Commit | Line | Data |
---|---|---|
47505b8b | 1 | // SPDX-License-Identifier: GPL-2.0-or-later |
60c778b2 | 2 | /* SCTP kernel implementation |
1da177e4 LT |
3 | * Copyright (c) 1999-2000 Cisco, Inc. |
4 | * Copyright (c) 1999-2001 Motorola, Inc. | |
5 | * Copyright (c) 2001-2003 International Business Machines, Corp. | |
6 | * Copyright (c) 2001 Intel Corp. | |
7 | * Copyright (c) 2001 Nokia, Inc. | |
8 | * Copyright (c) 2001 La Monte H.P. Yarroll | |
9 | * | |
60c778b2 | 10 | * This file is part of the SCTP kernel implementation |
1da177e4 LT |
11 | * |
12 | * These functions handle all input from the IP layer into SCTP. | |
13 | * | |
1da177e4 LT |
14 | * Please send any bug reports or fixes you make to the |
15 | * email address(es): | |
91705c61 | 16 | * lksctp developers <linux-sctp@vger.kernel.org> |
1da177e4 | 17 | * |
1da177e4 LT |
18 | * Written or modified by: |
19 | * La Monte H.P. Yarroll <piggy@acm.org> | |
20 | * Karl Knutson <karl@athena.chicago.il.us> | |
21 | * Xingang Guo <xingang.guo@intel.com> | |
22 | * Jon Grimm <jgrimm@us.ibm.com> | |
23 | * Hui Huang <hui.huang@nokia.com> | |
24 | * Daisy Chang <daisyc@us.ibm.com> | |
25 | * Sridhar Samudrala <sri@us.ibm.com> | |
26 | * Ardelle Fan <ardelle.fan@intel.com> | |
1da177e4 LT |
27 | */ |
28 | ||
29 | #include <linux/types.h> | |
30 | #include <linux/list.h> /* For struct list_head */ | |
31 | #include <linux/socket.h> | |
32 | #include <linux/ip.h> | |
33 | #include <linux/time.h> /* For struct timeval */ | |
5a0e3ad6 | 34 | #include <linux/slab.h> |
1da177e4 LT |
35 | #include <net/ip.h> |
36 | #include <net/icmp.h> | |
37 | #include <net/snmp.h> | |
38 | #include <net/sock.h> | |
39 | #include <net/xfrm.h> | |
40 | #include <net/sctp/sctp.h> | |
41 | #include <net/sctp/sm.h> | |
9ad0977f | 42 | #include <net/sctp/checksum.h> |
dcfc23ca | 43 | #include <net/net_namespace.h> |
0eb71a9d | 44 | #include <linux/rhashtable.h> |
532ae2f1 | 45 | #include <net/sock_reuseport.h> |
1da177e4 LT |
46 | |
47 | /* Forward declarations for internal helpers. */ | |
48 | static int sctp_rcv_ootb(struct sk_buff *); | |
4110cc25 EB |
49 | static struct sctp_association *__sctp_rcv_lookup(struct net *net, |
50 | struct sk_buff *skb, | |
1da177e4 | 51 | const union sctp_addr *paddr, |
57565993 | 52 | const union sctp_addr *laddr, |
1da177e4 | 53 | struct sctp_transport **transportp); |
532ae2f1 XL |
54 | static struct sctp_endpoint *__sctp_rcv_lookup_endpoint( |
55 | struct net *net, struct sk_buff *skb, | |
56 | const union sctp_addr *laddr, | |
57 | const union sctp_addr *daddr); | |
1da177e4 | 58 | static struct sctp_association *__sctp_lookup_association( |
4110cc25 | 59 | struct net *net, |
1da177e4 LT |
60 | const union sctp_addr *local, |
61 | const union sctp_addr *peer, | |
62 | struct sctp_transport **pt); | |
63 | ||
50b1a782 | 64 | static int sctp_add_backlog(struct sock *sk, struct sk_buff *skb); |
61c9fed4 | 65 | |
1da177e4 LT |
66 | |
67 | /* Calculate the SCTP checksum of an SCTP packet. */ | |
b01a2407 | 68 | static inline int sctp_rcv_checksum(struct net *net, struct sk_buff *skb) |
1da177e4 | 69 | { |
2c0fd387 | 70 | struct sctphdr *sh = sctp_hdr(skb); |
4458f04c | 71 | __le32 cmp = sh->checksum; |
024ec3de | 72 | __le32 val = sctp_compute_cksum(skb, 0); |
1da177e4 LT |
73 | |
74 | if (val != cmp) { | |
75 | /* CRC failure, dump it. */ | |
08e3baef | 76 | __SCTP_INC_STATS(net, SCTP_MIB_CHECKSUMERRORS); |
1da177e4 LT |
77 | return -1; |
78 | } | |
79 | return 0; | |
80 | } | |
81 | ||
1da177e4 LT |
82 | /* |
83 | * This is the routine which IP calls when receiving an SCTP packet. | |
84 | */ | |
85 | int sctp_rcv(struct sk_buff *skb) | |
86 | { | |
87 | struct sock *sk; | |
88 | struct sctp_association *asoc; | |
89 | struct sctp_endpoint *ep = NULL; | |
90 | struct sctp_ep_common *rcvr; | |
91 | struct sctp_transport *transport = NULL; | |
92 | struct sctp_chunk *chunk; | |
1da177e4 LT |
93 | union sctp_addr src; |
94 | union sctp_addr dest; | |
a20ea298 | 95 | int bound_dev_if; |
1da177e4 LT |
96 | int family; |
97 | struct sctp_af *af; | |
4cdadcbc | 98 | struct net *net = dev_net(skb->dev); |
1dd27cde | 99 | bool is_gso = skb_is_gso(skb) && skb_is_gso_sctp(skb); |
1da177e4 | 100 | |
cb3f837b | 101 | if (skb->pkt_type != PACKET_HOST) |
1da177e4 LT |
102 | goto discard_it; |
103 | ||
08e3baef | 104 | __SCTP_INC_STATS(net, SCTP_MIB_INSCTPPACKS); |
1da177e4 | 105 | |
3acb50c1 MRL |
106 | /* If packet is too small to contain a single chunk, let's not |
107 | * waste time on it anymore. | |
108 | */ | |
109 | if (skb->len < sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr) + | |
110 | skb_transport_offset(skb)) | |
28cd7752 HX |
111 | goto discard_it; |
112 | ||
4c2f2454 MRL |
113 | /* If the packet is fragmented and we need to do crc checking, |
114 | * it's better to just linearize it otherwise crc computing | |
115 | * takes longer. | |
116 | */ | |
1dd27cde | 117 | if ((!is_gso && skb_linearize(skb)) || |
4c2f2454 | 118 | !pskb_may_pull(skb, sizeof(struct sctphdr))) |
3acb50c1 | 119 | goto discard_it; |
1da177e4 | 120 | |
3acb50c1 | 121 | /* Pull up the IP header. */ |
ea2ae17d | 122 | __skb_pull(skb, skb_transport_offset(skb)); |
202863fe TH |
123 | |
124 | skb->csum_valid = 0; /* Previous value not applicable */ | |
125 | if (skb_csum_unnecessary(skb)) | |
126 | __skb_decr_checksum_unnecessary(skb); | |
90017acc | 127 | else if (!sctp_checksum_disable && |
1dd27cde | 128 | !is_gso && |
90017acc | 129 | sctp_rcv_checksum(net, skb) < 0) |
1da177e4 | 130 | goto discard_it; |
202863fe | 131 | skb->csum_valid = 1; |
1da177e4 | 132 | |
3acb50c1 | 133 | __skb_pull(skb, sizeof(struct sctphdr)); |
1da177e4 | 134 | |
eddc9ec5 | 135 | family = ipver2af(ip_hdr(skb)->version); |
1da177e4 LT |
136 | af = sctp_get_af_specific(family); |
137 | if (unlikely(!af)) | |
138 | goto discard_it; | |
e7487c86 | 139 | SCTP_INPUT_CB(skb)->af = af; |
1da177e4 LT |
140 | |
141 | /* Initialize local addresses for lookups. */ | |
142 | af->from_skb(&src, skb, 1); | |
143 | af->from_skb(&dest, skb, 0); | |
144 | ||
145 | /* If the packet is to or from a non-unicast address, | |
146 | * silently discard the packet. | |
147 | * | |
148 | * This is not clearly defined in the RFC except in section | |
149 | * 8.4 - OOTB handling. However, based on the book "Stream Control | |
150 | * Transmission Protocol" 2.1, "It is important to note that the | |
151 | * IP address of an SCTP transport address must be a routable | |
152 | * unicast address. In other words, IP multicast addresses and | |
153 | * IP broadcast addresses cannot be used in an SCTP transport | |
154 | * address." | |
155 | */ | |
5636bef7 VY |
156 | if (!af->addr_valid(&src, NULL, skb) || |
