blob: 007e2eb769d33ede3afdb4e4fc69c3e9de6db1b4 [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 Dumazetf84af322010-04-28 15:31:51 -0700295 if (ip_queue_rcv_skb(sk, skb) < 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700296 kfree_skb(skb);
297 return NET_RX_DROP;
298 }
299
300 return NET_RX_SUCCESS;
301}
302
303int raw_rcv(struct sock *sk, struct sk_buff *skb)
304{
305 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
Wang Chen33c732c2007-11-13 20:30:01 -0800306 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700307 kfree_skb(skb);
308 return NET_RX_DROP;
309 }
Patrick McHardyb59c2702006-01-06 23:06:10 -0800310 nf_reset(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700312 skb_push(skb, skb->data - skb_network_header(skb));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700313
314 raw_rcv_skb(sk, skb);
315 return 0;
316}
317
David S. Miller77968b72011-05-08 17:12:19 -0700318static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4,
319 void *from, size_t length,
320 struct rtable **rtp,
321 unsigned int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700322{
323 struct inet_sock *inet = inet_sk(sk);
Pavel Emelyanov0388b002008-07-14 23:00:43 -0700324 struct net *net = sock_net(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700325 struct iphdr *iph;
326 struct sk_buff *skb;
Herbert Xuf844c742008-01-05 23:14:44 -0800327 unsigned int iphlen;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700328 int err;
Eric Dumazet1789a642010-06-03 22:23:57 +0000329 struct rtable *rt = *rtp;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700330
Changli Gaod8d1f302010-06-10 23:31:35 -0700331 if (length > rt->dst.dev->mtu) {
David S. Miller77968b72011-05-08 17:12:19 -0700332 ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport,
Changli Gaod8d1f302010-06-10 23:31:35 -0700333 rt->dst.dev->mtu);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700334 return -EMSGSIZE;
335 }
336 if (flags&MSG_PROBE)
337 goto out;
338
Johannes Bergf5184d22008-05-12 20:48:31 -0700339 skb = sock_alloc_send_skb(sk,
Changli Gaod8d1f302010-06-10 23:31:35 -0700340 length + LL_ALLOCATED_SPACE(rt->dst.dev) + 15,
Johannes Bergf5184d22008-05-12 20:48:31 -0700341 flags & MSG_DONTWAIT, &err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700342 if (skb == NULL)
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900343 goto error;
Changli Gaod8d1f302010-06-10 23:31:35 -0700344 skb_reserve(skb, LL_RESERVED_SPACE(rt->dst.dev));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700345
346 skb->priority = sk->sk_priority;
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800347 skb->mark = sk->sk_mark;
Changli Gaod8d1f302010-06-10 23:31:35 -0700348 skb_dst_set(skb, &rt->dst);
Eric Dumazet1789a642010-06-03 22:23:57 +0000349 *rtp = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700350
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300351 skb_reset_network_header(skb);
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700352 iph = ip_hdr(skb);
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300353 skb_put(skb, length);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700354
355 skb->ip_summed = CHECKSUM_NONE;
356
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -0700357 skb->transport_header = skb->network_header;
Neil Horman55888df2009-10-28 08:59:47 +0000358 err = -EFAULT;
359 if (memcpy_fromiovecend((void *)iph, from, 0, length))
360 goto error_free;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700361
Herbert Xuf844c742008-01-05 23:14:44 -0800362 iphlen = iph->ihl * 4;
Neil Horman55888df2009-10-28 08:59:47 +0000363
364 /*
365 * We don't want to modify the ip header, but we do need to
366 * be sure that it won't cause problems later along the network
367 * stack. Specifically we want to make sure that iph->ihl is a
368 * sane value. If ihl points beyond the length of the buffer passed
369 * in, reject the frame as invalid
370 */
371 err = -EINVAL;
372 if (iphlen > length)
373 goto error_free;
374
375 if (iphlen >= sizeof(*iph)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376 if (!iph->saddr)
