blob: ab466305b629bae03bbc44767156d3fb467684cf [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
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 *
Jesper Juhl02c30a82005-05-05 16:16:16 -07008 * Authors: Ross Biro
Linus Torvalds1da177e2005-04-16 15:20:36 -07009 * 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 */
Dave Jonesbf0d5242006-09-22 14:00:29 -070039
Alan Cox715b49e2006-01-18 17:44:07 -080040#include <linux/types.h>
Arun Sharma600634972011-07-26 16:09:06 -070041#include <linux/atomic.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070042#include <asm/byteorder.h>
43#include <asm/current.h>
44#include <asm/uaccess.h>
45#include <asm/ioctls.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070046#include <linux/stddef.h>
47#include <linux/slab.h>
48#include <linux/errno.h>
49#include <linux/aio.h>
50#include <linux/kernel.h>
Paul Gortmakerbc3b2d72011-07-15 11:47:34 -040051#include <linux/export.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070052#include <linux/spinlock.h>
53#include <linux/sockios.h>
54#include <linux/socket.h>
55#include <linux/in.h>
56#include <linux/mroute.h>
57#include <linux/netdevice.h>
58#include <linux/in_route.h>
59#include <linux/route.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070060#include <linux/skbuff.h>
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020061#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070062#include <net/dst.h>
63#include <net/sock.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070064#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>
Arnaldo Carvalho de Meloc752f072005-08-09 20:08:28 -070071#include <net/tcp_states.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070072#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>
Eric W. Biederman709b46e2011-01-29 16:15:56 +000080#include <linux/compat.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070081
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080082static struct raw_hashinfo raw_v4_hashinfo = {
Robert P. J. Day938b93a2008-03-18 00:59:23 -070083 .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock),
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080084};
Linus Torvalds1da177e2005-04-16 15:20:36 -070085
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -070086void raw_hash_sk(struct sock *sk)
Linus Torvalds1da177e2005-04-16 15:20:36 -070087{
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -070088 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
Pavel Emelyanov65b4c502007-11-19 22:37:24 -080089 struct hlist_head *head;
Linus Torvalds1da177e2005-04-16 15:20:36 -070090
Eric Dumazetc720c7e2009-10-15 06:30:45 +000091 head = &h->ht[inet_sk(sk)->inet_num & (RAW_HTABLE_SIZE - 1)];
Pavel Emelyanov65b4c502007-11-19 22:37:24 -080092
93 write_lock_bh(&h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -070094 sk_add_node(sk, head);
Pavel Emelyanovc29a0bc2008-03-31 19:41:46 -070095 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
Pavel Emelyanov65b4c502007-11-19 22:37:24 -080096 write_unlock_bh(&h->lock);
97}
98EXPORT_SYMBOL_GPL(raw_hash_sk);
99
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700100void raw_unhash_sk(struct sock *sk)
Pavel Emelyanovab707682007-11-19 22:37:58 -0800101{
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700102 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
103
Pavel Emelyanovab707682007-11-19 22:37:58 -0800104 write_lock_bh(&h->lock);
105 if (sk_del_node_init(sk))
Pavel Emelyanovc29a0bc2008-03-31 19:41:46 -0700106 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
Pavel Emelyanovab707682007-11-19 22:37:58 -0800107 write_unlock_bh(&h->lock);
108}
109EXPORT_SYMBOL_GPL(raw_unhash_sk);
110
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800111static struct sock *__raw_v4_lookup(struct net *net, struct sock *sk,
112 unsigned short num, __be32 raddr, __be32 laddr, int dif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700113{
114 struct hlist_node *node;
115
116 sk_for_each_from(sk, node) {
117 struct inet_sock *inet = inet_sk(sk);
118
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000119 if (net_eq(sock_net(sk), net) && inet->inet_num == num &&
120 !(inet->inet_daddr && inet->inet_daddr != raddr) &&
121 !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) &&
Linus Torvalds1da177e2005-04-16 15:20:36 -0700122 !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
123 goto found; /* gotcha */
124 }
125 sk = NULL;
126found:
127 return sk;
128}
129
130/*
131 * 0 - deliver
132 * 1 - block
133 */
134static __inline__ int icmp_filter(struct sock *sk, struct sk_buff *skb)
135{
136 int type;
137
138 if (!pskb_may_pull(skb, sizeof(struct icmphdr)))
139 return 1;
140
Arnaldo Carvalho de Melo88c76642007-03-13 14:43:18 -0300141 type = icmp_hdr(skb)->type;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700142 if (type < 32) {
143 __u32 data = raw_sk(sk)->filter.data;
144
145 return ((1 << type) & data) != 0;
146 }
147
148 /* Do not block unknown ICMP types */
149 return 0;
150}
151
152/* IP input processing comes here for RAW socket delivery.
