init(newsk
)) {
635 sk_common_release(newsk
);
643 /* Map address, empty for v4 family */
644 static void sctp_v4_addr_v4map(struct sctp_sock
*sp
, union sctp_addr
*addr
)
649 /* Dump the v4 addr to the seq file. */
650 static void sctp_v4_seq_dump_addr(struct seq_file
*seq
, union sctp_addr
*addr
)
652 seq_printf(seq
, "%pI4 ", &addr
->v4
.sin_addr
);
655 static void sctp_v4_ecn_capable(struct sock
*sk
)
660 /* Event handler for inet address addition/deletion events.
661 * The sctp_local_addr_list needs to be protocted by a spin lock since
662 * multiple notifiers (say IPv4 and IPv6) may be running at the same
663 * time and thus corrupt the list.
664 * The reader side is protected with RCU.
666 static int sctp_inetaddr_event(struct notifier_block
*this, unsigned long ev
,
669 struct in_ifaddr
*ifa
= (struct in_ifaddr
*)ptr
;
670 struct sctp_sockaddr_entry
*addr
= NULL
;
671 struct sctp_sockaddr_entry
*temp
;
674 if (!net_eq(dev_net(ifa
->ifa_dev
->dev
), &init_net
))
679 addr
= kmalloc(sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
681 addr
->a
.v4
.sin_family
= AF_INET
;
682 addr
->a
.v4
.sin_port
= 0;
683 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
685 spin_lock_bh(&sctp_local_addr_lock
);
686 list_add_tail_rcu(&addr
->list
, &sctp_local_addr_list
);
687 spin_unlock_bh(&sctp_local_addr_lock
);
691 spin_lock_bh(&sctp_local_addr_lock
);
692 list_for_each_entry_safe(addr
, temp
,
693 &sctp_local_addr_list
, list
) {
694 if (addr
->a
.sa
.sa_family
== AF_INET
&&
695 addr
->a
.v4
.sin_addr
.s_addr
==
699 list_del_rcu(&addr
->list
);
703 spin_unlock_bh(&sctp_local_addr_lock
);
705 call_rcu(&addr
->rcu
, sctp_local_addr_free
);
713 * Initialize the control inode/socket with a control endpoint data
714 * structure. This endpoint is reserved exclusively for the OOTB processing.
716 static int sctp_ctl_sock_init(void)
721 if (sctp_get_pf_specific(PF_INET6
))
726 err
= inet_ctl_sock_create(&sctp_ctl_sock
, family
,
727 SOCK_SEQPACKET
, IPPROTO_SCTP
, &init_net
);
730 "SCTP: Failed to create the SCTP control socket.\n");
736 /* Register address family specific functions. */
737 int sctp_register_af(struct sctp_af
*af
)
739 switch (af
->sa_family
) {
741 if (sctp_af_v4_specific
)
743 sctp_af_v4_specific
= af
;
746 if (sctp_af_v6_specific
)
748 sctp_af_v6_specific
= af
;
754 INIT_LIST_HEAD(&af
->list
);
755 list_add_tail(&af
->list
, &sctp_address_families
);
759 /* Get the table of functions for manipulating a particular address
762 struct sctp_af
*sctp_get_af_specific(sa_family_t family
)
766 return sctp_af_v4_specific
;
768 return sctp_af_v6_specific
;
774 /* Common code to initialize a AF_INET msg_name. */
775 static void sctp_inet_msgname(char *msgname
, int *addr_len
)
777 struct sockaddr_in
*sin
;
779 sin
= (struct sockaddr_in
*)msgname
;
780 *addr_len
= sizeof(struct sockaddr_in
);
781 sin
->sin_family
= AF_INET
;
782 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
785 /* Copy the primary address of the peer primary address as the msg_name. */
786 static void sctp_inet_event_msgname(struct sctp_ulpevent
*event
, char *msgname
,
789 struct sockaddr_in
*sin
, *sinfrom
;
792 struct sctp_association
*asoc
;
795 sctp_inet_msgname(msgname
, addr_len
);
796 sin
= (struct sockaddr_in
*)msgname
;
797 sinfrom
= &asoc
->peer
.primary_addr
.v4
;
798 sin
->sin_port
= htons(asoc
->peer
.port
);
799 sin
->sin_addr
.s_addr
= sinfrom
->sin_addr
.s_addr
;
803 /* Initialize and copy out a msgname from an inbound skb. */
804 static void sctp_inet_skb_msgname(struct sk_buff
*skb
, char *msgname
, int *len
)
807 struct sctphdr
*sh
= sctp_hdr(skb
);
808 struct sockaddr_in
*sin
= (struct sockaddr_in
*)msgname
;
810 sctp_inet_msgname(msgname
, len
);
811 sin
->sin_port
= sh
->source
;
812 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
816 /* Do we support this AF? */
817 static int sctp_inet_af_supported(sa_family_t family
, struct sctp_sock
*sp
)
819 /* PF_INET only supports AF_INET addresses. */
820 return (AF_INET
== family
);
823 /* Address matching with wildcards allowed. */
824 static int sctp_inet_cmp_addr(const union sctp_addr
*addr1
,
825 const union sctp_addr
*addr2
,
826 struct sctp_sock
*opt
)
828 /* PF_INET only supports AF_INET addresses. */
829 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
831 if (htonl(INADDR_ANY
) == addr1
->v4
.sin_addr
.s_addr
||
832 htonl(INADDR_ANY
) == addr2
->v4
.sin_addr
.s_addr
)
834 if (addr1
->v4
.sin_addr
.s_addr
== addr2
->v4
.sin_addr
.s_addr
)
840 /* Verify that provided sockaddr looks bindable. Common verification has
841 * already been taken care of.