157 | !af->addr_valid(&dest, NULL, skb)) | |
1da177e4 LT |
158 | goto discard_it; |
159 | ||
4110cc25 | 160 | asoc = __sctp_rcv_lookup(net, skb, &src, &dest, &transport); |
1c7d1fc1 | 161 | |
0fd9a65a | 162 | if (!asoc) |
532ae2f1 | 163 | ep = __sctp_rcv_lookup_endpoint(net, skb, &dest, &src); |
0fd9a65a NH |
164 | |
165 | /* Retrieve the common input handling substructure. */ | |
166 | rcvr = asoc ? &asoc->base : &ep->base; | |
167 | sk = rcvr->sk; | |
168 | ||
169 | /* | |
170 | * If a frame arrives on an interface and the receiving socket is | |
171 | * bound to another interface, via SO_BINDTODEVICE, treat it as OOTB | |
172 | */ | |
a20ea298 ED |
173 | bound_dev_if = READ_ONCE(sk->sk_bound_dev_if); |
174 | if (bound_dev_if && (bound_dev_if != af->skb_iif(skb))) { | |
dae399d7 XL |
175 | if (transport) { |
176 | sctp_transport_put(transport); | |
0fd9a65a | 177 | asoc = NULL; |
dae399d7 | 178 | transport = NULL; |
0fd9a65a NH |
179 | } else { |
180 | sctp_endpoint_put(ep); | |
181 | ep = NULL; | |
182 | } | |
2ce95503 | 183 | sk = net->sctp.ctl_sock; |
0fd9a65a NH |
184 | ep = sctp_sk(sk)->ep; |
185 | sctp_endpoint_hold(ep); | |
0fd9a65a NH |
186 | rcvr = &ep->base; |
187 | } | |
188 | ||
1da177e4 LT |
189 | /* |
190 | * RFC 2960, 8.4 - Handle "Out of the blue" Packets. | |
191 | * An SCTP packet is called an "out of the blue" (OOTB) | |
192 | * packet if it is correctly formed, i.e., passed the | |
193 | * receiver's checksum check, but the receiver is not | |
194 | * able to identify the association to which this | |
195 | * packet belongs. | |
196 | */ | |
197 | if (!asoc) { | |
1da177e4 | 198 | if (sctp_rcv_ootb(skb)) { |
08e3baef | 199 | __SCTP_INC_STATS(net, SCTP_MIB_OUTOFBLUES); |
1da177e4 LT |
200 | goto discard_release; |
201 | } | |
202 | } | |
203 | ||
1da177e4 LT |
204 | if (!xfrm_policy_check(sk, XFRM_POLICY_IN, skb, family)) |
205 | goto discard_release; | |
895b5c9f | 206 | nf_reset_ct(skb); |
1da177e4 | 207 | |
fda9ef5d | 208 | if (sk_filter(sk, skb)) |
d808ad9a | 209 | goto discard_release; |
1da177e4 LT |
210 | |
211 | /* Create an SCTP packet structure. */ | |
cea8768f | 212 | chunk = sctp_chunkify(skb, asoc, sk, GFP_ATOMIC); |
2babf9da | 213 | if (!chunk) |
1da177e4 | 214 | goto discard_release; |
79af02c2 | 215 | SCTP_INPUT_CB(skb)->chunk = chunk; |
1da177e4 | 216 | |
1da177e4 LT |
217 | /* Remember what endpoint is to handle this packet. */ |
218 | chunk->rcvr = rcvr; | |
219 | ||
220 | /* Remember the SCTP header. */ | |
3acb50c1 | 221 | chunk->sctp_hdr = sctp_hdr(skb); |
1da177e4 LT |
222 | |
223 | /* Set the source and destination addresses of the incoming chunk. */ | |
d55c41b1 | 224 | sctp_init_addrs(chunk, &src, &dest); |
1da177e4 LT |
225 | |
226 | /* Remember where we came from. */ | |
227 | chunk->transport = transport; | |
228 | ||
229 | /* Acquire access to the sock lock. Note: We are safe from other | |
230 | * bottom halves on this lock, but a user may be in the lock too, | |
231 | * so check if it is busy. | |
232 | */ | |
5bc1d1b4 | 233 | bh_lock_sock(sk); |
1da177e4 | 234 | |
ae53b5bd VY |
235 | if (sk != rcvr->sk) { |
236 | /* Our cached sk is different from the rcvr->sk. This is | |
237 | * because migrate()/accept() may have moved the association | |
238 | * to a new socket and released all the sockets. So now we | |
239 | * are holding a lock on the old socket while the user may | |
240 | * be doing something with the new socket. Switch our veiw | |
241 | * of the current sk. | |
242 | */ | |
5bc1d1b4 | 243 | bh_unlock_sock(sk); |
ae53b5bd | 244 | sk = rcvr->sk; |
5bc1d1b4 | 245 | bh_lock_sock(sk); |
ae53b5bd VY |
246 | } |
247 | ||
819be810 | 248 | if (sock_owned_by_user(sk) || !sctp_newsk_ready(sk)) { |
50b1a782 | 249 | if (sctp_add_backlog(sk, skb)) { |
5bc1d1b4 | 250 | bh_unlock_sock(sk); |
50b1a782 ZY |
251 | sctp_chunk_free(chunk); |
252 | skb = NULL; /* sctp_chunk_free already freed the skb */ | |
253 | goto discard_release; | |
254 | } | |
08e3baef | 255 | __SCTP_INC_STATS(net, SCTP_MIB_IN_PKT_BACKLOG); |
ac0b0462 | 256 | } else { |
08e3baef | 257 | __SCTP_INC_STATS(net, SCTP_MIB_IN_PKT_SOFTIRQ); |
61c9fed4 | 258 | sctp_inq_push(&chunk->rcvr->inqueue, chunk); |
ac0b0462 | 259 | } |
1da177e4 | 260 | |
5bc1d1b4 | 261 | bh_unlock_sock(sk); |
61c9fed4 VY |
262 | |
263 | /* Release the asoc/ep ref we took in the lookup calls. */ | |
dae399d7 XL |
264 | if (transport) |
265 | sctp_transport_put(transport); | |
61c9fed4 VY |
266 | else |
267 | sctp_endpoint_put(ep); | |
7a48f923 | 268 | |
2babf9da | 269 | return 0; |
1da177e4 LT |
270 | |
271 | discard_it: | |
08e3baef | 272 | __SCTP_INC_STATS(net, SCTP_MIB_IN_PKT_DISCARDS); |
1da177e4 | 273 | kfree_skb(skb); |
2babf9da | 274 | return 0; |
1da177e4 LT |
275 | |
276 | discard_release: | |
61c9fed4 | 277 | /* Release the asoc/ep ref we took in the lookup calls. */ |
dae399d7 XL |
278 | if (transport) |
279 | sctp_transport_put(transport); | |
0fd9a65a | 280 | else |
1da177e4 | 281 | sctp_endpoint_put(ep); |
1da177e4 LT |
282 | |
283 | goto discard_it; | |
284 | } | |
285 | ||
61c9fed4 VY |
286 | /* Process the backlog queue of the socket. Every skb on |
287 | * the backlog holds a ref on an association or endpoint. | |
288 | * We hold this ref throughout the state machine to make | |
289 | * sure that the structure we need is still around. | |
1da177e4 LT |
290 | */ |
291 | int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb) | |
292 | { | |
79af02c2 | 293 | struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk; |
d808ad9a | 294 | struct sctp_inq *inqueue = &chunk->rcvr->inqueue; |
dae399d7 | 295 | struct sctp_transport *t = chunk->transport; |
d808ad9a | 296 | struct sctp_ep_common *rcvr = NULL; |
61c9fed4 | 297 | int backloged = 0; |
7a48f923 | 298 | |
d808ad9a | 299 | rcvr = chunk->rcvr; |
c4d2444e | 300 | |
61c9fed4 VY |
301 | /* If the rcvr is dead then the association or endpoint |
302 | * has been deleted and we can safely drop the chunk | |
303 | * and refs that we are holding. | |
304 | */ | |
305 | if (rcvr->dead) { | |
306 | sctp_chunk_free(chunk); | |
307 | goto done; | |
308 | } | |
309 | ||
310 | if (unlikely(rcvr->sk != sk)) { | |
311 | /* In this case, the association moved from one socket to | |
312 | * another. We are currently sitting on the backlog of the | |
313 | * old socket, so we need to move. | |
314 | * However, since we are here in the process context we | |
315 | * need to take make sure that the user doesn't own | |
316 | * the new socket when we process the packet. | |
317 | * If the new socket is user-owned, queue the chunk to the | |
318 | * backlog of the new socket without dropping any refs. | |
319 | * Otherwise, we can safely push the chunk on the inqueue. | |