153 * Caller owns SKB, so we must make clones.
154 *
155 * RFC 1122: SHOULD pass TOS value up to the transport layer.
156 * -> It does. And not only TOS, but all IP header.
157 */
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000158static int raw_v4_input(struct sk_buff *skb, const struct iphdr *iph, int hash)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700159{
160 struct sock *sk;
161 struct hlist_head *head;
Patrick McHardyd13964f2005-08-09 19:45:02 -0700162 int delivered = 0;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800163 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800165 read_lock(&raw_v4_hashinfo.lock);
166 head = &raw_v4_hashinfo.ht[hash];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700167 if (hlist_empty(head))
168 goto out;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800169
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900170 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800171 sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700172 iph->saddr, iph->daddr,
173 skb->dev->ifindex);
174
175 while (sk) {
Patrick McHardyd13964f2005-08-09 19:45:02 -0700176 delivered = 1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700177 if (iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) {
178 struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
179
180 /* Not releasing hash table! */
181 if (clone)
182 raw_rcv(sk, clone);
183 }
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800184 sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700185 iph->saddr, iph->daddr,
186 skb->dev->ifindex);
187 }
188out:
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800189 read_unlock(&raw_v4_hashinfo.lock);
Patrick McHardyd13964f2005-08-09 19:45:02 -0700190 return delivered;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700191}
192
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800193int raw_local_deliver(struct sk_buff *skb, int protocol)
194{
195 int hash;
196 struct sock *raw_sk;
197
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800198 hash = protocol & (RAW_HTABLE_SIZE - 1);
199 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800200
201 /* If there maybe a raw socket we must check - if not we
202 * don't care less
203 */
204 if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash))
205 raw_sk = NULL;
206
207 return raw_sk != NULL;
208
209}
210
211static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700212{
213 struct inet_sock *inet = inet_sk(sk);
Arnaldo Carvalho de Melo88c76642007-03-13 14:43:18 -0300214 const int type = icmp_hdr(skb)->type;
215 const int code = icmp_hdr(skb)->code;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700216 int err = 0;
217 int harderr = 0;
218
219 /* Report error on raw socket, if:
220 1. User requested ip_recverr.
221 2. Socket is connected (otherwise the error indication
222 is useless without ip_recverr and error is hard.
223 */
224 if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED)
225 return;
226
227 switch (type) {
228 default:
229 case ICMP_TIME_EXCEEDED:
230 err = EHOSTUNREACH;
231 break;
232 case ICMP_SOURCE_QUENCH:
233 return;
234 case ICMP_PARAMETERPROB:
235 err = EPROTO;
236 harderr = 1;
237 break;
238 case ICMP_DEST_UNREACH:
239 err = EHOSTUNREACH;
240 if (code > NR_ICMP_UNREACH)
241 break;
242 err = icmp_err_convert[code].errno;
243 harderr = icmp_err_convert[code].fatal;
244 if (code == ICMP_FRAG_NEEDED) {
245 harderr = inet->pmtudisc != IP_PMTUDISC_DONT;
246 err = EMSGSIZE;
247 }
248 }
249
250 if (inet->recverr) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000251 const struct iphdr *iph = (const struct iphdr *)skb->data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700252 u8 *payload = skb->data + (iph->ihl << 2);
253
254 if (inet->hdrincl)
255 payload = skb->data;
256 ip_icmp_error(sk, skb, err, 0, info, payload);
257 }
258
259 if (inet->recverr || harderr) {
260 sk->sk_err = err;
261 sk->sk_error_report(sk);
262 }
263}
264
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800265void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info)
266{
267 int hash;
268 struct sock *raw_sk;
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000269 const struct iphdr *iph;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800270 struct net *net;
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800271
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800272 hash = protocol & (RAW_HTABLE_SIZE - 1);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800273
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800274 read_lock(&raw_v4_hashinfo.lock);
275 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800276 if (raw_sk != NULL) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000277 iph = (const struct iphdr *)skb->data;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900278 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800279
280 while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol,
281 iph->daddr, iph->saddr,
