~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

TOMOYO Linux Cross Reference
Linux/net/sctp/protocol.c

Version: ~ [ linux-5.13-rc1 ] ~ [ linux-5.12.2 ] ~ [ linux-5.11.19 ] ~ [ linux-5.10.35 ] ~ [ linux-5.9.16 ] ~ [ linux-5.8.18 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.117 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.190 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.232 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.268 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.268 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.18.140 ] ~ [ linux-3.16.85 ] ~ [ linux-3.14.79 ] ~ [ linux-3.12.74 ] ~ [ linux-3.10.108 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.5 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  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
  8  *
  9  * This file is part of the SCTP kernel implementation
 10  *
 11  * Initialization/cleanup for SCTP protocol support.
 12  *
 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)
 17  * any later version.
 18  *
 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.
 24  *
 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/>.
 28  *
 29  * Please send any bug reports or fixes you make to the
 30  * email address(es):
 31  *    lksctp developers <linux-sctp@vger.kernel.org>
 32  *
 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>
 40  */
 41 
 42 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
 43 
 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/bootmem.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>
 55 #include <net/ip.h>
 56 #include <net/ipv6.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>
 62 
 63 #define MAX_SCTP_PORT_HASH_ENTRIES (64 * 1024)
 64 
 65 /* Global data structures. */
 66 struct sctp_globals sctp_globals __read_mostly;
 67 
 68 struct idr sctp_assocs_id;
 69 DEFINE_SPINLOCK(sctp_assocs_id_lock);
 70 
 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;
 75 
 76 struct kmem_cache *sctp_chunk_cachep __read_mostly;
 77 struct kmem_cache *sctp_bucket_cachep __read_mostly;
 78 
 79 long sysctl_sctp_mem[3];
 80 int sysctl_sctp_rmem[3];
 81 int sysctl_sctp_wmem[3];
 82 
 83 /* Set up the proc fs entry for the SCTP protocol. */
 84 static int __net_init sctp_proc_init(struct net *net)
 85 {
 86 #ifdef CONFIG_PROC_FS
 87         net->sctp.proc_net_sctp = proc_net_mkdir(net, "sctp", net->proc_net);
 88         if (!net->sctp.proc_net_sctp)
 89                 goto out_proc_net_sctp;
 90         if (sctp_snmp_proc_init(net))
 91                 goto out_snmp_proc_init;
 92         if (sctp_eps_proc_init(net))
 93                 goto out_eps_proc_init;
 94         if (sctp_assocs_proc_init(net))
 95                 goto out_assocs_proc_init;
 96         if (sctp_remaddr_proc_init(net))
 97                 goto out_remaddr_proc_init;
 98 
 99         return 0;
100 
101 out_remaddr_proc_init:
102         sctp_assocs_proc_exit(net);
103 out_assocs_proc_init:
104         sctp_eps_proc_exit(net);
105 out_eps_proc_init:
106         sctp_snmp_proc_exit(net);
107 out_snmp_proc_init:
108         remove_proc_entry("sctp", net->proc_net);
109         net->sctp.proc_net_sctp = NULL;
110 out_proc_net_sctp:
111         return -ENOMEM;
112 #endif /* CONFIG_PROC_FS */
113         return 0;
114 }
115 
116 /* Clean up the proc fs entry for the SCTP protocol.
117  * Note: Do not make this __exit as it is used in the init error
118  * path.
119  */
120 static void sctp_proc_exit(struct net *net)
121 {
122 #ifdef CONFIG_PROC_FS
123         sctp_snmp_proc_exit(net);
124         sctp_eps_proc_exit(net);
125         sctp_assocs_proc_exit(net);
126         sctp_remaddr_proc_exit(net);
127 
128         remove_proc_entry("sctp", net->proc_net);
129         net->sctp.proc_net_sctp = NULL;
130 #endif
131 }
132 
133 /* Private helper to extract ipv4 address and stash them in
134  * the protocol structure.
135  */
136 static void sctp_v4_copy_addrlist(struct list_head *addrlist,
137                                   struct net_device *dev)
138 {
139         struct in_device *in_dev;
140         struct in_ifaddr *ifa;
141         struct sctp_sockaddr_entry *addr;
142 
143         rcu_read_lock();
144         if ((in_dev = __in_dev_get_rcu(dev)) == NULL) {
145                 rcu_read_unlock();
146                 return;
147         }
148 
149         for (ifa = in_dev->ifa_list; ifa; ifa = ifa->ifa_next) {
150                 /* Add the address to the local list.  */
151                 addr = kzalloc(sizeof(*addr), GFP_ATOMIC);
152                 if (addr) {
153                         addr->a.v4.sin_family = AF_INET;
154                         addr->a.v4.sin_port = 0;
155                         addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
156                         addr->valid = 1;
157                         INIT_LIST_HEAD(&addr->list);
158                         list_add_tail(&addr->list, addrlist);
159                 }
160         }
161 
162         rcu_read_unlock();
163 }
164 
165 /* Extract our IP addresses from the system and stash them in the
166  * protocol structure.
167  */
168 static void sctp_get_local_addr_list(struct net *net)
169 {
170         struct net_device *dev;
171         struct list_head *pos;
172         struct sctp_af *af;
173 
174         rcu_read_lock();
175         for_each_netdev_rcu(net, dev) {
176                 list_for_each(pos, &sctp_address_families) {
177                         af = list_entry(pos, struct sctp_af, list);
178                         af->copy_addrlist(&net->sctp.local_addr_list, dev);
179                 }
180         }
181         rcu_read_unlock();
182 }
183 
184 /* Free the existing local addresses.  */
185 static void sctp_free_local_addr_list(struct net *net)
186 {
187         struct sctp_sockaddr_entry *addr;
188         struct list_head *pos, *temp;
189 
190         list_for_each_safe(pos, temp, &net->sctp.local_addr_list) {
191                 addr = list_entry(pos, struct sctp_sockaddr_entry, list);
192                 list_del(pos);
193                 kfree(addr);
194         }
195 }
196 
197 /* Copy the local addresses which are valid for 'scope' into 'bp'.  */
198 int sctp_copy_local_addr_list(struct net *net, struct sctp_bind_addr *bp,
199                               enum sctp_scope scope, gfp_t gfp, int copy_flags)
200 {
201         struct sctp_sockaddr_entry *addr;
202         union sctp_addr laddr;
203         int error = 0;
204 
205         rcu_read_lock();
206         list_for_each_entry_rcu(addr, &net->sctp.local_addr_list, list) {
207                 if (!addr->valid)
208                         continue;
209                 if (!sctp_in_scope(net, &addr->a, scope))
210                         continue;
211 
212                 /* Now that the address is in scope, check to see if
213                  * the address type is really supported by the local
214                  * sock as well as the remote peer.