843 static int sctp_inet_bind_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
845 return sctp_v4_available(addr
, opt
);
848 /* Verify that sockaddr looks sendable. Common verification has already
849 * been taken care of.
851 static int sctp_inet_send_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
856 /* Fill in Supported Address Type information for INIT and INIT-ACK
857 * chunks. Returns number of addresses supported.
859 static int sctp_inet_supported_addrs(const struct sctp_sock
*opt
,
862 types
[0] = SCTP_PARAM_IPV4_ADDRESS
;
866 /* Wrapper routine that calls the ip transmit routine. */
867 static inline int sctp_v4_xmit(struct sk_buff
*skb
,
868 struct sctp_transport
*transport
)
870 struct inet_sock
*inet
= inet_sk(skb
->sk
);
872 SCTP_DEBUG_PRINTK("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n",
873 __func__
, skb
, skb
->len
,
874 &skb
->rtable
->rt_src
,
875 &skb
->rtable
->rt_dst
);
877 inet
->pmtudisc
= transport
->param_flags
& SPP_PMTUD_ENABLE
?
878 IP_PMTUDISC_DO
: IP_PMTUDISC_DONT
;
880 SCTP_INC_STATS(SCTP_MIB_OUTSCTPPACKS
);
881 return ip_queue_xmit(skb
, 0);
884 static struct sctp_af sctp_af_inet
;
886 static struct sctp_pf sctp_pf_inet
= {
887 .event_msgname
= sctp_inet_event_msgname
,
888 .skb_msgname
= sctp_inet_skb_msgname
,
889 .af_supported
= sctp_inet_af_supported
,
890 .cmp_addr
= sctp_inet_cmp_addr
,
891 .bind_verify
= sctp_inet_bind_verify
,
892 .send_verify
= sctp_inet_send_verify
,
893 .supported_addrs
= sctp_inet_supported_addrs
,
894 .create_accept_sk
= sctp_v4_create_accept_sk
,
895 .addr_v4map
= sctp_v4_addr_v4map
,
899 /* Notifier for inetaddr addition/deletion events. */
900 static struct notifier_block sctp_inetaddr_notifier
= {
901 .notifier_call
= sctp_inetaddr_event
,
904 /* Socket operations. */
905 static const struct proto_ops inet_seqpacket_ops
= {
907 .owner
= THIS_MODULE
,
908 .release
= inet_release
, /* Needs to be wrapped... */
910 .connect
= inet_dgram_connect
,
911 .socketpair
= sock_no_socketpair
,
912 .accept
= inet_accept
,
913 .getname
= inet_getname
, /* Semantics are different. */
916 .listen
= sctp_inet_listen
,
917 .shutdown
= inet_shutdown
, /* Looks harmless. */
918 .setsockopt
= sock_common_setsockopt
, /* IP_SOL IP_OPTION is a problem */
919 .getsockopt
= sock_common_getsockopt
,
920 .sendmsg
= inet_sendmsg
,
921 .recvmsg
= sock_common_recvmsg
,
922 .mmap
= sock_no_mmap
,
923 .sendpage
= sock_no_sendpage
,
925 .compat_setsockopt
= compat_sock_common_setsockopt
,
926 .compat_getsockopt
= compat_sock_common_getsockopt
,
930 /* Registration with AF_INET family. */
931 static struct inet_protosw sctp_seqpacket_protosw
= {
932 .type
= SOCK_SEQPACKET
,
933 .protocol
= IPPROTO_SCTP
,
935 .ops
= &inet_seqpacket_ops
,
938 .flags
= SCTP_PROTOSW_FLAG
940 static struct inet_protosw sctp_stream_protosw
= {
942 .protocol
= IPPROTO_SCTP
,
944 .ops
