Commit | Line | Data |
---|---|---|
60c778b2 | 1 | /* SCTP kernel implementation |
1da177e4 LT |
2 | * Copyright (c) 1999-2000 Cisco, Inc. |
3 | * Copyright (c) 1999-2001 Motorola, Inc. | |
4 | * Copyright (c) 2001-2003 International Business Machines, Corp. | |
5 | * Copyright (c) 2001 Intel Corp. | |
6 | * Copyright (c) 2001 Nokia, Inc. | |
7 | * Copyright (c) 2001 La Monte H.P. Yarroll | |
8 | * | |
60c778b2 | 9 | * This file is part of the SCTP kernel implementation |
1da177e4 LT |
10 | * |
11 | * These functions handle all input from the IP layer into SCTP. | |
12 | * | |
60c778b2 | 13 | * This SCTP implementation is free software; |
1da177e4 LT |
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 | * | |
60c778b2 | 19 | * This SCTP implementation is distributed in the hope that it |
1da177e4 LT |
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 | |
4b2f13a2 JK |
26 | * along with GNU CC; see the file COPYING. If not, see |
27 | * <http://www.gnu.org/licenses/>. | |
1da177e4 LT |
28 | * |
29 | * Please send any bug reports or fixes you make to the | |
30 | * email address(es): | |
91705c61 | 31 | * lksctp developers <linux-sctp@vger.kernel.org> |
1da177e4 | 32 | * |
1da177e4 LT |
33 | * Written or modified by: |
34 | * La Monte H.P. Yarroll <piggy@acm.org> | |
35 | * Karl Knutson <karl@athena.chicago.il.us> | |
36 | * Xingang Guo <xingang.guo@intel.com> | |
37 | * Jon Grimm <jgrimm@us.ibm.com> | |
38 | * Hui Huang <hui.huang@nokia.com> | |
39 | * Daisy Chang <daisyc@us.ibm.com> | |
40 | * Sridhar Samudrala <sri@us.ibm.com> | |
41 | * Ardelle Fan <ardelle.fan@intel.com> | |
1da177e4 LT |
42 | */ |
43 | ||
44 | #include <linux/types.h> | |
45 | #include <linux/list.h> /* For struct list_head */ | |
46 | #include <linux/socket.h> | |
47 | #include <linux/ip.h> | |
48 | #include <linux/time.h> /* For struct timeval */ | |
5a0e3ad6 | 49 | #include <linux/slab.h> |
1da177e4 LT |
50 | #include <net/ip.h> |
51 | #include <net/icmp.h> | |
52 | #include <net/snmp.h> | |
53 | #include <net/sock.h> | |
54 | #include <net/xfrm.h> | |
55 | #include <net/sctp/sctp.h> | |
56 | #include <net/sctp/sm.h> | |
9ad0977f | 57 | #include <net/sctp/checksum.h> |
dcfc23ca | 58 | #include <net/net_namespace.h> |
0eb71a9d | 59 | #include <linux/rhashtable.h> |
532ae2f1 | 60 | #include <net/sock_reuseport.h> |
1da177e4 LT |
61 | |
62 | /* Forward declarations for internal helpers. */ | |
63 | static int sctp_rcv_ootb(struct sk_buff *); | |
4110cc25 EB |
64 | static struct sctp_association *__sctp_rcv_lookup(struct net *net, |
65 | struct sk_buff *skb, | |
1da177e4 | 66 | const union sctp_addr *paddr, |
57565993 | 67 | const union sctp_addr *laddr, |
1da177e4 | 68 | struct sctp_transport **transportp); |
532ae2f1 XL |
69 | static struct sctp_endpoint *__sctp_rcv_lookup_endpoint( |
70 | struct net *net, struct sk_buff *skb, | |
71 | const union sctp_addr *laddr, | |
72 | const union sctp_addr *daddr); | |
1da177e4 | 73 | static struct sctp_association *__sctp_lookup_association( |
4110cc25 | 74 | struct net *net, |
1da177e4 LT |
75 | const union sctp_addr *local, |
76 | const union sctp_addr *peer, | |
77 | struct sctp_transport **pt); | |
78 | ||
50b1a782 | 79 | static int sctp_add_backlog(struct sock *sk, struct sk_buff *skb); |
61c9fed4 | 80 | |
1da177e4 LT |
81 | |
82 | /* Calculate the SCTP checksum of an SCTP packet. */ | |
b01a2407 | 83 | static inline int sctp_rcv_checksum(struct net *net, struct sk_buff *skb) |
1da177e4 | 84 | { |
2c0fd387 | 85 | struct sctphdr *sh = sctp_hdr(skb); |
4458f04c | 86 | __le32 cmp = sh->checksum; |
024ec3de | 87 | __le32 val = sctp_compute_cksum(skb, 0); |
1da177e4 LT |
88 | |
89 | if (val != cmp) { | |
90 | /* CRC failure, dump it. */ | |
08e3baef | 91 | __SCTP_INC_STATS(net, SCTP_MIB_CHECKSUMERRORS); |
1da177e4 LT |
92 | return -1; |
93 | } | |
94 | return 0; | |
95 | } | |
96 | ||
1da177e4 LT |
97 | /* |
98 | * This is the routine which IP calls when receiving an SCTP packet. | |
99 | */ | |
100 | int sctp_rcv(struct sk_buff *skb) | |
101 | { | |
102 | struct sock *sk; | |
103 | struct sctp_association *asoc; | |
104 | struct sctp_endpoint *ep = NULL; | |
105 | struct sctp_ep_common *rcvr; | |
106 | struct sctp_transport *transport = NULL; | |
107 | struct sctp_chunk *chunk; | |
1da177e4 LT |
108 | union sctp_addr src; |
109 | union sctp_addr dest; | |
110 | int family; | |
111 | struct sctp_af *af; | |
4cdadcbc | 112 | struct net *net = dev_net(skb->dev); |
1dd27cde | 113 | bool is_gso = skb_is_gso(skb) && skb_is_gso_sctp(skb); |
1da177e4 | 114 | |
cb3f837b | 115 | if (skb->pkt_type != PACKET_HOST) |
1da177e4 LT |
116 | goto discard_it; |
117 | ||
08e3baef | 118 | __SCTP_INC_STATS(net, SCTP_MIB_INSCTPPACKS); |
1da177e4 | 119 | |
3acb50c1 MRL |
120 | /* If packet is too small to contain a single chunk, let's not |
121 | * waste time on it anymore. | |
122 | */ | |
123 | if (skb->len < sizeof(struct sctphdr) + sizeof(struct sctp_chunkhdr) + | |
124 | skb_transport_offset(skb)) | |
28cd7752 HX |
125 | goto discard_it; |
126 | ||
4c2f2454 MRL |
127 | /* If the packet is fragmented and we need to do crc checking, |
128 | * it's better to just linearize it otherwise crc computing | |
129 | * takes longer. | |
130 | */ | |
1dd27cde | 131 | if ((!is_gso && skb_linearize(skb)) || |
4c2f2454 | 132 | !pskb_may_pull(skb, sizeof(struct sctphdr))) |
3acb50c1 | 133 | goto discard_it; |
1da177e4 | 134 | |
3acb50c1 | 135 | /* Pull up the IP header. */ |
ea2ae17d | 136 | __skb_pull(skb, skb_transport_offset(skb)); |
202863fe TH |
137 | |
138 | skb->csum_valid = 0; /* Previous value not applicable */ | |
139 | if (skb_csum_unnecessary(skb)) | |
140 | __skb_decr_checksum_unnecessary(skb); | |
90017acc | 141 | else if (!sctp_checksum_disable && |
1dd27cde | 142 | !is_gso && |
90017acc | 143 | sctp_rcv_checksum(net, skb) < 0) |
1da177e4 | 144 | goto discard_it; |
202863fe | 145 | skb->csum_valid = 1; |
1da177e4 | 146 | |
3acb50c1 | 147 | __skb_pull(skb, sizeof(struct sctphdr)); |
1da177e4 | 148 | |
eddc9ec5 | 149 | family = ipver2af(ip_hdr(skb)->version); |
1da177e4 LT |
150 | af = sctp_get_af_specific(family); |
151 | if (unlikely(!af)) | |
152 | goto discard_it; | |
e7487c86 | 153 | SCTP_INPUT_CB(skb)->af = af; |
1da177e4 LT |
154 | |
155 | /* Initialize local addresses for lookups. */ | |
156 | af->from_skb(&src, skb, 1); | |
157 | af->from_skb(&dest, skb, 0); | |
158 | ||
159 | /* If the packet is to or from a non-unicast address, | |
160 | * silently discard the packet. | |
161 | * | |
162 | * This is not clearly defined in the RFC except in section | |