215                  */
216                 if (addr->a.sa.sa_family == AF_INET &&
217                     !(copy_flags & SCTP_ADDR4_PEERSUPP))
218                         continue;
219                 if (addr->a.sa.sa_family == AF_INET6 &&
220                     (!(copy_flags & SCTP_ADDR6_ALLOWED) ||
221                      !(copy_flags & SCTP_ADDR6_PEERSUPP)))
222                         continue;
223 
224                 laddr = addr->a;
225                 /* also works for setting ipv6 address port */
226                 laddr.v4.sin_port = htons(bp->port);
227                 if (sctp_bind_addr_state(bp, &laddr) != -1)
228                         continue;
229 
230                 error = sctp_add_bind_addr(bp, &addr->a, sizeof(addr->a),
231                                            SCTP_ADDR_SRC, GFP_ATOMIC);
232                 if (error)
233                         break;
234         }
235 
236         rcu_read_unlock();
237         return error;
238 }
239 
240 /* Initialize a sctp_addr from in incoming skb.  */
241 static void sctp_v4_from_skb(union sctp_addr *addr, struct sk_buff *skb,
242                              int is_saddr)
243 {
244         /* Always called on head skb, so this is safe */
245         struct sctphdr *sh = sctp_hdr(skb);
246         struct sockaddr_in *sa = &addr->v4;
247 
248         addr->v4.sin_family = AF_INET;
249 
250         if (is_saddr) {
251                 sa->sin_port = sh->source;
252                 sa->sin_addr.s_addr = ip_hdr(skb)->saddr;
253         } else {
254                 sa->sin_port = sh->dest;
255                 sa->sin_addr.s_addr = ip_hdr(skb)->daddr;
256         }
257 }
258 
259 /* Initialize an sctp_addr from a socket. */
260 static void sctp_v4_from_sk(union sctp_addr *addr, struct sock *sk)
261 {
262         addr->v4.sin_family = AF_INET;
263         addr->v4.sin_port = 0;
264         addr->v4.sin_addr.s_addr = inet_sk(sk)->inet_rcv_saddr;
265 }
266 
267 /* Initialize sk->sk_rcv_saddr from sctp_addr. */
268 static void sctp_v4_to_sk_saddr(union sctp_addr *addr, struct sock *sk)
269 {
270         inet_sk(sk)->inet_rcv_saddr = addr->v4.sin_addr.s_addr;
271 }
272 
273 /* Initialize sk->sk_daddr from sctp_addr. */
274 static void sctp_v4_to_sk_daddr(union sctp_addr *addr, struct sock *sk)
275 {
276         inet_sk(sk)->inet_daddr = addr->v4.sin_addr.s_addr;
277 }
278 
279 /* Initialize a sctp_addr from an address parameter. */
280 static void sctp_v4_from_addr_param(union sctp_addr *addr,
281                                     union sctp_addr_param *param,
282                                     __be16 port, int iif)
283 {
284         addr->v4.sin_family = AF_INET;
285         addr->v4.sin_port = port;
286         addr->v4.sin_addr.s_addr = param->v4.addr.s_addr;
287 }
288 
289 /* Initialize an address parameter from a sctp_addr and return the length
290  * of the address parameter.
291  */
292 static int sctp_v4_to_addr_param(const union sctp_addr *addr,
293                                  union sctp_addr_param *param)
294 {
295         int length = sizeof(struct sctp_ipv4addr_param);
296 
297         param->v4.param_hdr.type = SCTP_PARAM_IPV4_ADDRESS;
298         param->v4.param_hdr.length = htons(length);
299         param->v4.addr.s_addr = addr->v4.sin_addr.s_addr;
300 
301         return length;
302 }
303 
304 /* Initialize a sctp_addr from a dst_entry. */
305 static void sctp_v4_dst_saddr(union sctp_addr *saddr, struct flowi4 *fl4,
306                               __be16 port)
307 {
308         saddr->v4.sin_family = AF_INET;
309         saddr->v4.sin_port = port;
310         saddr->v4.sin_addr.s_addr = fl4->saddr;
311 }
312 
313 /* Compare two addresses exactly. */
314 static int sctp_v4_cmp_addr(const union sctp_addr *addr1,
315                             const union sctp_addr *addr2)
316 {
317         if (addr1->sa.sa_family != addr2->sa.sa_family)
318                 return 0;
319         if (addr1->v4.sin_port != addr2->v4.sin_port)
320                 return 0;
321         if (addr1->v4.sin_addr.s_addr != addr2->v4.sin_addr.s_addr)
322                 return 0;
323 
324         return 1;
325 }
326 
327 /* Initialize addr struct to INADDR_ANY. */
328 static void sctp_v4_inaddr_any(union sctp_addr *addr, __be16 port)
329 {
330         addr->v4.sin_family = AF_INET;
331         addr->v4.sin_addr.s_addr = htonl(INADDR_ANY);
332         addr->v4.sin_port = port;
333 }
334 
335 /* Is this a wildcard address? */
336 static int sctp_v4_is_any(const union sctp_addr *addr)
337 {
338         return htonl(INADDR_ANY) == addr->v4.sin_addr.s_addr;
339 }
340 
341 /* This function checks if the address is a valid address to be used for
342  * SCTP binding.
343  *
344  * Output:
345  * Return 0 - If the address is a non-unicast or an illegal address.
346  * Return 1 - If the address is a unicast.
347  */
348 static int sctp_v4_addr_valid(union sctp_addr *addr,
349                               struct sctp_sock *sp,
350                               const struct sk_buff *skb)
351 {
352         /* IPv4 addresses not allowed */
353         if (sp && ipv6_only_sock(sctp_opt2sk(sp)))
354                 return 0;
355 
356         /* Is this a non-unicast address or a unusable SCTP address? */
357         if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr))
358                 return 0;
359 
360         /* Is this a broadcast address? */
361         if (skb && skb_rtable(skb)->rt_flags & RTCF_BROADCAST)
362                 return 0;
363 
364         return 1;
365 }
366 
367 /* Should this be available for binding?   */
368 static int sctp_v4_available(union sctp_addr *addr, struct sctp_sock *sp)
369 {
370         struct net *net = sock_net(&sp->inet.sk);
371         int ret = inet_addr_type(net, addr->v4.sin_addr.s_addr);
372 
373 
374         if (addr->v4.sin_addr.s_addr != htonl(INADDR_ANY) &&
375            ret != RTN_LOCAL &&
376            !sp->inet.freebind &&
377            !net->ipv4.sysctl_ip_nonlocal_bind)
378                 return 0;
379 
380         if (ipv6_only_sock(sctp_opt2sk(sp)))
381                 return 0;
382 
383         return 1;
384 }
385 
386 /* Checking the loopback, private and other address scopes as defined in
387  * RFC 1918.   The IPv4 scoping is based on the draft for SCTP IPv4
388  * scoping <draft-stewart-tsvwg-sctp-ipv4-00.txt>.
389  *
390  * Level 0 - unusable SCTP addresses
391  * Level 1 - loopback address
392  * Level 2 - link-local addresses
393  * Level 3 - private addresses.
394  * Level 4 - global addresses
395  * For INIT and INIT-ACK address list, let L be the level of
396  * of requested destination address, sender and receiver
397  * SHOULD include all of its addresses with level greater
398  * than or equal to L.
399  *
400  * IPv4 scoping can be controlled through sysctl option
401  * net.sctp.addr_scope_policy
402  */
403 static enum sctp_scope sctp_v4_scope(union sctp_addr *addr)
404 {
405         enum sctp_scope retval;
406 
407         /* Check for unusable SCTP addresses. */
408         if (IS_IPV4_UNUSABLE_ADDRESS(addr->v4.sin_addr.s_addr)) {
409                 retval =  SCTP_SCOPE_UNUSABLE;
410         } else if (ipv4_is_loopback(addr->v4.sin_addr.s_addr)) {
411                 retval = SCTP_SCOPE_LOOPBACK;
412         } else if (ipv4_is_linklocal_169(addr->v4.sin_addr.s_addr)) {
413                 retval = SCTP_SCOPE_LINK;
414         } else if (ipv4_is_private_10(addr->v4.sin_addr.s_addr) ||
415                    ipv4_is_private_172(addr->v4.sin_addr.s_addr) ||
416                    ipv4_is_private_192(addr->v4.sin_addr.s_addr)) {
417                 retval = SCTP_SCOPE_PRIVATE;
418         } else {
419                 retval = SCTP_SCOPE_GLOBAL;
420         }
421 
422         return retval;
423 }
424 
425 /* Returns a valid dst cache entry for the given source and destination ip
426  * addresses. If an association is passed, trys to get a dst entry with a
427  * source address that matches an address in the bind address list.