= &inet_seqpacket_ops
,
947 .flags
= SCTP_PROTOSW_FLAG
950 /* Register with IP layer. */
951 static struct net_protocol sctp_protocol
= {
953 .err_handler
= sctp_v4_err
,
957 /* IPv4 address related functions. */
958 static struct sctp_af sctp_af_inet
= {
959 .sa_family
= AF_INET
,
960 .sctp_xmit
= sctp_v4_xmit
,
961 .setsockopt
= ip_setsockopt
,
962 .getsockopt
= ip_getsockopt
,
963 .get_dst
= sctp_v4_get_dst
,
964 .get_saddr
= sctp_v4_get_saddr
,
965 .copy_addrlist
= sctp_v4_copy_addrlist
,
966 .from_skb
= sctp_v4_from_skb
,
967 .from_sk
= sctp_v4_from_sk
,
968 .to_sk_saddr
= sctp_v4_to_sk_saddr
,
969 .to_sk_daddr
= sctp_v4_to_sk_daddr
,
970 .from_addr_param
= sctp_v4_from_addr_param
,
971 .to_addr_param
= sctp_v4_to_addr_param
,
972 .dst_saddr
= sctp_v4_dst_saddr
,
973 .cmp_addr
= sctp_v4_cmp_addr
,
974 .addr_valid
= sctp_v4_addr_valid
,
975 .inaddr_any
= sctp_v4_inaddr_any
,
976 .is_any
= sctp_v4_is_any
,
977 .available
= sctp_v4_available
,
978 .scope
= sctp_v4_scope
,
979 .skb_iif
= sctp_v4_skb_iif
,
980 .is_ce
= sctp_v4_is_ce
,
981 .seq_dump_addr
= sctp_v4_seq_dump_addr
,
982 .ecn_capable
= sctp_v4_ecn_capable
,
983 .net_header_len
= sizeof(struct iphdr
),
984 .sockaddr_len
= sizeof(struct sockaddr_in
),
986 .compat_setsockopt
= compat_ip_setsockopt
,
987 .compat_getsockopt
= compat_ip_getsockopt
,
991 struct sctp_pf
*sctp_get_pf_specific(sa_family_t family
) {
995 return sctp_pf_inet_specific
;
997 return sctp_pf_inet6_specific
;
1003 /* Register the PF specific function table. */
1004 int sctp_register_pf(struct sctp_pf
*pf
, sa_family_t family
)
1008 if (sctp_pf_inet_specific
)
1010 sctp_pf_inet_specific
= pf
;
1013 if (sctp_pf_inet6_specific
)
1015 sctp_pf_inet6_specific
= pf
;
1023 static inline int init_sctp_mibs(void)
1025 return snmp_mib_init((void**)sctp_statistics
, sizeof(struct sctp_mib
));
1028 static inline void cleanup_sctp_mibs(void)
1030 snmp_mib_free((void**)sctp_statistics
);
1033 static void sctp_v4_pf_init(void)
1035 /* Initialize the SCTP specific PF functions. */
1036 sctp_register_pf(&sctp_pf_inet
, PF_INET
);
1037 sctp_register_af(&sctp_af_inet
);
1040 static void sctp_v4_pf_exit(void)
1042 list_del(&sctp_af_inet
.list
);
1045 static int sctp_v4_protosw_init(void)
1049 rc
= proto_register(&sctp_prot
, 1);
1053 /* Register SCTP(UDP and TCP style) with socket layer. */
1054 inet_register_protosw(&sctp_seqpacket_protosw
);
1055 inet_register_protosw(&sctp_stream_protosw
);
1060 static void sctp_v4_protosw_exit(void)
1062 inet_unregister_protosw(&sctp_stream_protosw
);
1063 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1064 proto_unregister(&sctp_prot
);
1067 static int sctp_v4_add_protocol(void)
1069 /* Register notifier for inet address additions/deletions. */
1070 register_inetaddr_notifier(&sctp_inetaddr_notifier
);
1072 /* Register SCTP with inet layer. */