163 | * 8.4 - OOTB handling. However, based on the book "Stream Control | |
164 | * Transmission Protocol" 2.1, "It is important to note that the | |
165 | * IP address of an SCTP transport address must be a routable | |
166 | * unicast address. In other words, IP multicast addresses and | |
167 | * IP broadcast addresses cannot be used in an SCTP transport | |
168 | * address." | |
169 | */ | |
5636bef7 VY |
170 | if (!af->addr_valid(&src, NULL, skb) || |
171 | !af->addr_valid(&dest, NULL, skb)) | |
1da177e4 LT |
172 | goto discard_it; |
173 | ||
4110cc25 | 174 | asoc = __sctp_rcv_lookup(net, skb, &src, &dest, &transport); |
1c7d1fc1 | 175 | |
0fd9a65a | 176 | if (!asoc) |
532ae2f1 | 177 | ep = __sctp_rcv_lookup_endpoint(net, skb, &dest, &src); |
0fd9a65a NH |
178 | |
179 | /* Retrieve the common input handling substructure. */ | |
180 | rcvr = asoc ? &asoc->base : &ep->base; | |
181 | sk = rcvr->sk; | |
182 | ||
183 | /* | |
184 | * If a frame arrives on an interface and the receiving socket is | |
185 | * bound to another interface, via SO_BINDTODEVICE, treat it as OOTB | |
186 | */ | |
8d72651d | 187 | if (sk->sk_bound_dev_if && (sk->sk_bound_dev_if != af->skb_iif(skb))) { |
dae399d7 XL |
188 | if (transport) { |
189 | sctp_transport_put(transport); | |
0fd9a65a | 190 | asoc = NULL; |
dae399d7 | 191 | transport = NULL; |
0fd9a65a NH |
192 | } else { |
193 | sctp_endpoint_put(ep); | |
194 | ep = NULL; | |
195 | } | |
2ce95503 | 196 | sk = net->sctp.ctl_sock; |
0fd9a65a NH |
197 | ep = sctp_sk(sk)->ep; |
198 | sctp_endpoint_hold(ep); | |
0fd9a65a NH |
199 | rcvr = &ep->base; |
200 | } | |
201 | ||
1da177e4 LT |
202 | /* |
203 | * RFC 2960, 8.4 - Handle "Out of the blue" Packets. | |
204 | * An SCTP packet is called an "out of the blue" (OOTB) | |
205 | * packet if it is correctly formed, i.e., passed the | |
206 | * receiver's checksum check, but the receiver is not | |
207 | * able to identify the association to which this | |
208 | * packet belongs. | |
209 | */ | |
210 | if (!asoc) { | |
1da177e4 | 211 | if (sctp_rcv_ootb(skb)) { |
08e3baef | 212 | __SCTP_INC_STATS(net, SCTP_MIB_OUTOFBLUES); |
1da177e4 LT |
213 | goto discard_release; |
214 | } | |
215 | } | |
216 | ||
1da177e4 LT |
217 | if (!xfrm_policy_check(sk, XFRM_POLICY_IN, skb, family)) |
218 | goto discard_release; | |
b59c2701 | 219 | nf_reset(skb); |
1da177e4 | 220 | |
fda9ef5d | 221 | if (sk_filter(sk, skb)) |
d808ad9a | 222 | goto discard_release; |
1da177e4 LT |
223 | |
224 | /* Create an SCTP packet structure. */ | |
cea8768f | 225 | chunk = sctp_chunkify(skb, asoc, sk, GFP_ATOMIC); |
2babf9da | 226 | if (!chunk) |
1da177e4 | 227 | goto discard_release; |
79af02c2 | 228 | SCTP_INPUT_CB(skb)->chunk = chunk; |
1da177e4 | 229 | |
1da177e4 LT |
230 | /* Remember what endpoint is to handle this packet. */ |
231 | chunk->rcvr = rcvr; | |
232 | ||
233 | /* Remember the SCTP header. */ | |
3acb50c1 | 234 | chunk->sctp_hdr = sctp_hdr(skb); |
1da177e4 LT |
235 | |
236 | /* Set the source and destination addresses of the incoming chunk. */ | |
d55c41b1 | 237 | sctp_init_addrs(chunk, &src, &dest); |
1da177e4 LT |
238 | |
239 | /* Remember where we came from. */ | |
240 | chunk->transport = transport; | |
241 | ||
242 | /* Acquire access to the sock lock. Note: We are safe from other | |
243 | * bottom halves on this lock, but a user may be in the lock too, | |
244 | * so check if it is busy. | |
245 | */ | |
5bc1d1b4 | 246 | bh_lock_sock(sk); |
1da177e4 | 247 | |
ae53b5bd VY |
248 | if (sk != rcvr->sk) { |
249 | /* Our cached sk is different from the rcvr->sk. This is | |
250 | * because migrate()/accept() may have moved the association | |
251 | * to a new socket and released all the sockets. So now we | |
252 | * are holding a lock on the old socket while the user may | |
253 | * be doing something with the new socket. Switch our veiw | |
254 | * of the current sk. | |
255 | */ | |
5bc1d1b4 | 256 | bh_unlock_sock(sk); |
ae53b5bd | 257 | sk = rcvr->sk; |
5bc1d1b4 | 258 | bh_lock_sock(sk); |
ae53b5bd VY |
259 | } |
260 | ||
ac0b0462 | 261 | if (sock_owned_by_user(sk)) { |
50b1a782 | 262 | if (sctp_add_backlog(sk, skb)) { |
5bc1d1b4 | 263 | bh_unlock_sock(sk); |
50b1a782 ZY |
264 | sctp_chunk_free(chunk); |
265 | skb = NULL; /* sctp_chunk_free already freed the skb */ | |
266 | goto discard_release; | |
267 | } | |
08e3baef | 268 | __SCTP_INC_STATS(net, SCTP_MIB_IN_PKT_BACKLOG); |
ac0b0462 | 269 | } else { |
08e3baef | 270 | __SCTP_INC_STATS(net, SCTP_MIB_IN_PKT_SOFTIRQ); |
61c9fed4 | 271 | sctp_inq_push(&chunk->rcvr->inqueue, chunk); |
ac0b0462 | 272 | } |
1da177e4 | 273 | |
5bc1d1b4 | 274 | bh_unlock_sock(sk); |
61c9fed4 VY |
275 | |
276 | /* Release the asoc/ep ref we took in the lookup calls. */ | |
dae399d7 XL |
277 | if (transport) |
278 | sctp_transport_put(transport); | |
61c9fed4 VY |
279 | else |
280 | sctp_endpoint_put(ep); | |
7a48f923 | 281 | |
2babf9da | 282 | return 0; |
1da177e4 LT |
283 | |
284 | discard_it: | |
08e3baef | 285 | __SCTP_INC_STATS(net, SCTP_MIB_IN_PKT_DISCARDS); |
1da177e4 | 286 | kfree_skb(skb); |
2babf9da | 287 | return 0; |
1da177e4 LT |
288 | |
289 | discard_release: | |
61c9fed4 | 290 | /* Release the asoc/ep ref we took in the lookup calls. */ |
dae399d7 XL |
291 | if (transport) |
292 | sctp_transport_put(transport); | |
0fd9a65a | 293 | else |
1da177e4 | 294 | sctp_endpoint_put(ep); |
1da177e4 LT |
295 | |
296 | goto discard_it; | |
297 | } | |
298 | ||
61c9fed4 VY |
299 | /* Process the backlog queue of the socket. Every skb on |
300 | * the backlog holds a ref on an association or endpoint. | |
301 | * We hold this ref throughout the state machine to make | |
302 | * sure that the structure we need is still around. | |
1da177e4 LT |
303 | */ |
304 | int sctp_backlog_rcv(struct sock *sk, struct sk_buff *skb) | |
305 | { | |
79af02c2 | 306 | struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk; |
d808ad9a | 307 | struct sctp_inq *inqueue = &chunk->rcvr->inqueue; |
dae399d7 | 308 | struct sctp_transport *t = chunk->transport; |
d808ad9a | 309 | struct sctp_ep_common *rcvr = NULL; |
61c9fed4 | 310 | int backloged = 0; |
7a48f923 | 311 | |
d808ad9a | 312 | rcvr = chunk->rcvr; |
c4d2444e | 313 | |
61c9fed4 VY |
314 | /* If the rcvr is dead then the association or endpoint |
315 | * has been deleted and we can safely drop the chunk | |
316 | * and refs that we are holding. | |
317 | */ | |
318 | if (rcvr->dead) { | |
319 | sctp_chunk_free(chunk); | |
320 | goto done; | |
321 | } | |
322 | ||