David S. Miller77968b72011-05-08 17:12:19 -0700377 iph->saddr = fl4->saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700378 iph->check = 0;
379 iph->tot_len = htons(length);
380 if (!iph->id)
Changli Gaod8d1f302010-06-10 23:31:35 -0700381 ip_select_ident(iph, &rt->dst, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382
383 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
384 }
David L Stevens96793b42007-09-17 09:57:33 -0700385 if (iph->protocol == IPPROTO_ICMP)
Pavel Emelyanov0388b002008-07-14 23:00:43 -0700386 icmp_out_count(net, ((struct icmphdr *)
David L Stevens96793b42007-09-17 09:57:33 -0700387 skb_transport_header(skb))->type);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700388
Jan Engelhardt9bbc7682010-03-23 04:07:29 +0100389 err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, skb, NULL,
Changli Gaod8d1f302010-06-10 23:31:35 -0700390 rt->dst.dev, dst_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700391 if (err > 0)
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700392 err = net_xmit_errno(err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700393 if (err)
394 goto error;
395out:
396 return 0;
397
Neil Horman55888df2009-10-28 08:59:47 +0000398error_free:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700399 kfree_skb(skb);
400error:
Pavel Emelyanov5e38e272008-07-16 20:19:49 -0700401 IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700402 if (err == -ENOBUFS && !inet->recverr)
403 err = 0;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900404 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700405}
406
David S. Miller9d6ec932011-03-12 01:12:47 -0500407static int raw_probe_proto_opt(struct flowi4 *fl4, struct msghdr *msg)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700408{
409 struct iovec *iov;
410 u8 __user *type = NULL;
411 u8 __user *code = NULL;
412 int probed = 0;
Jesper Juhl93765d82005-06-18 23:00:15 -0700413 unsigned int i;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414
415 if (!msg->msg_iov)
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800416 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700417
418 for (i = 0; i < msg->msg_iovlen; i++) {
419 iov = &msg->msg_iov[i];
420 if (!iov)
421 continue;
422
David S. Miller9d6ec932011-03-12 01:12:47 -0500423 switch (fl4->flowi4_proto) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700424 case IPPROTO_ICMP:
425 /* check if one-byte field is readable or not. */
426 if (iov->iov_base && iov->iov_len < 1)
427 break;
428
429 if (!type) {
430 type = iov->iov_base;
431 /* check if code field is readable or not. */
432 if (iov->iov_len > 1)
433 code = type + 1;
434 } else if (!code)
435 code = iov->iov_base;
436
437 if (type && code) {
David S. Miller9cce96d2011-03-12 03:00:33 -0500438 if (get_user(fl4->fl4_icmp_type, type) ||
439 get_user(fl4->fl4_icmp_code, code))
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800440 return -EFAULT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700441 probed = 1;
442 }
443 break;
444 default:
445 probed = 1;
446 break;
447 }
448 if (probed)
449 break;
450 }
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800451 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700452}
453
454static int raw_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
455 size_t len)
456{
457 struct inet_sock *inet = inet_sk(sk);
458 struct ipcm_cookie ipc;
459 struct rtable *rt = NULL;
David S. Miller77968b72011-05-08 17:12:19 -0700460 struct flowi4 fl4;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700461 int free = 0;
Al Viro3ca3c682006-09-27 18:28:07 -0700462 __be32 daddr;
Al Viroc1d18f92006-09-27 18:28:28 -0700463 __be32 saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700464 u8 tos;
465 int err;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000466 struct ip_options_data opt_copy;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700467
468 err = -EMSGSIZE;
Jesper Juhl926d4b82005-06-18 23:00:00 -0700469 if (len > 0xFFFF)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700470 goto out;
471
472 /*
473 * Check the flags.
