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, see
27 * <http://www.gnu.org/licenses/>.
29 * Please send any bug reports or fixes you make to the
31 * lksctp developers <linux-sctp@vger.kernel.org>
33 * Written or modified by:
34 * La Monte H.P. Yarroll <piggy@acm.org>
35 * Karl Knutson <karl@athena.chicago.il.us>
36 * Jon Grimm <jgrimm@us.ibm.com>
37 * Sridhar Samudrala <sri@us.ibm.com>
38 * Daisy Chang <daisyc@us.ibm.com>
39 * Ardelle Fan <ardelle.fan@intel.com>
42 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
44 #include <linux/module.h>
45 #include <linux/init.h>
46 #include <linux/netdevice.h>
47 #include <linux/inetdevice.h>
48 #include <linux/seq_file.h>
49 #include <linux/memblock.h>
50 #include <linux/highmem.h>
51 #include <linux/swap.h>
52 #include <linux/slab.h>
53 #include <net/net_namespace.h>
54 #include <net/protocol.h>
57 #include <net/route.h>
58 #include <net/sctp/sctp.h>
59 #include <net/addrconf.h>
60 #include <net/inet_common.h>
61 #include <net/inet_ecn.h>
63 #define MAX_SCTP_PORT_HASH_ENTRIES (64 * 1024)
65 /* Global data structures. */
66 struct sctp_globals sctp_globals __read_mostly
;
68 struct idr sctp_assocs_id
;
69 DEFINE_SPINLOCK(sctp_assocs_id_lock
);
71 static struct sctp_pf
*sctp_pf_inet6_specific
;
72 static struct sctp_pf
*sctp_pf_inet_specific
;
73 static struct sctp_af
*sctp_af_v4_specific
;
74 static struct sctp_af
*sctp_af_v6_specific
;
76 struct kmem_cache
*sctp_chunk_cachep __read_mostly
;
77 struct kmem_cache
*sctp_bucket_cachep __read_mostly
;
79 long sysctl_sctp_mem
[3];
80 int sysctl_sctp_rmem
[3];
81 int sysctl_sctp_wmem
[3];
83 /* Private helper to extract ipv4 address and stash them in
84 * the protocol structure.
86 static void sctp_v4_copy_addrlist(struct list_head
*addrlist
,
87 struct net_device
*dev
)
89 struct in_device
*in_dev
;
90 struct in_ifaddr
*ifa
;
91 struct sctp_sockaddr_entry
*addr
;
94 if ((in_dev
= __in_dev_get_rcu(dev
)) == NULL
) {
99 for (ifa
= in_dev
->ifa_list
; ifa
; ifa
= ifa
->ifa_next
) {
100 /* Add the address to the local list. */
101 addr
= kzalloc(sizeof(*addr
), GFP_ATOMIC
);
103 addr
->a
.v4
.sin_family
= AF_INET
;
104 addr
->a
.v4
.sin_port
= 0;
105 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
107 INIT_LIST_HEAD(&addr
->list
);
108 list_add_tail(&addr
->list
, addrlist
);
115 /* Extract our IP addresses from the system and stash them in the
116 * protocol structure.
118 static void sctp_get_local_addr_list(struct net
*net
)
120 struct net_device
*dev
;
121 struct list_head
*pos
;
125 for_each_netdev_rcu(net
, dev
) {
126 list_for_each(pos
, &sctp_address_families
) {
127 af
= list_entry(pos
, struct sctp_af
, list
);
128 af
->copy_addrlist(&net
->sctp
.local_addr_list
, dev
);
134 /* Free the existing local addresses. */
135 static void sctp_free_local_addr_list(struct net
*net
)
137 struct sctp_sockaddr_entry
*addr
;
138 struct list_head
*pos
, *temp
;
140 list_for_each_safe(pos
, temp
, &net
->sctp
.local_addr_list
) {
141 addr
= list_entry(pos
, struct sctp_sockaddr_entry
, list
);
147 /* Copy the local addresses which are valid for 'scope' into 'bp'. */
148 int sctp_copy_local_addr_list(struct net
*net
, struct sctp_bind_addr
*bp
,
149 enum sctp_scope scope
, gfp_t gfp
, int copy_flags
)
151 struct sctp_sockaddr_entry
*addr
;
152 union sctp_addr laddr
;
156 list_for_each_entry_rcu(addr
, &net
->sctp
.local_addr_list
, list
) {
159 if (!sctp_in_scope(net
, &addr
->a
, scope
))
162 /* Now that the address is in scope, check to see if
163 * the address type is really supported by the local
164 * sock as well as the remote peer.
166 if (addr
->a
.sa
.sa_family
== AF_INET
&&
167 !(copy_flags
& SCTP_ADDR4_PEERSUPP
))
169 if (addr
->a
.sa
.sa_family
== AF_INET6
&&
170 (!(copy_flags
& SCTP_ADDR6_ALLOWED
) ||
171 !(copy_flags
& SCTP_ADDR6_PEERSUPP
)))
175 /* also works for setting ipv6 address port */
176 laddr
.v4
.sin_port
= htons(bp
->port
);
177 if (sctp_bind_addr_state(bp
, &laddr
) != -1)
180 error
= sctp_add_bind_addr(bp
, &addr
->a
, sizeof(addr
->a
),
181 SCTP_ADDR_SRC
, GFP_ATOMIC
);
190 /* Copy over any ip options */
191 static void sctp_v4_copy_ip_options(struct sock
*sk
, struct sock
*newsk
)
193 struct inet_sock
*newinet
, *inet
= inet_sk(sk
);
194 struct ip_options_rcu
*inet_opt
, *newopt
= NULL
;
196 newinet
= inet_sk(newsk
);
199 inet_opt
= rcu_dereference(inet
->inet_opt
);
201 newopt
= sock_kmalloc(newsk
, sizeof(*inet_opt
) +
202 inet_opt
->opt
.optlen
, GFP_ATOMIC
);
204 memcpy(newopt
, inet_opt
, sizeof(*inet_opt
) +
205 inet_opt
->opt
.optlen
);
207 pr_err("%s: Failed to copy ip options\n", __func__
);
209 RCU_INIT_POINTER(newinet
->inet_opt
, newopt
);
213 /* Account for the IP options */
214 static int sctp_v4_ip_options_len(struct sock
*sk
)
216 struct inet_sock
*inet
= inet_sk(sk
);
217 struct ip_options_rcu
*inet_opt
;
221 inet_opt
= rcu_dereference(inet
->inet_opt
);
223 len
= inet_opt
->opt
.optlen
;
229 /* Initialize a sctp_addr from in incoming skb. */
230 static void sctp_v4_from_skb(union sctp_addr
*addr
, struct sk_buff
*skb
,
233 /* Always called on head skb, so this is safe */
234 struct sctphdr
*sh
= sctp_hdr(skb
);
235 struct sockaddr_in
*sa
= &addr
->v4
;
237 addr
->v4
.sin_family
= AF_INET
;
240 sa
->sin_port
= sh
->source
;
241 sa
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
243 sa
->sin_port
= sh
->dest
;
244 sa
->sin_addr
.s_addr
= ip_hdr(skb
)->daddr
;
248 /* Initialize an sctp_addr from a socket. */
249 static void sctp_v4_from_sk(union sctp_addr
*addr
, struct sock
*sk
)
251 addr
->v4
.sin_family
= AF_INET
;
252 addr
->v4
.sin_port
= 0;
253 addr
->v4
.sin_addr
.s_addr
= inet_sk(sk
)->inet_rcv_saddr
;
256 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
257 static void sctp_v4_to_sk_saddr(union sctp_addr
*addr
, struct sock
*sk
)
259 inet_sk(sk
)->inet_rcv_saddr
= addr
->v4
.sin_addr
.s_addr
;
262 /* Initialize sk->sk_daddr from sctp_addr. */
263 static void sctp_v4_to_sk_daddr(union sctp_addr
*addr
, struct sock
*sk
)
265 inet_sk(sk
)->inet_daddr
= addr
->v4
.sin_addr
.s_addr
;
268 /* Initialize a sctp_addr from an address parameter. */
269 static void sctp_v4_from_addr_param(union sctp_addr
*addr
,
270 union sctp_addr_param
*param
,
271 __be16 port
, int iif
)
273 addr
->v4
.sin_family
= AF_INET
;
274 addr
->v4
.sin_port
= port
;
275 addr
->v4
.sin_addr
.s_addr
= param
->v4
.addr
.s_addr
;
278 /* Initialize an address parameter from a sctp_addr and return the length
279 * of the address parameter.
