Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * INET An implementation of the TCP/IP protocol suite for the LINUX | |
3 | * operating system. INET is implemented using the BSD Socket | |
4 | * interface as the means of communication with the user level. | |
5 | * | |
6 | * RAW - implementation of IP "raw" sockets. | |
7 | * | |
02c30a84 | 8 | * Authors: Ross Biro |
1da177e4 LT |
9 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
10 | * | |
11 | * Fixes: | |
12 | * Alan Cox : verify_area() fixed up | |
13 | * Alan Cox : ICMP error handling | |
14 | * Alan Cox : EMSGSIZE if you send too big a packet | |
15 | * Alan Cox : Now uses generic datagrams and shared | |
16 | * skbuff library. No more peek crashes, | |
17 | * no more backlogs | |
18 | * Alan Cox : Checks sk->broadcast. | |
19 | * Alan Cox : Uses skb_free_datagram/skb_copy_datagram | |
20 | * Alan Cox : Raw passes ip options too | |
21 | * Alan Cox : Setsocketopt added | |
22 | * Alan Cox : Fixed error return for broadcasts | |
23 | * Alan Cox : Removed wake_up calls | |
24 | * Alan Cox : Use ttl/tos | |
25 | * Alan Cox : Cleaned up old debugging | |
26 | * Alan Cox : Use new kernel side addresses | |
27 | * Arnt Gulbrandsen : Fixed MSG_DONTROUTE in raw sockets. | |
28 | * Alan Cox : BSD style RAW socket demultiplexing. | |
29 | * Alan Cox : Beginnings of mrouted support. | |
30 | * Alan Cox : Added IP_HDRINCL option. | |
31 | * Alan Cox : Skip broadcast check if BSDism set. | |
32 | * David S. Miller : New socket lookup architecture. | |
33 | * | |
34 | * This program is free software; you can redistribute it and/or | |
35 | * modify it under the terms of the GNU General Public License | |
36 | * as published by the Free Software Foundation; either version | |
37 | * 2 of the License, or (at your option) any later version. | |
38 | */ | |
bf0d5249 | 39 | |
715b49ef | 40 | #include <linux/types.h> |
60063497 | 41 | #include <linux/atomic.h> |
1da177e4 LT |
42 | #include <asm/byteorder.h> |
43 | #include <asm/current.h> | |
7c0f6ba6 | 44 | #include <linux/uaccess.h> |
1da177e4 | 45 | #include <asm/ioctls.h> |
1da177e4 LT |
46 | #include <linux/stddef.h> |
47 | #include <linux/slab.h> | |
48 | #include <linux/errno.h> | |
1da177e4 | 49 | #include <linux/kernel.h> |
bc3b2d7f | 50 | #include <linux/export.h> |
1da177e4 LT |
51 | #include <linux/spinlock.h> |
52 | #include <linux/sockios.h> | |
53 | #include <linux/socket.h> | |
54 | #include <linux/in.h> | |
55 | #include <linux/mroute.h> | |
56 | #include <linux/netdevice.h> | |
57 | #include <linux/in_route.h> | |
58 | #include <linux/route.h> | |
1da177e4 | 59 | #include <linux/skbuff.h> |
f5220d63 | 60 | #include <linux/igmp.h> |
457c4cbc | 61 | #include <net/net_namespace.h> |
1da177e4 LT |
62 | #include <net/dst.h> |
63 | #include <net/sock.h> | |
1da177e4 LT |
64 | #include <linux/ip.h> |
65 | #include <linux/net.h> | |
66 | #include <net/ip.h> | |
67 | #include <net/icmp.h> | |
68 | #include <net/udp.h> | |
69 | #include <net/raw.h> | |
70 | #include <net/snmp.h> | |
c752f073 | 71 | #include <net/tcp_states.h> |
1da177e4 LT |
72 | #include <net/inet_common.h> |
73 | #include <net/checksum.h> | |
74 | #include <net/xfrm.h> | |
75 | #include <linux/rtnetlink.h> | |
76 | #include <linux/proc_fs.h> | |
77 | #include <linux/seq_file.h> | |
78 | #include <linux/netfilter.h> | |
79 | #include <linux/netfilter_ipv4.h> | |
709b46e8 | 80 | #include <linux/compat.h> |
c008ba5b HX |
81 | #include <linux/uio.h> |
82 | ||
83 | struct raw_frag_vec { | |
b61e9dcc | 84 | struct msghdr *msg; |
c008ba5b HX |
85 | union { |
86 | struct icmphdr icmph; | |
87 | char c[1]; | |
88 | } hdr; | |
89 | int hlen; | |
90 | }; | |
1da177e4 | 91 | |
432490f9 | 92 | struct raw_hashinfo raw_v4_hashinfo = { |
938b93ad | 93 | .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock), |
b673e4df | 94 | }; |
432490f9 | 95 | EXPORT_SYMBOL_GPL(raw_v4_hashinfo); |
1da177e4 | 96 | |
086c653f | 97 | int raw_hash_sk(struct sock *sk) |
1da177e4 | 98 | { |
fc8717ba | 99 | struct raw_hashinfo *h = sk->sk_prot->h.raw_hash; |
65b4c50b | 100 | struct hlist_head *head; |
1da177e4 | 101 | |
c720c7e8 | 102 | head = &h->ht[inet_sk(sk)->inet_num & (RAW_HTABLE_SIZE - 1)]; |
65b4c50b PE |
103 | |
104 | write_lock_bh(&h->lock); | |
1da177e4 | 105 | sk_add_node(sk, head); |
c29a0bc4 | 106 | sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1); |
65b4c50b | 107 | write_unlock_bh(&h->lock); |
086c653f CG |
108 | |
109 | return 0; | |
65b4c50b PE |
110 | } |
111 | EXPORT_SYMBOL_GPL(raw_hash_sk); | |
112 | ||
fc8717ba | 113 | void raw_unhash_sk(struct sock *sk) |
ab70768e | 114 | { |
fc8717ba PE |
115 | struct raw_hashinfo *h = sk->sk_prot->h.raw_hash; |
116 | ||
ab70768e PE |
117 | write_lock_bh(&h->lock); |
118 | if (sk_del_node_init(sk)) | |
c29a0bc4 | 119 | sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1); |
ab70768e PE |
120 | write_unlock_bh(&h->lock); |
121 | } | |
122 | EXPORT_SYMBOL_GPL(raw_unhash_sk); | |
123 | ||
432490f9 | 124 | struct sock *__raw_v4_lookup(struct net *net, struct sock *sk, |
67359930 DA |
125 | unsigned short num, __be32 raddr, __be32 laddr, |
126 | int dif, int sdif) | |
1da177e4 | 127 | { |
b67bfe0d | 128 | sk_for_each_from(sk) { |
1da177e4 LT |
129 | struct inet_sock *inet = inet_sk(sk); |