320 | */ | |
321 | ||
322 | sk = rcvr->sk; | |
eefc1b1d | 323 | local_bh_disable(); |
5bc1d1b4 | 324 | bh_lock_sock(sk); |
61c9fed4 | 325 | |
819be810 | 326 | if (sock_owned_by_user(sk) || !sctp_newsk_ready(sk)) { |
8265792b | 327 | if (sk_add_backlog(sk, skb, READ_ONCE(sk->sk_rcvbuf))) |
50b1a782 ZY |
328 | sctp_chunk_free(chunk); |
329 | else | |
330 | backloged = 1; | |
61c9fed4 VY |
331 | } else |
332 | sctp_inq_push(inqueue, chunk); | |
333 | ||
5bc1d1b4 | 334 | bh_unlock_sock(sk); |
eefc1b1d | 335 | local_bh_enable(); |
61c9fed4 VY |
336 | |
337 | /* If the chunk was backloged again, don't drop refs */ | |
338 | if (backloged) | |
339 | return 0; | |
340 | } else { | |
819be810 | 341 | if (!sctp_newsk_ready(sk)) { |
8265792b | 342 | if (!sk_add_backlog(sk, skb, READ_ONCE(sk->sk_rcvbuf))) |
819be810 XL |
343 | return 0; |
344 | sctp_chunk_free(chunk); | |
345 | } else { | |
346 | sctp_inq_push(inqueue, chunk); | |
347 | } | |
61c9fed4 VY |
348 | } |
349 | ||
350 | done: | |
351 | /* Release the refs we took in sctp_add_backlog */ | |
352 | if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type) | |
dae399d7 | 353 | sctp_transport_put(t); |
61c9fed4 VY |
354 | else if (SCTP_EP_TYPE_SOCKET == rcvr->type) |
355 | sctp_endpoint_put(sctp_ep(rcvr)); | |
356 | else | |
357 | BUG(); | |
358 | ||
d808ad9a | 359 | return 0; |
1da177e4 LT |
360 | } |
361 | ||
50b1a782 | 362 | static int sctp_add_backlog(struct sock *sk, struct sk_buff *skb) |
c4d2444e | 363 | { |
61c9fed4 | 364 | struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk; |
dae399d7 | 365 | struct sctp_transport *t = chunk->transport; |
61c9fed4 | 366 | struct sctp_ep_common *rcvr = chunk->rcvr; |
50b1a782 | 367 | int ret; |
c4d2444e | 368 | |
8265792b | 369 | ret = sk_add_backlog(sk, skb, READ_ONCE(sk->sk_rcvbuf)); |
50b1a782 ZY |
370 | if (!ret) { |
371 | /* Hold the assoc/ep while hanging on the backlog queue. | |
372 | * This way, we know structures we need will not disappear | |
373 | * from us | |
374 | */ | |
375 | if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type) | |
dae399d7 | 376 | sctp_transport_hold(t); |
50b1a782 ZY |
377 | else if (SCTP_EP_TYPE_SOCKET == rcvr->type) |
378 | sctp_endpoint_hold(sctp_ep(rcvr)); | |
379 | else | |
380 | BUG(); | |
381 | } | |
382 | return ret; | |
61c9fed4 | 383 | |
c4d2444e SS |
384 | } |
385 | ||
1da177e4 LT |
386 | /* Handle icmp frag needed error. */ |
387 | void sctp_icmp_frag_needed(struct sock *sk, struct sctp_association *asoc, | |
388 | struct sctp_transport *t, __u32 pmtu) | |
389 | { | |
83696408 XL |
390 | if (!t || |
391 | (t->pathmtu <= pmtu && | |
392 | t->pl.probe_size + sctp_transport_pl_hlen(t) <= pmtu)) | |
52ccb8e9 FF |
393 | return; |
394 | ||
8a479491 | 395 | if (sock_owned_by_user(sk)) { |
d805397c | 396 | atomic_set(&t->mtu_info, pmtu); |
8a479491 VY |
397 | asoc->pmtu_pending = 1; |
398 | t->pmtu_pending = 1; | |
399 | return; | |
400 | } | |
401 | ||
cc35c3d1 MRL |
402 | if (!(t->param_flags & SPP_PMTUD_ENABLE)) |
403 | /* We can't allow retransmitting in such case, as the | |
404 | * retransmission would be sized just as before, and thus we | |
405 | * would get another icmp, and retransmit again. | |
406 | */ | |
407 | return; | |
1da177e4 | 408 | |
b6c5734d MRL |
409 | /* Update transports view of the MTU. Return if no update was needed. |
410 | * If an update wasn't needed/possible, it also doesn't make sense to | |
411 | * try to retransmit now. | |
412 | */ | |
413 | if (!sctp_transport_update_pmtu(t, pmtu)) | |
414 | return; | |
52ccb8e9 | 415 | |
cc35c3d1 MRL |
416 | /* Update association pmtu. */ |
417 | sctp_assoc_sync_pmtu(asoc); | |
418 | ||
419 | /* Retransmit with the new pmtu setting. */ | |
52ccb8e9 | 420 | sctp_retransmit(&asoc->outqueue, t, SCTP_RTXR_PMTUD); |
1da177e4 LT |
421 | } |
422 | ||
ec18d9a2 DM |
423 | void sctp_icmp_redirect(struct sock *sk, struct sctp_transport *t, |
424 | struct sk_buff *skb) | |
55be7a9c DM |
425 | { |
426 | struct dst_entry *dst; | |
427 | ||
1cc276ce | 428 | if (sock_owned_by_user(sk) || !t) |
55be7a9c DM |
429 | return; |
430 | dst = sctp_transport_dst_check(t); | |
1ed5c48f | 431 | if (dst) |
6700c270 | 432 | dst->ops->redirect(dst, sk, skb); |
55be7a9c DM |
433 | } |
434 | ||
1da177e4 LT |
435 | /* |
436 | * SCTP Implementer's Guide, 2.37 ICMP handling procedures | |
437 | * | |
438 | * ICMP8) If the ICMP code is a "Unrecognized next header type encountered" | |
439 | * or a "Protocol Unreachable" treat this message as an abort | |
440 | * with the T bit set. | |
441 | * | |
442 | * This function sends an event to the state machine, which will abort the | |
443 | * association. | |
444 | * | |
445 | */ | |
446 | void sctp_icmp_proto_unreachable(struct sock *sk, | |
d808ad9a YH |
447 | struct sctp_association *asoc, |
448 | struct sctp_transport *t) | |
1da177e4 | 449 | { |
50b5d6ad VY |
450 | if (sock_owned_by_user(sk)) { |
451 | if (timer_pending(&t->proto_unreach_timer)) | |
452 | return; | |
453 | else { | |
454 | if (!mod_timer(&t->proto_unreach_timer, | |
455 | jiffies + (HZ/20))) | |
057a10fa | 456 | sctp_transport_hold(t); |
50b5d6ad | 457 | } |
50b5d6ad | 458 | } else { |
55e26eb9 EB |
459 | struct net *net = sock_net(sk); |
460 | ||
bb33381d DB |
461 | pr_debug("%s: unrecognized next header type " |
462 | "encountered!\n", __func__); | |
463 | ||
25cc4ae9 | 464 | if (del_timer(&t->proto_unreach_timer)) |
057a10fa | 465 | sctp_transport_put(t); |
1da177e4 | 466 | |
55e26eb9 | 467 | sctp_do_sm(net, SCTP_EVENT_T_OTHER, |
50b5d6ad VY |
468 | SCTP_ST_OTHER(SCTP_EVENT_ICMP_PROTO_UNREACH), |
469 | asoc->state, asoc->ep, asoc, t, | |
470 | GFP_ATOMIC); | |
471 | } | |
1da177e4 LT |
472 | } |
473 | ||
474 | /* Common lookup code for icmp/icmpv6 error handler. */ | |
4110cc25 | 475 | struct sock *sctp_err_lookup(struct net *net, int family, struct sk_buff *skb, |
1da177e4 | 476 | struct sctphdr *sctphdr, |
1da177e4 LT |
477 | struct sctp_association **app, |
478 | struct sctp_transport **tpp) | |
479 | { | |
804ec7eb | 480 | struct sctp_init_chunk *chunkhdr, _chunkhdr; |
1da177e4 LT |
481 | union sctp_addr saddr; |
482 | union sctp_addr daddr; | |
483 | struct sctp_af *af; | |
484 | struct sock *sk = NULL; | |
8de8c873 | 485 | struct sctp_association *asoc; |
1da177e4 | 486 | struct sctp_transport *transport = NULL; |
7115e632 | 487 | __u32 vtag = ntohl(sctphdr->vtag); |
1da177e4 | 488 | |
d1ad1ff2 | 489 | *app = NULL; *tpp = NULL; |
1da177e4 LT |
490 | |
491 | af = sctp_get_af_specific(family); | |
492 | if (unlikely(!af)) { | |
493 | return NULL; | |
494 | } | |
495 | ||
496 | /* Initialize local addresses for lookups. */ | |
497 | af->from_skb(&saddr, skb, 1); | |