281 static int sctp_v4_to_addr_param(const union sctp_addr
*addr
,
282 union sctp_addr_param
*param
)
284 int length
= sizeof(struct sctp_ipv4addr_param
);
286 param
->v4
.param_hdr
.type
= SCTP_PARAM_IPV4_ADDRESS
;
287 param
->v4
.param_hdr
.length
= htons(length
);
288 param
->v4
.addr
.s_addr
= addr
->v4
.sin_addr
.s_addr
;
293 /* Initialize a sctp_addr from a dst_entry. */
294 static void sctp_v4_dst_saddr(union sctp_addr
*saddr
, struct flowi4
*fl4
,
297 saddr
->v4
.sin_family
= AF_INET
;
298 saddr
->v4
.sin_port
= port
;
299 saddr
->v4
.sin_addr
.s_addr
= fl4
->saddr
;
302 /* Compare two addresses exactly. */
303 static int sctp_v4_cmp_addr(const union sctp_addr
*addr1
,
304 const union sctp_addr
*addr2
)
306 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
308 if (addr1
->v4
.sin_port
!= addr2
->v4
.sin_port
)
310 if (addr1
->v4
.sin_addr
.s_addr
!= addr2
->v4
.sin_addr
.s_addr
)
316 /* Initialize addr struct to INADDR_ANY. */
317 static void sctp_v4_inaddr_any(union sctp_addr
*addr
, __be16 port
)
319 addr
->v4
.sin_family
= AF_INET
;
320 addr
->v4
.sin_addr
.s_addr
= htonl(INADDR_ANY
);
321 addr
->v4
.sin_port
= port
;
324 /* Is this a wildcard address? */
325 static int sctp_v4_is_any(const union sctp_addr
*addr
)
327 return htonl(INADDR_ANY
) == addr
->v4
.sin_addr
.s_addr
;
330 /* This function checks if the address is a valid address to be used for
334 * Return 0 - If the address is a non-unicast or an illegal address.
335 * Return 1 - If the address is a unicast.
337 static int sctp_v4_addr_valid(union sctp_addr
*addr
,
338 struct sctp_sock
*sp
,
339 const struct sk_buff
*skb
)
341 /* IPv4 addresses not allowed */
342 if (sp
&& ipv6_only_sock(sctp_opt2sk(sp
)))
345 /* Is this a non-unicast address or a unusable SCTP address? */
346 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
))
349 /* Is this a broadcast address? */
350 if (skb
&& skb_rtable(skb
)->rt_flags
& RTCF_BROADCAST
)
356 /* Should this be available for binding? */
357 static int sctp_v4_available(union sctp_addr
*addr
, struct sctp_sock
*sp
)
359 struct net
*net
= sock_net(&sp
->inet
.sk
);
360 int ret
= inet_addr_type(net
, addr
->v4
.sin_addr
.s_addr
);
363 if (addr
->v4
.sin_addr
.s_addr
!= htonl(INADDR_ANY
) &&
365 !sp
->inet
.freebind
&&
366 !net
->ipv4
.sysctl_ip_nonlocal_bind
)
369 if (ipv6_only_sock(sctp_opt2sk(sp
)))
375 /* Checking the loopback, private and other address scopes as defined in
376 * RFC 1918. The IPv4 scoping is based on the draft for SCTP IPv4
377 * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
379 * Level 0 - unusable SCTP addresses
380 * Level 1 - loopback address
381 * Level 2 - link-local addresses
382 * Level 3 - private addresses.
383 * Level 4 - global addresses
384 * For INIT and INIT-ACK address list, let L be the level of
385 * of requested destination address, sender and receiver
386 * SHOULD include all of its addresses with level greater
387 * than or equal to L.
389 * IPv4 scoping can be controlled through sysctl option
390 * net.sctp.addr_scope_policy
392 static enum sctp_scope
sctp_v4_scope(union sctp_addr
*addr
)
394 enum sctp_scope retval
;
396 /* Check for unusable SCTP addresses. */
397 if (IS_IPV4_UNUSABLE_ADDRESS(addr
->v4
.sin_addr
.s_addr
)) {
398 retval
= SCTP_SCOPE_UNUSABLE
;
399 } else if (ipv4_is_loopback(addr
->v4
.sin_addr
.s_addr
)) {
400 retval
= SCTP_SCOPE_LOOPBACK
;
401 } else if (ipv4_is_linklocal_169(addr
->v4
.sin_addr
.s_addr
)) {
402 retval
= SCTP_SCOPE_LINK
;
403 } else if (ipv4_is_private_10(addr
->v4
.sin_addr
.s_addr
) ||
404 ipv4_is_private_172(addr
->v4
.sin_addr
.s_addr
) ||
405 ipv4_is_private_192(addr
->v4
.sin_addr
.s_addr
)) {
406 retval
= SCTP_SCOPE_PRIVATE
;
408 retval
= SCTP_SCOPE_GLOBAL
;
414 /* Returns a valid dst cache entry for the given source and destination ip
415 * addresses. If an association is passed, trys to get a dst entry with a
416 * source address that matches an address in the bind address list.