428  */
429 static void sctp_v4_get_dst(struct sctp_transport *t, union sctp_addr *saddr,
430                                 struct flowi *fl, struct sock *sk)
431 {
432         struct sctp_association *asoc = t->asoc;
433         struct rtable *rt;
434         struct flowi4 *fl4 = &fl->u.ip4;
435         struct sctp_bind_addr *bp;
436         struct sctp_sockaddr_entry *laddr;
437         struct dst_entry *dst = NULL;
438         union sctp_addr *daddr = &t->ipaddr;
439         union sctp_addr dst_saddr;
440 
441         memset(fl4, 0x0, sizeof(struct flowi4));
442         fl4->daddr  = daddr->v4.sin_addr.s_addr;
443         fl4->fl4_dport = daddr->v4.sin_port;
444         fl4->flowi4_proto = IPPROTO_SCTP;
445         if (asoc) {
446                 fl4->flowi4_tos = RT_CONN_FLAGS(asoc->base.sk);
447                 fl4->flowi4_oif = asoc->base.sk->sk_bound_dev_if;
448                 fl4->fl4_sport = htons(asoc->base.bind_addr.port);
449         }
450         if (saddr) {
451                 fl4->saddr = saddr->v4.sin_addr.s_addr;
452                 fl4->fl4_sport = saddr->v4.sin_port;
453         }
454 
455         pr_debug("%s: dst:%pI4, src:%pI4 - ", __func__, &fl4->daddr,
456                  &fl4->saddr);
457 
458         rt = ip_route_output_key(sock_net(sk), fl4);
459         if (!IS_ERR(rt))
460                 dst = &rt->dst;
461 
462         /* If there is no association or if a source address is passed, no
463          * more validation is required.
464          */
465         if (!asoc || saddr)
466                 goto out;
467 
468         bp = &asoc->base.bind_addr;
469 
470         if (dst) {
471                 /* Walk through the bind address list and look for a bind
472                  * address that matches the source address of the returned dst.
473                  */
474                 sctp_v4_dst_saddr(&dst_saddr, fl4, htons(bp->port));
475                 rcu_read_lock();
476                 list_for_each_entry_rcu(laddr, &bp->address_list, list) {
477                         if (!laddr->valid || (laddr->state == SCTP_ADDR_DEL) ||
478                             (laddr->state != SCTP_ADDR_SRC &&
479                             !asoc->src_out_of_asoc_ok))
480                                 continue;
481                         if (sctp_v4_cmp_addr(&dst_saddr, &laddr->a))
482                                 goto out_unlock;
483                 }
484                 rcu_read_unlock();
485 
486                 /* None of the bound addresses match the source address of the
487                  * dst. So release it.
488                  */
489                 dst_release(dst);
490                 dst = NULL;
491         }
492 
493         /* Walk through the bind address list and try to get a dst that
494          * matches a bind address as the source address.
495          */
496         rcu_read_lock();
497         list_for_each_entry_rcu(laddr, &bp->address_list, list) {
498                 struct net_device *odev;
499 
500                 if (!laddr->valid)
501                         continue;
502                 if (laddr->state != SCTP_ADDR_SRC ||
503                     AF_INET != laddr->a.sa.sa_family)
504                         continue;
505 
506                 fl4->fl4_sport = laddr->a.v4.sin_port;
507                 flowi4_update_output(fl4,
508                                      asoc->base.sk->sk_bound_dev_if,
509                                      RT_CONN_FLAGS(asoc->base.sk),
510                                      daddr->v4.sin_addr.s_addr,
511                                      laddr->a.v4.sin_addr.s_addr);
512 
513                 rt = ip_route_output_key(sock_net(sk), fl4);
514                 if (IS_ERR(rt))
515                         continue;
516 
517                 /* Ensure the src address belongs to the output
518                  * interface.
519                  */
520                 odev = __ip_dev_find(sock_net(sk), laddr->a.v4.sin_addr.s_addr,
521                                      false);
522                 if (!odev || odev->ifindex != fl4->flowi4_oif) {
523                         if (!dst)
524                                 dst = &rt->dst;
525                         else
526                                 dst_release(&rt->dst);
527                         continue;
528                 }
529 
530                 dst_release(dst);
531                 dst = &rt->dst;
532                 break;
533         }
534 
535 out_unlock:
536         rcu_read_unlock();
537 out:
538         t->dst = dst;
539         if (dst)
540                 pr_debug("rt_dst:%pI4, rt_src:%pI4\n",
541                          &fl4->daddr, &fl4->saddr);
542         else
543                 pr_debug("no route\n");
544 }
545 
546 /* For v4, the source address is cached in the route entry(dst). So no need
547  * to cache it separately and hence this is an empty routine.
548  */
549 static void sctp_v4_get_saddr(struct sctp_sock *sk,
550                               struct sctp_transport *t,
551                               struct flowi *fl)
552 {
553         union sctp_addr *saddr = &t->saddr;
554         struct rtable *rt = (struct rtable *)t->dst;
555 
556         if (rt) {
557                 saddr->v4.sin_family = AF_INET;
558                 saddr->v4.sin_addr.s_addr = fl->u.ip4.saddr;
559         }
560 }
561 
562 /* What interface did this skb arrive on? */
563 static int sctp_v4_skb_iif(const struct sk_buff *skb)
564 {
565         return inet_iif(skb);
566 }
567 
568 /* Was this packet marked by Explicit Congestion Notification? */
569 static int sctp_v4_is_ce(const struct sk_buff *skb)
570 {
571         return INET_ECN_is_ce(ip_hdr(skb)->tos);
572 }
573 
574 /* Create and initialize a new sk for the socket returned by accept(). */
575 static struct sock *sctp_v4_create_accept_sk(struct sock *sk,
576                                              struct sctp_association *asoc,
577                                              bool kern)
578 {
579         struct sock *newsk = sk_alloc(sock_net(sk), PF_INET, GFP_KERNEL,
580                         sk->sk_prot, kern);
581         struct inet_sock *newinet;
582 
583         if (!newsk)
584                 goto out;
585 
586         sock_init_data(NULL, newsk);
587 
588         sctp_copy_sock(newsk, sk, asoc);
589         sock_reset_flag(newsk, SOCK_ZAPPED);
590 
591         newinet = inet_sk(newsk);
592 
593         newinet->inet_daddr = asoc->peer.primary_addr.v4.sin_addr.s_addr;
594 
595         sk_refcnt_debug_inc(newsk);
596 
597         if (newsk->sk_prot->init(newsk)) {
598                 sk_common_release(newsk);
599                 newsk = NULL;
600         }
601 
602 out:
603         return newsk;
604 }
605 
606 static int sctp_v4_addr_to_user(struct sctp_sock *sp, union sctp_addr *addr)
607 {
608         /* No address mapping for V4 sockets */
609         return sizeof(struct sockaddr_in);
610 }
611 
612 /* Dump the v4 addr to the seq file. */
613 static void sctp_v4_seq_dump_addr(struct seq_file *seq, union sctp_addr *addr)
614 {
615         seq_printf(seq, "%pI4 ", &addr->v4.sin_addr);
616 }
617 
618 static void sctp_v4_ecn_capable(struct sock *sk)
619 {
620         INET_ECN_xmit(sk);
621 }
622 
623 static void sctp_addr_wq_timeout_handler(struct timer_list *t)
624 {
625         struct net *net = from_timer(net, t, sctp.addr_wq_timer);
626         struct sctp_sockaddr_entry *addrw, *temp;
627         struct sctp_sock *sp;
628 
629         spin_lock_bh(&net->sctp.addr_wq_lock);
630 
631         list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) {
632                 pr_debug("%s: the first ent in wq:%p is addr:%pISc for cmd:%d at "
633                          "entry:%p\n", __func__, &net->sctp.addr_waitq, &addrw->a.sa,
634                          addrw->state, addrw);
635 
636 #if IS_ENABLED(CONFIG_IPV6)
637                 /* Now we send an ASCONF for each association */
638                 /* Note. we currently don't handle link local IPv6 addressees */
639                 if (addrw->a.sa.sa_family == AF_INET6) {
640                         struct in6_addr *in6;