130 | ||
c720c7e8 ED |
131 | if (net_eq(sock_net(sk), net) && inet->inet_num == num && |
132 | !(inet->inet_daddr && inet->inet_daddr != raddr) && | |
133 | !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) && | |
7055420f | 134 | raw_sk_bound_dev_eq(net, sk->sk_bound_dev_if, dif, sdif)) |
1da177e4 LT |
135 | goto found; /* gotcha */ |
136 | } | |
137 | sk = NULL; | |
138 | found: | |
139 | return sk; | |
140 | } | |
432490f9 | 141 | EXPORT_SYMBOL_GPL(__raw_v4_lookup); |
1da177e4 LT |
142 | |
143 | /* | |
144 | * 0 - deliver | |
145 | * 1 - block | |
146 | */ | |
ab43ed8b | 147 | static int icmp_filter(const struct sock *sk, const struct sk_buff *skb) |
1da177e4 | 148 | { |
ab43ed8b ED |
149 | struct icmphdr _hdr; |
150 | const struct icmphdr *hdr; | |
1da177e4 | 151 | |
ab43ed8b ED |
152 | hdr = skb_header_pointer(skb, skb_transport_offset(skb), |
153 | sizeof(_hdr), &_hdr); | |
154 | if (!hdr) | |
1da177e4 LT |
155 | return 1; |
156 | ||
ab43ed8b | 157 | if (hdr->type < 32) { |
1da177e4 LT |
158 | __u32 data = raw_sk(sk)->filter.data; |
159 | ||
ab43ed8b | 160 | return ((1U << hdr->type) & data) != 0; |
1da177e4 LT |
161 | } |
162 | ||
163 | /* Do not block unknown ICMP types */ | |
164 | return 0; | |
165 | } | |
166 | ||
167 | /* IP input processing comes here for RAW socket delivery. | |
168 | * Caller owns SKB, so we must make clones. | |
169 | * | |
170 | * RFC 1122: SHOULD pass TOS value up to the transport layer. | |
171 | * -> It does. And not only TOS, but all IP header. | |
172 | */ | |
b71d1d42 | 173 | static int raw_v4_input(struct sk_buff *skb, const struct iphdr *iph, int hash) |
1da177e4 | 174 | { |
67359930 | 175 | int sdif = inet_sdif(skb); |
1da177e4 LT |
176 | struct sock *sk; |
177 | struct hlist_head *head; | |
d13964f4 | 178 | int delivered = 0; |
be185884 | 179 | struct net *net; |
1da177e4 | 180 | |
b673e4df PE |
181 | read_lock(&raw_v4_hashinfo.lock); |
182 | head = &raw_v4_hashinfo.ht[hash]; | |
1da177e4 LT |
183 | if (hlist_empty(head)) |
184 | goto out; | |
be185884 | 185 | |
c346dca1 | 186 | net = dev_net(skb->dev); |
be185884 | 187 | sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol, |
1da177e4 | 188 | iph->saddr, iph->daddr, |
67359930 | 189 | skb->dev->ifindex, sdif); |
1da177e4 LT |
190 | |
191 | while (sk) { | |
d13964f4 | 192 | delivered = 1; |
f5220d63 QA |
193 | if ((iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) && |
194 | ip_mc_sf_allow(sk, iph->daddr, iph->saddr, | |
60d9b031 | 195 | skb->dev->ifindex, sdif)) { |
1da177e4 LT |
196 | struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC); |
197 | ||
198 | /* Not releasing hash table! */ | |
199 | if (clone) | |
200 | raw_rcv(sk, clone); | |
201 | } | |
be185884 | 202 | sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol, |
1da177e4 | 203 | iph->saddr, iph->daddr, |
67359930 | 204 | skb->dev->ifindex, sdif); |
1da177e4 LT |
205 | } |
206 | out: | |
b673e4df | 207 | read_unlock(&raw_v4_hashinfo.lock); |
d13964f4 | 208 | return delivered; |
1da177e4 LT |
209 | } |
210 | ||
7bc54c90 PE |
211 | int raw_local_deliver(struct sk_buff *skb, int protocol) |
212 | { | |
213 | int hash; | |
214 | struct sock *raw_sk; | |
215 | ||
b673e4df PE |
216 | hash = protocol & (RAW_HTABLE_SIZE - 1); |
217 | raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]); | |
7bc54c90 PE |
218 | |
219 | /* If there maybe a raw socket we must check - if not we | |
220 | * don't care less | |
221 | */ | |
222 | if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash)) | |
223 | raw_sk = NULL; | |
224 | ||
225 | return raw_sk != NULL; | |
226 | ||
227 | } | |
228 | ||
229 | static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info) | |
1da177e4 LT |
230 | { |
231 | struct inet_sock *inet = inet_sk(sk); | |
88c7664f ACM |
232 | const int type = icmp_hdr(skb)->type; |
233 | const int code = icmp_hdr(skb)->code; | |
1da177e4 LT |
234 | int err = 0; |
235 | int harderr = 0; | |
236 | ||
36393395 DM |
237 | if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) |
238 | ipv4_sk_update_pmtu(skb, sk, info); | |
8d65b119 | 239 | else if (type == ICMP_REDIRECT) { |
55be7a9c | 240 | ipv4_sk_redirect(skb, sk); |
8d65b119 DJ |
241 | return; |
242 | } | |
36393395 | 243 | |
1da177e4 LT |
244 | /* Report error on raw socket, if: |
245 | 1. User requested ip_recverr. | |
246 | 2. Socket is connected (otherwise the error indication | |
247 | is useless without ip_recverr and error is hard. | |
248 | */ | |
249 | if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED) | |
250 | return; | |
251 | ||
252 | switch (type) { | |
253 | default: | |
254 | case ICMP_TIME_EXCEEDED: | |
255 | err = EHOSTUNREACH; | |
256 | break; | |
257 | case ICMP_SOURCE_QUENCH: | |
258 | return; | |
259 | case ICMP_PARAMETERPROB: | |
260 | err = EPROTO; | |
261 | harderr = 1; | |
262 | break; | |
263 | case ICMP_DEST_UNREACH: | |
264 | err = EHOSTUNREACH; | |
265 | if (code > NR_ICMP_UNREACH) | |
266 | break; | |
267 | err = icmp_err_convert[code].errno; | |
268 | harderr = icmp_err_convert[code].fatal; | |
269 | if (code == ICMP_FRAG_NEEDED) { | |
270 | harderr = inet->pmtudisc != IP_PMTUDISC_DONT; | |
271 | err = EMSGSIZE; | |
272 | } | |
273 | } | |
274 | ||
275 | if (inet->recverr) { | |