323 | if (unlikely(rcvr->sk != sk)) { | |
324 | /* In this case, the association moved from one socket to | |
325 | * another. We are currently sitting on the backlog of the | |
326 | * old socket, so we need to move. | |
327 | * However, since we are here in the process context we | |
328 | * need to take make sure that the user doesn't own | |
329 | * the new socket when we process the packet. | |
330 | * If the new socket is user-owned, queue the chunk to the | |
331 | * backlog of the new socket without dropping any refs. | |
332 | * Otherwise, we can safely push the chunk on the inqueue. | |
333 | */ | |
334 | ||
335 | sk = rcvr->sk; | |
eefc1b1d | 336 | local_bh_disable(); |
5bc1d1b4 | 337 | bh_lock_sock(sk); |
61c9fed4 VY |
338 | |
339 | if (sock_owned_by_user(sk)) { | |
f545a38f | 340 | if (sk_add_backlog(sk, skb, sk->sk_rcvbuf)) |
50b1a782 ZY |
341 | sctp_chunk_free(chunk); |
342 | else | |
343 | backloged = 1; | |
61c9fed4 VY |
344 | } else |
345 | sctp_inq_push(inqueue, chunk); | |
346 | ||
5bc1d1b4 | 347 | bh_unlock_sock(sk); |
eefc1b1d | 348 | local_bh_enable(); |
61c9fed4 VY |
349 | |
350 | /* If the chunk was backloged again, don't drop refs */ | |
351 | if (backloged) | |
352 | return 0; | |
353 | } else { | |
354 | sctp_inq_push(inqueue, chunk); | |
355 | } | |
356 | ||
357 | done: | |
358 | /* Release the refs we took in sctp_add_backlog */ | |
359 | if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type) | |
dae399d7 | 360 | sctp_transport_put(t); |
61c9fed4 VY |
361 | else if (SCTP_EP_TYPE_SOCKET == rcvr->type) |
362 | sctp_endpoint_put(sctp_ep(rcvr)); | |
363 | else | |
364 | BUG(); | |
365 | ||
d808ad9a | 366 | return 0; |
1da177e4 LT |
367 | } |
368 | ||
50b1a782 | 369 | static int sctp_add_backlog(struct sock *sk, struct sk_buff *skb) |
c4d2444e | 370 | { |
61c9fed4 | 371 | struct sctp_chunk *chunk = SCTP_INPUT_CB(skb)->chunk; |
dae399d7 | 372 | struct sctp_transport *t = chunk->transport; |
61c9fed4 | 373 | struct sctp_ep_common *rcvr = chunk->rcvr; |
50b1a782 | 374 | int ret; |
c4d2444e | 375 | |
f545a38f | 376 | ret = sk_add_backlog(sk, skb, sk->sk_rcvbuf); |
50b1a782 ZY |
377 | if (!ret) { |
378 | /* Hold the assoc/ep while hanging on the backlog queue. | |
379 | * This way, we know structures we need will not disappear | |
380 | * from us | |
381 | */ | |
382 | if (SCTP_EP_TYPE_ASSOCIATION == rcvr->type) | |
dae399d7 | 383 | sctp_transport_hold(t); |
50b1a782 ZY |
384 | else if (SCTP_EP_TYPE_SOCKET == rcvr->type) |
385 | sctp_endpoint_hold(sctp_ep(rcvr)); | |
386 | else | |
387 | BUG(); | |
388 | } | |
389 | return ret; | |
61c9fed4 | 390 | |
c4d2444e SS |
391 | } |
392 | ||
1da177e4 LT |
393 | /* Handle icmp frag needed error. */ |
394 | void sctp_icmp_frag_needed(struct sock *sk, struct sctp_association *asoc, | |
395 | struct sctp_transport *t, __u32 pmtu) | |
396 | { | |
91bd6b1e | 397 | if (!t || (t->pathmtu <= pmtu)) |
52ccb8e9 FF |
398 | return; |
399 | ||
8a479491 | 400 | if (sock_owned_by_user(sk)) { |
d805397c | 401 | atomic_set(&t->mtu_info, pmtu); |
8a479491 VY |
402 | asoc->pmtu_pending = 1; |
403 | t->pmtu_pending = 1; | |
404 | return; | |
405 | } | |
406 | ||
cc35c3d1 MRL |
407 | if (!(t->param_flags & SPP_PMTUD_ENABLE)) |
408 | /* We can't allow retransmitting in such case, as the | |
409 | * retransmission would be sized just as before, and thus we | |
410 | * would get another icmp, and retransmit again. | |
411 | */ | |
412 | return; | |
1da177e4 | 413 | |
b6c5734d MRL |
414 | /* Update transports view of the MTU. Return if no update was needed. |
415 | * If an update wasn't needed/possible, it also doesn't make sense to | |
416 | * try to retransmit now. | |
417 | */ | |
418 | if (!sctp_transport_update_pmtu(t, pmtu)) | |
419 | return; | |
52ccb8e9 | 420 | |
cc35c3d1 MRL |
421 | /* Update association pmtu. */ |
422 | sctp_assoc_sync_pmtu(asoc); | |
423 | ||
424 | /* Retransmit with the new pmtu setting. */ | |
52ccb8e9 | 425 | sctp_retransmit(&asoc->outqueue, t, SCTP_RTXR_PMTUD); |
1da177e4 LT |
426 | } |
427 | ||
ec18d9a2 DM |
428 | void sctp_icmp_redirect(struct sock *sk, struct sctp_transport *t, |
429 | struct sk_buff *skb) | |
55be7a9c DM |
430 | { |
431 | struct dst_entry *dst; | |
432 | ||
1cc276ce | 433 | if (sock_owned_by_user(sk) || !t) |
55be7a9c DM |
434 | return; |
435 | dst = sctp_transport_dst_check(t); | |
1ed5c48f | 436 | if (dst) |
6700c270 | 437 | dst->ops->redirect(dst, sk, skb); |
55be7a9c DM |
438 | } |
439 | ||
1da177e4 LT |
440 | /* |
441 | * SCTP Implementer's Guide, 2.37 ICMP handling procedures | |
442 | * | |
443 | * ICMP8) If the ICMP code is a "Unrecognized next header type encountered" | |
444 | * or a "Protocol Unreachable" treat this message as an abort | |
445 | * with the T bit set. | |
446 | * | |
447 | * This function sends an event to the state machine, which will abort the | |
448 | * association. | |
449 | * | |
450 | */ | |
451 | void sctp_icmp_proto_unreachable(struct sock *sk, | |
d808ad9a YH |
452 | struct sctp_association *asoc, |
453 | struct sctp_transport *t) | |
1da177e4 | 454 | { |
50b5d6ad VY |
455 | if (sock_owned_by_user(sk)) { |
456 | if (timer_pending(&t->proto_unreach_timer)) | |
457 | return; | |
458 | else { | |
459 | if (!mod_timer(&t->proto_unreach_timer, | |
460 | jiffies + (HZ/20))) | |
461 | sctp_association_hold(asoc); | |
462 | } | |
50b5d6ad | 463 | } else { |
55e26eb9 EB |
464 | struct net *net = sock_net(sk); |
465 | ||
bb33381d DB |
466 | pr_debug("%s: unrecognized next header type " |
467 | "encountered!\n", __func__); | |
468 | ||
25cc4ae9 | 469 | if (del_timer(&t->proto_unreach_timer)) |
50b5d6ad | 470 | sctp_association_put(asoc); |
1da177e4 | 471 | |
55e26eb9 | 472 | sctp_do_sm(net, SCTP_EVENT_T_OTHER, |
50b5d6ad VY |
473 | SCTP_ST_OTHER(SCTP_EVENT_ICMP_PROTO_UNREACH), |
474 | asoc->state, asoc->ep, asoc, t, | |
475 | GFP_ATOMIC); | |
476 | } | |
1da177e4 LT |
477 | } |
478 | ||
479 | /* Common lookup code for icmp/icmpv6 error handler. */ | |
4110cc25 | 480 | struct sock *sctp_err_lookup(struct net *net, int family, struct sk_buff *skb, |
1da177e4 | 481 | struct sctphdr *sctphdr, |
1da177e4 LT |
482 | struct sctp_association **app, |
483 | struct sctp_transport **tpp) | |
484 | { | |
804ec7eb | 485 | struct sctp_init_chunk *chunkhdr, _chunkhdr; |
1da177e4 LT |
486 | union sctp_addr saddr; |
487 | union sctp_addr daddr; | |
488 | struct sctp_af *af; | |
489 | struct sock *sk = NULL; | |
8de8c873 | 490 | struct sctp_association *asoc; |
1da177e4 | 491 | struct sctp_transport *transport = NULL; |