418 static void sctp_v4_get_dst(struct sctp_transport
*t
, union sctp_addr
*saddr
,
419 struct flowi
*fl
, struct sock
*sk
)
421 struct sctp_association
*asoc
= t
->asoc
;
423 struct flowi4
*fl4
= &fl
->u
.ip4
;
424 struct sctp_bind_addr
*bp
;
425 struct sctp_sockaddr_entry
*laddr
;
426 struct dst_entry
*dst
= NULL
;
427 union sctp_addr
*daddr
= &t
->ipaddr
;
428 union sctp_addr dst_saddr
;
429 __u8 tos
= inet_sk(sk
)->tos
;
431 if (t
->dscp
& SCTP_DSCP_SET_MASK
)
432 tos
= t
->dscp
& SCTP_DSCP_VAL_MASK
;
433 memset(fl4
, 0x0, sizeof(struct flowi4
));
434 fl4
->daddr
= daddr
->v4
.sin_addr
.s_addr
;
435 fl4
->fl4_dport
= daddr
->v4
.sin_port
;
436 fl4
->flowi4_proto
= IPPROTO_SCTP
;
438 fl4
->flowi4_tos
= RT_CONN_FLAGS_TOS(asoc
->base
.sk
, tos
);
439 fl4
->flowi4_oif
= asoc
->base
.sk
->sk_bound_dev_if
;
440 fl4
->fl4_sport
= htons(asoc
->base
.bind_addr
.port
);
443 fl4
->saddr
= saddr
->v4
.sin_addr
.s_addr
;
444 fl4
->fl4_sport
= saddr
->v4
.sin_port
;
447 pr_debug("%s: dst:%pI4, src:%pI4 - ", __func__
, &fl4
->daddr
,
450 rt
= ip_route_output_key(sock_net(sk
), fl4
);
454 /* If there is no association or if a source address is passed, no
455 * more validation is required.
460 bp
= &asoc
->base
.bind_addr
;
463 /* Walk through the bind address list and look for a bind
464 * address that matches the source address of the returned dst.
466 sctp_v4_dst_saddr(&dst_saddr
, fl4
, htons(bp
->port
));
468 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
469 if (!laddr
->valid
|| (laddr
->state
== SCTP_ADDR_DEL
) ||
470 (laddr
->state
!= SCTP_ADDR_SRC
&&
471 !asoc
->src_out_of_asoc_ok
))
473 if (sctp_v4_cmp_addr(&dst_saddr
, &laddr
->a
))
478 /* None of the bound addresses match the source address of the
479 * dst. So release it.
485 /* Walk through the bind address list and try to get a dst that
486 * matches a bind address as the source address.
489 list_for_each_entry_rcu(laddr
, &bp
->address_list
, list
) {
490 struct net_device
*odev
;
494 if (laddr
->state
!= SCTP_ADDR_SRC
||
495 AF_INET
!= laddr
->a
.sa
.sa_family
)
498 fl4
->fl4_sport
= laddr
->a
.v4
.sin_port
;
499 flowi4_update_output(fl4
,
500 asoc
->base
.sk
->sk_bound_dev_if
,
501 RT_CONN_FLAGS_TOS(asoc
->base
.sk
, tos
),
502 daddr
->v4
.sin_addr
.s_addr
,
503 laddr
->a
.v4
.sin_addr
.s_addr
);
505 rt
= ip_route_output_key(sock_net(sk
), fl4
);
509 /* Ensure the src address belongs to the output
512 odev
= __ip_dev_find(sock_net(sk
), laddr
->a
.v4
.sin_addr
.s_addr
,
514 if (!odev
|| odev
->ifindex
!= fl4
->flowi4_oif
) {
518 dst_release(&rt
->dst
);
532 pr_debug("rt_dst:%pI4, rt_src:%pI4\n",
533 &fl4
->daddr
, &fl4
->saddr
);
535 pr_debug("no route\n");
538 /* For v4, the source address is cached in the route entry(dst). So no need
539 * to cache it separately and hence this is an empty routine.
541 static void sctp_v4_get_saddr(struct sctp_sock
*sk
,
542 struct sctp_transport
*t
,
545 union sctp_addr
*saddr
= &t
->saddr
;
546 struct rtable
*rt
= (struct rtable
*)t
->dst
;
549 saddr
->v4
.sin_family
= AF_INET
;
550 saddr
->v4
.sin_addr
.s_addr
= fl
->u
.ip4
.saddr
;
554 /* What interface did this skb arrive on? */
555 static int sctp_v4_skb_iif(const struct sk_buff
*skb
)
557 return inet_iif(skb
);
560 /* Was this packet marked by Explicit Congestion Notification? */
561 static int sctp_v4_is_ce(const struct sk_buff
*skb
)
563 return INET_ECN_is_ce(ip_hdr(skb
)->tos
);
566 /* Create and initialize a new sk for the socket returned by accept(). */
567 static struct sock
*sctp_v4_create_accept_sk(struct sock
*sk
,
568 struct sctp_association
*asoc
,
571 struct sock
*newsk
= sk_alloc(sock_net(sk
), PF_INET
, GFP_KERNEL
,
573 struct inet_sock
*newinet
;
578 sock_init_data(NULL
, newsk
);
580 sctp_copy_sock(newsk
, sk
, asoc
);
581 sock_reset_flag(newsk
, SOCK_ZAPPED
);
583 sctp_v4_copy_ip_options(sk
, newsk
);
585 newinet
= inet_sk(newsk
);
587 newinet
->inet_daddr
= asoc
->peer
.primary_addr
.v4
.sin_addr
.s_addr
;
589 sk_refcnt_debug_inc(newsk
);
591 if (newsk
->sk_prot
->init(newsk
)) {
592 sk_common_release(newsk
);
600 static int sctp_v4_addr_to_user(struct sctp_sock
*sp
, union sctp_addr
*addr
)
602 /* No address mapping for V4 sockets */
603 return sizeof(struct sockaddr_in
);
606 /* Dump the v4 addr to the seq file. */
607 static void sctp_v4_seq_dump_addr(struct seq_file
*seq
, union sctp_addr
*addr
)
609 seq_printf(seq
, "%pI4 ", &addr
->v4
.sin_addr
);
612 static void sctp_v4_ecn_capable(struct sock
*sk
)
617 static void sctp_addr_wq_timeout_handler(struct timer_list
*t
)
619 struct net
*net
= from_timer(net
, t
, sctp
.addr_wq_timer
);
620 struct sctp_sockaddr_entry
*addrw
, *temp
;
621 struct sctp_sock
*sp
;
623 spin_lock_bh(&net
->sctp
.addr_wq_lock
);
625 list_for_each_entry_safe(addrw
, temp
, &net
->sctp
.addr_waitq
, list
) {
626 pr_debug("%s: the first ent in wq:%p is addr:%pISc for cmd:%d at "
627 "entry:%p\n", __func__
, &net
->sctp
.addr_waitq
, &addrw
->a
.sa
,
628 addrw
->state
, addrw
);
630 #if IS_ENABLED(CONFIG_IPV6)
631 /* Now we send an ASCONF for each association */