b71d1d42 | 276 | const struct iphdr *iph = (const struct iphdr *)skb->data; |
1da177e4 LT |
277 | u8 *payload = skb->data + (iph->ihl << 2); |
278 | ||
279 | if (inet->hdrincl) | |
280 | payload = skb->data; | |
281 | ip_icmp_error(sk, skb, err, 0, info, payload); | |
282 | } | |
283 | ||
284 | if (inet->recverr || harderr) { | |
285 | sk->sk_err = err; | |
286 | sk->sk_error_report(sk); | |
287 | } | |
288 | } | |
289 | ||
7bc54c90 PE |
290 | void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info) |
291 | { | |
292 | int hash; | |
293 | struct sock *raw_sk; | |
b71d1d42 | 294 | const struct iphdr *iph; |
be185884 | 295 | struct net *net; |
7bc54c90 | 296 | |
b673e4df | 297 | hash = protocol & (RAW_HTABLE_SIZE - 1); |
7bc54c90 | 298 | |
b673e4df PE |
299 | read_lock(&raw_v4_hashinfo.lock); |
300 | raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]); | |
00db4124 | 301 | if (raw_sk) { |
67359930 DA |
302 | int dif = skb->dev->ifindex; |
303 | int sdif = inet_sdif(skb); | |
304 | ||
b71d1d42 | 305 | iph = (const struct iphdr *)skb->data; |
c346dca1 | 306 | net = dev_net(skb->dev); |
be185884 PE |
307 | |
308 | while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol, | |
309 | iph->daddr, iph->saddr, | |
67359930 | 310 | dif, sdif)) != NULL) { |
7bc54c90 PE |
311 | raw_err(raw_sk, skb, info); |
312 | raw_sk = sk_next(raw_sk); | |
b71d1d42 | 313 | iph = (const struct iphdr *)skb->data; |
7bc54c90 PE |
314 | } |
315 | } | |
b673e4df | 316 | read_unlock(&raw_v4_hashinfo.lock); |
7bc54c90 PE |
317 | } |
318 | ||
5e73ea1a | 319 | static int raw_rcv_skb(struct sock *sk, struct sk_buff *skb) |
1da177e4 LT |
320 | { |
321 | /* Charge it to the socket. */ | |
e905a9ed | 322 | |
fbf8866d | 323 | ipv4_pktinfo_prepare(sk, skb); |
d826eb14 | 324 | if (sock_queue_rcv_skb(sk, skb) < 0) { |
1da177e4 LT |
325 | kfree_skb(skb); |
326 | return NET_RX_DROP; | |
327 | } | |
328 | ||
329 | return NET_RX_SUCCESS; | |
330 | } | |
331 | ||
332 | int raw_rcv(struct sock *sk, struct sk_buff *skb) | |
333 | { | |
334 | if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) { | |
33c732c3 | 335 | atomic_inc(&sk->sk_drops); |
1da177e4 LT |
336 | kfree_skb(skb); |
337 | return NET_RX_DROP; | |
338 | } | |
b59c2701 | 339 | nf_reset(skb); |
1da177e4 | 340 | |
d56f90a7 | 341 | skb_push(skb, skb->data - skb_network_header(skb)); |
1da177e4 LT |
342 | |
343 | raw_rcv_skb(sk, skb); | |
344 | return 0; | |
345 | } | |
346 | ||
77968b78 | 347 | static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4, |
7ae9abfd | 348 | struct msghdr *msg, size_t length, |
c14ac945 SHY |
349 | struct rtable **rtp, unsigned int flags, |
350 | const struct sockcm_cookie *sockc) | |
1da177e4 LT |
351 | { |
352 | struct inet_sock *inet = inet_sk(sk); | |
0388b004 | 353 | struct net *net = sock_net(sk); |
1da177e4 LT |
354 | struct iphdr *iph; |
355 | struct sk_buff *skb; | |
f844c74f | 356 | unsigned int iphlen; |
1da177e4 | 357 | int err; |
1789a640 | 358 | struct rtable *rt = *rtp; |
66088243 | 359 | int hlen, tlen; |
1da177e4 | 360 | |
d8d1f30b | 361 | if (length > rt->dst.dev->mtu) { |
77968b78 | 362 | ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport, |
d8d1f30b | 363 | rt->dst.dev->mtu); |
1da177e4 LT |
364 | return -EMSGSIZE; |
365 | } | |
86f4c90a AP |
366 | if (length < sizeof(struct iphdr)) |
367 | return -EINVAL; | |
368 | ||
1da177e4 LT |
369 | if (flags&MSG_PROBE) |
370 | goto out; | |
371 | ||
66088243 HX |
372 | hlen = LL_RESERVED_SPACE(rt->dst.dev); |
373 | tlen = rt->dst.dev->needed_tailroom; | |
f5184d26 | 374 | skb = sock_alloc_send_skb(sk, |
66088243 | 375 | length + hlen + tlen + 15, |
f5184d26 | 376 | flags & MSG_DONTWAIT, &err); |
51456b29 | 377 | if (!skb) |
e905a9ed | 378 | goto error; |
66088243 | 379 | skb_reserve(skb, hlen); |
1da177e4 LT |
380 | |
381 | skb->priority = sk->sk_priority; | |
4a19ec58 | 382 | skb->mark = sk->sk_mark; |
bc969a97 | 383 | skb->tstamp = sockc->transmit_time; |
d8d1f30b | 384 | skb_dst_set(skb, &rt->dst); |
1789a640 | 385 | *rtp = NULL; |
1da177e4 | 386 | |
7e28ecc2 | 387 | skb_reset_network_header(skb); |
eddc9ec5 | 388 | iph = ip_hdr(skb); |
7e28ecc2 | 389 | skb_put(skb, length); |
1da177e4 LT |
390 | |
391 | skb->ip_summed = CHECKSUM_NONE; | |
392 | ||
c14ac945 | 393 | sock_tx_timestamp(sk, sockc->tsflags, &skb_shinfo(skb)->tx_flags); |
11878b40 | 394 | |
0dec879f JA |
395 | if (flags & MSG_CONFIRM) |
396 | skb_set_dst_pending_confirm(skb, 1); | |
397 | ||
b0e380b1 | 398 | skb->transport_header = skb->network_header; |
55888dfb | 399 | err = -EFAULT; |
21226abb | 400 | if (memcpy_from_msg(iph, msg, length)) |
55888dfb | 401 | goto error_free; |
1da177e4 | 402 | |
f844c74f | 403 | iphlen = iph->ihl * 4; |
55888dfb NH |
404 | |
405 | /* | |
406 | * We don't want to modify the ip header, but we do need to | |
407 | * be sure that it won't cause problems later along the network | |
408 | * stack. Specifically we want to make sure that iph->ihl is a | |
409 | * sane value. If ihl points beyond the length of the buffer passed | |
410 | * in, reject the frame as invalid | |
411 | */ | |
412 | err = -EINVAL; | |
413 | if (iphlen > length) | |
414 | goto error_free; | |
415 | ||
416 | if (iphlen >= sizeof(*iph)) { | |
1da177e4 | 417 | if (!iph->saddr) |
77968b78 | 418 | iph->saddr = fl4->saddr; |