7115e632 | 492 | __u32 vtag = ntohl(sctphdr->vtag); |
1da177e4 | 493 | |
d1ad1ff2 | 494 | *app = NULL; *tpp = NULL; |
1da177e4 LT |
495 | |
496 | af = sctp_get_af_specific(family); | |
497 | if (unlikely(!af)) { | |
498 | return NULL; | |
499 | } | |
500 | ||
501 | /* Initialize local addresses for lookups. */ | |
502 | af->from_skb(&saddr, skb, 1); | |
503 | af->from_skb(&daddr, skb, 0); | |
504 | ||
505 | /* Look for an association that matches the incoming ICMP error | |
506 | * packet. | |
507 | */ | |
4110cc25 | 508 | asoc = __sctp_lookup_association(net, &saddr, &daddr, &transport); |
d1ad1ff2 SS |
509 | if (!asoc) |
510 | return NULL; | |
1da177e4 | 511 | |
d1ad1ff2 | 512 | sk = asoc->base.sk; |
1da177e4 | 513 | |
7115e632 WY |
514 | /* RFC 4960, Appendix C. ICMP Handling |
515 | * | |
516 | * ICMP6) An implementation MUST validate that the Verification Tag | |
517 | * contained in the ICMP message matches the Verification Tag of | |
518 | * the peer. If the Verification Tag is not 0 and does NOT | |
519 | * match, discard the ICMP message. If it is 0 and the ICMP | |
520 | * message contains enough bytes to verify that the chunk type is | |
521 | * an INIT chunk and that the Initiate Tag matches the tag of the | |
522 | * peer, continue with ICMP7. If the ICMP message is too short | |
523 | * or the chunk type or the Initiate Tag does not match, silently | |
524 | * discard the packet. | |
525 | */ | |
526 | if (vtag == 0) { | |
804ec7eb DC |
527 | /* chunk header + first 4 octects of init header */ |
528 | chunkhdr = skb_header_pointer(skb, skb_transport_offset(skb) + | |
529 | sizeof(struct sctphdr), | |
530 | sizeof(struct sctp_chunkhdr) + | |
531 | sizeof(__be32), &_chunkhdr); | |
532 | if (!chunkhdr || | |
7115e632 | 533 | chunkhdr->chunk_hdr.type != SCTP_CID_INIT || |
804ec7eb | 534 | ntohl(chunkhdr->init_hdr.init_tag) != asoc->c.my_vtag) |
7115e632 | 535 | goto out; |
804ec7eb | 536 | |
7115e632 | 537 | } else if (vtag != asoc->c.peer_vtag) { |
d1ad1ff2 SS |
538 | goto out; |
539 | } | |
1da177e4 | 540 | |
5bc1d1b4 | 541 | bh_lock_sock(sk); |
1da177e4 LT |
542 | |
543 | /* If too many ICMPs get dropped on busy | |
544 | * servers this needs to be solved differently. | |
545 | */ | |
546 | if (sock_owned_by_user(sk)) | |
02a1d6e7 | 547 | __NET_INC_STATS(net, LINUX_MIB_LOCKDROPPEDICMPS); |
1da177e4 | 548 | |
1da177e4 LT |
549 | *app = asoc; |
550 | *tpp = transport; | |
551 | return sk; | |
552 | ||
553 | out: | |
dae399d7 | 554 | sctp_transport_put(transport); |
1da177e4 LT |
555 | return NULL; |
556 | } | |
557 | ||
558 | /* Common cleanup code for icmp/icmpv6 error handler. */ | |
dae399d7 | 559 | void sctp_err_finish(struct sock *sk, struct sctp_transport *t) |
1da177e4 | 560 | { |
5bc1d1b4 | 561 | bh_unlock_sock(sk); |
dae399d7 | 562 | sctp_transport_put(t); |
1da177e4 LT |
563 | } |
564 | ||
565 | /* | |
566 | * This routine is called by the ICMP module when it gets some | |
567 | * sort of error condition. If err < 0 then the socket should | |
568 | * be closed and the error returned to the user. If err > 0 | |
569 | * it's just the icmp type << 8 | icmp code. After adjustment | |
570 | * header points to the first 8 bytes of the sctp header. We need | |
571 | * to find the appropriate port. | |
572 | * | |
573 | * The locking strategy used here is very "optimistic". When | |
574 | * someone else accesses the socket the ICMP is just dropped | |
575 | * and for some paths there is no check at all. | |
576 | * A more general error queue to queue errors for later handling | |
577 | * is probably better. | |
578 | * | |
579 | */ | |
32bbd879 | 580 | int sctp_v4_err(struct sk_buff *skb, __u32 info) |
1da177e4 | 581 | { |
b71d1d42 | 582 | const struct iphdr *iph = (const struct iphdr *)skb->data; |
a27ef749 | 583 | const int ihlen = iph->ihl * 4; |
88c7664f ACM |
584 | const int type = icmp_hdr(skb)->type; |
585 | const int code = icmp_hdr(skb)->code; | |
1da177e4 | 586 | struct sock *sk; |
8de8c873 | 587 | struct sctp_association *asoc = NULL; |
1da177e4 LT |
588 | struct sctp_transport *transport; |
589 | struct inet_sock *inet; | |
aef6de51 | 590 | __u16 saveip, savesctp; |
1da177e4 | 591 | int err; |
4110cc25 | 592 | struct net *net = dev_net(skb->dev); |
1da177e4 | 593 | |
1da177e4 | 594 | /* Fix up skb to look at the embedded net header. */ |
b0e380b1 ACM |
595 | saveip = skb->network_header; |
596 | savesctp = skb->transport_header; | |
31c7711b | 597 | skb_reset_network_header(skb); |
a27ef749 | 598 | skb_set_transport_header(skb, ihlen); |
4110cc25 | 599 | sk = sctp_err_lookup(net, AF_INET, skb, sctp_hdr(skb), &asoc, &transport); |
2e07fa9c | 600 | /* Put back, the original values. */ |
b0e380b1 ACM |
601 | skb->network_header = saveip; |
602 | skb->transport_header = savesctp; | |
1da177e4 | 603 | if (!sk) { |
5d3848bc | 604 | __ICMP_INC_STATS(net, ICMP_MIB_INERRORS); |
32bbd879 | 605 | return -ENOENT; |
1da177e4 LT |
606 | } |
607 | /* Warning: The sock lock is held. Remember to call | |
608 | * sctp_err_finish! | |
609 | */ | |
610 | ||
611 | switch (type) { | |
612 | case ICMP_PARAMETERPROB: | |
613 | err = EPROTO; | |
614 | break; | |
615 | case ICMP_DEST_UNREACH: | |
616 | if (code > NR_ICMP_UNREACH) | |
617 | goto out_unlock; | |
618 | ||
619 | /* PMTU discovery (RFC1191) */ | |
620 | if (ICMP_FRAG_NEEDED == code) { | |
3822a5ff | 621 | sctp_icmp_frag_needed(sk, asoc, transport, |
e2f036a9 | 622 | SCTP_TRUNC4(info)); |
1da177e4 | 623 | goto out_unlock; |
8d72651d | 624 | } else { |
1da177e4 | 625 | if (ICMP_PROT_UNREACH == code) { |
d1ad1ff2 | 626 | sctp_icmp_proto_unreachable(sk, asoc, |
1da177e4 LT |
627 | transport); |
628 | goto out_unlock; | |
629 | } | |
630 | } | |
631 | err = icmp_err_convert[code].errno; | |
632 | break; | |
633 | case ICMP_TIME_EXCEEDED: | |
634 | /* Ignore any time exceeded errors due to fragment reassembly | |
635 | * timeouts. | |
636 | */ | |
637 | if (ICMP_EXC_FRAGTIME == code) | |
638 | goto out_unlock; | |
639 | ||
640 | err = EHOSTUNREACH; | |
641 | break; | |
55be7a9c DM |
642 | case ICMP_REDIRECT: |
643 | sctp_icmp_redirect(sk, transport, skb); | |
3f96a532 | 644 | /* Fall through to out_unlock. */ |
1da177e4 LT |
645 | default: |
646 | goto out_unlock; | |
647 | } | |
648 | ||
649 | inet = inet_sk(sk); | |
650 | if (!sock_owned_by_user(sk) && inet->recverr) { | |
651 | sk->sk_err = err; | |
652 | sk->sk_error_report(sk); | |
653 | } else { /* Only an error on timeout */ | |
654 | sk->sk_err_soft = err; | |
655 | } | |
656 | ||
657 | out_unlock: | |