474 */
475
476 err = -EOPNOTSUPP;
477 if (msg->msg_flags & MSG_OOB) /* Mirror BSD error message */
478 goto out; /* compatibility */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900479
Linus Torvalds1da177e2005-04-16 15:20:36 -0700480 /*
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900481 * Get and verify the address.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700482 */
483
484 if (msg->msg_namelen) {
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800485 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700486 err = -EINVAL;
487 if (msg->msg_namelen < sizeof(*usin))
488 goto out;
489 if (usin->sin_family != AF_INET) {
490 static int complained;
491 if (!complained++)
492 printk(KERN_INFO "%s forgot to set AF_INET in "
493 "raw sendmsg. Fix it!\n",
494 current->comm);
495 err = -EAFNOSUPPORT;
496 if (usin->sin_family)
497 goto out;
498 }
499 daddr = usin->sin_addr.s_addr;
500 /* ANK: I did not forget to get protocol from port field.
501 * I just do not know, who uses this weirdness.
502 * IP_HDRINCL is much more convenient.
503 */
504 } else {
505 err = -EDESTADDRREQ;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900506 if (sk->sk_state != TCP_ESTABLISHED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700507 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000508 daddr = inet->inet_daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700509 }
510
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000511 ipc.addr = inet->inet_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700512 ipc.opt = NULL;
Oliver Hartkopp2244d072010-08-17 08:59:14 +0000513 ipc.tx_flags = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700514 ipc.oif = sk->sk_bound_dev_if;
515
516 if (msg->msg_controllen) {
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900517 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700518 if (err)
519 goto out;
520 if (ipc.opt)
521 free = 1;
522 }
523
524 saddr = ipc.addr;
525 ipc.addr = daddr;
526
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000527 if (!ipc.opt) {
528 struct ip_options_rcu *inet_opt;
529
530 rcu_read_lock();
531 inet_opt = rcu_dereference(inet->inet_opt);
532 if (inet_opt) {
533 memcpy(&opt_copy, inet_opt,
534 sizeof(*inet_opt) + inet_opt->opt.optlen);
535 ipc.opt = &opt_copy.opt;
536 }
537 rcu_read_unlock();
538 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700539
540 if (ipc.opt) {
541 err = -EINVAL;
542 /* Linux does not mangle headers on raw sockets,
543 * so that IP options + IP_HDRINCL is non-sense.
544 */
545 if (inet->hdrincl)
546 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000547 if (ipc.opt->opt.srr) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700548 if (!daddr)
549 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000550 daddr = ipc.opt->opt.faddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700551 }
552 }
553 tos = RT_CONN_FLAGS(sk);
554 if (msg->msg_flags & MSG_DONTROUTE)
555 tos |= RTO_ONLINK;
556
Joe Perchesf97c1e02007-12-16 13:45:43 -0800557 if (ipv4_is_multicast(daddr)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700558 if (!ipc.oif)
559 ipc.oif = inet->mc_index;
560 if (!saddr)
561 saddr = inet->mc_addr;
562 }
563
David S. Miller77968b72011-05-08 17:12:19 -0700564 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
565 RT_SCOPE_UNIVERSE,
566 inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
Julian Anastasov47670b72011-08-07 09:16:09 +0000567 inet_sk_flowi_flags(sk) | FLOWI_FLAG_CAN_SLEEP,
568 daddr, saddr, 0, 0);
David S. Milleref164ae2011-03-31 04:53:51 -0700569
David S. Miller77968b72011-05-08 17:12:19 -0700570 if (!inet->hdrincl) {
571 err = raw_probe_proto_opt(&fl4, msg);
572 if (err)
David S. Millerb23dd4f2011-03-02 14:31:35 -0800573 goto done;
David S. Miller77968b72011-05-08 17:12:19 -0700574 }
575