641 
642                         if (ipv6_addr_type(&addrw->a.v6.sin6_addr) &
643                             IPV6_ADDR_LINKLOCAL)
644                                 goto free_next;
645 
646                         in6 = (struct in6_addr *)&addrw->a.v6.sin6_addr;
647                         if (ipv6_chk_addr(net, in6, NULL, 0) == 0 &&
648                             addrw->state == SCTP_ADDR_NEW) {
649                                 unsigned long timeo_val;
650 
651                                 pr_debug("%s: this is on DAD, trying %d sec "
652                                          "later\n", __func__,
653                                          SCTP_ADDRESS_TICK_DELAY);
654 
655                                 timeo_val = jiffies;
656                                 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
657                                 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
658                                 break;
659                         }
660                 }
661 #endif
662                 list_for_each_entry(sp, &net->sctp.auto_asconf_splist, auto_asconf_list) {
663                         struct sock *sk;
664 
665                         sk = sctp_opt2sk(sp);
666                         /* ignore bound-specific endpoints */
667                         if (!sctp_is_ep_boundall(sk))
668                                 continue;
669                         bh_lock_sock(sk);
670                         if (sctp_asconf_mgmt(sp, addrw) < 0)
671                                 pr_debug("%s: sctp_asconf_mgmt failed\n", __func__);
672                         bh_unlock_sock(sk);
673                 }
674 #if IS_ENABLED(CONFIG_IPV6)
675 free_next:
676 #endif
677                 list_del(&addrw->list);
678                 kfree(addrw);
679         }
680         spin_unlock_bh(&net->sctp.addr_wq_lock);
681 }
682 
683 static void sctp_free_addr_wq(struct net *net)
684 {
685         struct sctp_sockaddr_entry *addrw;
686         struct sctp_sockaddr_entry *temp;
687 
688         spin_lock_bh(&net->sctp.addr_wq_lock);
689         del_timer(&net->sctp.addr_wq_timer);
690         list_for_each_entry_safe(addrw, temp, &net->sctp.addr_waitq, list) {
691                 list_del(&addrw->list);
692                 kfree(addrw);
693         }
694         spin_unlock_bh(&net->sctp.addr_wq_lock);
695 }
696 
697 /* lookup the entry for the same address in the addr_waitq
698  * sctp_addr_wq MUST be locked
699  */
700 static struct sctp_sockaddr_entry *sctp_addr_wq_lookup(struct net *net,
701                                         struct sctp_sockaddr_entry *addr)
702 {
703         struct sctp_sockaddr_entry *addrw;
704 
705         list_for_each_entry(addrw, &net->sctp.addr_waitq, list) {
706                 if (addrw->a.sa.sa_family != addr->a.sa.sa_family)
707                         continue;
708                 if (addrw->a.sa.sa_family == AF_INET) {
709                         if (addrw->a.v4.sin_addr.s_addr ==
710                             addr->a.v4.sin_addr.s_addr)
711                                 return addrw;
712                 } else if (addrw->a.sa.sa_family == AF_INET6) {
713                         if (ipv6_addr_equal(&addrw->a.v6.sin6_addr,
714                             &addr->a.v6.sin6_addr))
715                                 return addrw;
716                 }
717         }
718         return NULL;
719 }
720 
721 void sctp_addr_wq_mgmt(struct net *net, struct sctp_sockaddr_entry *addr, int cmd)
722 {
723         struct sctp_sockaddr_entry *addrw;
724         unsigned long timeo_val;
725 
726         /* first, we check if an opposite message already exist in the queue.
727          * If we found such message, it is removed.
728          * This operation is a bit stupid, but the DHCP client attaches the
729          * new address after a couple of addition and deletion of that address
730          */
731 
732         spin_lock_bh(&net->sctp.addr_wq_lock);
733         /* Offsets existing events in addr_wq */
734         addrw = sctp_addr_wq_lookup(net, addr);
735         if (addrw) {
736                 if (addrw->state != cmd) {
737                         pr_debug("%s: offsets existing entry for %d, addr:%pISc "
738                                  "in wq:%p\n", __func__, addrw->state, &addrw->a.sa,
739                                  &net->sctp.addr_waitq);
740 
741                         list_del(&addrw->list);
742                         kfree(addrw);
743                 }
744                 spin_unlock_bh(&net->sctp.addr_wq_lock);
745                 return;
746         }
747 
748         /* OK, we have to add the new address to the wait queue */
749         addrw = kmemdup(addr, sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
750         if (addrw == NULL) {
751                 spin_unlock_bh(&net->sctp.addr_wq_lock);
752                 return;
753         }
754         addrw->state = cmd;
755         list_add_tail(&addrw->list, &net->sctp.addr_waitq);
756 
757         pr_debug("%s: add new entry for cmd:%d, addr:%pISc in wq:%p\n",
758                  __func__, addrw->state, &addrw->a.sa, &net->sctp.addr_waitq);
759 
760         if (!timer_pending(&net->sctp.addr_wq_timer)) {
761                 timeo_val = jiffies;
762                 timeo_val += msecs_to_jiffies(SCTP_ADDRESS_TICK_DELAY);
763                 mod_timer(&net->sctp.addr_wq_timer, timeo_val);
764         }
765         spin_unlock_bh(&net->sctp.addr_wq_lock);
766 }
767 
768 /* Event handler for inet address addition/deletion events.
769  * The sctp_local_addr_list needs to be protocted by a spin lock since
770  * multiple notifiers (say IPv4 and IPv6) may be running at the same
771  * time and thus corrupt the list.
772  * The reader side is protected with RCU.
773  */
774 static int sctp_inetaddr_event(struct notifier_block *this, unsigned long ev,
775                                void *ptr)
776 {
777         struct in_ifaddr *ifa = (struct in_ifaddr *)ptr;
778         struct sctp_sockaddr_entry *addr = NULL;
779         struct sctp_sockaddr_entry *temp;
780         struct net *net = dev_net(ifa->ifa_dev->dev);
781         int found = 0;
782 
783         switch (ev) {
784         case NETDEV_UP:
785                 addr = kmalloc(sizeof(struct sctp_sockaddr_entry), GFP_ATOMIC);
786                 if (addr) {
787                         addr->a.v4.sin_family = AF_INET;
788                         addr->a.v4.sin_port = 0;
789                         addr->a.v4.sin_addr.s_addr = ifa->ifa_local;
790                         addr->valid = 1;
791                         spin_lock_bh(&net->sctp.local_addr_lock);
792                         list_add_tail_rcu(&addr->list, &net->sctp.local_addr_list);
793                         sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_NEW);
794                         spin_unlock_bh(&net->sctp.local_addr_lock);
795                 }
796                 break;
797         case NETDEV_DOWN:
798                 spin_lock_bh(&net->sctp.local_addr_lock);
799                 list_for_each_entry_safe(addr, temp,
800                                         &net->sctp.local_addr_list, list) {
801                         if (addr->a.sa.sa_family == AF_INET &&
802                                         addr->a.v4.sin_addr.s_addr ==
803                                         ifa->ifa_local) {
804                                 sctp_addr_wq_mgmt(net, addr, SCTP_ADDR_DEL);
805                                 found = 1;
806                                 addr->valid = 0;
807                                 list_del_rcu(&addr->list);
808                                 break;
809                         }
810                 }
811                 spin_unlock_bh(&net->sctp.local_addr_lock);
812                 if (found)
813                         kfree_rcu(addr, rcu);
814                 break;
815         }
816 
817         return NOTIFY_DONE;
818 }
819 
820 /*
821  * Initialize the control inode/socket with a control endpoint data
822  * structure.  This endpoint is reserved exclusively for the OOTB processing.