632 /* Note. we currently don't handle link local IPv6 addressees */
633 if (addrw
->a
.sa
.sa_family
== AF_INET6
) {
634 struct in6_addr
*in6
;
636 if (ipv6_addr_type(&addrw
->a
.v6
.sin6_addr
) &
640 in6
= (struct in6_addr
*)&addrw
->a
.v6
.sin6_addr
;
641 if (ipv6_chk_addr(net
, in6
, NULL
, 0) == 0 &&
642 addrw
->state
== SCTP_ADDR_NEW
) {
643 unsigned long timeo_val
;
645 pr_debug("%s: this is on DAD, trying %d sec "
647 SCTP_ADDRESS_TICK_DELAY
);
650 timeo_val
+= msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY
);
651 mod_timer(&net
->sctp
.addr_wq_timer
, timeo_val
);
656 list_for_each_entry(sp
, &net
->sctp
.auto_asconf_splist
, auto_asconf_list
) {
659 sk
= sctp_opt2sk(sp
);
660 /* ignore bound-specific endpoints */
661 if (!sctp_is_ep_boundall(sk
))
664 if (sctp_asconf_mgmt(sp
, addrw
) < 0)
665 pr_debug("%s: sctp_asconf_mgmt failed\n", __func__
);
668 #if IS_ENABLED(CONFIG_IPV6)
671 list_del(&addrw
->list
);
674 spin_unlock_bh(&net
->sctp
.addr_wq_lock
);
677 static void sctp_free_addr_wq(struct net
*net
)
679 struct sctp_sockaddr_entry
*addrw
;
680 struct sctp_sockaddr_entry
*temp
;
682 spin_lock_bh(&net
->sctp
.addr_wq_lock
);
683 del_timer(&net
->sctp
.addr_wq_timer
);
684 list_for_each_entry_safe(addrw
, temp
, &net
->sctp
.addr_waitq
, list
) {
685 list_del(&addrw
->list
);
688 spin_unlock_bh(&net
->sctp
.addr_wq_lock
);
691 /* lookup the entry for the same address in the addr_waitq
692 * sctp_addr_wq MUST be locked
694 static struct sctp_sockaddr_entry
*sctp_addr_wq_lookup(struct net
*net
,
695 struct sctp_sockaddr_entry
*addr
)
697 struct sctp_sockaddr_entry
*addrw
;
699 list_for_each_entry(addrw
, &net
->sctp
.addr_waitq
, list
) {
700 if (addrw
->a
.sa
.sa_family
!= addr
->a
.sa
.sa_family
)
702 if (addrw
->a
.sa
.sa_family
== AF_INET
) {
703 if (addrw
->a
.v4
.sin_addr
.s_addr
==
704 addr
->a
.v4
.sin_addr
.s_addr
)
706 } else if (addrw
->a
.sa
.sa_family
== AF_INET6
) {
707 if (ipv6_addr_equal(&addrw
->a
.v6
.sin6_addr
,
708 &addr
->a
.v6
.sin6_addr
))
715 void sctp_addr_wq_mgmt(struct net
*net
, struct sctp_sockaddr_entry
*addr
, int cmd
)
717 struct sctp_sockaddr_entry
*addrw
;
718 unsigned long timeo_val
;
720 /* first, we check if an opposite message already exist in the queue.
721 * If we found such message, it is removed.
722 * This operation is a bit stupid, but the DHCP client attaches the
723 * new address after a couple of addition and deletion of that address
726 spin_lock_bh(&net
->sctp
.addr_wq_lock
);
727 /* Offsets existing events in addr_wq */
728 addrw
= sctp_addr_wq_lookup(net
, addr
);
730 if (addrw
->state
!= cmd
) {
731 pr_debug("%s: offsets existing entry for %d, addr:%pISc "
732 "in wq:%p\n", __func__
, addrw
->state
, &addrw
->a
.sa
,
733 &net
->sctp
.addr_waitq
);
735 list_del(&addrw
->list
);
738 spin_unlock_bh(&net
->sctp
.addr_wq_lock
);
742 /* OK, we have to add the new address to the wait queue */
743 addrw
= kmemdup(addr
, sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
745 spin_unlock_bh(&net
->sctp
.addr_wq_lock
);
749 list_add_tail(&addrw
->list
, &net
->sctp
.addr_waitq
);
751 pr_debug("%s: add new entry for cmd:%d, addr:%pISc in wq:%p\n",
752 __func__
, addrw
->state
, &addrw
->a
.sa
, &net
->sctp
.addr_waitq
);
754 if (!timer_pending(&net
->sctp
.addr_wq_timer
)) {
756 timeo_val
+= msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY
);
757 mod_timer(&net
->sctp
.addr_wq_timer
, timeo_val
);
759 spin_unlock_bh(&net
->sctp
.addr_wq_lock
);
762 /* Event handler for inet address addition/deletion events.
763 * The sctp_local_addr_list needs to be protocted by a spin lock since
764 * multiple notifiers (say IPv4 and IPv6) may be running at the same
765 * time and thus corrupt the list.
766 * The reader side is protected with RCU.
768 static int sctp_inetaddr_event(struct notifier_block
*this, unsigned long ev
,
771 struct in_ifaddr
*ifa
= (struct in_ifaddr
*)ptr
;
772 struct sctp_sockaddr_entry
*addr
= NULL
;
773 struct sctp_sockaddr_entry
*temp
;
774 struct net
*net
= dev_net(ifa
->ifa_dev
->dev
);
779 addr
= kmalloc(sizeof(struct sctp_sockaddr_entry
), GFP_ATOMIC
);
781 addr
->a
.v4
.sin_family
= AF_INET
;
782 addr
->a
.v4
.sin_port
= 0;
783 addr
->a
.v4
.sin_addr
.s_addr
= ifa
->ifa_local
;
785 spin_lock_bh(&net
->sctp
.local_addr_lock
);
786 list_add_tail_rcu(&addr
->list
, &net
->sctp
.local_addr_list
);
787 sctp_addr_wq_mgmt(net
, addr
, SCTP_ADDR_NEW
);
788 spin_unlock_bh(&net
->sctp
.local_addr_lock
);
792 spin_lock_bh(&net
->sctp
.local_addr_lock
);
793 list_for_each_entry_safe(addr
, temp
,
794 &net
->sctp
.local_addr_list
, list
) {
795 if (addr
->a
.sa
.sa_family
== AF_INET
&&
796 addr
->a
.v4
.sin_addr
.s_addr
==
798 sctp_addr_wq_mgmt(net
, addr
, SCTP_ADDR_DEL
);
801 list_del_rcu(&addr
->list
);
805 spin_unlock_bh(&net
->sctp
.local_addr_lock
);
807 kfree_rcu(addr
, rcu
);
815 * Initialize the control inode/socket with a control endpoint data
816 * structure. This endpoint is reserved exclusively for the OOTB processing.