1da177e4 LT |
419 | iph->check = 0; |
420 | iph->tot_len = htons(length); | |
421 | if (!iph->id) | |
b6a7719a | 422 | ip_select_ident(net, skb, NULL); |
1da177e4 LT |
423 | |
424 | iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl); | |
027ac58e BCC |
425 | skb->transport_header += iphlen; |
426 | if (iph->protocol == IPPROTO_ICMP && | |
427 | length >= iphlen + sizeof(struct icmphdr)) | |
428 | icmp_out_count(net, ((struct icmphdr *) | |
429 | skb_transport_header(skb))->type); | |
1da177e4 LT |
430 | } |
431 | ||
29a26a56 EB |
432 | err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, |
433 | net, sk, skb, NULL, rt->dst.dev, | |
13206b6b | 434 | dst_output); |
1da177e4 | 435 | if (err > 0) |
6ce9e7b5 | 436 | err = net_xmit_errno(err); |
1da177e4 LT |
437 | if (err) |
438 | goto error; | |
439 | out: | |
440 | return 0; | |
441 | ||
55888dfb | 442 | error_free: |
1da177e4 LT |
443 | kfree_skb(skb); |
444 | error: | |
5e38e270 | 445 | IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS); |
6ce9e7b5 ED |
446 | if (err == -ENOBUFS && !inet->recverr) |
447 | err = 0; | |
e905a9ed | 448 | return err; |
1da177e4 LT |
449 | } |
450 | ||
c008ba5b | 451 | static int raw_probe_proto_opt(struct raw_frag_vec *rfv, struct flowi4 *fl4) |
1da177e4 | 452 | { |
32b5913a | 453 | int err; |
1da177e4 | 454 | |
32b5913a | 455 | if (fl4->flowi4_proto != IPPROTO_ICMP) |
a27b58fe | 456 | return 0; |
1da177e4 | 457 | |
32b5913a | 458 | /* We only need the first two bytes. */ |
c008ba5b HX |
459 | rfv->hlen = 2; |
460 | ||
b61e9dcc | 461 | err = memcpy_from_msg(rfv->hdr.c, rfv->msg, rfv->hlen); |
32b5913a HX |
462 | if (err) |
463 | return err; | |
464 | ||
c008ba5b HX |
465 | fl4->fl4_icmp_type = rfv->hdr.icmph.type; |
466 | fl4->fl4_icmp_code = rfv->hdr.icmph.code; | |
32b5913a | 467 | |
a27b58fe | 468 | return 0; |
1da177e4 LT |
469 | } |
470 | ||
c008ba5b HX |
471 | static int raw_getfrag(void *from, char *to, int offset, int len, int odd, |
472 | struct sk_buff *skb) | |
473 | { | |
474 | struct raw_frag_vec *rfv = from; | |
475 | ||
476 | if (offset < rfv->hlen) { | |
477 | int copy = min(rfv->hlen - offset, len); | |
478 | ||
479 | if (skb->ip_summed == CHECKSUM_PARTIAL) | |
480 | memcpy(to, rfv->hdr.c + offset, copy); | |
481 | else | |
482 | skb->csum = csum_block_add( | |
483 | skb->csum, | |
484 | csum_partial_copy_nocheck(rfv->hdr.c + offset, | |
485 | to, copy, 0), | |
486 | odd); | |
487 | ||
488 | odd = 0; | |
489 | offset += copy; | |
490 | to += copy; | |
491 | len -= copy; | |
492 | ||
493 | if (!len) | |
494 | return 0; | |
495 | } | |
496 | ||
497 | offset -= rfv->hlen; | |
498 | ||
f69e6d13 | 499 | return ip_generic_getfrag(rfv->msg, to, offset, len, odd, skb); |
c008ba5b HX |
500 | } |
501 | ||
1b784140 | 502 | static int raw_sendmsg(struct sock *sk, struct msghdr *msg, size_t len) |
1da177e4 LT |
503 | { |
504 | struct inet_sock *inet = inet_sk(sk); | |
bb191c3e | 505 | struct net *net = sock_net(sk); |
1da177e4 LT |
506 | struct ipcm_cookie ipc; |
507 | struct rtable *rt = NULL; | |
77968b78 | 508 | struct flowi4 fl4; |
1da177e4 | 509 | int free = 0; |
3ca3c68e | 510 | __be32 daddr; |
c1d18f9f | 511 | __be32 saddr; |
1da177e4 LT |
512 | u8 tos; |
513 | int err; | |
f6d8bd05 | 514 | struct ip_options_data opt_copy; |
c008ba5b | 515 | struct raw_frag_vec rfv; |
8f659a03 | 516 | int hdrincl; |
1da177e4 LT |
517 | |
518 | err = -EMSGSIZE; | |
926d4b81 | 519 | if (len > 0xFFFF) |
1da177e4 LT |
520 | goto out; |
521 | ||
8f659a03 | 522 | /* hdrincl should be READ_ONCE(inet->hdrincl) |
20b50d79 NS |
523 | * but READ_ONCE() doesn't work with bit fields. |
524 | * Doing this indirectly yields the same result. | |
8f659a03 MG |
525 | */ |
526 | hdrincl = inet->hdrincl; | |
20b50d79 | 527 | hdrincl = READ_ONCE(hdrincl); |
1da177e4 LT |
528 | /* |
529 | * Check the flags. | |
530 | */ | |
531 | ||
532 | err = -EOPNOTSUPP; | |
533 | if (msg->msg_flags & MSG_OOB) /* Mirror BSD error message */ | |
534 | goto out; /* compatibility */ | |
e905a9ed | 535 | |
1da177e4 | 536 | /* |
e905a9ed | 537 | * Get and verify the address. |
1da177e4 LT |
538 | */ |
539 | ||
540 | if (msg->msg_namelen) { | |
342dfc30 | 541 | DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name); |
1da177e4 LT |
542 | err = -EINVAL; |
543 | if (msg->msg_namelen < sizeof(*usin)) | |
544 | goto out; | |
545 | if (usin->sin_family != AF_INET) { | |
058bd4d2 JP |
546 | pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n", |
547 | __func__, current->comm); | |
1da177e4 LT |
548 | err = -EAFNOSUPPORT; |
549 | if (usin->sin_family) | |
550 | goto out; | |
551 | } | |
552 | daddr = usin->sin_addr.s_addr; | |
553 | /* ANK: I did not forget to get protocol from port field. | |
554 | * I just do not know, who uses this weirdness. | |
555 | * IP_HDRINCL is much more convenient. | |
556 | */ | |
557 | } else { | |
558 | err = -EDESTADDRREQ; | |
e905a9ed | 559 | if (sk->sk_state != TCP_ESTABLISHED) |
1da177e4 | 560 | goto out; |
c720c7e8 | 561 | daddr = inet->inet_daddr; |
1da177e4 LT |
562 | } |
563 | ||
35178206 | 564 | ipcm_init_sk(&ipc, inet); |
1da177e4 LT |
565 | |
566 | if (msg->msg_controllen) { | |
24025c46 | 567 | err = ip_cmsg_send(sk, msg, &ipc, false); |
91948309 ED |
568 | if (unlikely(err)) { |