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800282 skb->dev->ifindex)) != NULL) {
283 raw_err(raw_sk, skb, info);
284 raw_sk = sk_next(raw_sk);
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000285 iph = (const struct iphdr *)skb->data;
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800286 }
287 }
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800288 read_unlock(&raw_v4_hashinfo.lock);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800289}
290
Linus Torvalds1da177e2005-04-16 15:20:36 -0700291static int raw_rcv_skb(struct sock * sk, struct sk_buff * skb)
292{
293 /* Charge it to the socket. */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900294
Eric Dumazetd826eb12011-11-09 07:24:35 +0000295 ipv4_pktinfo_prepare(skb);
296 if (sock_queue_rcv_skb(sk, skb) < 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700297 kfree_skb(skb);
298 return NET_RX_DROP;
299 }
300
301 return NET_RX_SUCCESS;
302}
303
304int raw_rcv(struct sock *sk, struct sk_buff *skb)
305{
306 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
Wang Chen33c732c2007-11-13 20:30:01 -0800307 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700308 kfree_skb(skb);
309 return NET_RX_DROP;
310 }
Patrick McHardyb59c2702006-01-06 23:06:10 -0800311 nf_reset(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700312
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700313 skb_push(skb, skb->data - skb_network_header(skb));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700314
315 raw_rcv_skb(sk, skb);
316 return 0;
317}
318
David S. Miller77968b72011-05-08 17:12:19 -0700319static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4,
320 void *from, size_t length,
321 struct rtable **rtp,
322 unsigned int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700323{
324 struct inet_sock *inet = inet_sk(sk);
Pavel Emelyanov0388b002008-07-14 23:00:43 -0700325 struct net *net = sock_net(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700326 struct iphdr *iph;
327 struct sk_buff *skb;
Herbert Xuf844c742008-01-05 23:14:44 -0800328 unsigned int iphlen;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700329 int err;
Eric Dumazet1789a642010-06-03 22:23:57 +0000330 struct rtable *rt = *rtp;
Herbert Xu66088242011-11-18 02:20:04 +0000331 int hlen, tlen;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700332
Changli Gaod8d1f302010-06-10 23:31:35 -0700333 if (length > rt->dst.dev->mtu) {
David S. Miller77968b72011-05-08 17:12:19 -0700334 ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport,
Changli Gaod8d1f302010-06-10 23:31:35 -0700335 rt->dst.dev->mtu);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700336 return -EMSGSIZE;
337 }
338 if (flags&MSG_PROBE)
339 goto out;
340
Herbert Xu66088242011-11-18 02:20:04 +0000341 hlen = LL_RESERVED_SPACE(rt->dst.dev);
342 tlen = rt->dst.dev->needed_tailroom;
Johannes Bergf5184d22008-05-12 20:48:31 -0700343 skb = sock_alloc_send_skb(sk,
Herbert Xu66088242011-11-18 02:20:04 +0000344 length + hlen + tlen + 15,
Johannes Bergf5184d22008-05-12 20:48:31 -0700345 flags & MSG_DONTWAIT, &err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700346 if (skb == NULL)
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900347 goto error;
Herbert Xu66088242011-11-18 02:20:04 +0000348 skb_reserve(skb, hlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700349
350 skb->priority = sk->sk_priority;
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800351 skb->mark = sk->sk_mark;
Changli Gaod8d1f302010-06-10 23:31:35 -0700352 skb_dst_set(skb, &rt->dst);
Eric Dumazet1789a642010-06-03 22:23:57 +0000353 *rtp = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300355 skb_reset_network_header(skb);
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700356 iph = ip_hdr(skb);
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300357 skb_put(skb, length);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700358
359 skb->ip_summed = CHECKSUM_NONE;
360
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -0700361 skb->transport_header = skb->network_header;
Neil Horman55888df2009-10-28 08:59:47 +0000362 err = -EFAULT;
363 if (memcpy_fromiovecend((void *)iph, from, 0, length))
364 goto error_free;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700365
Herbert Xuf844c742008-01-05 23:14:44 -0800366 iphlen = iph->ihl * 4;
Neil Horman55888df2009-10-28 08:59:47 +0000367
368 /*
369 * We don't want to modify the ip header, but we do need to
370 * be sure that it won't cause problems later along the network
371 * stack. Specifically we want to make sure that iph->ihl is a
372 * sane value. If ihl points beyond the length of the buffer passed
373 * in, reject the frame as invalid
374 */
375 err = -EINVAL;
376 if (iphlen > length)
377 goto error_free;
378
379 if (iphlen >= sizeof(*iph)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700380 if (!iph->saddr)
David S. Miller77968b72011-05-08 17:12:19 -0700381 iph->saddr = fl4->saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382 iph->check = 0;