818 static int sctp_ctl_sock_init(struct net
*net
)
821 sa_family_t family
= PF_INET
;
823 if (sctp_get_pf_specific(PF_INET6
))
826 err
= inet_ctl_sock_create(&net
->sctp
.ctl_sock
, family
,
827 SOCK_SEQPACKET
, IPPROTO_SCTP
, net
);
829 /* If IPv6 socket could not be created, try the IPv4 socket */
830 if (err
< 0 && family
== PF_INET6
)
831 err
= inet_ctl_sock_create(&net
->sctp
.ctl_sock
, AF_INET
,
832 SOCK_SEQPACKET
, IPPROTO_SCTP
,
836 pr_err("Failed to create the SCTP control socket\n");
842 /* Register address family specific functions. */
843 int sctp_register_af(struct sctp_af
*af
)
845 switch (af
->sa_family
) {
847 if (sctp_af_v4_specific
)
849 sctp_af_v4_specific
= af
;
852 if (sctp_af_v6_specific
)
854 sctp_af_v6_specific
= af
;
860 INIT_LIST_HEAD(&af
->list
);
861 list_add_tail(&af
->list
, &sctp_address_families
);
865 /* Get the table of functions for manipulating a particular address
868 struct sctp_af
*sctp_get_af_specific(sa_family_t family
)
872 return sctp_af_v4_specific
;
874 return sctp_af_v6_specific
;
880 /* Common code to initialize a AF_INET msg_name. */
881 static void sctp_inet_msgname(char *msgname
, int *addr_len
)
883 struct sockaddr_in
*sin
;
885 sin
= (struct sockaddr_in
*)msgname
;
886 *addr_len
= sizeof(struct sockaddr_in
);
887 sin
->sin_family
= AF_INET
;
888 memset(sin
->sin_zero
, 0, sizeof(sin
->sin_zero
));
891 /* Copy the primary address of the peer primary address as the msg_name. */
892 static void sctp_inet_event_msgname(struct sctp_ulpevent
*event
, char *msgname
,
895 struct sockaddr_in
*sin
, *sinfrom
;
898 struct sctp_association
*asoc
;
901 sctp_inet_msgname(msgname
, addr_len
);
902 sin
= (struct sockaddr_in
*)msgname
;
903 sinfrom
= &asoc
->peer
.primary_addr
.v4
;
904 sin
->sin_port
= htons(asoc
->peer
.port
);
905 sin
->sin_addr
.s_addr
= sinfrom
->sin_addr
.s_addr
;
909 /* Initialize and copy out a msgname from an inbound skb. */
910 static void sctp_inet_skb_msgname(struct sk_buff
*skb
, char *msgname
, int *len
)
913 struct sctphdr
*sh
= sctp_hdr(skb
);
914 struct sockaddr_in
*sin
= (struct sockaddr_in
*)msgname
;
916 sctp_inet_msgname(msgname
, len
);
917 sin
->sin_port
= sh
->source
;
918 sin
->sin_addr
.s_addr
= ip_hdr(skb
)->saddr
;
922 /* Do we support this AF? */
923 static int sctp_inet_af_supported(sa_family_t family
, struct sctp_sock
*sp
)
925 /* PF_INET only supports AF_INET addresses. */
926 return AF_INET
== family
;
929 /* Address matching with wildcards allowed. */
930 static int sctp_inet_cmp_addr(const union sctp_addr
*addr1
,
931 const union sctp_addr
*addr2
,
932 struct sctp_sock
*opt
)
934 /* PF_INET only supports AF_INET addresses. */
935 if (addr1
->sa
.sa_family
!= addr2
->sa
.sa_family
)
937 if (htonl(INADDR_ANY
) == addr1
->v4
.sin_addr
.s_addr
||
938 htonl(INADDR_ANY
) == addr2
->v4
.sin_addr
.s_addr
)
940 if (addr1
->v4
.sin_addr
.s_addr
== addr2
->v4
.sin_addr
.s_addr
)
946 /* Verify that provided sockaddr looks bindable. Common verification has
947 * already been taken care of.
949 static int sctp_inet_bind_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
951 return sctp_v4_available(addr
, opt
);
954 /* Verify that sockaddr looks sendable. Common verification has already
955 * been taken care of.
957 static int sctp_inet_send_verify(struct sctp_sock
*opt
, union sctp_addr
*addr
)
962 /* Fill in Supported Address Type information for INIT and INIT-ACK
963 * chunks. Returns number of addresses supported.
965 static int sctp_inet_supported_addrs(const struct sctp_sock
*opt
,
968 types
[0] = SCTP_PARAM_IPV4_ADDRESS
;
972 /* Wrapper routine that calls the ip transmit routine. */
973 static inline int sctp_v4_xmit(struct sk_buff
*skb
,
974 struct sctp_transport
*transport
)
976 struct inet_sock
*inet
= inet_sk(skb
->sk
);
977 __u8 dscp
= inet
->tos
;
979 pr_debug("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n", __func__
, skb
,
980 skb
->len
, &transport
->fl
.u
.ip4
.saddr
,
981 &transport
->fl
.u
.ip4
.daddr
);
983 if (transport
->dscp
& SCTP_DSCP_SET_MASK
)
984 dscp
= transport
->dscp
& SCTP_DSCP_VAL_MASK
;
986 inet
->pmtudisc
= transport
->param_flags
& SPP_PMTUD_ENABLE
?
987 IP_PMTUDISC_DO
: IP_PMTUDISC_DONT
;
989 SCTP_INC_STATS(sock_net(&inet
->sk
), SCTP_MIB_OUTSCTPPACKS
);
991 return __ip_queue_xmit(&inet
->sk
, skb
, &transport
->fl
, dscp
);
994 static struct sctp_af sctp_af_inet
;
996 static struct sctp_pf sctp_pf_inet
= {
997 .event_msgname
= sctp_inet_event_msgname
,
998 .skb_msgname
= sctp_inet_skb_msgname
,
999 .af_supported
= sctp_inet_af_supported
,
1000 .cmp_addr
= sctp_inet_cmp_addr
,
1001 .bind_verify
= sctp_inet_bind_verify
,
1002 .send_verify
= sctp_inet_send_verify
,
1003 .supported_addrs
= sctp_inet_supported_addrs
,
1004 .create_accept_sk
= sctp_v4_create_accept_sk
,
1005 .addr_to_user
= sctp_v4_addr_to_user
,
1006 .to_sk_saddr
= sctp_v4_to_sk_saddr
,
1007 .to_sk_daddr
= sctp_v4_to_sk_daddr
,
1008 .copy_ip_options
= sctp_v4_copy_ip_options
,
1012 /* Notifier for inetaddr addition/deletion events. */
1013 static struct notifier_block sctp_inetaddr_notifier
= {
1014 .notifier_call
= sctp_inetaddr_event
,
1017 /* Socket operations. */
1018 static const struct proto_ops inet_seqpacket_ops
= {
1020 .owner
= THIS_MODULE
,
1021 .release
= inet_release
