blob: 63e5be0abd86ddab6476221161c48aacef34d8c4 [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>
Linus Torvalds1da177e2005-04-16 15:20:36 -070049#include <linux/kernel.h>
Paul Gortmakerbc3b2d72011-07-15 11:47:34 -040050#include <linux/export.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070051#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>
Linus Torvalds1da177e2005-04-16 15:20:36 -070059#include <linux/skbuff.h>
Quentin Armitagef5220d62014-07-23 09:58:01 +010060#include <linux/igmp.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>
Herbert Xuc008ba5b2014-11-07 21:27:09 +080081#include <linux/uio.h>
82
83struct raw_frag_vec {
Al Virob61e9dc2014-11-24 10:52:29 -050084 struct msghdr *msg;
Herbert Xuc008ba5b2014-11-07 21:27:09 +080085 union {
86 struct icmphdr icmph;
87 char c[1];
88 } hdr;
89 int hlen;
90};
Linus Torvalds1da177e2005-04-16 15:20:36 -070091
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080092static struct raw_hashinfo raw_v4_hashinfo = {
Robert P. J. Day938b93a2008-03-18 00:59:23 -070093 .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock),
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080094};
Linus Torvalds1da177e2005-04-16 15:20:36 -070095
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -070096void raw_hash_sk(struct sock *sk)
Linus Torvalds1da177e2005-04-16 15:20:36 -070097{
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -070098 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
Pavel Emelyanov65b4c502007-11-19 22:37:24 -080099 struct hlist_head *head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000101 head = &h->ht[inet_sk(sk)->inet_num & (RAW_HTABLE_SIZE - 1)];
Pavel Emelyanov65b4c502007-11-19 22:37:24 -0800102
103 write_lock_bh(&h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700104 sk_add_node(sk, head);
Pavel Emelyanovc29a0bc2008-03-31 19:41:46 -0700105 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
Pavel Emelyanov65b4c502007-11-19 22:37:24 -0800106 write_unlock_bh(&h->lock);
107}
108EXPORT_SYMBOL_GPL(raw_hash_sk);
109
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700110void raw_unhash_sk(struct sock *sk)
Pavel Emelyanovab707682007-11-19 22:37:58 -0800111{
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700112 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
113
Pavel Emelyanovab707682007-11-19 22:37:58 -0800114 write_lock_bh(&h->lock);
115 if (sk_del_node_init(sk))
Pavel Emelyanovc29a0bc2008-03-31 19:41:46 -0700116 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
Pavel Emelyanovab707682007-11-19 22:37:58 -0800117 write_unlock_bh(&h->lock);
118}
119EXPORT_SYMBOL_GPL(raw_unhash_sk);
120
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800121static struct sock *__raw_v4_lookup(struct net *net, struct sock *sk,
122 unsigned short num, __be32 raddr, __be32 laddr, int dif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700123{
Sasha Levinb67bfe02013-02-27 17:06:00 -0800124 sk_for_each_from(sk) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125 struct inet_sock *inet = inet_sk(sk);
126
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000127 if (net_eq(sock_net(sk), net) && inet->inet_num == num &&
128 !(inet->inet_daddr && inet->inet_daddr != raddr) &&
129 !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) &&
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130 !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
131 goto found; /* gotcha */
132 }
133 sk = NULL;
134found:
135 return sk;
136}
137
138/*
139 * 0 - deliver
140 * 1 - block
141 */
Eric Dumazetab43ed82012-09-22 00:08:29 +0000142static int icmp_filter(const struct sock *sk, const struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700143{
Eric Dumazetab43ed82012-09-22 00:08:29 +0000144 struct icmphdr _hdr;
145 const struct icmphdr *hdr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700146
Eric Dumazetab43ed82012-09-22 00:08:29 +0000147 hdr = skb_header_pointer(skb, skb_transport_offset(skb),
148 sizeof(_hdr), &_hdr);
149 if (!hdr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150 return 1;
151
Eric Dumazetab43ed82012-09-22 00:08:29 +0000152 if (hdr->type < 32) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700153 __u32 data = raw_sk(sk)->filter.data;
154
Eric Dumazetab43ed82012-09-22 00:08:29 +0000155 return ((1U << hdr->type) & data) != 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700156 }
157
158 /* Do not block unknown ICMP types */
159 return 0;
160}
161
162/* IP input processing comes here for RAW socket delivery.
163 * Caller owns SKB, so we must make clones.