576 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
577 rt = ip_route_output_flow(sock_net(sk), &fl4, sk);
578 if (IS_ERR(rt)) {
579 err = PTR_ERR(rt);
580 rt = NULL;
581 goto done;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700582 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700583
584 err = -EACCES;
585 if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
586 goto done;
587
588 if (msg->msg_flags & MSG_CONFIRM)
589 goto do_confirm;
590back_from_confirm:
591
592 if (inet->hdrincl)
David S. Miller77968b72011-05-08 17:12:19 -0700593 err = raw_send_hdrinc(sk, &fl4, msg->msg_iov, len,
594 &rt, msg->msg_flags);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900595
Linus Torvalds1da177e2005-04-16 15:20:36 -0700596 else {
597 if (!ipc.addr)
David S. Miller77968b72011-05-08 17:12:19 -0700598 ipc.addr = fl4.daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700599 lock_sock(sk);
David S. Millerf5fca602011-05-08 17:24:10 -0700600 err = ip_append_data(sk, &fl4, ip_generic_getfrag,
David S. Miller77968b72011-05-08 17:12:19 -0700601 msg->msg_iov, len, 0,
602 &ipc, &rt, msg->msg_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700603 if (err)
604 ip_flush_pending_frames(sk);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700605 else if (!(msg->msg_flags & MSG_MORE)) {
David S. Miller77968b72011-05-08 17:12:19 -0700606 err = ip_push_pending_frames(sk, &fl4);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700607 if (err == -ENOBUFS && !inet->recverr)
608 err = 0;
609 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700610 release_sock(sk);
611 }
612done:
613 if (free)
614 kfree(ipc.opt);
615 ip_rt_put(rt);
616
Jesper Juhl5418c6922005-06-18 22:59:45 -0700617out:
618 if (err < 0)
619 return err;
620 return len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700621
622do_confirm:
Changli Gaod8d1f302010-06-10 23:31:35 -0700623 dst_confirm(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700624 if (!(msg->msg_flags & MSG_PROBE) || len)
625 goto back_from_confirm;
626 err = 0;
627 goto done;
628}
629
630static void raw_close(struct sock *sk, long timeout)
631{
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900632 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300633 * Raw sockets may have direct kernel references. Kill them.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700634 */
635 ip_ra_control(sk, 0, NULL);
636
637 sk_common_release(sk);
638}
639
Brian Haley7d06b2e2008-06-14 17:04:49 -0700640static void raw_destroy(struct sock *sk)
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700641{
642 lock_sock(sk);
643 ip_flush_pending_frames(sk);
644 release_sock(sk);
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700645}
646
Linus Torvalds1da177e2005-04-16 15:20:36 -0700647/* This gets rid of all the nasties in af_inet. -DaveM */
648static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
649{
650 struct inet_sock *inet = inet_sk(sk);
651 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
652 int ret = -EINVAL;
653 int chk_addr_ret;
654
655 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
656 goto out;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900657 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700658 ret = -EADDRNOTAVAIL;
659 if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
660 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
661 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000662 inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700663 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000664 inet->inet_saddr = 0; /* Use device */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700665 sk_dst_reset(sk);
666 ret = 0;
667out: return ret;
668}
669
670/*
671 * This should be easy, if there is something there
672 * we return it, otherwise we block.