1073 if (inet_add_protocol(&sctp_protocol
, IPPROTO_SCTP
) < 0)
1079 static void sctp_v4_del_protocol(void)
1081 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1082 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1085 /* Initialize the universe into something sensible. */
1086 SCTP_STATIC __init
int sctp_init(void)
1089 int status
= -EINVAL
;
1091 unsigned long limit
;
1092 unsigned long nr_pages
;
1096 /* SCTP_DEBUG sanity check. */
1097 if (!sctp_sanity_check())
1100 /* Allocate bind_bucket and chunk caches. */
1102 sctp_bucket_cachep
= kmem_cache_create("sctp_bind_bucket",
1103 sizeof(struct sctp_bind_bucket
),
1104 0, SLAB_HWCACHE_ALIGN
,
1106 if (!sctp_bucket_cachep
)
1109 sctp_chunk_cachep
= kmem_cache_create("sctp_chunk",
1110 sizeof(struct sctp_chunk
),
1111 0, SLAB_HWCACHE_ALIGN
,
1113 if (!sctp_chunk_cachep
)
1114 goto err_chunk_cachep
;
1116 /* Allocate and initialise sctp mibs. */
1117 status
= init_sctp_mibs();
1121 /* Initialize proc fs directory. */
1122 status
= sctp_proc_init();
1126 /* Initialize object count debugging. */
1127 sctp_dbg_objcnt_init();
1130 * 14. Suggested SCTP Protocol Parameter Values
1132 /* The following protocol parameters are RECOMMENDED: */
1133 /* RTO.Initial - 3 seconds */
1134 sctp_rto_initial
= SCTP_RTO_INITIAL
;
1135 /* RTO.Min - 1 second */
1136 sctp_rto_min
= SCTP_RTO_MIN
;
1137 /* RTO.Max - 60 seconds */
1138 sctp_rto_max
= SCTP_RTO_MAX
;
1139 /* RTO.Alpha - 1/8 */
1140 sctp_rto_alpha
= SCTP_RTO_ALPHA
;
1141 /* RTO.Beta - 1/4 */
1142 sctp_rto_beta
= SCTP_RTO_BETA
;
1144 /* Valid.Cookie.Life - 60 seconds */
1145 sctp_valid_cookie_life
= SCTP_DEFAULT_COOKIE_LIFE
;
1147 /* Whether Cookie Preservative is enabled(1) or not(0) */
1148 sctp_cookie_preserve_enable
= 1;
1151 sctp_max_burst
= SCTP_DEFAULT_MAX_BURST
;
1153 /* Association.Max.Retrans - 10 attempts
1154 * Path.Max.Retrans - 5 attempts (per destination address)
1155 * Max.Init.Retransmits - 8 attempts
1157 sctp_max_retrans_association
= 10;
1158 sctp_max_retrans_path
= 5;
1159 sctp_max_retrans_init
= 8;
1161 /* Sendbuffer growth - do per-socket accounting */
1162 sctp_sndbuf_policy
= 0;
1164 /* Rcvbuffer growth - do per-socket accounting */
1165 sctp_rcvbuf_policy
= 0;
1167 /* HB.interval - 30 seconds */
1168 sctp_hb_interval
= SCTP_DEFAULT_TIMEOUT_HEARTBEAT
;
1170 /* delayed SACK timeout */
1171 sctp_sack_timeout
= SCTP_DEFAULT_TIMEOUT_SACK
;
1173 /* Implementation specific variables. */
1175 /* Initialize default stream count setup information. */
1176 sctp_max_instreams
= SCTP_DEFAULT_INSTREAMS
;
1177 sctp_max_outstreams
= SCTP_DEFAULT_OUTSTREAMS
;
1179 /* Initialize handle used for association ids. */
1180 idr_init(&sctp_assocs_id
);
1182 /* Set the pressure threshold to be a fraction of global memory that
1183 * is up to 1/2 at 256 MB, decreasing toward zero with the amount of
1184 * memory, with a floor of 128 pages.