498 | af->from_skb(&daddr, skb, 0); | |
499 | ||
500 | /* Look for an association that matches the incoming ICMP error | |
501 | * packet. | |
502 | */ | |
4110cc25 | 503 | asoc = __sctp_lookup_association(net, &saddr, &daddr, &transport); |
d1ad1ff2 SS |
504 | if (!asoc) |
505 | return NULL; | |
1da177e4 | 506 | |
d1ad1ff2 | 507 | sk = asoc->base.sk; |
1da177e4 | 508 | |
7115e632 WY |
509 | /* RFC 4960, Appendix C. ICMP Handling |
510 | * | |
511 | * ICMP6) An implementation MUST validate that the Verification Tag | |
512 | * contained in the ICMP message matches the Verification Tag of | |
513 | * the peer. If the Verification Tag is not 0 and does NOT | |
514 | * match, discard the ICMP message. If it is 0 and the ICMP | |
515 | * message contains enough bytes to verify that the chunk type is | |
516 | * an INIT chunk and that the Initiate Tag matches the tag of the | |
517 | * peer, continue with ICMP7. If the ICMP message is too short | |
518 | * or the chunk type or the Initiate Tag does not match, silently | |
519 | * discard the packet. | |
520 | */ | |
521 | if (vtag == 0) { | |
804ec7eb DC |
522 | /* chunk header + first 4 octects of init header */ |
523 | chunkhdr = skb_header_pointer(skb, skb_transport_offset(skb) + | |
524 | sizeof(struct sctphdr), | |
525 | sizeof(struct sctp_chunkhdr) + | |
526 | sizeof(__be32), &_chunkhdr); | |
527 | if (!chunkhdr || | |
7115e632 | 528 | chunkhdr->chunk_hdr.type != SCTP_CID_INIT || |
804ec7eb | 529 | ntohl(chunkhdr->init_hdr.init_tag) != asoc->c.my_vtag) |
7115e632 | 530 | goto out; |
804ec7eb | 531 | |
7115e632 | 532 | } else if (vtag != asoc->c.peer_vtag) { |
d1ad1ff2 SS |
533 | goto out; |
534 | } | |
1da177e4 | 535 | |
5bc1d1b4 | 536 | bh_lock_sock(sk); |
1da177e4 LT |
537 | |
538 | /* If too many ICMPs get dropped on busy | |
539 | * servers this needs to be solved differently. | |
540 | */ | |
541 | if (sock_owned_by_user(sk)) | |
02a1d6e7 | 542 | __NET_INC_STATS(net, LINUX_MIB_LOCKDROPPEDICMPS); |
1da177e4 | 543 | |
1da177e4 LT |
544 | *app = asoc; |
545 | *tpp = transport; | |
546 | return sk; | |
547 | ||
548 | out: | |
dae399d7 | 549 | sctp_transport_put(transport); |
1da177e4 LT |
550 | return NULL; |
551 | } | |
552 | ||
553 | /* Common cleanup code for icmp/icmpv6 error handler. */ | |
dae399d7 | 554 | void sctp_err_finish(struct sock *sk, struct sctp_transport *t) |
887cf3d1 | 555 | __releases(&((__sk)->sk_lock.slock)) |
1da177e4 | 556 | { |
5bc1d1b4 | 557 | bh_unlock_sock(sk); |
dae399d7 | 558 | sctp_transport_put(t); |
1da177e4 LT |
559 | } |
560 | ||
d8306075 XL |
561 | static void sctp_v4_err_handle(struct sctp_transport *t, struct sk_buff *skb, |
562 | __u8 type, __u8 code, __u32 info) | |
563 | { | |
564 | struct sctp_association *asoc = t->asoc; | |
565 | struct sock *sk = asoc->base.sk; | |
566 | int err = 0; | |
567 | ||
568 | switch (type) { | |
569 | case ICMP_PARAMETERPROB: | |
570 | err = EPROTO; | |
571 | break; | |
572 | case ICMP_DEST_UNREACH: | |
573 | if (code > NR_ICMP_UNREACH) | |
574 | return; | |
575 | if (code == ICMP_FRAG_NEEDED) { | |
576 | sctp_icmp_frag_needed(sk, asoc, t, SCTP_TRUNC4(info)); | |
577 | return; | |
578 | } | |
579 | if (code == ICMP_PROT_UNREACH) { | |
580 | sctp_icmp_proto_unreachable(sk, asoc, t); | |
581 | return; | |
582 | } | |
583 | err = icmp_err_convert[code].errno; | |
584 | break; | |
585 | case ICMP_TIME_EXCEEDED: | |
586 | if (code == ICMP_EXC_FRAGTIME) | |
587 | return; | |
588 | ||
589 | err = EHOSTUNREACH; | |
590 | break; | |
591 | case ICMP_REDIRECT: | |
592 | sctp_icmp_redirect(sk, t, skb); | |
dbe69e43 | 593 | return; |
d8306075 XL |
594 | default: |
595 | return; | |
596 | } | |
597 | if (!sock_owned_by_user(sk) && inet_sk(sk)->recverr) { | |
598 | sk->sk_err = err; | |
e3ae2365 | 599 | sk_error_report(sk); |
d8306075 XL |
600 | } else { /* Only an error on timeout */ |
601 | sk->sk_err_soft = err; | |
602 | } | |
603 | } | |
604 | ||
1da177e4 LT |
605 | /* |
606 | * This routine is called by the ICMP module when it gets some | |
607 | * sort of error condition. If err < 0 then the socket should | |
608 | * be closed and the error returned to the user. If err > 0 | |
609 | * it's just the icmp type << 8 | icmp code. After adjustment | |
610 | * header points to the first 8 bytes of the sctp header. We need | |
611 | * to find the appropriate port. | |
612 | * | |
613 | * The locking strategy used here is very "optimistic". When | |
614 | * someone else accesses the socket the ICMP is just dropped | |
615 | * and for some paths there is no check at all. | |
616 | * A more general error queue to queue errors for later handling | |
617 | * is probably better. | |
618 | * | |
619 | */ | |
32bbd879 | 620 | int sctp_v4_err(struct sk_buff *skb, __u32 info) |
1da177e4 | 621 | { |
b71d1d42 | 622 | const struct iphdr *iph = (const struct iphdr *)skb->data; |
88c7664f ACM |
623 | const int type = icmp_hdr(skb)->type; |
624 | const int code = icmp_hdr(skb)->code; | |
d8306075 | 625 | struct net *net = dev_net(skb->dev); |
1da177e4 | 626 | struct sctp_transport *transport; |
d8306075 | 627 | struct sctp_association *asoc; |
aef6de51 | 628 | __u16 saveip, savesctp; |
d8306075 | 629 | struct sock *sk; |
1da177e4 | 630 | |
1da177e4 | 631 | /* Fix up skb to look at the embedded net header. */ |
b0e380b1 ACM |
632 | saveip = skb->network_header; |
633 | savesctp = skb->transport_header; | |
31c7711b | 634 | skb_reset_network_header(skb); |
d8306075 | 635 | skb_set_transport_header(skb, iph->ihl * 4); |
4110cc25 | 636 | sk = sctp_err_lookup(net, AF_INET, skb, sctp_hdr(skb), &asoc, &transport); |
2e07fa9c | 637 | /* Put back, the original values. */ |
b0e380b1 ACM |
638 | skb->network_header = saveip; |
639 | skb->transport_header = savesctp; | |
1da177e4 | 640 | if (!sk) { |
5d3848bc | 641 | __ICMP_INC_STATS(net, ICMP_MIB_INERRORS); |
32bbd879 | 642 | return -ENOENT; |
1da177e4 | 643 | } |
1da177e4 | 644 | |
d8306075 | 645 | sctp_v4_err_handle(transport, skb, type, code, info); |
dae399d7 | 646 | sctp_err_finish(sk, transport); |
1da177e4 | 647 | |
32bbd879 | 648 | return 0; |
1da177e4 | 649 | } |
1da177e4 | 650 | |
9e47df00 XL |
651 | int sctp_udp_v4_err(struct sock *sk, struct sk_buff *skb) |
652 | { | |
653 | struct net *net = dev_net(skb->dev); | |
654 | struct sctp_association *asoc; | |
655 | struct sctp_transport *t; | |
656 | struct icmphdr *hdr; | |
657 | __u32 info = 0; | |
658 | ||
659 | skb->transport_header += sizeof(struct udphdr); | |
660 | sk = sctp_err_lookup(net, AF_INET, skb, sctp_hdr(skb), &asoc, &t); | |
661 | if (!sk) { | |
662 | __ICMP_INC_STATS(net, ICMP_MIB_INERRORS); | |
663 | return -ENOENT; | |
1da177e4 LT |
664 | } |
665 | ||
9e47df00 XL |