673 */
674
675static int raw_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
676 size_t len, int noblock, int flags, int *addr_len)
677{
678 struct inet_sock *inet = inet_sk(sk);
679 size_t copied = 0;
680 int err = -EOPNOTSUPP;
681 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
682 struct sk_buff *skb;
683
684 if (flags & MSG_OOB)
685 goto out;
686
687 if (addr_len)
688 *addr_len = sizeof(*sin);
689
690 if (flags & MSG_ERRQUEUE) {
691 err = ip_recv_error(sk, msg, len);
692 goto out;
693 }
694
695 skb = skb_recv_datagram(sk, flags, noblock, &err);
696 if (!skb)
697 goto out;
698
699 copied = skb->len;
700 if (len < copied) {
701 msg->msg_flags |= MSG_TRUNC;
702 copied = len;
703 }
704
705 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
706 if (err)
707 goto done;
708
Neil Horman3b885782009-10-12 13:26:31 -0700709 sock_recv_ts_and_drops(msg, sk, skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700710
711 /* Copy the address. */
712 if (sin) {
713 sin->sin_family = AF_INET;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700714 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
Tetsuo Handaf59fc7f2006-07-25 17:05:35 -0700715 sin->sin_port = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700716 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
717 }
718 if (inet->cmsg_flags)
719 ip_cmsg_recv(msg, skb);
720 if (flags & MSG_TRUNC)
721 copied = skb->len;
722done:
723 skb_free_datagram(sk, skb);
Jesper Juhl5418c6922005-06-18 22:59:45 -0700724out:
725 if (err)
726 return err;
727 return copied;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700728}
729
730static int raw_init(struct sock *sk)
731{
732 struct raw_sock *rp = raw_sk(sk);
733
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000734 if (inet_sk(sk)->inet_num == IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700735 memset(&rp->filter, 0, sizeof(rp->filter));
736 return 0;
737}
738
739static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
740{
741 if (optlen > sizeof(struct icmp_filter))
742 optlen = sizeof(struct icmp_filter);
743 if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
744 return -EFAULT;
745 return 0;
746}
747
748static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
749{
750 int len, ret = -EFAULT;
751
752 if (get_user(len, optlen))
753 goto out;
754 ret = -EINVAL;
755 if (len < 0)
756 goto out;
757 if (len > sizeof(struct icmp_filter))
758 len = sizeof(struct icmp_filter);
759 ret = -EFAULT;
760 if (put_user(len, optlen) ||
761 copy_to_user(optval, &raw_sk(sk)->filter, len))
762 goto out;
763 ret = 0;
764out: return ret;
765}
766
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800767static int do_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700768 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700769{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700770 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000771 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700772 return -EOPNOTSUPP;
773 else
774 return raw_seticmpfilter(sk, optval, optlen);
775 }
776 return -ENOPROTOOPT;
777}
778
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800779static int raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700780 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700781{
782 if (level != SOL_RAW)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800783 return ip_setsockopt(sk, level, optname, optval, optlen);
784 return do_raw_setsockopt(sk, level, optname, optval, optlen);
785}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700786
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800787#ifdef CONFIG_COMPAT
788static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700789 char __user *optval, unsigned int optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800790{
791 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800792 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800793 return do_raw_setsockopt(sk, level, optname, optval, optlen);
794}
795#endif
796
797static int do_raw_getsockopt(struct sock *sk, int level, int optname,
798 char __user *optval, int __user *optlen)
799{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700800 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000801 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700802 return -EOPNOTSUPP;
803 else
804 return raw_geticmpfilter(sk, optval, optlen);
805 }
806 return -ENOPROTOOPT;
807}
808
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800809static int raw_getsockopt(struct sock *sk, int level, int optname,
810 char __user *optval, int __user *optlen)
811{
812 if (level != SOL_RAW)
813 return ip_getsockopt(sk, level, optname, optval, optlen);
814 return do_raw_getsockopt(sk, level, optname, optval, optlen);
815}
816
817#ifdef CONFIG_COMPAT
818static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800819 char __user *optval, int __user *optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800820{
821 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800822 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800823 return do_raw_getsockopt(sk, level, optname, optval, optlen);
824}
825#endif
826
Linus Torvalds1da177e2005-04-16 15:20:36 -0700827static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