1185 * Note this initalizes the data in sctpv6_prot too
1186 * Unabashedly stolen from tcp_init
1188 nr_pages
= totalram_pages
- totalhigh_pages
;
1189 limit
= min(nr_pages
, 1UL<<(28-PAGE_SHIFT
)) >> (20-PAGE_SHIFT
);
1190 limit
= (limit
* (nr_pages
>> (20-PAGE_SHIFT
))) >> (PAGE_SHIFT
-11);
1191 limit
= max(limit
, 128UL);
1192 sysctl_sctp_mem
[0] = limit
/ 4 * 3;
1193 sysctl_sctp_mem
[1] = limit
;
1194 sysctl_sctp_mem
[2] = sysctl_sctp_mem
[0] * 2;
1196 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1197 limit
= (sysctl_sctp_mem
[1]) << (PAGE_SHIFT
- 7);
1198 max_share
= min(4UL*1024*1024, limit
);
1200 sysctl_sctp_rmem
[0] = SK_MEM_QUANTUM
; /* give each asoc 1 page min */
1201 sysctl_sctp_rmem
[1] = (1500 *(sizeof(struct sk_buff
) + 1));
1202 sysctl_sctp_rmem
[2] = max(sysctl_sctp_rmem
[1], max_share
);
1204 sysctl_sctp_wmem
[0] = SK_MEM_QUANTUM
;
1205 sysctl_sctp_wmem
[1] = 16*1024;
1206 sysctl_sctp_wmem
[2] = max(64*1024, max_share
);
1208 /* Size and allocate the association hash table.
1209 * The methodology is similar to that of the tcp hash tables.
1211 if (num_physpages
>= (128 * 1024))
1212 goal
= num_physpages
>> (22 - PAGE_SHIFT
);
1214 goal
= num_physpages
>> (24 - PAGE_SHIFT
);
1216 for (order
= 0; (1UL << order
) < goal
; order
++)
1220 sctp_assoc_hashsize
= (1UL << order
) * PAGE_SIZE
/
1221 sizeof(struct sctp_hashbucket
);
1222 if ((sctp_assoc_hashsize
> (64 * 1024)) && order
> 0)
1224 sctp_assoc_hashtable
= (struct sctp_hashbucket
*)
1225 __get_free_pages(GFP_ATOMIC
, order
);
1226 } while (!sctp_assoc_hashtable
&& --order
> 0);
1227 if (!sctp_assoc_hashtable
) {
1228 printk(KERN_ERR
"SCTP: Failed association hash alloc.\n");
1230 goto err_ahash_alloc
;
1232 for (i
= 0; i
< sctp_assoc_hashsize
; i
++) {
1233 rwlock_init(&sctp_assoc_hashtable
[i
].lock
);
1234 INIT_HLIST_HEAD(&sctp_assoc_hashtable
[i
].chain
);
1237 /* Allocate and initialize the endpoint hash table. */
1238 sctp_ep_hashsize
= 64;
1239 sctp_ep_hashtable
= (struct sctp_hashbucket
*)
1240 kmalloc(64 * sizeof(struct sctp_hashbucket
), GFP_KERNEL
);
1241 if (!sctp_ep_hashtable
) {
1242 printk(KERN_ERR
"SCTP: Failed endpoint_hash alloc.\n");
1244 goto err_ehash_alloc
;
1246 for (i
= 0; i
< sctp_ep_hashsize
; i
++) {
1247 rwlock_init(&sctp_ep_hashtable
[i
].lock
);
1248 INIT_HLIST_HEAD(&sctp_ep_hashtable
[i
].chain
);
1251 /* Allocate and initialize the SCTP port hash table. */
1253 sctp_port_hashsize
= (1UL << order
) * PAGE_SIZE
/
1254 sizeof(struct sctp_bind_hashbucket
);
1255 if ((sctp_port_hashsize
> (64 * 1024)) && order
> 0)
1257 sctp_port_hashtable
= (struct sctp_bind_hashbucket
*)
1258 __get_free_pages(GFP_ATOMIC
, order
);
1259 } while (!sctp_port_hashtable
&& --order
> 0);
1260 if (!sctp_port_hashtable
) {
1261 printk(KERN_ERR
"SCTP: Failed bind hash alloc.");
1263 goto err_bhash_alloc
;
1265 for (i
= 0; i
< sctp_port_hashsize
; i
++) {
1266 spin_lock_init(&sctp_port_hashtable
[i
].lock
);
1267 INIT_HLIST_HEAD(&sctp_port_hashtable
[i
].chain
);
1270 printk(KERN_INFO
"SCTP: Hash tables configured "
1271 "(established %d bind %d)\n",
1272 sctp_assoc_hashsize
, sctp_port_hashsize
);
1274 /* Disable ADDIP by default. */
1275 sctp_addip_enable
= 0;
1276 sctp_addip_noauth