dae399d7 | 658 | sctp_err_finish(sk, transport); |
32bbd879 | 659 | return 0; |
1da177e4 LT |
660 | } |
661 | ||
662 | /* | |
663 | * RFC 2960, 8.4 - Handle "Out of the blue" Packets. | |
664 | * | |
665 | * This function scans all the chunks in the OOTB packet to determine if | |
666 | * the packet should be discarded right away. If a response might be needed | |
667 | * for this packet, or, if further processing is possible, the packet will | |
668 | * be queued to a proper inqueue for the next phase of handling. | |
669 | * | |
670 | * Output: | |
671 | * Return 0 - If further processing is needed. | |
672 | * Return 1 - If the packet can be discarded right away. | |
673 | */ | |
04675210 | 674 | static int sctp_rcv_ootb(struct sk_buff *skb) |
1da177e4 | 675 | { |
922dbc5b | 676 | struct sctp_chunkhdr *ch, _ch; |
3acb50c1 | 677 | int ch_end, offset = 0; |
1da177e4 LT |
678 | |
679 | /* Scan through all the chunks in the packet. */ | |
a7d1f1b6 | 680 | do { |
3acb50c1 | 681 | /* Make sure we have at least the header there */ |
922dbc5b | 682 | if (offset + sizeof(_ch) > skb->len) |
3acb50c1 MRL |
683 | break; |
684 | ||
685 | ch = skb_header_pointer(skb, offset, sizeof(*ch), &_ch); | |
686 | ||
a7d1f1b6 | 687 | /* Break out if chunk length is less then minimal. */ |
922dbc5b | 688 | if (ntohs(ch->length) < sizeof(_ch)) |
a7d1f1b6 TF |
689 | break; |
690 | ||
e2f036a9 | 691 | ch_end = offset + SCTP_PAD4(ntohs(ch->length)); |
3acb50c1 | 692 | if (ch_end > skb->len) |
a7d1f1b6 | 693 | break; |
1da177e4 LT |
694 | |
695 | /* RFC 8.4, 2) If the OOTB packet contains an ABORT chunk, the | |
696 | * receiver MUST silently discard the OOTB packet and take no | |
697 | * further action. | |
698 | */ | |
699 | if (SCTP_CID_ABORT == ch->type) | |
700 | goto discard; | |
701 | ||
702 | /* RFC 8.4, 6) If the packet contains a SHUTDOWN COMPLETE | |
703 | * chunk, the receiver should silently discard the packet | |
704 | * and take no further action. | |
705 | */ | |
706 | if (SCTP_CID_SHUTDOWN_COMPLETE == ch->type) | |
707 | goto discard; | |
708 | ||
3c77f961 VY |
709 | /* RFC 4460, 2.11.2 |
710 | * This will discard packets with INIT chunk bundled as | |
711 | * subsequent chunks in the packet. When INIT is first, | |
712 | * the normal INIT processing will discard the chunk. | |
713 | */ | |
714 | if (SCTP_CID_INIT == ch->type && (void *)ch != skb->data) | |
715 | goto discard; | |
716 | ||
3acb50c1 MRL |
717 | offset = ch_end; |
718 | } while (ch_end < skb->len); | |
1da177e4 LT |
719 | |
720 | return 0; | |
721 | ||
722 | discard: | |
723 | return 1; | |
724 | } | |
725 | ||
726 | /* Insert endpoint into the hash table. */ | |
76c6d988 | 727 | static int __sctp_hash_endpoint(struct sctp_endpoint *ep) |
1da177e4 | 728 | { |
76c6d988 XL |
729 | struct sock *sk = ep->base.sk; |
730 | struct net *net = sock_net(sk); | |
1da177e4 | 731 | struct sctp_hashbucket *head; |
76c6d988 | 732 | struct sctp_ep_common *epb; |
1da177e4 LT |
733 | |
734 | epb = &ep->base; | |
4cdadcbc | 735 | epb->hashent = sctp_ep_hashfn(net, epb->bind_addr.port); |
1da177e4 LT |
736 | head = &sctp_ep_hashtable[epb->hashent]; |
737 | ||
76c6d988 XL |
738 | if (sk->sk_reuseport) { |
739 | bool any = sctp_is_ep_boundall(sk); | |
740 | struct sctp_ep_common *epb2; | |
741 | struct list_head *list; | |
742 | int cnt = 0, err = 1; | |
743 | ||
744 | list_for_each(list, &ep->base.bind_addr.address_list) | |
745 | cnt++; | |
746 | ||
747 | sctp_for_each_hentry(epb2, &head->chain) { | |
748 | struct sock *sk2 = epb2->sk; | |
749 | ||
750 | if (!net_eq(sock_net(sk2), net) || sk2 == sk || | |
751 | !uid_eq(sock_i_uid(sk2), sock_i_uid(sk)) || | |
752 | !sk2->sk_reuseport) | |
753 | continue; | |
754 | ||
755 | err = sctp_bind_addrs_check(sctp_sk(sk2), | |
756 | sctp_sk(sk), cnt); | |
757 | if (!err) { | |
758 | err = reuseport_add_sock(sk, sk2, any); | |
759 | if (err) | |
760 | return err; | |
761 | break; | |
762 | } else if (err < 0) { | |
763 | return err; | |
764 | } | |
765 | } | |
766 | ||
767 | if (err) { | |
768 | err = reuseport_alloc(sk, any); | |
769 | if (err) | |
770 | return err; | |
771 | } | |
772 | } | |
773 | ||
387602df | 774 | write_lock(&head->lock); |
d970dbf8 | 775 | hlist_add_head(&epb->node, &head->chain); |
387602df | 776 | write_unlock(&head->lock); |
76c6d988 | 777 | return 0; |
1da177e4 LT |
778 | } |
779 | ||
780 | /* Add an endpoint to the hash. Local BH-safe. */ | |
76c6d988 | 781 | int sctp_hash_endpoint(struct sctp_endpoint *ep) |
1da177e4 | 782 | { |
76c6d988 XL |
783 | int err; |
784 | ||
79b91130 | 785 | local_bh_disable(); |
76c6d988 | 786 | err = __sctp_hash_endpoint(ep); |
79b91130 | 787 | local_bh_enable(); |
76c6d988 XL |
788 | |
789 | return err; | |
1da177e4 LT |
790 | } |
791 | ||
792 | /* Remove endpoint from the hash table. */ | |
793 | static void __sctp_unhash_endpoint(struct sctp_endpoint *ep) | |
794 | { | |
76c6d988 | 795 | struct sock *sk = ep->base.sk; |
1da177e4 LT |
796 | struct sctp_hashbucket *head; |
797 | struct sctp_ep_common *epb; | |
798 | ||
799 | epb = &ep->base; | |
800 | ||
76c6d988 | 801 | epb->hashent = sctp_ep_hashfn(sock_net(sk), epb->bind_addr.port); |
1da177e4 LT |
802 | |
803 | head = &sctp_ep_hashtable[epb->hashent]; | |
804 | ||
76c6d988 XL |
805 | if (rcu_access_pointer(sk->sk_reuseport_cb)) |
806 | reuseport_detach_sock(sk); | |
807 | ||
387602df | 808 | write_lock(&head->lock); |
2eebc1e1 | 809 | hlist_del_init(&epb->node); |
387602df | 810 | write_unlock(&head->lock); |
1da177e4 LT |
811 | } |
812 | ||
813 | /* Remove endpoint from the hash. Local BH-safe. */ | |
814 | void sctp_unhash_endpoint(struct sctp_endpoint *ep) | |
815 | { | |
79b91130 | 816 | local_bh_disable(); |
1da177e4 | 817 | __sctp_unhash_endpoint(ep); |
79b91130 | 818 | local_bh_enable(); |
1da177e4 LT |
819 | } |
820 | ||
532ae2f1 XL |
821 | static inline __u32 sctp_hashfn(const struct net *net, __be16 lport, |
822 | const union sctp_addr *paddr, __u32 seed) | |
823 | { | |
824 | __u32 addr; | |
825 | ||
826 | if (paddr->sa.sa_family == AF_INET6) | |
827 | addr = jhash(&paddr->v6.sin6_addr, 16, seed); | |
828 | else | |
829 | addr = (__force __u32)paddr->v4.sin_addr.s_addr; | |
830 | ||
831 | return jhash_3words(addr, ((__force __u32)paddr->v4.sin_port) << 16 | | |
832 | (__force __u32)lport, net_hash_mix(net), seed); | |
833 | } | |
834 | ||
1da177e4 | 835 | /* Look up an endpoint. */ |
532ae2f1 XL |
836 | static struct sctp_endpoint *__sctp_rcv_lookup_endpoint( |
837 | struct net *net, struct sk_buff *skb, | |
838 | const union sctp_addr *laddr, | |