828{
829 switch (cmd) {
Joe Perches4500ebf2011-07-01 09:43:07 +0000830 case SIOCOUTQ: {
831 int amount = sk_wmem_alloc_get(sk);
Eric Dumazet31e6d362009-06-17 19:05:41 -0700832
Joe Perches4500ebf2011-07-01 09:43:07 +0000833 return put_user(amount, (int __user *)arg);
834 }
835 case SIOCINQ: {
836 struct sk_buff *skb;
837 int amount = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700838
Joe Perches4500ebf2011-07-01 09:43:07 +0000839 spin_lock_bh(&sk->sk_receive_queue.lock);
840 skb = skb_peek(&sk->sk_receive_queue);
841 if (skb != NULL)
842 amount = skb->len;
843 spin_unlock_bh(&sk->sk_receive_queue.lock);
844 return put_user(amount, (int __user *)arg);
845 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700846
Joe Perches4500ebf2011-07-01 09:43:07 +0000847 default:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700848#ifdef CONFIG_IP_MROUTE
Joe Perches4500ebf2011-07-01 09:43:07 +0000849 return ipmr_ioctl(sk, cmd, (void __user *)arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700850#else
Joe Perches4500ebf2011-07-01 09:43:07 +0000851 return -ENOIOCTLCMD;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700852#endif
853 }
854}
855
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000856#ifdef CONFIG_COMPAT
857static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg)
858{
859 switch (cmd) {
860 case SIOCOUTQ:
861 case SIOCINQ:
862 return -ENOIOCTLCMD;
863 default:
864#ifdef CONFIG_IP_MROUTE
865 return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg));
866#else
867 return -ENOIOCTLCMD;
868#endif
869 }
870}
871#endif
872
Linus Torvalds1da177e2005-04-16 15:20:36 -0700873struct proto raw_prot = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800874 .name = "RAW",
875 .owner = THIS_MODULE,
876 .close = raw_close,
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700877 .destroy = raw_destroy,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800878 .connect = ip4_datagram_connect,
879 .disconnect = udp_disconnect,
880 .ioctl = raw_ioctl,
881 .init = raw_init,
882 .setsockopt = raw_setsockopt,
883 .getsockopt = raw_getsockopt,
884 .sendmsg = raw_sendmsg,
885 .recvmsg = raw_recvmsg,
886 .bind = raw_bind,
887 .backlog_rcv = raw_rcv_skb,
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700888 .hash = raw_hash_sk,
889 .unhash = raw_unhash_sk,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800890 .obj_size = sizeof(struct raw_sock),
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700891 .h.raw_hash = &raw_v4_hashinfo,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800892#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800893 .compat_setsockopt = compat_raw_setsockopt,
894 .compat_getsockopt = compat_raw_getsockopt,
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000895 .compat_ioctl = compat_raw_ioctl,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800896#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700897};
898
899#ifdef CONFIG_PROC_FS
Linus Torvalds1da177e2005-04-16 15:20:36 -0700900static struct sock *raw_get_first(struct seq_file *seq)
901{
902 struct sock *sk;
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800903 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700904
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800905 for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
906 ++state->bucket) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700907 struct hlist_node *node;
908
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800909 sk_for_each(sk, node, &state->h->ht[state->bucket])
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900910 if (sock_net(sk) == seq_file_net(seq))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700911 goto found;
912 }
913 sk = NULL;
914found:
915 return sk;
916}
917
918static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
919{
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800920 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700921
922 do {
923 sk = sk_next(sk);
924try_again:
925 ;
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900926 } while (sk && sock_net(sk) != seq_file_net(seq));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700927
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800928 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800929 sk = sk_head(&state->h->ht[state->bucket]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700930 goto try_again;
931 }
932 return sk;
933}
934
935static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
936{
937 struct sock *sk = raw_get_first(seq);
938
939 if (sk)
940 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
941 --pos;
942 return pos ? NULL : sk;
943}
944
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800945void *raw_seq_start(struct seq_file *seq, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700946{
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800947 struct raw_iter_state *state = raw_seq_private(seq);
948
949 read_lock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700950 return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
951}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800952EXPORT_SYMBOL_GPL(raw_seq_start);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700953