, /* Needs to be wrapped... */
1023 .connect
= sctp_inet_connect
,
1024 .socketpair
= sock_no_socketpair
,
1025 .accept
= inet_accept
,
1026 .getname
= inet_getname
, /* Semantics are different. */
1028 .ioctl
= inet_ioctl
,
1029 .listen
= sctp_inet_listen
,
1030 .shutdown
= inet_shutdown
, /* Looks harmless. */
1031 .setsockopt
= sock_common_setsockopt
, /* IP_SOL IP_OPTION is a problem */
1032 .getsockopt
= sock_common_getsockopt
,
1033 .sendmsg
= inet_sendmsg
,
1034 .recvmsg
= inet_recvmsg
,
1035 .mmap
= sock_no_mmap
,
1036 .sendpage
= sock_no_sendpage
,
1037 #ifdef CONFIG_COMPAT
1038 .compat_setsockopt
= compat_sock_common_setsockopt
,
1039 .compat_getsockopt
= compat_sock_common_getsockopt
,
1043 /* Registration with AF_INET family. */
1044 static struct inet_protosw sctp_seqpacket_protosw
= {
1045 .type
= SOCK_SEQPACKET
,
1046 .protocol
= IPPROTO_SCTP
,
1048 .ops
= &inet_seqpacket_ops
,
1049 .flags
= SCTP_PROTOSW_FLAG
1051 static struct inet_protosw sctp_stream_protosw
= {
1052 .type
= SOCK_STREAM
,
1053 .protocol
= IPPROTO_SCTP
,
1055 .ops
= &inet_seqpacket_ops
,
1056 .flags
= SCTP_PROTOSW_FLAG
1059 /* Register with IP layer. */
1060 static const struct net_protocol sctp_protocol
= {
1061 .handler
= sctp_rcv
,
1062 .err_handler
= sctp_v4_err
,
1065 .icmp_strict_tag_validation
= 1,
1068 /* IPv4 address related functions. */
1069 static struct sctp_af sctp_af_inet
= {
1070 .sa_family
= AF_INET
,
1071 .sctp_xmit
= sctp_v4_xmit
,
1072 .setsockopt
= ip_setsockopt
,
1073 .getsockopt
= ip_getsockopt
,
1074 .get_dst
= sctp_v4_get_dst
,
1075 .get_saddr
= sctp_v4_get_saddr
,
1076 .copy_addrlist
= sctp_v4_copy_addrlist
,
1077 .from_skb
= sctp_v4_from_skb
,
1078 .from_sk
= sctp_v4_from_sk
,
1079 .from_addr_param
= sctp_v4_from_addr_param
,
1080 .to_addr_param
= sctp_v4_to_addr_param
,
1081 .cmp_addr
= sctp_v4_cmp_addr
,
1082 .addr_valid
= sctp_v4_addr_valid
,
1083 .inaddr_any
= sctp_v4_inaddr_any
,
1084 .is_any
= sctp_v4_is_any
,
1085 .available
= sctp_v4_available
,
1086 .scope
= sctp_v4_scope
,
1087 .skb_iif
= sctp_v4_skb_iif
,
1088 .is_ce
= sctp_v4_is_ce
,
1089 .seq_dump_addr
= sctp_v4_seq_dump_addr
,
1090 .ecn_capable
= sctp_v4_ecn_capable
,
1091 .net_header_len
= sizeof(struct iphdr
),
1092 .sockaddr_len
= sizeof(struct sockaddr_in
),
1093 .ip_options_len
= sctp_v4_ip_options_len
,
1094 #ifdef CONFIG_COMPAT
1095 .compat_setsockopt
= compat_ip_setsockopt
,
1096 .compat_getsockopt
= compat_ip_getsockopt
,
1100 struct sctp_pf
*sctp_get_pf_specific(sa_family_t family
)
1104 return sctp_pf_inet_specific
;
1106 return sctp_pf_inet6_specific
;
1112 /* Register the PF specific function table. */
1113 int sctp_register_pf(struct sctp_pf
*pf
, sa_family_t family
)
1117 if (sctp_pf_inet_specific
)
1119 sctp_pf_inet_specific
= pf
;
1122 if (sctp_pf_inet6_specific
)
1124 sctp_pf_inet6_specific
= pf
;
1132 static inline int init_sctp_mibs(struct net
*net
)
1134 net
->sctp
.sctp_statistics
= alloc_percpu(struct sctp_mib
);
1135 if (!net
->sctp
.sctp_statistics
)
1140 static inline void cleanup_sctp_mibs(struct net
*net
)
1142 free_percpu(net
->sctp
.sctp_statistics
);
1145 static void sctp_v4_pf_init(void)
1147 /* Initialize the SCTP specific PF functions. */
1148 sctp_register_pf(&sctp_pf_inet
, PF_INET
);
1149 sctp_register_af(&sctp_af_inet
);
1152 static void sctp_v4_pf_exit(void)
1154 list_del(&sctp_af_inet
.list
);
1157 static int sctp_v4_protosw_init(void)
1161 rc
= proto_register(&sctp_prot
, 1);
1165 /* Register SCTP(UDP and TCP style) with socket layer. */
1166 inet_register_protosw(&sctp_seqpacket_protosw
);
1167 inet_register_protosw(&sctp_stream_protosw
);
1172 static void sctp_v4_protosw_exit(void)
1174 inet_unregister_protosw(&sctp_stream_protosw
);
1175 inet_unregister_protosw(&sctp_seqpacket_protosw
);
1176 proto_unregister(&sctp_prot
);
1179 static int sctp_v4_add_protocol(void)
1181 /* Register notifier for inet address additions/deletions. */
1182 register_inetaddr_notifier(&sctp_inetaddr_notifier
);
1184 /* Register SCTP with inet layer. */
1185 if (inet_add_protocol(&sctp_protocol
, IPPROTO_SCTP
) < 0)
1191 static void sctp_v4_del_protocol(void)
1193 inet_del_protocol(&sctp_protocol
, IPPROTO_SCTP
);
1194 unregister_inetaddr_notifier(&sctp_inetaddr_notifier
);
1197 static int __net_init
sctp_defaults_init(struct net
*net
)
1202 * 14. Suggested SCTP Protocol Parameter Values
1204 /* The following protocol parameters are RECOMMENDED: */
1205 /* RTO.Initial - 3 seconds */
1206 net
->sctp
.rto_initial
= SCTP_RTO_INITIAL
;
1207 /* RTO.Min - 1 second */
1208 net
->sctp
.rto_min
= SCTP_RTO_MIN
;
1209 /* RTO.Max - 60 seconds */
1210 net
->sctp
.rto_max
= SCTP_RTO_MAX
;
1211 /* RTO.Alpha - 1/8 */
1212 net
->sctp
.rto_alpha
= SCTP_RTO_ALPHA
;
1213 /* RTO.Beta - 1/4 */
1214 net
->sctp
.rto_beta
= SCTP_RTO_BETA
;
1216 /* Valid.Cookie.Life - 60 seconds */
1217 net
->sctp
.valid_cookie_life
= SCTP_DEFAULT_COOKIE_LIFE
;
1219 /* Whether Cookie Preservative is enabled(1) or not(0) */
1220 net
->sctp
.cookie_preserve_enable
= 1;
1222 /* Default sctp sockets to use md5 as their hmac alg */
1223 #if defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_MD5)
1224 net
->sctp
.sctp_hmac_alg
= "md5";
1225 #elif defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_SHA1)
1226 net
->sctp
.sctp_hmac_alg