383 iph->tot_len = htons(length);
384 if (!iph->id)
Changli Gaod8d1f302010-06-10 23:31:35 -0700385 ip_select_ident(iph, &rt->dst, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386
387 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
388 }
David L Stevens96793b42007-09-17 09:57:33 -0700389 if (iph->protocol == IPPROTO_ICMP)
Pavel Emelyanov0388b002008-07-14 23:00:43 -0700390 icmp_out_count(net, ((struct icmphdr *)
David L Stevens96793b42007-09-17 09:57:33 -0700391 skb_transport_header(skb))->type);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700392
Jan Engelhardt9bbc7682010-03-23 04:07:29 +0100393 err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, skb, NULL,
Changli Gaod8d1f302010-06-10 23:31:35 -0700394 rt->dst.dev, dst_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395 if (err > 0)
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700396 err = net_xmit_errno(err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700397 if (err)
398 goto error;
399out:
400 return 0;
401
Neil Horman55888df2009-10-28 08:59:47 +0000402error_free:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700403 kfree_skb(skb);
404error:
Pavel Emelyanov5e38e272008-07-16 20:19:49 -0700405 IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700406 if (err == -ENOBUFS && !inet->recverr)
407 err = 0;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900408 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700409}
410
David S. Miller9d6ec932011-03-12 01:12:47 -0500411static int raw_probe_proto_opt(struct flowi4 *fl4, struct msghdr *msg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700412{
413 struct iovec *iov;
414 u8 __user *type = NULL;
415 u8 __user *code = NULL;
416 int probed = 0;
Jesper Juhl93765d82005-06-18 23:00:15 -0700417 unsigned int i;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700418
419 if (!msg->msg_iov)
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800420 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700421
422 for (i = 0; i < msg->msg_iovlen; i++) {
423 iov = &msg->msg_iov[i];
424 if (!iov)
425 continue;
426
David S. Miller9d6ec932011-03-12 01:12:47 -0500427 switch (fl4->flowi4_proto) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700428 case IPPROTO_ICMP:
429 /* check if one-byte field is readable or not. */
430 if (iov->iov_base && iov->iov_len < 1)
431 break;
432
433 if (!type) {
434 type = iov->iov_base;
435 /* check if code field is readable or not. */
436 if (iov->iov_len > 1)
437 code = type + 1;
438 } else if (!code)
439 code = iov->iov_base;
440
441 if (type && code) {
David S. Miller9cce96d2011-03-12 03:00:33 -0500442 if (get_user(fl4->fl4_icmp_type, type) ||
443 get_user(fl4->fl4_icmp_code, code))
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800444 return -EFAULT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700445 probed = 1;
446 }
447 break;
448 default:
449 probed = 1;
450 break;
451 }
452 if (probed)
453 break;
454 }
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800455 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700456}
457
458static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
459 size_t len)
460{
461 struct inet_sock *inet = inet_sk(sk);
462 struct ipcm_cookie ipc;
463 struct rtable *rt = NULL;
David S. Miller77968b72011-05-08 17:12:19 -0700464 struct flowi4 fl4;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700465 int free = 0;
Al Viro3ca3c682006-09-27 18:28:07 -0700466 __be32 daddr;
Al Viroc1d18f92006-09-27 18:28:28 -0700467 __be32 saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700468 u8 tos;
469 int err;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000470 struct ip_options_data opt_copy;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700471
472 err = -EMSGSIZE;
Jesper Juhl926d4b82005-06-18 23:00:00 -0700473 if (len > 0xFFFF)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700474 goto out;
475
476 /*
477 * Check the flags.
478 */
479
480 err = -EOPNOTSUPP;
481 if (msg->msg_flags & MSG_OOB) /* Mirror BSD error message */
482 goto out; /* compatibility */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900483
Linus Torvalds1da177e2005-04-16 15:20:36 -0700484 /*
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900485 * Get and verify the address.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486 */
487
488 if (msg->msg_namelen) {
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800489 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490 err = -EINVAL;
491 if (msg->msg_namelen < sizeof(*usin))
492 goto out;
493 if (usin->sin_family != AF_INET) {
494 static int complained;
495 if (!complained++)
496 printk(KERN_INFO "%s forgot to set AF_INET in "
497 "raw sendmsg. Fix it!\n",
498 current->comm);
499 err = -EAFNOSUPPORT;
500 if (usin->sin_family)
501 goto out;
502 }
503 daddr = usin->sin_addr.s_addr;
504 /* ANK: I did not forget to get protocol from port field.
505 * I just do not know, who uses this weirdness.
506 * IP_HDRINCL is much more convenient.