823  */
824 static int sctp_ctl_sock_init(struct net *net)
825 {
826         int err;
827         sa_family_t family = PF_INET;
828 
829         if (sctp_get_pf_specific(PF_INET6))
830                 family = PF_INET6;
831 
832         err = inet_ctl_sock_create(&net->sctp.ctl_sock, family,
833                                    SOCK_SEQPACKET, IPPROTO_SCTP, net);
834 
835         /* If IPv6 socket could not be created, try the IPv4 socket */
836         if (err < 0 && family == PF_INET6)
837                 err = inet_ctl_sock_create(&net->sctp.ctl_sock, AF_INET,
838                                            SOCK_SEQPACKET, IPPROTO_SCTP,
839                                            net);
840 
841         if (err < 0) {
842                 pr_err("Failed to create the SCTP control socket\n");
843                 return err;
844         }
845         return 0;
846 }
847 
848 /* Register address family specific functions. */
849 int sctp_register_af(struct sctp_af *af)
850 {
851         switch (af->sa_family) {
852         case AF_INET:
853                 if (sctp_af_v4_specific)
854                         return 0;
855                 sctp_af_v4_specific = af;
856                 break;
857         case AF_INET6:
858                 if (sctp_af_v6_specific)
859                         return 0;
860                 sctp_af_v6_specific = af;
861                 break;
862         default:
863                 return 0;
864         }
865 
866         INIT_LIST_HEAD(&af->list);
867         list_add_tail(&af->list, &sctp_address_families);
868         return 1;
869 }
870 
871 /* Get the table of functions for manipulating a particular address
872  * family.
873  */
874 struct sctp_af *sctp_get_af_specific(sa_family_t family)
875 {
876         switch (family) {
877         case AF_INET:
878                 return sctp_af_v4_specific;
879         case AF_INET6:
880                 return sctp_af_v6_specific;
881         default:
882                 return NULL;
883         }
884 }
885 
886 /* Common code to initialize a AF_INET msg_name. */
887 static void sctp_inet_msgname(char *msgname, int *addr_len)
888 {
889         struct sockaddr_in *sin;
890 
891         sin = (struct sockaddr_in *)msgname;
892         *addr_len = sizeof(struct sockaddr_in);
893         sin->sin_family = AF_INET;
894         memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
895 }
896 
897 /* Copy the primary address of the peer primary address as the msg_name. */
898 static void sctp_inet_event_msgname(struct sctp_ulpevent *event, char *msgname,
899                                     int *addr_len)
900 {
901         struct sockaddr_in *sin, *sinfrom;
902 
903         if (msgname) {
904                 struct sctp_association *asoc;
905 
906                 asoc = event->asoc;
907                 sctp_inet_msgname(msgname, addr_len);
908                 sin = (struct sockaddr_in *)msgname;
909                 sinfrom = &asoc->peer.primary_addr.v4;
910                 sin->sin_port = htons(asoc->peer.port);
911                 sin->sin_addr.s_addr = sinfrom->sin_addr.s_addr;
912         }
913 }
914 
915 /* Initialize and copy out a msgname from an inbound skb. */
916 static void sctp_inet_skb_msgname(struct sk_buff *skb, char *msgname, int *len)
917 {
918         if (msgname) {
919                 struct sctphdr *sh = sctp_hdr(skb);
920                 struct sockaddr_in *sin = (struct sockaddr_in *)msgname;
921 
922                 sctp_inet_msgname(msgname, len);
923                 sin->sin_port = sh->source;
924                 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
925         }
926 }
927 
928 /* Do we support this AF? */
929 static int sctp_inet_af_supported(sa_family_t family, struct sctp_sock *sp)
930 {
931         /* PF_INET only supports AF_INET addresses. */
932         return AF_INET == family;
933 }
934 
935 /* Address matching with wildcards allowed. */
936 static int sctp_inet_cmp_addr(const union sctp_addr *addr1,
937                               const union sctp_addr *addr2,
938                               struct sctp_sock *opt)
939 {
940         /* PF_INET only supports AF_INET addresses. */
941         if (addr1->sa.sa_family != addr2->sa.sa_family)
942                 return 0;
943         if (htonl(INADDR_ANY) == addr1->v4.sin_addr.s_addr ||
944             htonl(INADDR_ANY) == addr2->v4.sin_addr.s_addr)
945                 return 1;
946         if (addr1->v4.sin_addr.s_addr == addr2->v4.sin_addr.s_addr)
947                 return 1;
948 
949         return 0;
950 }
951 
952 /* Verify that provided sockaddr looks bindable.  Common verification has
953  * already been taken care of.
954  */
955 static int sctp_inet_bind_verify(struct sctp_sock *opt, union sctp_addr *addr)
956 {
957         return sctp_v4_available(addr, opt);
958 }
959 
960 /* Verify that sockaddr looks sendable.  Common verification has already
961  * been taken care of.
962  */
963 static int sctp_inet_send_verify(struct sctp_sock *opt, union sctp_addr *addr)
964 {
965         return 1;
966 }
967 
968 /* Fill in Supported Address Type information for INIT and INIT-ACK
969  * chunks.  Returns number of addresses supported.
970  */
971 static int sctp_inet_supported_addrs(const struct sctp_sock *opt,
972                                      __be16 *types)
973 {
974         types[0] = SCTP_PARAM_IPV4_ADDRESS;
975         return 1;
976 }
977 
978 /* Wrapper routine that calls the ip transmit routine. */
979 static inline int sctp_v4_xmit(struct sk_buff *skb,
980                                struct sctp_transport *transport)
981 {
982         struct inet_sock *inet = inet_sk(skb->sk);
983 
984         pr_debug("%s: skb:%p, len:%d, src:%pI4, dst:%pI4\n", __func__, skb,
985                  skb->len, &transport->fl.u.ip4.saddr, &transport->fl.u.ip4.daddr);
986 
987         inet->pmtudisc = transport->param_flags & SPP_PMTUD_ENABLE ?