666 | skb->transport_header -= sizeof(struct udphdr); |
667 | hdr = (struct icmphdr *)(skb_network_header(skb) - sizeof(struct icmphdr)); | |
668 | if (hdr->type == ICMP_REDIRECT) { | |
669 | /* can't be handled without outer iphdr known, leave it to udp_err */ | |
670 | sctp_err_finish(sk, t); | |
671 | return 0; | |
1da177e4 | 672 | } |
9e47df00 XL |
673 | if (hdr->type == ICMP_DEST_UNREACH && hdr->code == ICMP_FRAG_NEEDED) |
674 | info = ntohs(hdr->un.frag.mtu); | |
675 | sctp_v4_err_handle(t, skb, hdr->type, hdr->code, info); | |
1da177e4 | 676 | |
9e47df00 XL |
677 | sctp_err_finish(sk, t); |
678 | return 1; | |
1da177e4 LT |
679 | } |
680 | ||
681 | /* | |
682 | * RFC 2960, 8.4 - Handle "Out of the blue" Packets. | |
683 | * | |
684 | * This function scans all the chunks in the OOTB packet to determine if | |
685 | * the packet should be discarded right away. If a response might be needed | |
686 | * for this packet, or, if further processing is possible, the packet will | |
687 | * be queued to a proper inqueue for the next phase of handling. | |
688 | * | |
689 | * Output: | |
690 | * Return 0 - If further processing is needed. | |
691 | * Return 1 - If the packet can be discarded right away. | |
692 | */ | |
04675210 | 693 | static int sctp_rcv_ootb(struct sk_buff *skb) |
1da177e4 | 694 | { |
922dbc5b | 695 | struct sctp_chunkhdr *ch, _ch; |
3acb50c1 | 696 | int ch_end, offset = 0; |
1da177e4 LT |
697 | |
698 | /* Scan through all the chunks in the packet. */ | |
a7d1f1b6 | 699 | do { |
3acb50c1 | 700 | /* Make sure we have at least the header there */ |
922dbc5b | 701 | if (offset + sizeof(_ch) > skb->len) |
3acb50c1 MRL |
702 | break; |
703 | ||
704 | ch = skb_header_pointer(skb, offset, sizeof(*ch), &_ch); | |
705 | ||
a7d1f1b6 | 706 | /* Break out if chunk length is less then minimal. */ |
f7e745f8 | 707 | if (!ch || ntohs(ch->length) < sizeof(_ch)) |
a7d1f1b6 TF |
708 | break; |
709 | ||
e2f036a9 | 710 | ch_end = offset + SCTP_PAD4(ntohs(ch->length)); |
3acb50c1 | 711 | if (ch_end > skb->len) |
a7d1f1b6 | 712 | break; |
1da177e4 LT |
713 | |
714 | /* RFC 8.4, 2) If the OOTB packet contains an ABORT chunk, the | |
715 | * receiver MUST silently discard the OOTB packet and take no | |
716 | * further action. | |
717 | */ | |
718 | if (SCTP_CID_ABORT == ch->type) | |
719 | goto discard; | |
720 | ||
721 | /* RFC 8.4, 6) If the packet contains a SHUTDOWN COMPLETE | |
722 | * chunk, the receiver should silently discard the packet | |
723 | * and take no further action. | |
724 | */ | |
725 | if (SCTP_CID_SHUTDOWN_COMPLETE == ch->type) | |
726 | goto discard; | |
727 | ||
3c77f961 VY |
728 | /* RFC 4460, 2.11.2 |
729 | * This will discard packets with INIT chunk bundled as | |
730 | * subsequent chunks in the packet. When INIT is first, | |
731 | * the normal INIT processing will discard the chunk. | |
732 | */ | |
733 | if (SCTP_CID_INIT == ch->type && (void *)ch != skb->data) | |
734 | goto discard; | |
735 | ||
3acb50c1 MRL |
736 | offset = ch_end; |
737 | } while (ch_end < skb->len); | |
1da177e4 LT |
738 | |
739 | return 0; | |
740 | ||
741 | discard: | |
742 | return 1; | |
743 | } | |
744 | ||
745 | /* Insert endpoint into the hash table. */ | |
76c6d988 | 746 | static int __sctp_hash_endpoint(struct sctp_endpoint *ep) |
1da177e4 | 747 | { |
76c6d988 XL |
748 | struct sock *sk = ep->base.sk; |
749 | struct net *net = sock_net(sk); | |
1da177e4 LT |
750 | struct sctp_hashbucket *head; |
751 | ||
3d3b2f57 XL |
752 | ep->hashent = sctp_ep_hashfn(net, ep->base.bind_addr.port); |
753 | head = &sctp_ep_hashtable[ep->hashent]; | |
1da177e4 | 754 | |
76c6d988 XL |
755 | if (sk->sk_reuseport) { |
756 | bool any = sctp_is_ep_boundall(sk); | |
3d3b2f57 | 757 | struct sctp_endpoint *ep2; |
76c6d988 XL |
758 | struct list_head *list; |
759 | int cnt = 0, err = 1; | |
760 | ||
761 | list_for_each(list, &ep->base.bind_addr.address_list) | |
762 | cnt++; | |
763 | ||
3d3b2f57 XL |
764 | sctp_for_each_hentry(ep2, &head->chain) { |
765 | struct sock *sk2 = ep2->base.sk; | |
76c6d988 XL |
766 | |
767 | if (!net_eq(sock_net(sk2), net) || sk2 == sk || | |
768 | !uid_eq(sock_i_uid(sk2), sock_i_uid(sk)) || | |
769 | !sk2->sk_reuseport) | |
770 | continue; | |
771 | ||
772 | err = sctp_bind_addrs_check(sctp_sk(sk2), | |
773 | sctp_sk(sk), cnt); | |
774 | if (!err) { | |
775 | err = reuseport_add_sock(sk, sk2, any); | |
776 | if (err) | |
777 | return err; | |
778 | break; | |
779 | } else if (err < 0) { | |
780 | return err; | |
781 | } | |
782 | } | |
783 | ||
784 | if (err) { | |
785 | err = reuseport_alloc(sk, any); | |
786 | if (err) | |
787 | return err; | |
788 | } | |
789 | } | |
790 | ||
387602df | 791 | write_lock(&head->lock); |
3d3b2f57 | 792 | hlist_add_head(&ep->node, &head->chain); |
387602df | 793 | write_unlock(&head->lock); |
76c6d988 | 794 | return 0; |
1da177e4 LT |
795 | } |
796 | ||
797 | /* Add an endpoint to the hash. Local BH-safe. */ | |
76c6d988 | 798 | int sctp_hash_endpoint(struct sctp_endpoint *ep) |
1da177e4 | 799 | { |
76c6d988 XL |
800 | int err; |
801 | ||
79b91130 | 802 | local_bh_disable(); |
76c6d988 | 803 | err = __sctp_hash_endpoint(ep); |
79b91130 | 804 | local_bh_enable(); |
76c6d988 XL |
805 | |
806 | return err; | |
1da177e4 LT |
807 | } |
808 | ||
809 | /* Remove endpoint from the hash table. */ | |
810 | static void __sctp_unhash_endpoint(struct sctp_endpoint *ep) | |
811 | { | |
76c6d988 | 812 | struct sock *sk = ep->base.sk; |
1da177e4 | 813 | struct sctp_hashbucket *head; |
1da177e4 | 814 | |
3d3b2f57 | 815 | ep->hashent = sctp_ep_hashfn(sock_net(sk), ep->base.bind_addr.port); |
1da177e4 | 816 | |
3d3b2f57 | 817 | head = &sctp_ep_hashtable[ep->hashent]; |
1da177e4 | 818 | |
76c6d988 XL |
819 | if (rcu_access_pointer(sk->sk_reuseport_cb)) |
820 | reuseport_detach_sock(sk); | |
821 | ||
387602df | 822 | write_lock(&head->lock); |
3d3b2f57 | 823 | hlist_del_init(&ep->node); |
387602df | 824 | write_unlock(&head->lock); |
1da177e4 LT |
825 | } |
826 | ||
827 | /* Remove endpoint from the hash. Local BH-safe. */ | |
828 | void sctp_unhash_endpoint(struct sctp_endpoint *ep) | |
829 | { | |
79b91130 | 830 | local_bh_disable(); |
1da177e4 | 831 | __sctp_unhash_endpoint(ep); |
79b91130 | 832 | local_bh_enable(); |
1da177e4 LT |
833 | } |
834 | ||
532ae2f1 XL |
835 | static inline __u32 sctp_hashfn(const struct net *net, __be16 lport, |
836 | const union sctp_addr *paddr, __u32 seed) | |
837 | { | |
838 | __u32 addr; | |
839 | ||
840 | if (paddr->sa.sa_family == AF_INET6) | |
841 | addr = jhash(&paddr->v6.sin6_addr, 16, seed); | |
842 | else | |
843 | addr = (__force __u32)paddr->v4.sin_addr.s_addr; | |
844 | ||
845 | return jhash_3words(addr, ((__force __u32)paddr->v4.sin_port) << 16 | | |