507 */
508 } else {
509 err = -EDESTADDRREQ;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900510 if (sk->sk_state != TCP_ESTABLISHED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700511 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000512 daddr = inet->inet_daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700513 }
514
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000515 ipc.addr = inet->inet_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700516 ipc.opt = NULL;
Oliver Hartkopp2244d072010-08-17 08:59:14 +0000517 ipc.tx_flags = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700518 ipc.oif = sk->sk_bound_dev_if;
519
520 if (msg->msg_controllen) {
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900521 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700522 if (err)
523 goto out;
524 if (ipc.opt)
525 free = 1;
526 }
527
528 saddr = ipc.addr;
529 ipc.addr = daddr;
530
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000531 if (!ipc.opt) {
532 struct ip_options_rcu *inet_opt;
533
534 rcu_read_lock();
535 inet_opt = rcu_dereference(inet->inet_opt);
536 if (inet_opt) {
537 memcpy(&opt_copy, inet_opt,
538 sizeof(*inet_opt) + inet_opt->opt.optlen);
539 ipc.opt = &opt_copy.opt;
540 }
541 rcu_read_unlock();
542 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700543
544 if (ipc.opt) {
545 err = -EINVAL;
546 /* Linux does not mangle headers on raw sockets,
547 * so that IP options + IP_HDRINCL is non-sense.
548 */
549 if (inet->hdrincl)
550 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000551 if (ipc.opt->opt.srr) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700552 if (!daddr)
553 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000554 daddr = ipc.opt->opt.faddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700555 }
556 }
557 tos = RT_CONN_FLAGS(sk);
558 if (msg->msg_flags & MSG_DONTROUTE)
559 tos |= RTO_ONLINK;
560
Joe Perchesf97c1e02007-12-16 13:45:43 -0800561 if (ipv4_is_multicast(daddr)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700562 if (!ipc.oif)
563 ipc.oif = inet->mc_index;
564 if (!saddr)
565 saddr = inet->mc_addr;
Erich E. Hoover76e21052012-02-08 09:11:07 +0000566 } else if (!ipc.oif)
567 ipc.oif = inet->uc_index;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700568
David S. Miller77968b72011-05-08 17:12:19 -0700569 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
570 RT_SCOPE_UNIVERSE,
571 inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
Julian Anastasov47670b72011-08-07 09:16:09 +0000572 inet_sk_flowi_flags(sk) | FLOWI_FLAG_CAN_SLEEP,
573 daddr, saddr, 0, 0);
David S. Milleref164ae2011-03-31 04:53:51 -0700574
David S. Miller77968b72011-05-08 17:12:19 -0700575 if (!inet->hdrincl) {
576 err = raw_probe_proto_opt(&fl4, msg);
577 if (err)
David S. Millerb23dd4f2011-03-02 14:31:35 -0800578 goto done;
David S. Miller77968b72011-05-08 17:12:19 -0700579 }
580
581 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
582 rt = ip_route_output_flow(sock_net(sk), &fl4, sk);
583 if (IS_ERR(rt)) {
584 err = PTR_ERR(rt);
585 rt = NULL;
586 goto done;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700587 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700588
589 err = -EACCES;
590 if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
591 goto done;
592
593 if (msg->msg_flags & MSG_CONFIRM)
594 goto do_confirm;
595back_from_confirm:
596
597 if (inet->hdrincl)
David S. Miller77968b72011-05-08 17:12:19 -0700598 err = raw_send_hdrinc(sk, &fl4, msg->msg_iov, len,
599 &rt, msg->msg_flags);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900600
Linus Torvalds1da177e2005-04-16 15:20:36 -0700601 else {
602 if (!ipc.addr)
David S. Miller77968b72011-05-08 17:12:19 -0700603 ipc.addr = fl4.daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700604 lock_sock(sk);
David S. Millerf5fca602011-05-08 17:24:10 -0700605 err = ip_append_data(sk, &fl4, ip_generic_getfrag,
David S. Miller77968b72011-05-08 17:12:19 -0700606 msg->msg_iov, len, 0,
607 &ipc, &rt, msg->msg_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700608 if (err)
609 ip_flush_pending_frames(sk);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700610 else if (!(msg->msg_flags & MSG_MORE)) {
David S. Miller77968b72011-05-08 17:12:19 -0700611 err = ip_push_pending_frames(sk, &fl4);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700612 if (err == -ENOBUFS && !inet->recverr)
613 err = 0;
614 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700615 release_sock(sk);
616 }
617done:
618 if (free)
619 kfree(ipc.opt);
620 ip_rt_put(rt);
621
Jesper Juhl5418c6922005-06-18 22:59:45 -0700622out:
623 if (err < 0)
624 return err;
625 return len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700626
627do_confirm:
Changli Gaod8d1f302010-06-10 23:31:35 -0700628 dst_confirm(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700629 if (!(msg->msg_flags & MSG_PROBE) || len)
630 goto back_from_confirm;
631 err = 0;
632 goto done;
633}
634
635static void raw_close(struct sock *sk, long timeout)
636{
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900637 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300638 * Raw sockets may have direct kernel references. Kill them.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 */
640 ip_ra_control(sk, 0, NULL);
641
642 sk_common_release(sk);
643}
644
Brian Haley7d06b2e2008-06-14 17:04:49 -0700645static void raw_destroy(struct sock *sk)
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700646{
647 lock_sock(sk);
648 ip_flush_pending_frames(sk);
649 release_sock(sk);
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700650}
651
Linus Torvalds1da177e2005-04-16 15:20:36 -0700652/* This gets rid of all the nasties in af_inet. -DaveM */
653static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
654{
655 struct inet_sock *inet = inet_sk(sk);
656 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
657 int ret = -EINVAL;
658 int chk_addr_ret;
659
660 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
661 goto out;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900662 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700663 ret = -EADDRNOTAVAIL;
664 if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
665 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
666 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000667 inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700668 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000669 inet->inet_saddr = 0; /* Use device */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700670 sk_dst_reset(sk);
671 ret = 0;
672out: return ret;
673}
674
675/*
676 * This should be easy, if there is something there
677 * we return it, otherwise we block.