569 | kfree(ipc.opt); | |
1da177e4 | 570 | goto out; |
91948309 | 571 | } |
1da177e4 LT |
572 | if (ipc.opt) |
573 | free = 1; | |
574 | } | |
575 | ||
576 | saddr = ipc.addr; | |
577 | ipc.addr = daddr; | |
578 | ||
f6d8bd05 ED |
579 | if (!ipc.opt) { |
580 | struct ip_options_rcu *inet_opt; | |
581 | ||
582 | rcu_read_lock(); | |
583 | inet_opt = rcu_dereference(inet->inet_opt); | |
584 | if (inet_opt) { | |
585 | memcpy(&opt_copy, inet_opt, | |
586 | sizeof(*inet_opt) + inet_opt->opt.optlen); | |
587 | ipc.opt = &opt_copy.opt; | |
588 | } | |
589 | rcu_read_unlock(); | |
590 | } | |
1da177e4 LT |
591 | |
592 | if (ipc.opt) { | |
593 | err = -EINVAL; | |
594 | /* Linux does not mangle headers on raw sockets, | |
595 | * so that IP options + IP_HDRINCL is non-sense. | |
596 | */ | |
8f659a03 | 597 | if (hdrincl) |
1da177e4 | 598 | goto done; |
f6d8bd05 | 599 | if (ipc.opt->opt.srr) { |
1da177e4 LT |
600 | if (!daddr) |
601 | goto done; | |
f6d8bd05 | 602 | daddr = ipc.opt->opt.faddr; |
1da177e4 LT |
603 | } |
604 | } | |
aa661581 | 605 | tos = get_rtconn_flags(&ipc, sk); |
1da177e4 LT |
606 | if (msg->msg_flags & MSG_DONTROUTE) |
607 | tos |= RTO_ONLINK; | |
608 | ||
f97c1e0c | 609 | if (ipv4_is_multicast(daddr)) { |
854da991 | 610 | if (!ipc.oif || netif_index_is_l3_master(sock_net(sk), ipc.oif)) |
1da177e4 LT |
611 | ipc.oif = inet->mc_index; |
612 | if (!saddr) | |
613 | saddr = inet->mc_addr; | |
9515a2e0 | 614 | } else if (!ipc.oif) { |
76e21053 | 615 | ipc.oif = inet->uc_index; |
9515a2e0 DA |
616 | } else if (ipv4_is_lbcast(daddr) && inet->uc_index) { |
617 | /* oif is set, packet is to local broadcast and | |
618 | * and uc_index is set. oif is most likely set | |
619 | * by sk_bound_dev_if. If uc_index != oif check if the | |
620 | * oif is an L3 master and uc_index is an L3 slave. | |
621 | * If so, we want to allow the send using the uc_index. | |
622 | */ | |
623 | if (ipc.oif != inet->uc_index && | |
624 | ipc.oif == l3mdev_master_ifindex_by_index(sock_net(sk), | |
625 | inet->uc_index)) { | |
626 | ipc.oif = inet->uc_index; | |
627 | } | |
628 | } | |
1da177e4 | 629 | |
77968b78 DM |
630 | flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos, |
631 | RT_SCOPE_UNIVERSE, | |
8f659a03 | 632 | hdrincl ? IPPROTO_RAW : sk->sk_protocol, |
0e0d44ab | 633 | inet_sk_flowi_flags(sk) | |
8f659a03 | 634 | (hdrincl ? FLOWI_FLAG_KNOWN_NH : 0), |
e2d118a1 | 635 | daddr, saddr, 0, 0, sk->sk_uid); |
ef164ae3 | 636 | |
8f659a03 | 637 | if (!hdrincl) { |
b61e9dcc | 638 | rfv.msg = msg; |
c008ba5b HX |
639 | rfv.hlen = 0; |
640 | ||
641 | err = raw_probe_proto_opt(&rfv, &fl4); | |
77968b78 | 642 | if (err) |
b23dd4fe | 643 | goto done; |
77968b78 DM |
644 | } |
645 | ||
646 | security_sk_classify_flow(sk, flowi4_to_flowi(&fl4)); | |
bb191c3e | 647 | rt = ip_route_output_flow(net, &fl4, sk); |
77968b78 DM |
648 | if (IS_ERR(rt)) { |
649 | err = PTR_ERR(rt); | |
650 | rt = NULL; | |
651 | goto done; | |
1da177e4 | 652 | } |
1da177e4 LT |
653 | |
654 | err = -EACCES; | |
655 | if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST)) | |
656 | goto done; | |
657 | ||
658 | if (msg->msg_flags & MSG_CONFIRM) | |
659 | goto do_confirm; | |
660 | back_from_confirm: | |
661 | ||
8f659a03 | 662 | if (hdrincl) |
7ae9abfd | 663 | err = raw_send_hdrinc(sk, &fl4, msg, len, |
c14ac945 | 664 | &rt, msg->msg_flags, &ipc.sockc); |
e905a9ed | 665 | |
1da177e4 LT |
666 | else { |
667 | if (!ipc.addr) | |
77968b78 | 668 | ipc.addr = fl4.daddr; |
1da177e4 | 669 | lock_sock(sk); |
c008ba5b HX |
670 | err = ip_append_data(sk, &fl4, raw_getfrag, |
671 | &rfv, len, 0, | |
77968b78 | 672 | &ipc, &rt, msg->msg_flags); |
1da177e4 LT |
673 | if (err) |
674 | ip_flush_pending_frames(sk); | |
6ce9e7b5 | 675 | else if (!(msg->msg_flags & MSG_MORE)) { |
77968b78 | 676 | err = ip_push_pending_frames(sk, &fl4); |
6ce9e7b5 ED |
677 | if (err == -ENOBUFS && !inet->recverr) |
678 | err = 0; | |
679 | } | |
1da177e4 LT |
680 | release_sock(sk); |
681 | } | |
682 | done: | |
683 | if (free) | |
684 | kfree(ipc.opt); | |
685 | ip_rt_put(rt); | |
686 | ||
5418c692 JJ |
687 | out: |
688 | if (err < 0) | |
689 | return err; | |
690 | return len; | |
1da177e4 LT |
691 | |
692 | do_confirm: | |
0dec879f JA |
693 | if (msg->msg_flags & MSG_PROBE) |
694 | dst_confirm_neigh(&rt->dst, &fl4.daddr); | |
1da177e4 LT |
695 | if (!(msg->msg_flags & MSG_PROBE) || len) |
696 | goto back_from_confirm; | |
697 | err = 0; | |
698 | goto done; | |
699 | } | |
700 | ||
701 | static void raw_close(struct sock *sk, long timeout) | |
702 | { | |
e905a9ed | 703 | /* |
25985edc | 704 | * Raw sockets may have direct kernel references. Kill them. |
1da177e4 LT |
705 | */ |
706 | ip_ra_control(sk, 0, NULL); | |
707 | ||
708 | sk_common_release(sk); | |
709 | } | |
710 | ||
7d06b2e0 | 711 | static void raw_destroy(struct sock *sk) |
22dd4850 DL |
712 | { |
713 | lock_sock(sk); | |
714 | ip_flush_pending_frames(sk); | |
715 | release_sock(sk); | |
22dd4850 DL |
716 | } |
717 | ||
1da177e4 LT |
718 | /* This gets rid of all the nasties in af_inet. -DaveM */ |
719 | static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len) | |
720 | { | |
721 | struct inet_sock *inet = inet_sk(sk); | |
722 | struct sockaddr_in *addr = (struct sockaddr_in *) uaddr; | |