846 | (__force __u32)lport, net_hash_mix(net), seed); | |
847 | } | |
848 | ||
1da177e4 | 849 | /* Look up an endpoint. */ |
532ae2f1 XL |
850 | static struct sctp_endpoint *__sctp_rcv_lookup_endpoint( |
851 | struct net *net, struct sk_buff *skb, | |
852 | const union sctp_addr *laddr, | |
853 | const union sctp_addr *paddr) | |
1da177e4 LT |
854 | { |
855 | struct sctp_hashbucket *head; | |
1da177e4 | 856 | struct sctp_endpoint *ep; |
532ae2f1 XL |
857 | struct sock *sk; |
858 | __be16 lport; | |
1da177e4 LT |
859 | int hash; |
860 | ||
532ae2f1 XL |
861 | lport = laddr->v4.sin_port; |
862 | hash = sctp_ep_hashfn(net, ntohs(lport)); | |
1da177e4 LT |
863 | head = &sctp_ep_hashtable[hash]; |
864 | read_lock(&head->lock); | |
3d3b2f57 | 865 | sctp_for_each_hentry(ep, &head->chain) { |
4cdadcbc | 866 | if (sctp_endpoint_is_match(ep, net, laddr)) |
1da177e4 LT |
867 | goto hit; |
868 | } | |
869 | ||
2ce95503 | 870 | ep = sctp_sk(net->sctp.ctl_sock)->ep; |
1da177e4 LT |
871 | |
872 | hit: | |
532ae2f1 XL |
873 | sk = ep->base.sk; |
874 | if (sk->sk_reuseport) { | |
875 | __u32 phash = sctp_hashfn(net, lport, paddr, 0); | |
876 | ||
877 | sk = reuseport_select_sock(sk, phash, skb, | |
878 | sizeof(struct sctphdr)); | |
879 | if (sk) | |
880 | ep = sctp_sk(sk)->ep; | |
881 | } | |
1da177e4 | 882 | sctp_endpoint_hold(ep); |
1da177e4 LT |
883 | read_unlock(&head->lock); |
884 | return ep; | |
885 | } | |
886 | ||
d6c0256a XL |
887 | /* rhashtable for transport */ |
888 | struct sctp_hash_cmp_arg { | |
7fda702f XL |
889 | const union sctp_addr *paddr; |
890 | const struct net *net; | |
8d32503e | 891 | __be16 lport; |
d6c0256a XL |
892 | }; |
893 | ||
894 | static inline int sctp_hash_cmp(struct rhashtable_compare_arg *arg, | |
895 | const void *ptr) | |
896 | { | |
715f5552 | 897 | struct sctp_transport *t = (struct sctp_transport *)ptr; |
d6c0256a | 898 | const struct sctp_hash_cmp_arg *x = arg->key; |
715f5552 | 899 | int err = 1; |
d6c0256a | 900 | |
d6c0256a | 901 | if (!sctp_cmp_addr_exact(&t->ipaddr, x->paddr)) |
715f5552 XL |
902 | return err; |
903 | if (!sctp_transport_hold(t)) | |
904 | return err; | |
905 | ||
31243461 | 906 | if (!net_eq(t->asoc->base.net, x->net)) |
7fda702f XL |
907 | goto out; |
908 | if (x->lport != htons(t->asoc->base.bind_addr.port)) | |
715f5552 | 909 | goto out; |
d6c0256a | 910 | |
715f5552 XL |
911 | err = 0; |
912 | out: | |
913 | sctp_transport_put(t); | |
914 | return err; | |
d6c0256a XL |
915 | } |
916 | ||
8d32503e | 917 | static inline __u32 sctp_hash_obj(const void *data, u32 len, u32 seed) |
d6c0256a XL |
918 | { |
919 | const struct sctp_transport *t = data; | |
d6c0256a | 920 | |
31243461 | 921 | return sctp_hashfn(t->asoc->base.net, |
532ae2f1 XL |
922 | htons(t->asoc->base.bind_addr.port), |
923 | &t->ipaddr, seed); | |
d6c0256a XL |
924 | } |
925 | ||
8d32503e | 926 | static inline __u32 sctp_hash_key(const void *data, u32 len, u32 seed) |
d6c0256a XL |
927 | { |
928 | const struct sctp_hash_cmp_arg *x = data; | |
d6c0256a | 929 | |
532ae2f1 | 930 | return sctp_hashfn(x->net, x->lport, x->paddr, seed); |
d6c0256a XL |
931 | } |
932 | ||
933 | static const struct rhashtable_params sctp_hash_params = { | |
934 | .head_offset = offsetof(struct sctp_transport, node), | |
935 | .hashfn = sctp_hash_key, | |
936 | .obj_hashfn = sctp_hash_obj, | |
937 | .obj_cmpfn = sctp_hash_cmp, | |
938 | .automatic_shrinking = true, | |
939 | }; | |
940 | ||
941 | int sctp_transport_hashtable_init(void) | |
942 | { | |
7fda702f | 943 | return rhltable_init(&sctp_transport_hashtable, &sctp_hash_params); |
d6c0256a XL |
944 | } |
945 | ||
946 | void sctp_transport_hashtable_destroy(void) | |
947 | { | |
7fda702f | 948 | rhltable_destroy(&sctp_transport_hashtable); |
d6c0256a XL |
949 | } |
950 | ||
7fda702f | 951 | int sctp_hash_transport(struct sctp_transport *t) |
d6c0256a | 952 | { |
cd2b7087 XL |
953 | struct sctp_transport *transport; |
954 | struct rhlist_head *tmp, *list; | |
d6c0256a | 955 | struct sctp_hash_cmp_arg arg; |
7fda702f | 956 | int err; |
d6c0256a | 957 | |
dd7445ad | 958 | if (t->asoc->temp) |
7fda702f | 959 | return 0; |
dd7445ad | 960 | |
4e7696d9 | 961 | arg.net = t->asoc->base.net; |
7fda702f XL |
962 | arg.paddr = &t->ipaddr; |
963 | arg.lport = htons(t->asoc->base.bind_addr.port); | |
d6c0256a | 964 | |
5179b266 | 965 | rcu_read_lock(); |
cd2b7087 XL |
966 | list = rhltable_lookup(&sctp_transport_hashtable, &arg, |
967 | sctp_hash_params); | |
968 | ||
969 | rhl_for_each_entry_rcu(transport, tmp, list, node) | |
970 | if (transport->asoc->ep == t->asoc->ep) { | |
5179b266 | 971 | rcu_read_unlock(); |
27af86bb | 972 | return -EEXIST; |
cd2b7087 | 973 | } |
5179b266 | 974 | rcu_read_unlock(); |
cd2b7087 | 975 | |
7fda702f XL |
976 | err = rhltable_insert_key(&sctp_transport_hashtable, &arg, |
977 | &t->node, sctp_hash_params); | |
978 | if (err) | |
979 | pr_err_once("insert transport fail, errno %d\n", err); | |
980 | ||
981 | return err; | |
d6c0256a XL |
982 | } |
983 | ||
984 | void sctp_unhash_transport(struct sctp_transport *t) | |
985 | { | |
dd7445ad XL |
986 | if (t->asoc->temp) |
987 | return; | |
988 | ||
7fda702f XL |
989 | rhltable_remove(&sctp_transport_hashtable, &t->node, |
990 | sctp_hash_params); | |
d6c0256a XL |
991 | } |
992 | ||
7fda702f | 993 | /* return a transport with holding it */ |
d6c0256a XL |
994 | struct sctp_transport *sctp_addrs_lookup_transport( |
995 | struct net *net, | |
996 | const union sctp_addr *laddr, | |
997 | const union sctp_addr *paddr) | |
998 | { | |
7fda702f XL |
999 | struct rhlist_head *tmp, *list; |
1000 | struct sctp_transport *t; | |
d6c0256a | 1001 | struct sctp_hash_cmp_arg arg = { |
d6c0256a XL |
1002 | .paddr = paddr, |
1003 | .net = net, | |
7fda702f | 1004 | .lport = laddr->v4.sin_port, |
d6c0256a XL |
1005 | }; |
1006 | ||
7fda702f XL |
1007 | list = rhltable_lookup(&sctp_transport_hashtable, &arg, |
1008 | sctp_hash_params); | |
1009 | ||
1010 | rhl_for_each_entry_rcu(t, tmp, list, node) { | |
1011 | if (!sctp_transport_hold(t)) | |
1012 | continue; | |
1013 | ||
1014 | if (sctp_bind_addr_match(&t->asoc->base.bind_addr, | |
1015 | laddr, sctp_sk(t->asoc->base.sk))) | |
1016 | return t; | |
1017 | sctp_transport_put(t); | |
1018 | } | |
1019 | ||
1020 | return NULL; | |
d6c0256a XL |
1021 | } |
1022 | ||
7fda702f | 1023 | /* return a transport without holding it, as it's only used under sock lock */ |
d6c0256a XL |
1024 | struct sctp_transport *sctp_epaddr_lookup_transport( |
1025 | const struct sctp_endpoint *ep, | |
1026 | const union sctp_addr *paddr) | |
1027 | { | |
7fda702f XL |
1028 | struct rhlist_head *tmp, *list; |
1029 | struct sctp_transport *t; | |