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800954void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700955{
956 struct sock *sk;
957
958 if (v == SEQ_START_TOKEN)
959 sk = raw_get_first(seq);
960 else
961 sk = raw_get_next(seq, v);
962 ++*pos;
963 return sk;
964}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800965EXPORT_SYMBOL_GPL(raw_seq_next);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800967void raw_seq_stop(struct seq_file *seq, void *v)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700968{
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800969 struct raw_iter_state *state = raw_seq_private(seq);
970
971 read_unlock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700972}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800973EXPORT_SYMBOL_GPL(raw_seq_stop);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700974
Denis V. Lunev8cd850e2008-01-31 03:46:43 -0800975static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700976{
977 struct inet_sock *inet = inet_sk(sp);
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000978 __be32 dest = inet->inet_daddr,
979 src = inet->inet_rcv_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700980 __u16 destp = 0,
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000981 srcp = inet->inet_num;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700982
Denis V. Lunev8cd850e2008-01-31 03:46:43 -0800983 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
Dan Rosenberg71338aa2011-05-23 12:17:35 +0000984 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d\n",
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900985 i, src, srcp, dest, destp, sp->sk_state,
Eric Dumazet31e6d362009-06-17 19:05:41 -0700986 sk_wmem_alloc_get(sp),
987 sk_rmem_alloc_get(sp),
Linus Torvalds1da177e2005-04-16 15:20:36 -0700988 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
Wang Chen33c732c2007-11-13 20:30:01 -0800989 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700990}
991
992static int raw_seq_show(struct seq_file *seq, void *v)
993{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700994 if (v == SEQ_START_TOKEN)
Denis V. Lunev8cd850e2008-01-31 03:46:43 -0800995 seq_printf(seq, " sl local_address rem_address st tx_queue "
996 "rx_queue tr tm->when retrnsmt uid timeout "
Eric Dumazetcb61cb92008-06-17 21:04:56 -0700997 "inode ref pointer drops\n");
Denis V. Lunev8cd850e2008-01-31 03:46:43 -0800998 else
999 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001000 return 0;
1001}
1002
Stephen Hemmingerf6908082007-03-12 14:34:29 -07001003static const struct seq_operations raw_seq_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001004 .start = raw_seq_start,
1005 .next = raw_seq_next,
1006 .stop = raw_seq_stop,
1007 .show = raw_seq_show,
1008};
1009
Denis V. Lunev3046d762008-01-31 03:48:55 -08001010int raw_seq_open(struct inode *ino, struct file *file,
1011 struct raw_hashinfo *h, const struct seq_operations *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001012{
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001013 int err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001014 struct raw_iter_state *i;
1015
Denis V. Lunev3046d762008-01-31 03:48:55 -08001016 err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001017 if (err < 0)
1018 return err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001019
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001020 i = raw_seq_private((struct seq_file *)file->private_data);
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001021 i->h = h;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001022 return 0;
1023}
1024EXPORT_SYMBOL_GPL(raw_seq_open);
1025
1026static int raw_v4_seq_open(struct inode *inode, struct file *file)
1027{
Denis V. Lunev3046d762008-01-31 03:48:55 -08001028 return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001029}
1030
Arjan van de Ven9a321442007-02-12 00:55:35 -08001031static const struct file_operations raw_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001032 .owner = THIS_MODULE,
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001033 .open = raw_v4_seq_open,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001034 .read = seq_read,
1035 .llseek = seq_lseek,
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001036 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001037};
1038
Pavel Emelyanova308da12008-01-14 05:36:50 -08001039static __net_init int raw_init_net(struct net *net)
1040{
1041 if (!proc_net_fops_create(net, "raw", S_IRUGO, &raw_seq_fops))
1042 return -ENOMEM;
1043
1044 return 0;
1045}
1046
1047static __net_exit void raw_exit_net(struct net *net)
1048{
1049 proc_net_remove(net, "raw");
1050}
1051
1052static __net_initdata struct pernet_operations raw_net_ops = {
1053 .init = raw_init_net,
1054 .exit = raw_exit_net,
1055};
1056
Linus Torvalds1da177e2005-04-16 15:20:36 -07001057int __init raw_proc_init(void)
1058{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001059 return register_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001060}
1061
1062void __init raw_proc_exit(void)
1063{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001064 unregister_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001065}
1066#endif /* CONFIG_PROC_FS */