cd2c0f45 | 723 | u32 tb_id = RT_TABLE_LOCAL; |
1da177e4 LT |
724 | int ret = -EINVAL; |
725 | int chk_addr_ret; | |
726 | ||
727 | if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in)) | |
728 | goto out; | |
cd2c0f45 DA |
729 | |
730 | if (sk->sk_bound_dev_if) | |
731 | tb_id = l3mdev_fib_table_by_index(sock_net(sk), | |
732 | sk->sk_bound_dev_if) ? : tb_id; | |
733 | ||
734 | chk_addr_ret = inet_addr_type_table(sock_net(sk), addr->sin_addr.s_addr, | |
735 | tb_id); | |
736 | ||
1da177e4 LT |
737 | ret = -EADDRNOTAVAIL; |
738 | if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL && | |
739 | chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST) | |
740 | goto out; | |
c720c7e8 | 741 | inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr; |
1da177e4 | 742 | if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST) |
c720c7e8 | 743 | inet->inet_saddr = 0; /* Use device */ |
1da177e4 LT |
744 | sk_dst_reset(sk); |
745 | ret = 0; | |
746 | out: return ret; | |
747 | } | |
748 | ||
749 | /* | |
750 | * This should be easy, if there is something there | |
751 | * we return it, otherwise we block. | |
752 | */ | |
753 | ||
1b784140 YX |
754 | static int raw_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, |
755 | int noblock, int flags, int *addr_len) | |
1da177e4 LT |
756 | { |
757 | struct inet_sock *inet = inet_sk(sk); | |
758 | size_t copied = 0; | |
759 | int err = -EOPNOTSUPP; | |
342dfc30 | 760 | DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name); |
1da177e4 LT |
761 | struct sk_buff *skb; |
762 | ||
763 | if (flags & MSG_OOB) | |
764 | goto out; | |
765 | ||
1da177e4 | 766 | if (flags & MSG_ERRQUEUE) { |
85fbaa75 | 767 | err = ip_recv_error(sk, msg, len, addr_len); |
1da177e4 LT |
768 | goto out; |
769 | } | |
770 | ||
771 | skb = skb_recv_datagram(sk, flags, noblock, &err); | |
772 | if (!skb) | |
773 | goto out; | |
774 | ||
775 | copied = skb->len; | |
776 | if (len < copied) { | |
777 | msg->msg_flags |= MSG_TRUNC; | |
778 | copied = len; | |
779 | } | |
780 | ||
51f3d02b | 781 | err = skb_copy_datagram_msg(skb, 0, msg, copied); |
1da177e4 LT |
782 | if (err) |
783 | goto done; | |
784 | ||
3b885787 | 785 | sock_recv_ts_and_drops(msg, sk, skb); |
1da177e4 LT |
786 | |
787 | /* Copy the address. */ | |
788 | if (sin) { | |
789 | sin->sin_family = AF_INET; | |
eddc9ec5 | 790 | sin->sin_addr.s_addr = ip_hdr(skb)->saddr; |
f59fc7f3 | 791 | sin->sin_port = 0; |
1da177e4 | 792 | memset(&sin->sin_zero, 0, sizeof(sin->sin_zero)); |
bceaa902 | 793 | *addr_len = sizeof(*sin); |
1da177e4 LT |
794 | } |
795 | if (inet->cmsg_flags) | |
796 | ip_cmsg_recv(msg, skb); | |
797 | if (flags & MSG_TRUNC) | |
798 | copied = skb->len; | |
799 | done: | |
800 | skb_free_datagram(sk, skb); | |
5418c692 JJ |
801 | out: |
802 | if (err) | |
803 | return err; | |
804 | return copied; | |
1da177e4 LT |
805 | } |
806 | ||
6897445f | 807 | static int raw_sk_init(struct sock *sk) |
1da177e4 LT |
808 | { |
809 | struct raw_sock *rp = raw_sk(sk); | |
810 | ||
c720c7e8 | 811 | if (inet_sk(sk)->inet_num == IPPROTO_ICMP) |
1da177e4 LT |
812 | memset(&rp->filter, 0, sizeof(rp->filter)); |
813 | return 0; | |
814 | } | |
815 | ||
816 | static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen) | |
817 | { | |
818 | if (optlen > sizeof(struct icmp_filter)) | |
819 | optlen = sizeof(struct icmp_filter); | |
820 | if (copy_from_user(&raw_sk(sk)->filter, optval, optlen)) | |
821 | return -EFAULT; | |
822 | return 0; | |
823 | } | |
824 | ||
825 | static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen) | |
826 | { | |
827 | int len, ret = -EFAULT; | |
828 | ||
829 | if (get_user(len, optlen)) | |
830 | goto out; | |
831 | ret = -EINVAL; | |
832 | if (len < 0) | |
833 | goto out; | |
834 | if (len > sizeof(struct icmp_filter)) | |
835 | len = sizeof(struct icmp_filter); | |
836 | ret = -EFAULT; | |
837 | if (put_user(len, optlen) || | |
838 | copy_to_user(optval, &raw_sk(sk)->filter, len)) | |
839 | goto out; | |
840 | ret = 0; | |
841 | out: return ret; | |
842 | } | |
843 | ||
3fdadf7d | 844 | static int do_raw_setsockopt(struct sock *sk, int level, int optname, |
b7058842 | 845 | char __user *optval, unsigned int optlen) |
1da177e4 | 846 | { |
1da177e4 | 847 | if (optname == ICMP_FILTER) { |
c720c7e8 | 848 | if (inet_sk(sk)->inet_num != IPPROTO_ICMP) |
1da177e4 LT |
849 | return -EOPNOTSUPP; |
850 | else | |
851 | return raw_seticmpfilter(sk, optval, optlen); | |
852 | } | |
853 | return -ENOPROTOOPT; | |
854 | } | |
855 | ||
3fdadf7d | 856 | static int raw_setsockopt(struct sock *sk, int level, int optname, |
b7058842 | 857 | char __user *optval, unsigned int optlen) |
1da177e4 LT |
858 | { |
859 | if (level != SOL_RAW) | |
3fdadf7d DM |
860 | return ip_setsockopt(sk, level, optname, optval, optlen); |
861 | return do_raw_setsockopt(sk, level, optname, optval, optlen); | |
862 | } | |
1da177e4 | 863 | |
3fdadf7d DM |
864 | #ifdef CONFIG_COMPAT |
865 | static int compat_raw_setsockopt(struct sock *sk, int level, int optname, | |
b7058842 | 866 | char __user *optval, unsigned int optlen) |
3fdadf7d DM |
867 | { |
868 | if (level != SOL_RAW) | |
543d9cfe | 869 | return compat_ip_setsockopt(sk, level, optname, optval, optlen); |
3fdadf7d DM |