= 0;
1278 /* Enable PR-SCTP by default. */
1279 sctp_prsctp_enable
= 1;
1281 /* Disable AUTH by default. */
1282 sctp_auth_enable
= 0;
1284 sctp_sysctl_register();
1286 INIT_LIST_HEAD(&sctp_address_families
);
1290 /* Initialize the local address list. */
1291 INIT_LIST_HEAD(&sctp_local_addr_list
);
1292 spin_lock_init(&sctp_local_addr_lock
);
1293 sctp_get_local_addr_list();
1295 status
= sctp_v4_protosw_init();
1298 goto err_protosw_init
;
1300 status
= sctp_v6_protosw_init();
1302 goto err_v6_protosw_init
;
1304 /* Initialize the control inode/socket for handling OOTB packets. */
1305 if ((status
= sctp_ctl_sock_init())) {
1307 "SCTP: Failed to initialize the SCTP control sock.\n");
1308 goto err_ctl_sock_init
;
1311 status
= sctp_v4_add_protocol();
1313 goto err_add_protocol
;
1315 /* Register SCTP with inet6 layer. */
1316 status
= sctp_v6_add_protocol();
1318 goto err_v6_add_protocol
;
1323 err_v6_add_protocol
:
1324 sctp_v6_del_protocol();
1326 sctp_v4_del_protocol();
1327 inet_ctl_sock_destroy(sctp_ctl_sock
);
1329 sctp_v6_protosw_exit();
1330 err_v6_protosw_init
:
1331 sctp_v4_protosw_exit();
1333 sctp_free_local_addr_list();
1336 sctp_sysctl_unregister();
1337 list_del(&sctp_af_inet
.list
);
1338 free_pages((unsigned long)sctp_port_hashtable
,
1339 get_order(sctp_port_hashsize
*
1340 sizeof(struct sctp_bind_hashbucket
)));
1342 kfree(sctp_ep_hashtable
);
1344 free_pages((unsigned long)sctp_assoc_hashtable
,
1345 get_order(sctp_assoc_hashsize
*
1346 sizeof(struct sctp_hashbucket
)));
1348 sctp_dbg_objcnt_exit();
1351 cleanup_sctp_mibs();
1353 kmem_cache_destroy(sctp_chunk_cachep
);
1355 kmem_cache_destroy(sctp_bucket_cachep
);
1359 /* Exit handler for the SCTP protocol. */
1360 SCTP_STATIC __exit
void sctp_exit(void)
1362 /* BUG. This should probably do something useful like clean
1363 * up all the remaining associations and all that memory.
1366 /* Unregister with inet6/inet layers. */
1367 sctp_v6_del_protocol();
1368 sctp_v4_del_protocol();
1370 /* Free the control endpoint. */
1371 inet_ctl_sock_destroy(sctp_ctl_sock
);
1373 /* Free protosw registrations */
1374 sctp_v6_protosw_exit();
1375 sctp_v4_protosw_exit();
1377 /* Free the local address list. */
1378 sctp_free_local_addr_list();
1380 /* Unregister with socket layer. */
1384 sctp_sysctl_unregister();
1385 list_del(&sctp_af_inet
.list
);
1387 free_pages((unsigned long)sctp_assoc_hashtable
,
1388 get_order(sctp_assoc_hashsize
*
1389 sizeof(struct sctp_hashbucket
)));
1390 kfree(sctp_ep_hashtable
);
1391 free_pages((unsigned long)sctp_port_hashtable
,
1392 get_order(sctp_port_hashsize
*
1393 sizeof(struct sctp_bind_hashbucket
)));
1395 sctp_dbg_objcnt_exit();
1397 cleanup_sctp_mibs();
1399 kmem_cache_destroy(sctp_chunk_cachep
);
1400 kmem_cache_destroy(sctp_bucket_cachep
);
1403 module_init(sctp_init
);
1404 module_exit(sctp_exit
);
1407 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1409 MODULE_ALIAS("net-pf-" __stringify(PF_INET
) "-proto-132");
1410 MODULE_ALIAS("net-pf-" __stringify(PF_INET6
) "-proto-132");
1411 MODULE_AUTHOR("Linux Kernel SCTP developers <lksctp-developers@lists.sourceforge.net>");
1412 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1413 MODULE_LICENSE("GPL");