988                          IP_PMTUDISC_DO : IP_PMTUDISC_DONT;
989 
990         SCTP_INC_STATS(sock_net(&inet->sk), SCTP_MIB_OUTSCTPPACKS);
991 
992         return ip_queue_xmit(&inet->sk, skb, &transport->fl);
993 }
994 
995 static struct sctp_af sctp_af_inet;
996 
997 static struct sctp_pf sctp_pf_inet = {
998         .event_msgname = sctp_inet_event_msgname,
999         .skb_msgname   = sctp_inet_skb_msgname,
1000         .af_supported  = sctp_inet_af_supported,
1001         .cmp_addr      = sctp_inet_cmp_addr,
1002         .bind_verify   = sctp_inet_bind_verify,
1003         .send_verify   = sctp_inet_send_verify,
1004         .supported_addrs = sctp_inet_supported_addrs,
1005         .create_accept_sk = sctp_v4_create_accept_sk,
1006         .addr_to_user  = sctp_v4_addr_to_user,
1007         .to_sk_saddr   = sctp_v4_to_sk_saddr,
1008         .to_sk_daddr   = sctp_v4_to_sk_daddr,
1009         .af            = &sctp_af_inet
1010 };
1011 
1012 /* Notifier for inetaddr addition/deletion events.  */
1013 static struct notifier_block sctp_inetaddr_notifier = {
1014         .notifier_call = sctp_inetaddr_event,
1015 };
1016 
1017 /* Socket operations.  */
1018 static const struct proto_ops inet_seqpacket_ops = {
1019         .family            = PF_INET,
1020         .owner             = THIS_MODULE,
1021         .release           = inet_release,      /* Needs to be wrapped... */
1022         .bind              = inet_bind,
1023         .connect           = inet_dgram_connect,
1024         .socketpair        = sock_no_socketpair,
1025         .accept            = inet_accept,
1026         .getname           = inet_getname,      /* Semantics are different.  */
1027         .poll              = sctp_poll,
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,
1040 #endif
1041 };
1042 
1043 /* Registration with AF_INET family.  */
1044 static struct inet_protosw sctp_seqpacket_protosw = {
1045         .type       = SOCK_SEQPACKET,
1046         .protocol   = IPPROTO_SCTP,
1047         .prot       = &sctp_prot,
1048         .ops        = &inet_seqpacket_ops,
1049         .flags      = SCTP_PROTOSW_FLAG
1050 };
1051 static struct inet_protosw sctp_stream_protosw = {
1052         .type       = SOCK_STREAM,
1053         .protocol   = IPPROTO_SCTP,
1054         .prot       = &sctp_prot,
1055         .ops        = &inet_seqpacket_ops,
1056         .flags      = SCTP_PROTOSW_FLAG
1057 };
1058 
1059 /* Register with IP layer.  */
1060 static const struct net_protocol sctp_protocol = {
1061         .handler     = sctp_rcv,
1062         .err_handler = sctp_v4_err,
1063         .no_policy   = 1,
1064         .netns_ok    = 1,
1065         .icmp_strict_tag_validation = 1,
1066 };
1067 
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 #ifdef CONFIG_COMPAT
1094         .compat_setsockopt = compat_ip_setsockopt,
1095         .compat_getsockopt = compat_ip_getsockopt,
1096 #endif
1097 };
1098 
1099 struct sctp_pf *sctp_get_pf_specific(sa_family_t family)
1100 {
1101         switch (family) {
1102         case PF_INET:
1103                 return sctp_pf_inet_specific;
1104         case PF_INET6:
1105                 return sctp_pf_inet6_specific;
1106         default:
1107                 return NULL;
1108         }
1109 }
1110 
1111 /* Register the PF specific function table.  */
1112 int sctp_register_pf(struct sctp_pf *pf, sa_family_t family)
1113 {
1114         switch (family) {
1115         case PF_INET:
1116                 if (sctp_pf_inet_specific)
1117                         return 0;
1118                 sctp_pf_inet_specific = pf;
1119                 break;
1120         case PF_INET6:
1121                 if (sctp_pf_inet6_specific)
1122                         return 0;
1123                 sctp_pf_inet6_specific = pf;
1124                 break;
1125         default:
1126                 return 0;
1127         }
1128         return 1;
1129 }
1130 
1131 static inline int init_sctp_mibs(struct net *net)
1132 {
1133         net->sctp.sctp_statistics = alloc_percpu(struct sctp_mib);
1134         if (!net->sctp.sctp_statistics)
1135                 return -ENOMEM;
1136         return 0;
1137 }
1138 
1139 static inline void cleanup_sctp_mibs(struct net *net)
1140 {
1141         free_percpu(net->sctp.sctp_statistics);
1142 }
1143 
1144 static void sctp_v4_pf_init(void)
1145 {
1146         /* Initialize the SCTP specific PF functions. */
1147         sctp_register_pf(&sctp_pf_inet, PF_INET);
1148         sctp_register_af(&sctp_af_inet);
1149 }
1150 
1151 static void sctp_v4_pf_exit(void)
1152 {
1153         list_del(&sctp_af_inet.list);
1154 }
1155 
1156 static int sctp_v4_protosw_init(void)
1157 {
1158         int rc;
1159 
1160         rc = proto_register(&sctp_prot, 1);
1161         if (rc)
1162                 return rc;
1163 
1164         /* Register SCTP(UDP and TCP style) with socket layer.  */
1165         inet_register_protosw(&sctp_seqpacket_protosw);
1166         inet_register_protosw(&sctp_stream_protosw);
1167 
1168         return 0;
1169 }
1170 
1171 static void sctp_v4_protosw_exit(void)
1172 {
1173         inet_unregister_protosw(&sctp_stream_protosw);
1174         inet_unregister_protosw(&sctp_seqpacket_protosw);
1175         proto_unregister(&sctp_prot);
1176 }
1177 
1178 static int sctp_v4_add_protocol(void)
1179 {
1180         /* Register notifier for inet address additions/deletions. */
1181         register_inetaddr_notifier(&sctp_inetaddr_notifier);
1182 
1183         /* Register SCTP with inet layer.  */
1184         if (inet_add_protocol(&sctp_protocol, IPPROTO_SCTP) < 0)
1185                 return -EAGAIN;
1186 
1187         return 0;
1188 }
1189 
1190 static void sctp_v4_del_protocol(void)
1191 {
1192         inet_del_protocol(&sctp_protocol, IPPROTO_SCTP);
1193         unregister_inetaddr_notifier(&sctp_inetaddr_notifier);
1194 }
1195 
1196 static int __net_init sctp_defaults_init(struct net *net)
1197 {
1198         int status;
1199 
1200         /*
1201          * 14. Suggested SCTP Protocol Parameter Values
1202          */
1203         /* The following protocol parameters are RECOMMENDED:  */
1204         /* RTO.Initial              - 3  seconds */
1205         net->sctp.rto_initial                   = SCTP_RTO_INITIAL;
1206         /* RTO.Min                  - 1  second */
1207         net->sctp.rto_min                       = SCTP_RTO_MIN;
1208         /* RTO.Max                 -  60 seconds */
1209         net->sctp.rto_max                       = SCTP_RTO_MAX;
1210         /* RTO.Alpha                - 1/8 */
1211         net->sctp.rto_alpha                     = SCTP_RTO_ALPHA;
1212         /* RTO.Beta                 - 1/4 */
1213         net->sctp.rto_beta                      = SCTP_RTO_BETA;
1214 
1215         /* Valid.Cookie.Life        - 60  seconds */
1216         net->sctp.valid_cookie_life             = SCTP_DEFAULT_COOKIE_LIFE;
1217 
1218         /* Whether Cookie Preservative is enabled(1) or not(0) */
1219         net->sctp.cookie_preserve_enable        = 1;
1220 
1221         /* Default sctp sockets to use md5 as their hmac alg */
1222 #if defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_MD5)
1223         net->sctp.sctp_hmac_alg                 = "md5";
1224 #elif defined (CONFIG_SCTP_DEFAULT_COOKIE_HMAC_SHA1)
1225         net->sctp.sctp_hmac_alg                 = "sha1";
1226 #else