65a5124a | 1030 | struct sctp_hash_cmp_arg arg = { |
65a5124a | 1031 | .paddr = paddr, |
4e7696d9 | 1032 | .net = ep->base.net, |
7fda702f | 1033 | .lport = htons(ep->base.bind_addr.port), |
65a5124a | 1034 | }; |
d6c0256a | 1035 | |
7fda702f XL |
1036 | list = rhltable_lookup(&sctp_transport_hashtable, &arg, |
1037 | sctp_hash_params); | |
1038 | ||
1039 | rhl_for_each_entry_rcu(t, tmp, list, node) | |
1040 | if (ep == t->asoc->ep) | |
1041 | return t; | |
1042 | ||
1043 | return NULL; | |
d6c0256a XL |
1044 | } |
1045 | ||
1da177e4 LT |
1046 | /* Look up an association. */ |
1047 | static struct sctp_association *__sctp_lookup_association( | |
4110cc25 | 1048 | struct net *net, |
1da177e4 LT |
1049 | const union sctp_addr *local, |
1050 | const union sctp_addr *peer, | |
1051 | struct sctp_transport **pt) | |
1052 | { | |
4f008781 | 1053 | struct sctp_transport *t; |
1eed6779 | 1054 | struct sctp_association *asoc = NULL; |
1da177e4 | 1055 | |
4f008781 | 1056 | t = sctp_addrs_lookup_transport(net, local, peer); |
7fda702f | 1057 | if (!t) |
1eed6779 | 1058 | goto out; |
1da177e4 | 1059 | |
1eed6779 | 1060 | asoc = t->asoc; |
4f008781 | 1061 | *pt = t; |
1da177e4 | 1062 | |
1eed6779 | 1063 | out: |
1eed6779 | 1064 | return asoc; |
1da177e4 LT |
1065 | } |
1066 | ||
4f008781 | 1067 | /* Look up an association. protected by RCU read lock */ |
dda91928 | 1068 | static |
4110cc25 EB |
1069 | struct sctp_association *sctp_lookup_association(struct net *net, |
1070 | const union sctp_addr *laddr, | |
1da177e4 | 1071 | const union sctp_addr *paddr, |
dda91928 | 1072 | struct sctp_transport **transportp) |
1da177e4 LT |
1073 | { |
1074 | struct sctp_association *asoc; | |
1075 | ||
f46c7011 | 1076 | rcu_read_lock(); |
4110cc25 | 1077 | asoc = __sctp_lookup_association(net, laddr, paddr, transportp); |
f46c7011 | 1078 | rcu_read_unlock(); |
1da177e4 LT |
1079 | |
1080 | return asoc; | |
1081 | } | |
1082 | ||
1083 | /* Is there an association matching the given local and peer addresses? */ | |
53066538 XL |
1084 | bool sctp_has_association(struct net *net, |
1085 | const union sctp_addr *laddr, | |
1086 | const union sctp_addr *paddr) | |
1da177e4 | 1087 | { |
1da177e4 LT |
1088 | struct sctp_transport *transport; |
1089 | ||
53066538 | 1090 | if (sctp_lookup_association(net, laddr, paddr, &transport)) { |
dae399d7 | 1091 | sctp_transport_put(transport); |
53066538 | 1092 | return true; |
1da177e4 LT |
1093 | } |
1094 | ||
53066538 | 1095 | return false; |
1da177e4 LT |
1096 | } |
1097 | ||
1098 | /* | |
1099 | * SCTP Implementors Guide, 2.18 Handling of address | |
1100 | * parameters within the INIT or INIT-ACK. | |
1101 | * | |
1102 | * D) When searching for a matching TCB upon reception of an INIT | |
1103 | * or INIT-ACK chunk the receiver SHOULD use not only the | |
1104 | * source address of the packet (containing the INIT or | |
1105 | * INIT-ACK) but the receiver SHOULD also use all valid | |
1106 | * address parameters contained within the chunk. | |
1107 | * | |
1108 | * 2.18.3 Solution description | |
1109 | * | |
1110 | * This new text clearly specifies to an implementor the need | |
1111 | * to look within the INIT or INIT-ACK. Any implementation that | |
1112 | * does not do this, may not be able to establish associations | |
1113 | * in certain circumstances. | |
1114 | * | |
1115 | */ | |
4110cc25 EB |
1116 | static struct sctp_association *__sctp_rcv_init_lookup(struct net *net, |
1117 | struct sk_buff *skb, | |
1da177e4 LT |
1118 | const union sctp_addr *laddr, struct sctp_transport **transportp) |
1119 | { | |
1120 | struct sctp_association *asoc; | |
1121 | union sctp_addr addr; | |
1122 | union sctp_addr *paddr = &addr; | |
2c0fd387 | 1123 | struct sctphdr *sh = sctp_hdr(skb); |
1da177e4 | 1124 | union sctp_params params; |
01a992be | 1125 | struct sctp_init_chunk *init; |
1da177e4 LT |
1126 | struct sctp_af *af; |
1127 | ||
1da177e4 LT |
1128 | /* |
1129 | * This code will NOT touch anything inside the chunk--it is | |
1130 | * strictly READ-ONLY. | |
1131 | * | |
1132 | * RFC 2960 3 SCTP packet Format | |
1133 | * | |
1134 | * Multiple chunks can be bundled into one SCTP packet up to | |
1135 | * the MTU size, except for the INIT, INIT ACK, and SHUTDOWN | |
1136 | * COMPLETE chunks. These chunks MUST NOT be bundled with any | |
1137 | * other chunk in a packet. See Section 6.10 for more details | |
1138 | * on chunk bundling. | |
1139 | */ | |
1140 | ||
1141 | /* Find the start of the TLVs and the end of the chunk. This is | |
1142 | * the region we search for address parameters. | |
1143 | */ | |
01a992be | 1144 | init = (struct sctp_init_chunk *)skb->data; |
1da177e4 LT |
1145 | |
1146 | /* Walk the parameters looking for embedded addresses. */ | |
1147 | sctp_walk_params(params, init, init_hdr.params) { | |
1148 | ||
1149 | /* Note: Ignoring hostname addresses. */ | |
1150 | af = sctp_get_af_specific(param_type2af(params.p->type)); | |
1151 | if (!af) | |
1152 | continue; | |
1153 | ||
0c5dc070 MRL |
1154 | if (!af->from_addr_param(paddr, params.addr, sh->source, 0)) |
1155 | continue; | |
1da177e4 | 1156 | |
7c17fcc7 | 1157 | asoc = __sctp_lookup_association(net, laddr, paddr, transportp); |
1da177e4 LT |
1158 | if (asoc) |
1159 | return asoc; | |
1160 | } | |
1161 | ||
1162 | return NULL; | |
1163 | } | |
1164 | ||
df218577 VY |
1165 | /* ADD-IP, Section 5.2 |
1166 | * When an endpoint receives an ASCONF Chunk from the remote peer | |
1167 | * special procedures may be needed to identify the association the | |
1168 | * ASCONF Chunk is associated with. To properly find the association | |
1169 | * the following procedures SHOULD be followed: | |
1170 | * | |
1171 | * D2) If the association is not found, use the address found in the | |
1172 | * Address Parameter TLV combined with the port number found in the | |
1173 | * SCTP common header. If found proceed to rule D4. | |
1174 | * | |
1175 | * D2-ext) If more than one ASCONF Chunks are packed together, use the | |
1176 | * address found in the ASCONF Address Parameter TLV of each of the | |
1177 | * subsequent ASCONF Chunks. If found, proceed to rule D4. | |
1178 | */ | |
1179 | static struct sctp_association *__sctp_rcv_asconf_lookup( | |
4110cc25 | 1180 | struct net *net, |
922dbc5b | 1181 | struct sctp_chunkhdr *ch, |
df218577 | 1182 | const union sctp_addr *laddr, |
bc92dd19 | 1183 | __be16 peer_port, |
df218577 VY |
1184 | struct sctp_transport **transportp) |
1185 | { | |
68d75469 | 1186 | struct sctp_addip_chunk *asconf = (struct sctp_addip_chunk *)ch; |
df218577 VY |
1187 | struct sctp_af *af; |
1188 | union sctp_addr_param *param; | |
1189 | union sctp_addr paddr; | |
1190 | ||
b6ffe767 MRL |