678 */
679
680static int raw_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
681 size_t len, int noblock, int flags, int *addr_len)
682{
683 struct inet_sock *inet = inet_sk(sk);
684 size_t copied = 0;
685 int err = -EOPNOTSUPP;
686 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
687 struct sk_buff *skb;
688
689 if (flags & MSG_OOB)
690 goto out;
691
692 if (addr_len)
693 *addr_len = sizeof(*sin);
694
695 if (flags & MSG_ERRQUEUE) {
696 err = ip_recv_error(sk, msg, len);
697 goto out;
698 }
699
700 skb = skb_recv_datagram(sk, flags, noblock, &err);
701 if (!skb)
702 goto out;
703
704 copied = skb->len;
705 if (len < copied) {
706 msg->msg_flags |= MSG_TRUNC;
707 copied = len;
708 }
709
710 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
711 if (err)
712 goto done;
713
Neil Horman3b885782009-10-12 13:26:31 -0700714 sock_recv_ts_and_drops(msg, sk, skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700715
716 /* Copy the address. */
717 if (sin) {
718 sin->sin_family = AF_INET;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700719 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
Tetsuo Handaf59fc7f2006-07-25 17:05:35 -0700720 sin->sin_port = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700721 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
722 }
723 if (inet->cmsg_flags)
724 ip_cmsg_recv(msg, skb);
725 if (flags & MSG_TRUNC)
726 copied = skb->len;
727done:
728 skb_free_datagram(sk, skb);
Jesper Juhl5418c6922005-06-18 22:59:45 -0700729out:
730 if (err)
731 return err;
732 return copied;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700733}
734
735static int raw_init(struct sock *sk)
736{
737 struct raw_sock *rp = raw_sk(sk);
738
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000739 if (inet_sk(sk)->inet_num == IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700740 memset(&rp->filter, 0, sizeof(rp->filter));
741 return 0;
742}
743
744static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
745{
746 if (optlen > sizeof(struct icmp_filter))
747 optlen = sizeof(struct icmp_filter);
748 if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
749 return -EFAULT;
750 return 0;
751}
752
753static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
754{
755 int len, ret = -EFAULT;
756
757 if (get_user(len, optlen))
758 goto out;
759 ret = -EINVAL;
760 if (len < 0)
761 goto out;
762 if (len > sizeof(struct icmp_filter))
763 len = sizeof(struct icmp_filter);
764 ret = -EFAULT;
765 if (put_user(len, optlen) ||
766 copy_to_user(optval, &raw_sk(sk)->filter, len))
767 goto out;
768 ret = 0;
769out: return ret;
770}
771
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800772static int do_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700773 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700774{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700775 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000776 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700777 return -EOPNOTSUPP;
778 else
779 return raw_seticmpfilter(sk, optval, optlen);
780 }
781 return -ENOPROTOOPT;
782}
783
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800784static int raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700785 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786{
787 if (level != SOL_RAW)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800788 return ip_setsockopt(sk, level, optname, optval, optlen);
789 return do_raw_setsockopt(sk, level, optname, optval, optlen);
790}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700791
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800792#ifdef CONFIG_COMPAT
793static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700794 char __user *optval, unsigned int optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800795{
796 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800797 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800798 return do_raw_setsockopt(sk, level, optname, optval, optlen);
799}
800#endif
801
802static int do_raw_getsockopt(struct sock *sk, int level, int optname,
803 char __user *optval, int __user *optlen)
804{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700805 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000806 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700807 return -EOPNOTSUPP;
808 else
809 return raw_geticmpfilter(sk, optval, optlen);
810 }
811 return -ENOPROTOOPT;
812}
813
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800814static int raw_getsockopt(struct sock *sk, int level, int optname,
815 char __user *optval, int __user *optlen)
816{
817 if (level != SOL_RAW)
818 return ip_getsockopt(sk, level, optname, optval, optlen);
819 return do_raw_getsockopt(sk, level, optname, optval, optlen);
820}
821
822#ifdef CONFIG_COMPAT
823static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800824 char __user *optval, int __user *optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800825{
826 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800827 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800828 return do_raw_getsockopt(sk, level, optname, optval, optlen);
829}
830#endif
831
Linus Torvalds1da177e2005-04-16 15:20:36 -0700832static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
833{
834 switch (cmd) {