1227         net->sctp.sctp_hmac_alg                 = NULL;
1228 #endif
1229 
1230         /* Max.Burst                - 4 */
1231         net->sctp.max_burst                     = SCTP_DEFAULT_MAX_BURST;
1232 
1233         /* Enable pf state by default */
1234         net->sctp.pf_enable = 1;
1235 
1236         /* Association.Max.Retrans  - 10 attempts
1237          * Path.Max.Retrans         - 5  attempts (per destination address)
1238          * Max.Init.Retransmits     - 8  attempts
1239          */
1240         net->sctp.max_retrans_association       = 10;
1241         net->sctp.max_retrans_path              = 5;
1242         net->sctp.max_retrans_init              = 8;
1243 
1244         /* Sendbuffer growth        - do per-socket accounting */
1245         net->sctp.sndbuf_policy                 = 0;
1246 
1247         /* Rcvbuffer growth         - do per-socket accounting */
1248         net->sctp.rcvbuf_policy                 = 0;
1249 
1250         /* HB.interval              - 30 seconds */
1251         net->sctp.hb_interval                   = SCTP_DEFAULT_TIMEOUT_HEARTBEAT;
1252 
1253         /* delayed SACK timeout */
1254         net->sctp.sack_timeout                  = SCTP_DEFAULT_TIMEOUT_SACK;
1255 
1256         /* Disable ADDIP by default. */
1257         net->sctp.addip_enable = 0;
1258         net->sctp.addip_noauth = 0;
1259         net->sctp.default_auto_asconf = 0;
1260 
1261         /* Enable PR-SCTP by default. */
1262         net->sctp.prsctp_enable = 1;
1263 
1264         /* Disable RECONF by default. */
1265         net->sctp.reconf_enable = 0;
1266 
1267         /* Disable AUTH by default. */
1268         net->sctp.auth_enable = 0;
1269 
1270         /* Set SCOPE policy to enabled */
1271         net->sctp.scope_policy = SCTP_SCOPE_POLICY_ENABLE;
1272 
1273         /* Set the default rwnd update threshold */
1274         net->sctp.rwnd_upd_shift = SCTP_DEFAULT_RWND_SHIFT;
1275 
1276         /* Initialize maximum autoclose timeout. */
1277         net->sctp.max_autoclose         = INT_MAX / HZ;
1278 
1279         status = sctp_sysctl_net_register(net);
1280         if (status)
1281                 goto err_sysctl_register;
1282 
1283         /* Allocate and initialise sctp mibs.  */
1284         status = init_sctp_mibs(net);
1285         if (status)
1286                 goto err_init_mibs;
1287 
1288         /* Initialize proc fs directory.  */
1289         status = sctp_proc_init(net);
1290         if (status)
1291                 goto err_init_proc;
1292 
1293         sctp_dbg_objcnt_init(net);
1294 
1295         /* Initialize the local address list. */
1296         INIT_LIST_HEAD(&net->sctp.local_addr_list);
1297         spin_lock_init(&net->sctp.local_addr_lock);
1298         sctp_get_local_addr_list(net);
1299 
1300         /* Initialize the address event list */
1301         INIT_LIST_HEAD(&net->sctp.addr_waitq);
1302         INIT_LIST_HEAD(&net->sctp.auto_asconf_splist);
1303         spin_lock_init(&net->sctp.addr_wq_lock);
1304         net->sctp.addr_wq_timer.expires = 0;
1305         timer_setup(&net->sctp.addr_wq_timer, sctp_addr_wq_timeout_handler, 0);
1306 
1307         return 0;
1308 
1309 err_init_proc:
1310         cleanup_sctp_mibs(net);
1311 err_init_mibs:
1312         sctp_sysctl_net_unregister(net);
1313 err_sysctl_register:
1314         return status;
1315 }
1316 
1317 static void __net_exit sctp_defaults_exit(struct net *net)
1318 {
1319         /* Free the local address list */
1320         sctp_free_addr_wq(net);
1321         sctp_free_local_addr_list(net);
1322 
1323         sctp_dbg_objcnt_exit(net);
1324 
1325         sctp_proc_exit(net);
1326         cleanup_sctp_mibs(net);
1327         sctp_sysctl_net_unregister(net);
1328 }
1329 
1330 static struct pernet_operations sctp_defaults_ops = {
1331         .init = sctp_defaults_init,
1332         .exit = sctp_defaults_exit,
1333 };
1334 
1335 static int __net_init sctp_ctrlsock_init(struct net *net)
1336 {
1337         int status;
1338 
1339         /* Initialize the control inode/socket for handling OOTB packets.  */
1340         status = sctp_ctl_sock_init(net);
1341         if (status)
1342                 pr_err("Failed to initialize the SCTP control sock\n");
1343 
1344         return status;
1345 }
1346 
1347 static void __net_init sctp_ctrlsock_exit(struct net *net)
1348 {
1349         /* Free the control endpoint.  */
1350         inet_ctl_sock_destroy(net->sctp.ctl_sock);
1351 }
1352 
1353 static struct pernet_operations sctp_ctrlsock_ops = {
1354         .init = sctp_ctrlsock_init,
1355         .exit = sctp_ctrlsock_exit,
1356 };
1357 
1358 /* Initialize the universe into something sensible.  */
1359 static __init int sctp_init(void)
1360 {
1361         int i;
1362         int status = -EINVAL;
1363         unsigned long goal;
1364         unsigned long limit;
1365         int max_share;
1366         int order;
1367         int num_entries;
1368         int max_entry_order;
1369 
1370         sock_skb_cb_check_size(sizeof(struct sctp_ulpevent));
1371 
1372         /* Allocate bind_bucket and chunk caches. */
1373         status = -ENOBUFS;
1374         sctp_bucket_cachep = kmem_cache_create("sctp_bind_bucket",
1375                                                sizeof(struct sctp_bind_bucket),
1376                                                0, SLAB_HWCACHE_ALIGN,
1377                                                NULL);
1378         if (!sctp_bucket_cachep)
1379                 goto out;
1380 
1381         sctp_chunk_cachep = kmem_cache_create("sctp_chunk",
1382                                                sizeof(struct sctp_chunk),
1383                                                0, SLAB_HWCACHE_ALIGN,
1384                                                NULL);
1385         if (!sctp_chunk_cachep)
1386                 goto err_chunk_cachep;
1387 
1388         status = percpu_counter_init(&sctp_sockets_allocated, 0, GFP_KERNEL);
1389         if (status)
1390                 goto err_percpu_counter_init;
1391 
1392         /* Implementation specific variables. */
1393 
1394         /* Initialize default stream count setup information. */
1395         sctp_max_instreams              = SCTP_DEFAULT_INSTREAMS;
1396         sctp_max_outstreams             = SCTP_DEFAULT_OUTSTREAMS;
1397 
1398         /* Initialize handle used for association ids. */
1399         idr_init(&sctp_assocs_id);
1400 
1401         limit = nr_free_buffer_pages() / 8;
1402         limit = max(limit, 128UL);
1403         sysctl_sctp_mem[0] = limit / 4 * 3;
1404         sysctl_sctp_mem[1] = limit;
1405         sysctl_sctp_mem[2] = sysctl_sctp_mem[0] * 2;
1406 
1407         /* Set per-socket limits to no more than 1/128 the pressure threshold*/
1408         limit = (sysctl_sctp_mem[1]) << (PAGE_SHIFT - 7);
1409         max_share = min(4UL*1024*1024, limit);
1410 
1411         sysctl_sctp_rmem[0] = SK_MEM_QUANTUM; /* give each asoc 1 page min */
1412         sysctl_sctp_rmem[1] = 1500 * SKB_TRUESIZE(1);
1413         sysctl_sctp_rmem[2] = max(sysctl_sctp_rmem[1], max_share);
1414 
1415         sysctl_sctp_wmem[0] = SK_MEM_QUANTUM;
1416         sysctl_sctp_wmem[1] = 16*1024;
1417         sysctl_sctp_wmem[2] = max(64*1024, max_share);
1418 
1419         /* Size and allocate the association hash table.
1420          * The methodology is similar to that of the tcp hash tables.