1191 | if (ntohs(ch->length) < sizeof(*asconf) + sizeof(struct sctp_paramhdr)) |
1192 | return NULL; | |
1193 | ||
df218577 VY |
1194 | /* Skip over the ADDIP header and find the Address parameter */ |
1195 | param = (union sctp_addr_param *)(asconf + 1); | |
1196 | ||
6a435732 | 1197 | af = sctp_get_af_specific(param_type2af(param->p.type)); |
df218577 VY |
1198 | if (unlikely(!af)) |
1199 | return NULL; | |
1200 | ||
557fb586 | 1201 | if (!af->from_addr_param(&paddr, param, peer_port, 0)) |
0c5dc070 | 1202 | return NULL; |
df218577 | 1203 | |
4110cc25 | 1204 | return __sctp_lookup_association(net, laddr, &paddr, transportp); |
df218577 VY |
1205 | } |
1206 | ||
1207 | ||
bbd0d598 VY |
1208 | /* SCTP-AUTH, Section 6.3: |
1209 | * If the receiver does not find a STCB for a packet containing an AUTH | |
1210 | * chunk as the first chunk and not a COOKIE-ECHO chunk as the second | |
1211 | * chunk, it MUST use the chunks after the AUTH chunk to look up an existing | |
1212 | * association. | |
1213 | * | |
1214 | * This means that any chunks that can help us identify the association need | |
25985edc | 1215 | * to be looked at to find this association. |
bbd0d598 | 1216 | */ |
4110cc25 EB |
1217 | static struct sctp_association *__sctp_rcv_walk_lookup(struct net *net, |
1218 | struct sk_buff *skb, | |
bbd0d598 VY |
1219 | const union sctp_addr *laddr, |
1220 | struct sctp_transport **transportp) | |
1221 | { | |
df218577 | 1222 | struct sctp_association *asoc = NULL; |
922dbc5b | 1223 | struct sctp_chunkhdr *ch; |
df218577 VY |
1224 | int have_auth = 0; |
1225 | unsigned int chunk_num = 1; | |
1226 | __u8 *ch_end; | |
1227 | ||
1228 | /* Walk through the chunks looking for AUTH or ASCONF chunks | |
1229 | * to help us find the association. | |
bbd0d598 | 1230 | */ |
922dbc5b | 1231 | ch = (struct sctp_chunkhdr *)skb->data; |
df218577 VY |
1232 | do { |
1233 | /* Break out if chunk length is less then minimal. */ | |
922dbc5b | 1234 | if (ntohs(ch->length) < sizeof(*ch)) |
df218577 VY |
1235 | break; |
1236 | ||
e2f036a9 | 1237 | ch_end = ((__u8 *)ch) + SCTP_PAD4(ntohs(ch->length)); |
df218577 VY |
1238 | if (ch_end > skb_tail_pointer(skb)) |
1239 | break; | |
1240 | ||
cb3f837b | 1241 | switch (ch->type) { |
f7010e61 | 1242 | case SCTP_CID_AUTH: |
1243 | have_auth = chunk_num; | |
1244 | break; | |
1245 | ||
1246 | case SCTP_CID_COOKIE_ECHO: | |
1247 | /* If a packet arrives containing an AUTH chunk as | |
1248 | * a first chunk, a COOKIE-ECHO chunk as the second | |
1249 | * chunk, and possibly more chunks after them, and | |
1250 | * the receiver does not have an STCB for that | |
1251 | * packet, then authentication is based on | |
1252 | * the contents of the COOKIE- ECHO chunk. | |
1253 | */ | |
1254 | if (have_auth == 1 && chunk_num == 2) | |
1255 | return NULL; | |
1256 | break; | |
1257 | ||
1258 | case SCTP_CID_ASCONF: | |
1259 | if (have_auth || net->sctp.addip_noauth) | |
1260 | asoc = __sctp_rcv_asconf_lookup( | |
1261 | net, ch, laddr, | |
1262 | sctp_hdr(skb)->source, | |
1263 | transportp); | |
0572b37b | 1264 | break; |
f7010e61 | 1265 | default: |
1266 | break; | |
df218577 VY |
1267 | } |
1268 | ||
1269 | if (asoc) | |
1270 | break; | |
1271 | ||
922dbc5b | 1272 | ch = (struct sctp_chunkhdr *)ch_end; |
df218577 | 1273 | chunk_num++; |
50619dbf | 1274 | } while (ch_end + sizeof(*ch) < skb_tail_pointer(skb)); |
df218577 VY |
1275 | |
1276 | return asoc; | |
bbd0d598 VY |
1277 | } |
1278 | ||
1279 | /* | |
1280 | * There are circumstances when we need to look inside the SCTP packet | |
1281 | * for information to help us find the association. Examples | |
1282 | * include looking inside of INIT/INIT-ACK chunks or after the AUTH | |
1283 | * chunks. | |
1284 | */ | |
4110cc25 EB |
1285 | static struct sctp_association *__sctp_rcv_lookup_harder(struct net *net, |
1286 | struct sk_buff *skb, | |
bbd0d598 VY |
1287 | const union sctp_addr *laddr, |
1288 | struct sctp_transport **transportp) | |
1289 | { | |
922dbc5b | 1290 | struct sctp_chunkhdr *ch; |
bbd0d598 | 1291 | |
90017acc MRL |
1292 | /* We do not allow GSO frames here as we need to linearize and |
1293 | * then cannot guarantee frame boundaries. This shouldn't be an | |
1294 | * issue as packets hitting this are mostly INIT or INIT-ACK and | |
1295 | * those cannot be on GSO-style anyway. | |
1296 | */ | |
1dd27cde | 1297 | if (skb_is_gso(skb) && skb_is_gso_sctp(skb)) |
90017acc MRL |
1298 | return NULL; |
1299 | ||
922dbc5b | 1300 | ch = (struct sctp_chunkhdr *)skb->data; |
bbd0d598 | 1301 | |
df218577 VY |
1302 | /* The code below will attempt to walk the chunk and extract |
1303 | * parameter information. Before we do that, we need to verify | |
1304 | * that the chunk length doesn't cause overflow. Otherwise, we'll | |
1305 | * walk off the end. | |
1306 | */ | |
e2f036a9 | 1307 | if (SCTP_PAD4(ntohs(ch->length)) > skb->len) |
df218577 VY |
1308 | return NULL; |
1309 | ||
bbd0d598 | 1310 | /* If this is INIT/INIT-ACK look inside the chunk too. */ |
f482f2fc | 1311 | if (ch->type == SCTP_CID_INIT || ch->type == SCTP_CID_INIT_ACK) |
4110cc25 | 1312 | return __sctp_rcv_init_lookup(net, skb, laddr, transportp); |
bbd0d598 | 1313 | |
f482f2fc | 1314 | return __sctp_rcv_walk_lookup(net, skb, laddr, transportp); |
bbd0d598 VY |
1315 | } |
1316 | ||
1da177e4 | 1317 | /* Lookup an association for an inbound skb. */ |
4110cc25 EB |
1318 | static struct sctp_association *__sctp_rcv_lookup(struct net *net, |
1319 | struct sk_buff *skb, | |
1da177e4 LT |
1320 | const union sctp_addr *paddr, |
1321 | const union sctp_addr *laddr, | |
1322 | struct sctp_transport **transportp) | |
1323 | { | |
1324 | struct sctp_association *asoc; | |
1325 | ||
4110cc25 | 1326 | asoc = __sctp_lookup_association(net, laddr, paddr, transportp); |
b77b7565 MRL |
1327 | if (asoc) |
1328 | goto out; | |
1da177e4 LT |
1329 | |
1330 | /* Further lookup for INIT/INIT-ACK packets. | |
1331 | * SCTP Implementors Guide, 2.18 Handling of address | |
1332 | * parameters within the INIT or INIT-ACK. | |
1333 | */ | |
b77b7565 MRL |
1334 | asoc = __sctp_rcv_lookup_harder(net, skb, laddr, transportp); |
1335 | if (asoc) | |
1336 | goto out; | |
1da177e4 | 1337 | |
b77b7565 MRL |
1338 | if (paddr->sa.sa_family == AF_INET) |
1339 | pr_debug("sctp: asoc not found for src:%pI4:%d dst:%pI4:%d\n", | |
1340 | &laddr->v4.sin_addr, ntohs(laddr->v4.sin_port), | |
1341 | &paddr->v4.sin_addr, ntohs(paddr->v4.sin_port)); | |
1342 | else | |
1343 | pr_debug("sctp: asoc not found for src:%pI6:%d dst:%pI6:%d\n", | |
1344 | &laddr->v6.sin6_addr, ntohs(laddr->v6.sin6_port), | |
1345 | &paddr->v6.sin6_addr, ntohs(paddr->v6.sin6_port)); | |
1346 | ||
1347 | out: | |
1da177e4 LT |
1348 | return asoc; |
1349 | } |