= "sha1";
1228 net
->sctp
.sctp_hmac_alg
= NULL
;
1232 net
->sctp
.max_burst
= SCTP_DEFAULT_MAX_BURST
;
1234 /* Enable pf state by default */
1235 net
->sctp
.pf_enable
= 1;
1237 /* Association.Max.Retrans - 10 attempts
1238 * Path.Max.Retrans - 5 attempts (per destination address)
1239 * Max.Init.Retransmits - 8 attempts
1241 net
->sctp
.max_retrans_association
= 10;
1242 net
->sctp
.max_retrans_path
= 5;
1243 net
->sctp
.max_retrans_init
= 8;
1245 /* Sendbuffer growth - do per-socket accounting */
1246 net
->sctp
.sndbuf_policy
= 0;
1248 /* Rcvbuffer growth - do per-socket accounting */
1249 net
->sctp
.rcvbuf_policy
= 0;
1251 /* HB.interval - 30 seconds */
1252 net
->sctp
.hb_interval
= SCTP_DEFAULT_TIMEOUT_HEARTBEAT
;
1254 /* delayed SACK timeout */
1255 net
->sctp
.sack_timeout
= SCTP_DEFAULT_TIMEOUT_SACK
;
1257 /* Disable ADDIP by default. */
1258 net
->sctp
.addip_enable
= 0;
1259 net
->sctp
.addip_noauth
= 0;
1260 net
->sctp
.default_auto_asconf
= 0;
1262 /* Enable PR-SCTP by default. */
1263 net
->sctp
.prsctp_enable
= 1;
1265 /* Disable RECONF by default. */
1266 net
->sctp
.reconf_enable
= 0;
1268 /* Disable AUTH by default. */
1269 net
->sctp
.auth_enable
= 0;
1271 /* Set SCOPE policy to enabled */
1272 net
->sctp
.scope_policy
= SCTP_SCOPE_POLICY_ENABLE
;
1274 /* Set the default rwnd update threshold */
1275 net
->sctp
.rwnd_upd_shift
= SCTP_DEFAULT_RWND_SHIFT
;
1277 /* Initialize maximum autoclose timeout. */
1278 net
->sctp
.max_autoclose
= INT_MAX
/ HZ
;
1280 status
= sctp_sysctl_net_register(net
);
1282 goto err_sysctl_register
;
1284 /* Allocate and initialise sctp mibs. */
1285 status
= init_sctp_mibs(net
);
1289 #ifdef CONFIG_PROC_FS
1290 /* Initialize proc fs directory. */
1291 status
= sctp_proc_init(net
);
1296 sctp_dbg_objcnt_init(net
);
1298 /* Initialize the local address list. */
1299 INIT_LIST_HEAD(&net
->sctp
.local_addr_list
);
1300 spin_lock_init(&net
->sctp
.local_addr_lock
);
1301 sctp_get_local_addr_list(net
);
1303 /* Initialize the address event list */
1304 INIT_LIST_HEAD(&net
->sctp
.addr_waitq
);
1305 INIT_LIST_HEAD(&net
->sctp
.auto_asconf_splist
);
1306 spin_lock_init(&net
->sctp
.addr_wq_lock
);
1307 net
->sctp
.addr_wq_timer
.expires
= 0;
1308 timer_setup(&net
->sctp
.addr_wq_timer
, sctp_addr_wq_timeout_handler
, 0);
1312 #ifdef CONFIG_PROC_FS
1314 cleanup_sctp_mibs(net
);
1317 sctp_sysctl_net_unregister(net
);
1318 err_sysctl_register
:
1322 static void __net_exit
sctp_defaults_exit(struct net
*net
)
1324 /* Free the local address list */
1325 sctp_free_addr_wq(net
);
1326 sctp_free_local_addr_list(net
);
1328 #ifdef CONFIG_PROC_FS
1329 remove_proc_subtree("sctp", net
->proc_net
);
1330 net
->sctp
.proc_net_sctp
= NULL
;
1332 cleanup_sctp_mibs(net
);
1333 sctp_sysctl_net_unregister(net
);
1336 static struct pernet_operations sctp_defaults_ops
= {
1337 .init
= sctp_defaults_init
,
1338 .exit
= sctp_defaults_exit
,
1341 static int __net_init
sctp_ctrlsock_init(struct net
*net
)
1345 /* Initialize the control inode/socket for handling OOTB packets. */
1346 status
= sctp_ctl_sock_init(net
);
1348 pr_err("Failed to initialize the SCTP control sock\n");
1353 static void __net_init
sctp_ctrlsock_exit(struct net
*net
)
1355 /* Free the control endpoint. */
1356 inet_ctl_sock_destroy(net
->sctp
.ctl_sock
);
1359 static struct pernet_operations sctp_ctrlsock_ops
= {
1360 .init
= sctp_ctrlsock_init
,
1361 .exit
= sctp_ctrlsock_exit
,
1364 /* Initialize the universe into something sensible. */
1365 static __init
int sctp_init(void)
1368 int status
= -EINVAL
;
1370 unsigned long limit
;
1371 unsigned long nr_pages
= totalram_pages();
1375 int max_entry_order
;
1377 sock_skb_cb_check_size(sizeof(struct sctp_ulpevent
));
1379 /* Allocate bind_bucket and chunk caches. */
1381 sctp_bucket_cachep
= kmem_cache_create("sctp_bind_bucket",
1382 sizeof(struct sctp_bind_bucket
),
1383 0, SLAB_HWCACHE_ALIGN
,
1385 if (!sctp_bucket_cachep
)
1388 sctp_chunk_cachep
= kmem_cache_create("sctp_chunk",
1389 sizeof(struct sctp_chunk
),
1390 0, SLAB_HWCACHE_ALIGN
,
1392 if (!sctp_chunk_cachep
)
1393 goto err_chunk_cachep
;
1395 status
= percpu_counter_init(&sctp_sockets_allocated
, 0, GFP_KERNEL
);
1397 goto err_percpu_counter_init
;
1399 /* Implementation specific variables. */
1401 /* Initialize default stream count setup information. */
1402 sctp_max_instreams
= SCTP_DEFAULT_INSTREAMS
;
1403 sctp_max_outstreams
= SCTP_DEFAULT_OUTSTREAMS
;
1405 /* Initialize handle used for association ids. */
1406 idr_init(&sctp_assocs_id
);
1408 limit
= nr_free_buffer_pages() / 8;
1409 limit
= max(limit
, 128UL);
1410 sysctl_sctp_mem
[0] = limit
/ 4 * 3;
1411 sysctl_sctp_mem
[1] = limit
;
1412 sysctl_sctp_mem
[2] = sysctl_sctp_mem
[0] * 2;
1414 /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1415 limit
= (sysctl_sctp_mem
[1]) << (PAGE_SHIFT
- 7);
1416 max_share
= min(4UL*1024*1024, limit
);
1418 sysctl_sctp_rmem
[0] = SK_MEM_QUANTUM
; /* give each asoc 1 page min */
1419 sysctl_sctp_rmem
[1] = 1500 * SKB_TRUESIZE(1);
1420 sysctl_sctp_rmem
[2] = max(sysctl_sctp_rmem
[1], max_share
);
1422 sysctl_sctp_wmem
[0] = SK_MEM_QUANTUM
;
1423 sysctl_sctp_wmem
[1] = 16*1024;
1424 sysctl_sctp_wmem
[2] = max(64*1024, max_share
);
1426 /* Size and allocate the association hash table.