Joe Perches4500ebf2011-07-01 09:43:07 +0000835 case SIOCOUTQ: {
836 int amount = sk_wmem_alloc_get(sk);
Eric Dumazet31e6d362009-06-17 19:05:41 -0700837
Joe Perches4500ebf2011-07-01 09:43:07 +0000838 return put_user(amount, (int __user *)arg);
839 }
840 case SIOCINQ: {
841 struct sk_buff *skb;
842 int amount = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700843
Joe Perches4500ebf2011-07-01 09:43:07 +0000844 spin_lock_bh(&sk->sk_receive_queue.lock);
845 skb = skb_peek(&sk->sk_receive_queue);
846 if (skb != NULL)
847 amount = skb->len;
848 spin_unlock_bh(&sk->sk_receive_queue.lock);
849 return put_user(amount, (int __user *)arg);
850 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700851
Joe Perches4500ebf2011-07-01 09:43:07 +0000852 default:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700853#ifdef CONFIG_IP_MROUTE
Joe Perches4500ebf2011-07-01 09:43:07 +0000854 return ipmr_ioctl(sk, cmd, (void __user *)arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700855#else
Joe Perches4500ebf2011-07-01 09:43:07 +0000856 return -ENOIOCTLCMD;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700857#endif
858 }
859}
860
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000861#ifdef CONFIG_COMPAT
862static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg)
863{
864 switch (cmd) {
865 case SIOCOUTQ:
866 case SIOCINQ:
867 return -ENOIOCTLCMD;
868 default:
869#ifdef CONFIG_IP_MROUTE
870 return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg));
871#else
872 return -ENOIOCTLCMD;
873#endif
874 }
875}
876#endif
877
Linus Torvalds1da177e2005-04-16 15:20:36 -0700878struct proto raw_prot = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800879 .name = "RAW",
880 .owner = THIS_MODULE,
881 .close = raw_close,
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700882 .destroy = raw_destroy,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800883 .connect = ip4_datagram_connect,
884 .disconnect = udp_disconnect,
885 .ioctl = raw_ioctl,
886 .init = raw_init,
887 .setsockopt = raw_setsockopt,
888 .getsockopt = raw_getsockopt,
889 .sendmsg = raw_sendmsg,
890 .recvmsg = raw_recvmsg,
891 .bind = raw_bind,
892 .backlog_rcv = raw_rcv_skb,
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700893 .hash = raw_hash_sk,
894 .unhash = raw_unhash_sk,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800895 .obj_size = sizeof(struct raw_sock),
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700896 .h.raw_hash = &raw_v4_hashinfo,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800897#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800898 .compat_setsockopt = compat_raw_setsockopt,
899 .compat_getsockopt = compat_raw_getsockopt,
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000900 .compat_ioctl = compat_raw_ioctl,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800901#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700902};
903
904#ifdef CONFIG_PROC_FS
Linus Torvalds1da177e2005-04-16 15:20:36 -0700905static struct sock *raw_get_first(struct seq_file *seq)
906{
907 struct sock *sk;
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800908 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700909
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800910 for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
911 ++state->bucket) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700912 struct hlist_node *node;
913
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800914 sk_for_each(sk, node, &state->h->ht[state->bucket])
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900915 if (sock_net(sk) == seq_file_net(seq))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700916 goto found;
917 }
918 sk = NULL;
919found:
920 return sk;
921}
922
923static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
924{
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800925 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700926
927 do {
928 sk = sk_next(sk);
929try_again:
930 ;
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900931 } while (sk && sock_net(sk) != seq_file_net(seq));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700932
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800933 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800934 sk = sk_head(&state->h->ht[state->bucket]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700935 goto try_again;
936 }
937 return sk;
938}
939
940static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
941{
942 struct sock *sk = raw_get_first(seq);
943
944 if (sk)
945 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
946 --pos;
947 return pos ? NULL : sk;
948}
949
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800950void *raw_seq_start(struct seq_file *seq, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700951{
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800952 struct raw_iter_state *state = raw_seq_private(seq);
953
954 read_lock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700955 return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
956}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800957EXPORT_SYMBOL_GPL(raw_seq_start);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700958
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800959void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700960{