839 | const union sctp_addr *paddr) | |
1da177e4 LT |
840 | { |
841 | struct sctp_hashbucket *head; | |
842 | struct sctp_ep_common *epb; | |
843 | struct sctp_endpoint *ep; | |
532ae2f1 XL |
844 | struct sock *sk; |
845 | __be16 lport; | |
1da177e4 LT |
846 | int hash; |
847 | ||
532ae2f1 XL |
848 | lport = laddr->v4.sin_port; |
849 | hash = sctp_ep_hashfn(net, ntohs(lport)); | |
1da177e4 LT |
850 | head = &sctp_ep_hashtable[hash]; |
851 | read_lock(&head->lock); | |
b67bfe0d | 852 | sctp_for_each_hentry(epb, &head->chain) { |
1da177e4 | 853 | ep = sctp_ep(epb); |
4cdadcbc | 854 | if (sctp_endpoint_is_match(ep, net, laddr)) |
1da177e4 LT |
855 | goto hit; |
856 | } | |
857 | ||
2ce95503 | 858 | ep = sctp_sk(net->sctp.ctl_sock)->ep; |
1da177e4 LT |
859 | |
860 | hit: | |
532ae2f1 XL |
861 | sk = ep->base.sk; |
862 | if (sk->sk_reuseport) { | |
863 | __u32 phash = sctp_hashfn(net, lport, paddr, 0); | |
864 | ||
865 | sk = reuseport_select_sock(sk, phash, skb, | |
866 | sizeof(struct sctphdr)); | |
867 | if (sk) | |
868 | ep = sctp_sk(sk)->ep; | |
869 | } | |
1da177e4 | 870 | sctp_endpoint_hold(ep); |
1da177e4 LT |
871 | read_unlock(&head->lock); |
872 | return ep; | |
873 | } | |
874 | ||
d6c0256a XL |
875 | /* rhashtable for transport */ |
876 | struct sctp_hash_cmp_arg { | |
7fda702f XL |
877 | const union sctp_addr *paddr; |
878 | const struct net *net; | |
8d32503e | 879 | __be16 lport; |
d6c0256a XL |
880 | }; |
881 | ||
882 | static inline int sctp_hash_cmp(struct rhashtable_compare_arg *arg, | |
883 | const void *ptr) | |
884 | { | |
715f5552 | 885 | struct sctp_transport *t = (struct sctp_transport *)ptr; |
d6c0256a | 886 | const struct sctp_hash_cmp_arg *x = arg->key; |
715f5552 | 887 | int err = 1; |
d6c0256a | 888 | |
d6c0256a | 889 | if (!sctp_cmp_addr_exact(&t->ipaddr, x->paddr)) |
715f5552 XL |
890 | return err; |
891 | if (!sctp_transport_hold(t)) | |
892 | return err; | |
893 | ||
7fda702f XL |
894 | if (!net_eq(sock_net(t->asoc->base.sk), x->net)) |
895 | goto out; | |
896 | if (x->lport != htons(t->asoc->base.bind_addr.port)) | |
715f5552 | 897 | goto out; |
d6c0256a | 898 | |
715f5552 XL |
899 | err = 0; |
900 | out: | |
901 | sctp_transport_put(t); | |
902 | return err; | |
d6c0256a XL |
903 | } |
904 | ||
8d32503e | 905 | static inline __u32 sctp_hash_obj(const void *data, u32 len, u32 seed) |
d6c0256a XL |
906 | { |
907 | const struct sctp_transport *t = data; | |
d6c0256a | 908 | |
532ae2f1 XL |
909 | return sctp_hashfn(sock_net(t->asoc->base.sk), |
910 | htons(t->asoc->base.bind_addr.port), | |
911 | &t->ipaddr, seed); | |
d6c0256a XL |
912 | } |
913 | ||
8d32503e | 914 | static inline __u32 sctp_hash_key(const void *data, u32 len, u32 seed) |
d6c0256a XL |
915 | { |
916 | const struct sctp_hash_cmp_arg *x = data; | |
d6c0256a | 917 | |
532ae2f1 | 918 | return sctp_hashfn(x->net, x->lport, x->paddr, seed); |
d6c0256a XL |
919 | } |
920 | ||
921 | static const struct rhashtable_params sctp_hash_params = { | |
922 | .head_offset = offsetof(struct sctp_transport, node), | |
923 | .hashfn = sctp_hash_key, | |
924 | .obj_hashfn = sctp_hash_obj, | |
925 | .obj_cmpfn = sctp_hash_cmp, | |
926 | .automatic_shrinking = true, | |
927 | }; | |
928 | ||
929 | int sctp_transport_hashtable_init(void) | |
930 | { | |
7fda702f | 931 | return rhltable_init(&sctp_transport_hashtable, &sctp_hash_params); |
d6c0256a XL |
932 | } |
933 | ||
934 | void sctp_transport_hashtable_destroy(void) | |
935 | { | |
7fda702f | 936 | rhltable_destroy(&sctp_transport_hashtable); |
d6c0256a XL |
937 | } |
938 | ||
7fda702f | 939 | int sctp_hash_transport(struct sctp_transport *t) |
d6c0256a | 940 | { |
cd2b7087 XL |
941 | struct sctp_transport *transport; |
942 | struct rhlist_head *tmp, *list; | |
d6c0256a | 943 | struct sctp_hash_cmp_arg arg; |
7fda702f | 944 | int err; |
d6c0256a | 945 | |
dd7445ad | 946 | if (t->asoc->temp) |
7fda702f | 947 | return 0; |
dd7445ad | 948 | |
d6c0256a | 949 | arg.net = sock_net(t->asoc->base.sk); |
7fda702f XL |
950 | arg.paddr = &t->ipaddr; |
951 | arg.lport = htons(t->asoc->base.bind_addr.port); | |
d6c0256a | 952 | |
5179b266 | 953 | rcu_read_lock(); |
cd2b7087 XL |
954 | list = rhltable_lookup(&sctp_transport_hashtable, &arg, |
955 | sctp_hash_params); | |
956 | ||
957 | rhl_for_each_entry_rcu(transport, tmp, list, node) | |
958 | if (transport->asoc->ep == t->asoc->ep) { | |
5179b266 | 959 | rcu_read_unlock(); |
27af86bb | 960 | return -EEXIST; |
cd2b7087 | 961 | } |
5179b266 | 962 | rcu_read_unlock(); |
cd2b7087 | 963 | |
7fda702f XL |
964 | err = rhltable_insert_key(&sctp_transport_hashtable, &arg, |
965 | &t->node, sctp_hash_params); | |
966 | if (err) | |
967 | pr_err_once("insert transport fail, errno %d\n", err); | |
968 | ||
969 | return err; | |
d6c0256a XL |
970 | } |
971 | ||
972 | void sctp_unhash_transport(struct sctp_transport *t) | |
973 | { | |
dd7445ad XL |
974 | if (t->asoc->temp) |
975 | return; | |
976 | ||
7fda702f XL |
977 | rhltable_remove(&sctp_transport_hashtable, &t->node, |
978 | sctp_hash_params); | |
d6c0256a XL |
979 | } |
980 | ||
7fda702f | 981 | /* return a transport with holding it */ |
d6c0256a XL |
982 | struct sctp_transport *sctp_addrs_lookup_transport( |
983 | struct net *net, | |
984 | const union sctp_addr *laddr, | |
985 | const union sctp_addr *paddr) | |
986 | { | |
7fda702f XL |
987 | struct rhlist_head *tmp, *list; |
988 | struct sctp_transport *t; | |
d6c0256a | 989 | struct sctp_hash_cmp_arg arg = { |
d6c0256a XL |
990 | .paddr = paddr, |
991 | .net = net, | |
7fda702f | 992 | .lport = laddr->v4.sin_port, |
d6c0256a XL |
993 | }; |
994 | ||
7fda702f XL |
995 | list = rhltable_lookup(&sctp_transport_hashtable, &arg, |
996 | sctp_hash_params); | |
997 | ||
998 | rhl_for_each_entry_rcu(t, tmp, list, node) { | |
999 | if (!sctp_transport_hold(t)) | |
1000 | continue; | |
1001 | ||
1002 | if (sctp_bind_addr_match(&t->asoc->base.bind_addr, | |
1003 | laddr, sctp_sk(t->asoc->base.sk))) | |
1004 | return t; | |
1005 | sctp_transport_put(t); | |
1006 | } | |
1007 | ||
1008 | return NULL; | |
d6c0256a XL |
1009 | } |
1010 | ||
7fda702f | 1011 | /* return a transport without holding it, as it's only used under sock lock */ |
d6c0256a XL |
1012 | struct sctp_transport *sctp_epaddr_lookup_transport( |
1013 | const struct sctp_endpoint *ep, | |
1014 | const union sctp_addr *paddr) | |
1015 | { | |
d6c0256a | 1016 | struct net *net = sock_net(ep->base.sk); |
7fda702f XL |
1017 | struct rhlist_head *tmp, *list; |
1018 | struct sctp_transport *t; | |
65a5124a | 1019 | struct sctp_hash_cmp_arg arg = { |