1421          * Though not identical.  Start by getting a goal size
1422          */
1423         if (totalram_pages >= (128 * 1024))
1424                 goal = totalram_pages >> (22 - PAGE_SHIFT);
1425         else
1426                 goal = totalram_pages >> (24 - PAGE_SHIFT);
1427 
1428         /* Then compute the page order for said goal */
1429         order = get_order(goal);
1430 
1431         /* Now compute the required page order for the maximum sized table we
1432          * want to create
1433          */
1434         max_entry_order = get_order(MAX_SCTP_PORT_HASH_ENTRIES *
1435                                     sizeof(struct sctp_bind_hashbucket));
1436 
1437         /* Limit the page order by that maximum hash table size */
1438         order = min(order, max_entry_order);
1439 
1440         /* Allocate and initialize the endpoint hash table.  */
1441         sctp_ep_hashsize = 64;
1442         sctp_ep_hashtable =
1443                 kmalloc(64 * sizeof(struct sctp_hashbucket), GFP_KERNEL);
1444         if (!sctp_ep_hashtable) {
1445                 pr_err("Failed endpoint_hash alloc\n");
1446                 status = -ENOMEM;
1447                 goto err_ehash_alloc;
1448         }
1449         for (i = 0; i < sctp_ep_hashsize; i++) {
1450                 rwlock_init(&sctp_ep_hashtable[i].lock);
1451                 INIT_HLIST_HEAD(&sctp_ep_hashtable[i].chain);
1452         }
1453 
1454         /* Allocate and initialize the SCTP port hash table.
1455          * Note that order is initalized to start at the max sized
1456          * table we want to support.  If we can't get that many pages
1457          * reduce the order and try again
1458          */
1459         do {
1460                 sctp_port_hashtable = (struct sctp_bind_hashbucket *)
1461                         __get_free_pages(GFP_KERNEL | __GFP_NOWARN, order);
1462         } while (!sctp_port_hashtable && --order > 0);
1463 
1464         if (!sctp_port_hashtable) {
1465                 pr_err("Failed bind hash alloc\n");
1466                 status = -ENOMEM;
1467                 goto err_bhash_alloc;
1468         }
1469 
1470         /* Now compute the number of entries that will fit in the
1471          * port hash space we allocated
1472          */
1473         num_entries = (1UL << order) * PAGE_SIZE /
1474                       sizeof(struct sctp_bind_hashbucket);
1475 
1476         /* And finish by rounding it down to the nearest power of two
1477          * this wastes some memory of course, but its needed because
1478          * the hash function operates based on the assumption that
1479          * that the number of entries is a power of two
1480          */
1481         sctp_port_hashsize = rounddown_pow_of_two(num_entries);
1482 
1483         for (i = 0; i < sctp_port_hashsize; i++) {
1484                 spin_lock_init(&sctp_port_hashtable[i].lock);
1485                 INIT_HLIST_HEAD(&sctp_port_hashtable[i].chain);
1486         }
1487 
1488         status = sctp_transport_hashtable_init();
1489         if (status)
1490                 goto err_thash_alloc;
1491 
1492         pr_info("Hash tables configured (bind %d/%d)\n", sctp_port_hashsize,
1493                 num_entries);
1494 
1495         sctp_sysctl_register();
1496 
1497         INIT_LIST_HEAD(&sctp_address_families);
1498         sctp_v4_pf_init();
1499         sctp_v6_pf_init();
1500         sctp_sched_ops_init();
1501 
1502         status = register_pernet_subsys(&sctp_defaults_ops);
1503         if (status)
1504                 goto err_register_defaults;
1505 
1506         status = sctp_v4_protosw_init();
1507         if (status)
1508                 goto err_protosw_init;
1509 
1510         status = sctp_v6_protosw_init();
1511         if (status)
1512                 goto err_v6_protosw_init;
1513 
1514         status = register_pernet_subsys(&sctp_ctrlsock_ops);
1515         if (status)
1516                 goto err_register_ctrlsock;
1517 
1518         status = sctp_v4_add_protocol();
1519         if (status)
1520                 goto err_add_protocol;
1521 
1522         /* Register SCTP with inet6 layer.  */
1523         status = sctp_v6_add_protocol();
1524         if (status)
1525                 goto err_v6_add_protocol;
1526 
1527         if (sctp_offload_init() < 0)
1528                 pr_crit("%s: Cannot add SCTP protocol offload\n", __func__);
1529 
1530 out:
1531         return status;
1532 err_v6_add_protocol:
1533         sctp_v4_del_protocol();
1534 err_add_protocol:
1535         unregister_pernet_subsys(&sctp_ctrlsock_ops);
1536 err_register_ctrlsock:
1537         sctp_v6_protosw_exit();
1538 err_v6_protosw_init:
1539         sctp_v4_protosw_exit();
1540 err_protosw_init:
1541         unregister_pernet_subsys(&sctp_defaults_ops);
1542 err_register_defaults:
1543         sctp_v4_pf_exit();
1544         sctp_v6_pf_exit();
1545         sctp_sysctl_unregister();
1546         free_pages((unsigned long)sctp_port_hashtable,
1547                    get_order(sctp_port_hashsize *
1548                              sizeof(struct sctp_bind_hashbucket)));
1549 err_bhash_alloc:
1550         sctp_transport_hashtable_destroy();
1551 err_thash_alloc:
1552         kfree(sctp_ep_hashtable);
1553 err_ehash_alloc:
1554         percpu_counter_destroy(&sctp_sockets_allocated);
1555 err_percpu_counter_init:
1556         kmem_cache_destroy(sctp_chunk_cachep);
1557 err_chunk_cachep:
1558         kmem_cache_destroy(sctp_bucket_cachep);
1559         goto out;
1560 }
1561 
1562 /* Exit handler for the SCTP protocol.  */
1563 static __exit void sctp_exit(void)
1564 {
1565         /* BUG.  This should probably do something useful like clean
1566          * up all the remaining associations and all that memory.
1567          */
1568 
1569         /* Unregister with inet6/inet layers. */
1570         sctp_v6_del_protocol();
1571         sctp_v4_del_protocol();
1572 
1573         unregister_pernet_subsys(&sctp_ctrlsock_ops);
1574 
1575         /* Free protosw registrations */
1576         sctp_v6_protosw_exit();
1577         sctp_v4_protosw_exit();
1578 
1579         unregister_pernet_subsys(&sctp_defaults_ops);
1580 
1581         /* Unregister with socket layer. */
1582         sctp_v6_pf_exit();
1583         sctp_v4_pf_exit();
1584 
1585         sctp_sysctl_unregister();
1586 
1587         free_pages((unsigned long)sctp_port_hashtable,
1588                    get_order(sctp_port_hashsize *
1589                              sizeof(struct sctp_bind_hashbucket)));
1590         kfree(sctp_ep_hashtable);
1591         sctp_transport_hashtable_destroy();
1592 
1593         percpu_counter_destroy(&sctp_sockets_allocated);
1594 
1595         rcu_barrier(); /* Wait for completion of call_rcu()'s */
1596 
1597         kmem_cache_destroy(sctp_chunk_cachep);
1598         kmem_cache_destroy(sctp_bucket_cachep);
1599 }
1600 
1601 module_init(sctp_init);
1602 module_exit(sctp_exit);
1603 
1604 /*
1605  * __stringify doesn't likes enums, so use IPPROTO_SCTP value (132) directly.
1606  */
1607 MODULE_ALIAS("net-pf-" __stringify(PF_INET) "-proto-132");
1608 MODULE_ALIAS("net-pf-" __stringify(PF_INET6) "-proto-132");
1609 MODULE_AUTHOR("Linux Kernel SCTP developers <linux-sctp@vger.kernel.org>");
1610 MODULE_DESCRIPTION("Support for the SCTP protocol (RFC2960)");
1611 module_param_named(no_checksums, sctp_checksum_disable, bool, 0644);
1612 MODULE_PARM_DESC(no_checksums, "Disable checksums computing and verification");
1613 MODULE_LICENSE("GPL");
1614 

~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

kernel.org | git.kernel.org | LWN.net | Project Home | Wiki (Japanese) | Wiki (English) | SVN repository | Mail admin

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

osdn.jp