961 struct sock *sk;
962
963 if (v == SEQ_START_TOKEN)
964 sk = raw_get_first(seq);
965 else
966 sk = raw_get_next(seq, v);
967 ++*pos;
968 return sk;
969}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800970EXPORT_SYMBOL_GPL(raw_seq_next);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700971
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800972void raw_seq_stop(struct seq_file *seq, void *v)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973{
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800974 struct raw_iter_state *state = raw_seq_private(seq);
975
976 read_unlock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700977}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800978EXPORT_SYMBOL_GPL(raw_seq_stop);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700979
Denis V. Lunev8cd850e2008-01-31 03:46:43 -0800980static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700981{
982 struct inet_sock *inet = inet_sk(sp);
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000983 __be32 dest = inet->inet_daddr,
984 src = inet->inet_rcv_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700985 __u16 destp = 0,
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000986 srcp = inet->inet_num;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700987
Denis V. Lunev8cd850e2008-01-31 03:46:43 -0800988 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
Dan Rosenberg71338aa2011-05-23 12:17:35 +0000989 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d\n",
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900990 i, src, srcp, dest, destp, sp->sk_state,
Eric Dumazet31e6d362009-06-17 19:05:41 -0700991 sk_wmem_alloc_get(sp),
992 sk_rmem_alloc_get(sp),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700993 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
Wang Chen33c732c2007-11-13 20:30:01 -0800994 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700995}
996
997static int raw_seq_show(struct seq_file *seq, void *v)
998{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700999 if (v == SEQ_START_TOKEN)
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001000 seq_printf(seq, " sl local_address rem_address st tx_queue "
1001 "rx_queue tr tm->when retrnsmt uid timeout "
Eric Dumazetcb61cb92008-06-17 21:04:56 -07001002 "inode ref pointer drops\n");
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001003 else
1004 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001005 return 0;
1006}
1007
Stephen Hemmingerf6908082007-03-12 14:34:29 -07001008static const struct seq_operations raw_seq_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001009 .start = raw_seq_start,
1010 .next = raw_seq_next,
1011 .stop = raw_seq_stop,
1012 .show = raw_seq_show,
1013};
1014
Denis V. Lunev3046d762008-01-31 03:48:55 -08001015int raw_seq_open(struct inode *ino, struct file *file,
1016 struct raw_hashinfo *h, const struct seq_operations *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001017{
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001018 int err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001019 struct raw_iter_state *i;
1020
Denis V. Lunev3046d762008-01-31 03:48:55 -08001021 err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001022 if (err < 0)
1023 return err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001024
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001025 i = raw_seq_private((struct seq_file *)file->private_data);
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001026 i->h = h;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001027 return 0;
1028}
1029EXPORT_SYMBOL_GPL(raw_seq_open);
1030
1031static int raw_v4_seq_open(struct inode *inode, struct file *file)
1032{
Denis V. Lunev3046d762008-01-31 03:48:55 -08001033 return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001034}
1035
Arjan van de Ven9a321442007-02-12 00:55:35 -08001036static const struct file_operations raw_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001037 .owner = THIS_MODULE,
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001038 .open = raw_v4_seq_open,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001039 .read = seq_read,
1040 .llseek = seq_lseek,
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001041 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001042};
1043
Pavel Emelyanova308da12008-01-14 05:36:50 -08001044static __net_init int raw_init_net(struct net *net)
1045{
1046 if (!proc_net_fops_create(net, "raw", S_IRUGO, &raw_seq_fops))
1047 return -ENOMEM;
1048
1049 return 0;
1050}
1051
1052static __net_exit void raw_exit_net(struct net *net)
1053{
1054 proc_net_remove(net, "raw");
1055}
1056
1057static __net_initdata struct pernet_operations raw_net_ops = {
1058 .init = raw_init_net,
1059 .exit = raw_exit_net,
1060};
1061
Linus Torvalds1da177e2005-04-16 15:20:36 -07001062int __init raw_proc_init(void)
1063{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001064 return register_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001065}
1066
1067void __init raw_proc_exit(void)
1068{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001069 unregister_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001070}
1071#endif /* CONFIG_PROC_FS */