65a5124a XL |
1020 | .paddr = paddr, |
1021 | .net = net, | |
7fda702f | 1022 | .lport = htons(ep->base.bind_addr.port), |
65a5124a | 1023 | }; |
d6c0256a | 1024 | |
7fda702f XL |
1025 | list = rhltable_lookup(&sctp_transport_hashtable, &arg, |
1026 | sctp_hash_params); | |
1027 | ||
1028 | rhl_for_each_entry_rcu(t, tmp, list, node) | |
1029 | if (ep == t->asoc->ep) | |
1030 | return t; | |
1031 | ||
1032 | return NULL; | |
d6c0256a XL |
1033 | } |
1034 | ||
1da177e4 LT |
1035 | /* Look up an association. */ |
1036 | static struct sctp_association *__sctp_lookup_association( | |
4110cc25 | 1037 | struct net *net, |
1da177e4 LT |
1038 | const union sctp_addr *local, |
1039 | const union sctp_addr *peer, | |
1040 | struct sctp_transport **pt) | |
1041 | { | |
4f008781 | 1042 | struct sctp_transport *t; |
1eed6779 | 1043 | struct sctp_association *asoc = NULL; |
1da177e4 | 1044 | |
4f008781 | 1045 | t = sctp_addrs_lookup_transport(net, local, peer); |
7fda702f | 1046 | if (!t) |
1eed6779 | 1047 | goto out; |
1da177e4 | 1048 | |
1eed6779 | 1049 | asoc = t->asoc; |
4f008781 | 1050 | *pt = t; |
1da177e4 | 1051 | |
1eed6779 | 1052 | out: |
1eed6779 | 1053 | return asoc; |
1da177e4 LT |
1054 | } |
1055 | ||
4f008781 | 1056 | /* Look up an association. protected by RCU read lock */ |
dda91928 | 1057 | static |
4110cc25 EB |
1058 | struct sctp_association *sctp_lookup_association(struct net *net, |
1059 | const union sctp_addr *laddr, | |
1da177e4 | 1060 | const union sctp_addr *paddr, |
dda91928 | 1061 | struct sctp_transport **transportp) |
1da177e4 LT |
1062 | { |
1063 | struct sctp_association *asoc; | |
1064 | ||
f46c7011 | 1065 | rcu_read_lock(); |
4110cc25 | 1066 | asoc = __sctp_lookup_association(net, laddr, paddr, transportp); |
f46c7011 | 1067 | rcu_read_unlock(); |
1da177e4 LT |
1068 | |
1069 | return asoc; | |
1070 | } | |
1071 | ||
1072 | /* Is there an association matching the given local and peer addresses? */ | |
53066538 XL |
1073 | bool sctp_has_association(struct net *net, |
1074 | const union sctp_addr *laddr, | |
1075 | const union sctp_addr *paddr) | |
1da177e4 | 1076 | { |
1da177e4 LT |
1077 | struct sctp_transport *transport; |
1078 | ||
53066538 | 1079 | if (sctp_lookup_association(net, laddr, paddr, &transport)) { |
dae399d7 | 1080 | sctp_transport_put(transport); |
53066538 | 1081 | return true; |
1da177e4 LT |
1082 | } |
1083 | ||
53066538 | 1084 | return false; |
1da177e4 LT |
1085 | } |
1086 | ||
1087 | /* | |
1088 | * SCTP Implementors Guide, 2.18 Handling of address | |
1089 | * parameters within the INIT or INIT-ACK. | |
1090 | * | |
1091 | * D) When searching for a matching TCB upon reception of an INIT | |
1092 | * or INIT-ACK chunk the receiver SHOULD use not only the | |
1093 | * source address of the packet (containing the INIT or | |
1094 | * INIT-ACK) but the receiver SHOULD also use all valid | |
1095 | * address parameters contained within the chunk. | |
1096 | * | |
1097 | * 2.18.3 Solution description | |
1098 | * | |
1099 | * This new text clearly specifies to an implementor the need | |
1100 | * to look within the INIT or INIT-ACK. Any implementation that | |
1101 | * does not do this, may not be able to establish associations | |
1102 | * in certain circumstances. | |
1103 | * | |
1104 | */ | |
4110cc25 EB |
1105 | static struct sctp_association *__sctp_rcv_init_lookup(struct net *net, |
1106 | struct sk_buff *skb, | |
1da177e4 LT |
1107 | const union sctp_addr *laddr, struct sctp_transport **transportp) |
1108 | { | |
1109 | struct sctp_association *asoc; | |
1110 | union sctp_addr addr; | |
1111 | union sctp_addr *paddr = &addr; | |
2c0fd387 | 1112 | struct sctphdr *sh = sctp_hdr(skb); |
1da177e4 | 1113 | union sctp_params params; |
01a992be | 1114 | struct sctp_init_chunk *init; |
1da177e4 LT |
1115 | struct sctp_af *af; |
1116 | ||
1da177e4 LT |
1117 | /* |
1118 | * This code will NOT touch anything inside the chunk--it is | |
1119 | * strictly READ-ONLY. | |
1120 | * | |
1121 | * RFC 2960 3 SCTP packet Format | |
1122 | * | |
1123 | * Multiple chunks can be bundled into one SCTP packet up to | |
1124 | * the MTU size, except for the INIT, INIT ACK, and SHUTDOWN | |
1125 | * COMPLETE chunks. These chunks MUST NOT be bundled with any | |
1126 | * other chunk in a packet. See Section 6.10 for more details | |
1127 | * on chunk bundling. | |
1128 | */ | |
1129 | ||
1130 | /* Find the start of the TLVs and the end of the chunk. This is | |
1131 | * the region we search for address parameters. | |
1132 | */ | |
01a992be | 1133 | init = (struct sctp_init_chunk *)skb->data; |
1da177e4 LT |
1134 | |
1135 | /* Walk the parameters looking for embedded addresses. */ | |
1136 | sctp_walk_params(params, init, init_hdr.params) { | |
1137 | ||
1138 | /* Note: Ignoring hostname addresses. */ | |
1139 | af = sctp_get_af_specific(param_type2af(params.p->type)); | |
1140 | if (!af) | |
1141 | continue; | |
1142 | ||
dd86d136 | 1143 | af->from_addr_param(paddr, params.addr, sh->source, 0); |
1da177e4 | 1144 | |
7c17fcc7 | 1145 | asoc = __sctp_lookup_association(net, laddr, paddr, transportp); |
1da177e4 LT |
1146 | if (asoc) |
1147 | return asoc; | |
1148 | } | |
1149 | ||
1150 | return NULL; | |
1151 | } | |
1152 | ||
df218577 VY |
1153 | /* ADD-IP, Section 5.2 |
1154 | * When an endpoint receives an ASCONF Chunk from the remote peer | |
1155 | * special procedures may be needed to identify the association the | |
1156 | * ASCONF Chunk is associated with. To properly find the association | |
1157 | * the following procedures SHOULD be followed: | |
1158 | * | |
1159 | * D2) If the association is not found, use the address found in the | |
1160 | * Address Parameter TLV combined with the port number found in the | |
1161 | * SCTP common header. If found proceed to rule D4. | |
1162 | * | |
1163 | * D2-ext) If more than one ASCONF Chunks are packed together, use the | |
1164 | * address found in the ASCONF Address Parameter TLV of each of the | |
1165 | * subsequent ASCONF Chunks. If found, proceed to rule D4. | |
1166 | */ | |
1167 | static struct sctp_association *__sctp_rcv_asconf_lookup( | |
4110cc25 | 1168 | struct net *net, |
922dbc5b | 1169 | struct sctp_chunkhdr *ch, |
df218577 | 1170 | const union sctp_addr *laddr, |
bc92dd19 | 1171 | __be16 peer_port, |
df218577 VY |
1172 | struct sctp_transport **transportp) |
1173 | { | |
68d75469 | 1174 | struct sctp_addip_chunk *asconf = (struct sctp_addip_chunk *)ch; |
df218577 VY |
1175 | struct sctp_af *af; |
1176 | union sctp_addr_param *param; | |