870 | return do_raw_setsockopt(sk, level, optname, optval, optlen); |
871 | } | |
872 | #endif | |
873 | ||
874 | static int do_raw_getsockopt(struct sock *sk, int level, int optname, | |
875 | char __user *optval, int __user *optlen) | |
876 | { | |
1da177e4 | 877 | if (optname == ICMP_FILTER) { |
c720c7e8 | 878 | if (inet_sk(sk)->inet_num != IPPROTO_ICMP) |
1da177e4 LT |
879 | return -EOPNOTSUPP; |
880 | else | |
881 | return raw_geticmpfilter(sk, optval, optlen); | |
882 | } | |
883 | return -ENOPROTOOPT; | |
884 | } | |
885 | ||
3fdadf7d DM |
886 | static int raw_getsockopt(struct sock *sk, int level, int optname, |
887 | char __user *optval, int __user *optlen) | |
888 | { | |
889 | if (level != SOL_RAW) | |
890 | return ip_getsockopt(sk, level, optname, optval, optlen); | |
891 | return do_raw_getsockopt(sk, level, optname, optval, optlen); | |
892 | } | |
893 | ||
894 | #ifdef CONFIG_COMPAT | |
895 | static int compat_raw_getsockopt(struct sock *sk, int level, int optname, | |
543d9cfe | 896 | char __user *optval, int __user *optlen) |
3fdadf7d DM |
897 | { |
898 | if (level != SOL_RAW) | |
543d9cfe | 899 | return compat_ip_getsockopt(sk, level, optname, optval, optlen); |
3fdadf7d DM |
900 | return do_raw_getsockopt(sk, level, optname, optval, optlen); |
901 | } | |
902 | #endif | |
903 | ||
1da177e4 LT |
904 | static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg) |
905 | { | |
906 | switch (cmd) { | |
4500ebf8 JP |
907 | case SIOCOUTQ: { |
908 | int amount = sk_wmem_alloc_get(sk); | |
31e6d363 | 909 | |
4500ebf8 JP |
910 | return put_user(amount, (int __user *)arg); |
911 | } | |
912 | case SIOCINQ: { | |
913 | struct sk_buff *skb; | |
914 | int amount = 0; | |
915 | ||
916 | spin_lock_bh(&sk->sk_receive_queue.lock); | |
917 | skb = skb_peek(&sk->sk_receive_queue); | |
00db4124 | 918 | if (skb) |
4500ebf8 JP |
919 | amount = skb->len; |
920 | spin_unlock_bh(&sk->sk_receive_queue.lock); | |
921 | return put_user(amount, (int __user *)arg); | |
922 | } | |
1da177e4 | 923 | |
4500ebf8 | 924 | default: |
1da177e4 | 925 | #ifdef CONFIG_IP_MROUTE |
4500ebf8 | 926 | return ipmr_ioctl(sk, cmd, (void __user *)arg); |
1da177e4 | 927 | #else |
4500ebf8 | 928 | return -ENOIOCTLCMD; |
1da177e4 LT |
929 | #endif |
930 | } | |
931 | } | |
932 | ||
709b46e8 EB |
933 | #ifdef CONFIG_COMPAT |
934 | static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg) | |
935 | { | |
936 | switch (cmd) { | |
937 | case SIOCOUTQ: | |
938 | case SIOCINQ: | |
939 | return -ENOIOCTLCMD; | |
940 | default: | |
941 | #ifdef CONFIG_IP_MROUTE | |
942 | return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg)); | |
943 | #else | |
944 | return -ENOIOCTLCMD; | |
945 | #endif | |
946 | } | |
947 | } | |
948 | #endif | |
949 | ||
432490f9 CG |
950 | int raw_abort(struct sock *sk, int err) |
951 | { | |
952 | lock_sock(sk); | |
953 | ||
954 | sk->sk_err = err; | |
955 | sk->sk_error_report(sk); | |
3de864f8 | 956 | __udp_disconnect(sk, 0); |
432490f9 CG |
957 | |
958 | release_sock(sk); | |
959 | ||
960 | return 0; | |
961 | } | |
962 | EXPORT_SYMBOL_GPL(raw_abort); | |
963 | ||
1da177e4 | 964 | struct proto raw_prot = { |
543d9cfe ACM |
965 | .name = "RAW", |
966 | .owner = THIS_MODULE, | |
967 | .close = raw_close, | |
22dd4850 | 968 | .destroy = raw_destroy, |
543d9cfe | 969 | .connect = ip4_datagram_connect, |
286c72de | 970 | .disconnect = __udp_disconnect, |
543d9cfe | 971 | .ioctl = raw_ioctl, |
6897445f | 972 | .init = raw_sk_init, |
543d9cfe ACM |
973 | .setsockopt = raw_setsockopt, |
974 | .getsockopt = raw_getsockopt, | |
975 | .sendmsg = raw_sendmsg, | |
976 | .recvmsg = raw_recvmsg, | |
977 | .bind = raw_bind, | |
978 | .backlog_rcv = raw_rcv_skb, | |
8141ed9f | 979 | .release_cb = ip4_datagram_release_cb, |
fc8717ba PE |
980 | .hash = raw_hash_sk, |
981 | .unhash = raw_unhash_sk, | |
543d9cfe | 982 | .obj_size = sizeof(struct raw_sock), |
8c2bc895 DW |
983 | .useroffset = offsetof(struct raw_sock, filter), |
984 | .usersize = sizeof_field(struct raw_sock, filter), | |
fc8717ba | 985 | .h.raw_hash = &raw_v4_hashinfo, |
3fdadf7d | 986 | #ifdef CONFIG_COMPAT |
543d9cfe ACM |
987 | .compat_setsockopt = compat_raw_setsockopt, |
988 | .compat_getsockopt = compat_raw_getsockopt, | |
709b46e8 | 989 | .compat_ioctl = compat_raw_ioctl, |
3fdadf7d | 990 | #endif |
432490f9 | 991 | .diag_destroy = raw_abort, |
1da177e4 LT |
992 | }; |
993 | ||
994 | #ifdef CONFIG_PROC_FS | |
1da177e4 LT |
995 | static struct sock *raw_get_first(struct seq_file *seq) |
996 | { | |
997 | struct sock *sk; | |
93cb5a1f | 998 | struct raw_hashinfo *h = PDE_DATA(file_inode(seq->file)); |
5a5f3a8d | 999 | struct raw_iter_state *state = raw_seq_private(seq); |
1da177e4 | 1000 | |
b673e4df PE |
1001 | for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE; |
1002 | ++state->bucket) { | |
93cb5a1f | 1003 | sk_for_each(sk, &h->ht[state->bucket]) |
1218854a | 1004 | if (sock_net(sk) == seq_file_net(seq)) |
1da177e4 LT |
1005 | goto found; |
1006 | } | |
1007 | sk = NULL; | |
1008 | found: | |
1009 | return sk; | |
1010 | } | |
1011 | ||
1012 | static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk) | |
1013 | { | |
93cb5a1f | 1014 | struct raw_hashinfo *h = PDE_DATA(file_inode(seq->file)); |