1427 * The methodology is similar to that of the tcp hash tables.
1428 * Though not identical. Start by getting a goal size
1430 if (nr_pages
>= (128 * 1024))
1431 goal
= nr_pages
>> (22 - PAGE_SHIFT
);
1433 goal
= nr_pages
>> (24 - PAGE_SHIFT
);
1435 /* Then compute the page order for said goal */
1436 order
= get_order(goal
);
1438 /* Now compute the required page order for the maximum sized table we
1441 max_entry_order
= get_order(MAX_SCTP_PORT_HASH_ENTRIES
*
1442 sizeof(struct sctp_bind_hashbucket
));
1444 /* Limit the page order by that maximum hash table size */
1445 order
= min(order
, max_entry_order
);
1447 /* Allocate and initialize the endpoint hash table. */
1448 sctp_ep_hashsize
= 64;
1450 kmalloc_array(64, sizeof(struct sctp_hashbucket
), GFP_KERNEL
);
1451 if (!sctp_ep_hashtable
) {
1452 pr_err("Failed endpoint_hash alloc\n");
1454 goto err_ehash_alloc
;
1456 for (i
= 0; i
< sctp_ep_hashsize
; i
++) {
1457 rwlock_init(&sctp_ep_hashtable
[i
].lock
);
1458 INIT_HLIST_HEAD(&sctp_ep_hashtable
[i
].chain
);
1461 /* Allocate and initialize the SCTP port hash table.
1462 * Note that order is initalized to start at the max sized
1463 * table we want to support. If we can't get that many pages
1464 * reduce the order and try again
1467 sctp_port_hashtable
= (struct sctp_bind_hashbucket
*)
1468 __get_free_pages(GFP_KERNEL
| __GFP_NOWARN
, order
);
1469 } while (!sctp_port_hashtable
&& --order
> 0);
1471 if (!sctp_port_hashtable
) {
1472 pr_err("Failed bind hash alloc\n");
1474 goto err_bhash_alloc
;
1477 /* Now compute the number of entries that will fit in the
1478 * port hash space we allocated
1480 num_entries
= (1UL << order
) * PAGE_SIZE
/
1481 sizeof(struct sctp_bind_hashbucket
);
1483 /* And finish by rounding it down to the nearest power of two
1484 * this wastes some memory of course, but its needed because
1485 * the hash function operates based on the assumption that
1486 * that the number of entries is a power of two
1488 sctp_port_hashsize
= rounddown_pow_of_two(num_entries
);
1490 for (i
= 0; i
< sctp_port_hashsize
; i
++) {
1491 spin_lock_init(&sctp_port_hashtable
[i
].lock
);
1492 INIT_HLIST_HEAD(&sctp_port_hashtable
[i
].chain
);
1495 status
= sctp_transport_hashtable_init();
1497 goto err_thash_alloc
;
1499 pr_info("Hash tables configured (bind %d/%d)\n", sctp_port_hashsize
,
1502 sctp_sysctl_register();
1504 INIT_LIST_HEAD(&sctp_address_families
);
1507 sctp_sched_ops_init();
1509 status
= register_pernet_subsys(&sctp_defaults_ops
);
1511 goto err_register_defaults
;
1513 status
= sctp_v4_protosw_init();
1515 goto err_protosw_init
;
1517 status
= sctp_v6_protosw_init();
1519 goto err_v6_protosw_init
;
1521 status
= register_pernet_subsys(&sctp_ctrlsock_ops
);
1523 goto err_register_ctrlsock
;
1525 status
= sctp_v4_add_protocol();
1527 goto err_add_protocol
;
1529 /* Register SCTP with inet6 layer. */
1530 status
= sctp_v6_add_protocol();
1532 goto err_v6_add_protocol
;
1534 if (sctp_offload_init() < 0)
1535 pr_crit("%s: Cannot add SCTP protocol offload\n", __func__
);
1539 err_v6_add_protocol
:
1540 sctp_v4_del_protocol();
1542 unregister_pernet_subsys(&sctp_ctrlsock_ops
);
1543 err_register_ctrlsock
:
1544 sctp_v6_protosw_exit();
1545 err_v6_protosw_init
:
1546 sctp_v4_protosw_exit();
1548 unregister_pernet_subsys(&sctp_defaults_ops
);
1549 err_register_defaults
:
1552 sctp_sysctl_unregister();
1553 free_pages((unsigned long)sctp_port_hashtable
,
1554 get_order(sctp_port_hashsize
*
1555 sizeof(struct sctp_bind_hashbucket
)));
1557 sctp_transport_hashtable_destroy();
1559 kfree(sctp_ep_hashtable
);
1561 percpu_counter_destroy(&sctp_sockets_allocated
);
1562 err_percpu_counter_init
:
1563 kmem_cache_destroy(sctp_chunk_cachep
);
1565 kmem_cache_destroy(sctp_bucket_cachep
);
1569 /* Exit handler for the SCTP protocol. */
1570 static __exit
void sctp_exit(void)
1572 /* BUG. This should probably do something useful like clean
1573 * up all the remaining associations and all that memory.
1576 /* Unregister with inet6/inet layers. */
1577 sctp_v6_del_protocol();
1578 sctp_v4_del_protocol();
1580 unregister_pernet_subsys(&sctp_ctrlsock_ops
);
1582 /* Free protosw registrations */
1583 sctp_v6_protosw_exit();
1584 sctp_v4_protosw_exit();
1586 unregister_pernet_subsys(&sctp_defaults_ops
);
1588 /* Unregister with socket layer. */
1592 sctp_sysctl_unregister();
1594 free_pages((unsigned long)sctp_port_hashtable
,
1595 get_order(sctp_port_hashsize
*
1596 sizeof(struct sctp_bind_hashbucket
)));
1597 kfree(sctp_ep_hashtable
);
1598 sctp_transport_hashtable_destroy();
1600 percpu_counter_destroy(&sctp_sockets_allocated
);
1602 rcu_barrier(); /* Wait for completion of call_rcu()'s */
1604 kmem_cache_destroy(sctp_chunk_cachep
);
1605 kmem_cache_destroy(sctp_bucket_cachep
);
1608 module_init(sctp_init
);
1609 module_exit(sctp_exit
);
1612 * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1614 MODULE_ALIAS("net-pf-" __stringify(PF_INET
) "-proto-132");
1615 MODULE_ALIAS("net-pf-" __stringify(PF_INET6
) "-proto-132");
1616 MODULE_AUTHOR("Linux Kernel SCTP developers <linux-sctp@vger.kernel.org>");
1617 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1618 module_param_named(no_checksums
, sctp_checksum_disable
, bool, 0644);
1619 MODULE_PARM_DESC(no_checksums
, "Disable checksums computing and verification");
1620 MODULE_LICENSE("GPL");