1177 | union sctp_addr paddr; | |
1178 | ||
1179 | /* Skip over the ADDIP header and find the Address parameter */ | |
1180 | param = (union sctp_addr_param *)(asconf + 1); | |
1181 | ||
6a435732 | 1182 | af = sctp_get_af_specific(param_type2af(param->p.type)); |
df218577 VY |
1183 | if (unlikely(!af)) |
1184 | return NULL; | |
1185 | ||
1186 | af->from_addr_param(&paddr, param, peer_port, 0); | |
1187 | ||
4110cc25 | 1188 | return __sctp_lookup_association(net, laddr, &paddr, transportp); |
df218577 VY |
1189 | } |
1190 | ||
1191 | ||
bbd0d598 VY |
1192 | /* SCTP-AUTH, Section 6.3: |
1193 | * If the receiver does not find a STCB for a packet containing an AUTH | |
1194 | * chunk as the first chunk and not a COOKIE-ECHO chunk as the second | |
1195 | * chunk, it MUST use the chunks after the AUTH chunk to look up an existing | |
1196 | * association. | |
1197 | * | |
1198 | * This means that any chunks that can help us identify the association need | |
25985edc | 1199 | * to be looked at to find this association. |
bbd0d598 | 1200 | */ |
4110cc25 EB |
1201 | static struct sctp_association *__sctp_rcv_walk_lookup(struct net *net, |
1202 | struct sk_buff *skb, | |
bbd0d598 VY |
1203 | const union sctp_addr *laddr, |
1204 | struct sctp_transport **transportp) | |
1205 | { | |
df218577 | 1206 | struct sctp_association *asoc = NULL; |
922dbc5b | 1207 | struct sctp_chunkhdr *ch; |
df218577 VY |
1208 | int have_auth = 0; |
1209 | unsigned int chunk_num = 1; | |
1210 | __u8 *ch_end; | |
1211 | ||
1212 | /* Walk through the chunks looking for AUTH or ASCONF chunks | |
1213 | * to help us find the association. | |
bbd0d598 | 1214 | */ |
922dbc5b | 1215 | ch = (struct sctp_chunkhdr *)skb->data; |
df218577 VY |
1216 | do { |
1217 | /* Break out if chunk length is less then minimal. */ | |
922dbc5b | 1218 | if (ntohs(ch->length) < sizeof(*ch)) |
df218577 VY |
1219 | break; |
1220 | ||
e2f036a9 | 1221 | ch_end = ((__u8 *)ch) + SCTP_PAD4(ntohs(ch->length)); |
df218577 VY |
1222 | if (ch_end > skb_tail_pointer(skb)) |
1223 | break; | |
1224 | ||
cb3f837b | 1225 | switch (ch->type) { |
f7010e61 | 1226 | case SCTP_CID_AUTH: |
1227 | have_auth = chunk_num; | |
1228 | break; | |
1229 | ||
1230 | case SCTP_CID_COOKIE_ECHO: | |
1231 | /* If a packet arrives containing an AUTH chunk as | |
1232 | * a first chunk, a COOKIE-ECHO chunk as the second | |
1233 | * chunk, and possibly more chunks after them, and | |
1234 | * the receiver does not have an STCB for that | |
1235 | * packet, then authentication is based on | |
1236 | * the contents of the COOKIE- ECHO chunk. | |
1237 | */ | |
1238 | if (have_auth == 1 && chunk_num == 2) | |
1239 | return NULL; | |
1240 | break; | |
1241 | ||
1242 | case SCTP_CID_ASCONF: | |
1243 | if (have_auth || net->sctp.addip_noauth) | |
1244 | asoc = __sctp_rcv_asconf_lookup( | |
1245 | net, ch, laddr, | |
1246 | sctp_hdr(skb)->source, | |
1247 | transportp); | |
1248 | default: | |
1249 | break; | |
df218577 VY |
1250 | } |
1251 | ||
1252 | if (asoc) | |
1253 | break; | |
1254 | ||
922dbc5b | 1255 | ch = (struct sctp_chunkhdr *)ch_end; |
df218577 VY |
1256 | chunk_num++; |
1257 | } while (ch_end < skb_tail_pointer(skb)); | |
1258 | ||
1259 | return asoc; | |
bbd0d598 VY |
1260 | } |
1261 | ||
1262 | /* | |
1263 | * There are circumstances when we need to look inside the SCTP packet | |
1264 | * for information to help us find the association. Examples | |
1265 | * include looking inside of INIT/INIT-ACK chunks or after the AUTH | |
1266 | * chunks. | |
1267 | */ | |
4110cc25 EB |
1268 | static struct sctp_association *__sctp_rcv_lookup_harder(struct net *net, |
1269 | struct sk_buff *skb, | |
bbd0d598 VY |
1270 | const union sctp_addr *laddr, |
1271 | struct sctp_transport **transportp) | |
1272 | { | |
922dbc5b | 1273 | struct sctp_chunkhdr *ch; |
bbd0d598 | 1274 | |
90017acc MRL |
1275 | /* We do not allow GSO frames here as we need to linearize and |
1276 | * then cannot guarantee frame boundaries. This shouldn't be an | |
1277 | * issue as packets hitting this are mostly INIT or INIT-ACK and | |
1278 | * those cannot be on GSO-style anyway. | |
1279 | */ | |
1dd27cde | 1280 | if (skb_is_gso(skb) && skb_is_gso_sctp(skb)) |
90017acc MRL |
1281 | return NULL; |
1282 | ||
922dbc5b | 1283 | ch = (struct sctp_chunkhdr *)skb->data; |
bbd0d598 | 1284 | |
df218577 VY |
1285 | /* The code below will attempt to walk the chunk and extract |
1286 | * parameter information. Before we do that, we need to verify | |
1287 | * that the chunk length doesn't cause overflow. Otherwise, we'll | |
1288 | * walk off the end. | |
1289 | */ | |
e2f036a9 | 1290 | if (SCTP_PAD4(ntohs(ch->length)) > skb->len) |
df218577 VY |
1291 | return NULL; |
1292 | ||
bbd0d598 | 1293 | /* If this is INIT/INIT-ACK look inside the chunk too. */ |
f482f2fc | 1294 | if (ch->type == SCTP_CID_INIT || ch->type == SCTP_CID_INIT_ACK) |
4110cc25 | 1295 | return __sctp_rcv_init_lookup(net, skb, laddr, transportp); |
bbd0d598 | 1296 | |
f482f2fc | 1297 | return __sctp_rcv_walk_lookup(net, skb, laddr, transportp); |
bbd0d598 VY |
1298 | } |
1299 | ||
1da177e4 | 1300 | /* Lookup an association for an inbound skb. */ |
4110cc25 EB |
1301 | static struct sctp_association *__sctp_rcv_lookup(struct net *net, |
1302 | struct sk_buff *skb, | |
1da177e4 LT |
1303 | const union sctp_addr *paddr, |
1304 | const union sctp_addr *laddr, | |
1305 | struct sctp_transport **transportp) | |
1306 | { | |
1307 | struct sctp_association *asoc; | |
1308 | ||
4110cc25 | 1309 | asoc = __sctp_lookup_association(net, laddr, paddr, transportp); |
b77b7565 MRL |
1310 | if (asoc) |
1311 | goto out; | |
1da177e4 LT |
1312 | |
1313 | /* Further lookup for INIT/INIT-ACK packets. | |
1314 | * SCTP Implementors Guide, 2.18 Handling of address | |
1315 | * parameters within the INIT or INIT-ACK. | |
1316 | */ | |
b77b7565 MRL |
1317 | asoc = __sctp_rcv_lookup_harder(net, skb, laddr, transportp); |
1318 | if (asoc) | |
1319 | goto out; | |
1da177e4 | 1320 | |
b77b7565 MRL |
1321 | if (paddr->sa.sa_family == AF_INET) |
1322 | pr_debug("sctp: asoc not found for src:%pI4:%d dst:%pI4:%d\n", | |
1323 | &laddr->v4.sin_addr, ntohs(laddr->v4.sin_port), | |
1324 | &paddr->v4.sin_addr, ntohs(paddr->v4.sin_port)); | |
1325 | else | |
1326 | pr_debug("sctp: asoc not found for src:%pI6:%d dst:%pI6:%d\n", | |
1327 | &laddr->v6.sin6_addr, ntohs(laddr->v6.sin6_port), | |
1328 | &paddr->v6.sin6_addr, ntohs(paddr->v6.sin6_port)); | |
1329 | ||
1330 | out: | |
1da177e4 LT |
1331 | return asoc; |
1332 | } |