5a5f3a8d | 1015 | struct raw_iter_state *state = raw_seq_private(seq); |
1da177e4 LT |
1016 | |
1017 | do { | |
1018 | sk = sk_next(sk); | |
1019 | try_again: | |
1020 | ; | |
1218854a | 1021 | } while (sk && sock_net(sk) != seq_file_net(seq)); |
1da177e4 | 1022 | |
b673e4df | 1023 | if (!sk && ++state->bucket < RAW_HTABLE_SIZE) { |
93cb5a1f | 1024 | sk = sk_head(&h->ht[state->bucket]); |
1da177e4 LT |
1025 | goto try_again; |
1026 | } | |
1027 | return sk; | |
1028 | } | |
1029 | ||
1030 | static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos) | |
1031 | { | |
1032 | struct sock *sk = raw_get_first(seq); | |
1033 | ||
1034 | if (sk) | |
1035 | while (pos && (sk = raw_get_next(seq, sk)) != NULL) | |
1036 | --pos; | |
1037 | return pos ? NULL : sk; | |
1038 | } | |
1039 | ||
42a73808 | 1040 | void *raw_seq_start(struct seq_file *seq, loff_t *pos) |
1da177e4 | 1041 | { |
93cb5a1f | 1042 | struct raw_hashinfo *h = PDE_DATA(file_inode(seq->file)); |
42a73808 | 1043 | |
93cb5a1f | 1044 | read_lock(&h->lock); |
1da177e4 LT |
1045 | return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN; |
1046 | } | |
42a73808 | 1047 | EXPORT_SYMBOL_GPL(raw_seq_start); |
1da177e4 | 1048 | |
42a73808 | 1049 | void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos) |
1da177e4 LT |
1050 | { |
1051 | struct sock *sk; | |
1052 | ||
1053 | if (v == SEQ_START_TOKEN) | |
1054 | sk = raw_get_first(seq); | |
1055 | else | |
1056 | sk = raw_get_next(seq, v); | |
1057 | ++*pos; | |
1058 | return sk; | |
1059 | } | |
42a73808 | 1060 | EXPORT_SYMBOL_GPL(raw_seq_next); |
1da177e4 | 1061 | |
42a73808 | 1062 | void raw_seq_stop(struct seq_file *seq, void *v) |
1da177e4 | 1063 | { |
93cb5a1f | 1064 | struct raw_hashinfo *h = PDE_DATA(file_inode(seq->file)); |
42a73808 | 1065 | |
93cb5a1f | 1066 | read_unlock(&h->lock); |
1da177e4 | 1067 | } |
42a73808 | 1068 | EXPORT_SYMBOL_GPL(raw_seq_stop); |
1da177e4 | 1069 | |
8cd850ef | 1070 | static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i) |
1da177e4 LT |
1071 | { |
1072 | struct inet_sock *inet = inet_sk(sp); | |
c720c7e8 ED |
1073 | __be32 dest = inet->inet_daddr, |
1074 | src = inet->inet_rcv_saddr; | |
1da177e4 | 1075 | __u16 destp = 0, |
c720c7e8 | 1076 | srcp = inet->inet_num; |
1da177e4 | 1077 | |
8cd850ef | 1078 | seq_printf(seq, "%4d: %08X:%04X %08X:%04X" |
d14c5ab6 | 1079 | " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d\n", |
e905a9ed | 1080 | i, src, srcp, dest, destp, sp->sk_state, |
31e6d363 ED |
1081 | sk_wmem_alloc_get(sp), |
1082 | sk_rmem_alloc_get(sp), | |
a7cb5a49 EB |
1083 | 0, 0L, 0, |
1084 | from_kuid_munged(seq_user_ns(seq), sock_i_uid(sp)), | |
1085 | 0, sock_i_ino(sp), | |
41c6d650 | 1086 | refcount_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops)); |
1da177e4 LT |
1087 | } |
1088 | ||
1089 | static int raw_seq_show(struct seq_file *seq, void *v) | |
1090 | { | |
1da177e4 | 1091 | if (v == SEQ_START_TOKEN) |
8cd850ef DL |
1092 | seq_printf(seq, " sl local_address rem_address st tx_queue " |
1093 | "rx_queue tr tm->when retrnsmt uid timeout " | |
cb61cb9b | 1094 | "inode ref pointer drops\n"); |
8cd850ef DL |
1095 | else |
1096 | raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket); | |
1da177e4 LT |
1097 | return 0; |
1098 | } | |
1099 | ||
f690808e | 1100 | static const struct seq_operations raw_seq_ops = { |
1da177e4 LT |
1101 | .start = raw_seq_start, |
1102 | .next = raw_seq_next, | |
1103 | .stop = raw_seq_stop, | |
1104 | .show = raw_seq_show, | |
1105 | }; | |
1106 | ||
a308da16 | 1107 | static __net_init int raw_init_net(struct net *net) |
1da177e4 | 1108 | { |
c3506372 CH |
1109 | if (!proc_create_net_data("raw", 0444, net->proc_net, &raw_seq_ops, |
1110 | sizeof(struct raw_iter_state), &raw_v4_hashinfo)) | |
1da177e4 | 1111 | return -ENOMEM; |
a308da16 | 1112 | |
1da177e4 LT |
1113 | return 0; |
1114 | } | |
1115 | ||
a308da16 PE |
1116 | static __net_exit void raw_exit_net(struct net *net) |
1117 | { | |
ece31ffd | 1118 | remove_proc_entry("raw", net->proc_net); |
a308da16 PE |
1119 | } |
1120 | ||
1121 | static __net_initdata struct pernet_operations raw_net_ops = { | |
1122 | .init = raw_init_net, | |
1123 | .exit = raw_exit_net, | |
1124 | }; | |
1125 | ||
1126 | int __init raw_proc_init(void) | |
1127 | { | |
1128 | return register_pernet_subsys(&raw_net_ops); | |
1129 | } | |
1130 | ||
1da177e4 LT |
1131 | void __init raw_proc_exit(void) |
1132 | { | |
a308da16 | 1133 | unregister_pernet_subsys(&raw_net_ops); |
1da177e4 | 1134 | } |
6897445f MM |
1135 | |
1136 | static void raw_sysctl_init_net(struct net *net) | |
1137 | { | |
1138 | #ifdef CONFIG_NET_L3_MASTER_DEV | |
1139 | net->ipv4.sysctl_raw_l3mdev_accept = 1; | |
1140 | #endif | |
1141 | } | |
1142 | ||
1143 | static int __net_init raw_sysctl_init(struct net *net) | |
1144 | { | |
1145 | raw_sysctl_init_net(net); | |
1146 | return 0; | |
1147 | } | |
1148 | ||
1149 | static struct pernet_operations __net_initdata raw_sysctl_ops = { | |
1150 | .init = raw_sysctl_init, | |
1151 | }; | |
1152 | ||
1153 | void __init raw_init(void) | |
1154 | { | |
1155 | raw_sysctl_init_net(&init_net); | |
1156 | if (register_pernet_subsys(&raw_sysctl_ops)) | |
1157 | panic("RAW: failed to init sysctl parameters.\n"); | |
1158 | } | |
1da177e4 | 1159 | #endif /* CONFIG_PROC_FS */ |