164 *
165 * RFC 1122: SHOULD pass TOS value up to the transport layer.
166 * -> It does. And not only TOS, but all IP header.
167 */
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000168static int raw_v4_input(struct sk_buff *skb, const struct iphdr *iph, int hash)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700169{
170 struct sock *sk;
171 struct hlist_head *head;
Patrick McHardyd13964f2005-08-09 19:45:02 -0700172 int delivered = 0;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800173 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700174
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800175 read_lock(&raw_v4_hashinfo.lock);
176 head = &raw_v4_hashinfo.ht[hash];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700177 if (hlist_empty(head))
178 goto out;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800179
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900180 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800181 sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700182 iph->saddr, iph->daddr,
183 skb->dev->ifindex);
184
185 while (sk) {
Patrick McHardyd13964f2005-08-09 19:45:02 -0700186 delivered = 1;
Quentin Armitagef5220d62014-07-23 09:58:01 +0100187 if ((iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) &&
188 ip_mc_sf_allow(sk, iph->daddr, iph->saddr,
189 skb->dev->ifindex)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700190 struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
191
192 /* Not releasing hash table! */
193 if (clone)
194 raw_rcv(sk, clone);
195 }
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800196 sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700197 iph->saddr, iph->daddr,
198 skb->dev->ifindex);
199 }
200out:
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800201 read_unlock(&raw_v4_hashinfo.lock);
Patrick McHardyd13964f2005-08-09 19:45:02 -0700202 return delivered;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700203}
204
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800205int raw_local_deliver(struct sk_buff *skb, int protocol)
206{
207 int hash;
208 struct sock *raw_sk;
209
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800210 hash = protocol & (RAW_HTABLE_SIZE - 1);
211 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800212
213 /* If there maybe a raw socket we must check - if not we
214 * don't care less
215 */
216 if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash))
217 raw_sk = NULL;
218
219 return raw_sk != NULL;
220
221}
222
223static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700224{
225 struct inet_sock *inet = inet_sk(sk);
Arnaldo Carvalho de Melo88c76642007-03-13 14:43:18 -0300226 const int type = icmp_hdr(skb)->type;
227 const int code = icmp_hdr(skb)->code;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700228 int err = 0;
229 int harderr = 0;
230
David S. Miller36393392012-06-14 22:21:46 -0700231 if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED)
232 ipv4_sk_update_pmtu(skb, sk, info);
Duan Jiong8d65b112013-09-20 18:21:25 +0800233 else if (type == ICMP_REDIRECT) {
David S. Miller55be7a92012-07-11 21:27:49 -0700234 ipv4_sk_redirect(skb, sk);
Duan Jiong8d65b112013-09-20 18:21:25 +0800235 return;
236 }
David S. Miller36393392012-06-14 22:21:46 -0700237
Linus Torvalds1da177e2005-04-16 15:20:36 -0700238 /* Report error on raw socket, if:
239 1. User requested ip_recverr.
240 2. Socket is connected (otherwise the error indication
241 is useless without ip_recverr and error is hard.
242 */
243 if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED)
244 return;
245
246 switch (type) {
247 default:
248 case ICMP_TIME_EXCEEDED:
249 err = EHOSTUNREACH;
250 break;
251 case ICMP_SOURCE_QUENCH:
252 return;
253 case ICMP_PARAMETERPROB:
254 err = EPROTO;
255 harderr = 1;
256 break;
257 case ICMP_DEST_UNREACH:
258 err = EHOSTUNREACH;
259 if (code > NR_ICMP_UNREACH)
260 break;
261 err = icmp_err_convert[code].errno;
262 harderr = icmp_err_convert[code].fatal;
263 if (code == ICMP_FRAG_NEEDED) {
264 harderr = inet->pmtudisc != IP_PMTUDISC_DONT;
265 err = EMSGSIZE;
266 }
267 }
268
269 if (inet->recverr) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000270 const struct iphdr *iph = (const struct iphdr *)skb->data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700271 u8 *payload = skb->data + (iph->ihl << 2);
272
273 if (inet->hdrincl)
274 payload = skb->data;
275 ip_icmp_error(sk, skb, err, 0, info, payload);
276 }
277
278 if (inet->recverr || harderr) {
279 sk->sk_err = err;
280 sk->sk_error_report(sk);
281 }
282}
283
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800284void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info)
285{
286 int hash;
287 struct sock *raw_sk;
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000288 const struct iphdr *iph;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800289 struct net *net;
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800290
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800291 hash = protocol & (RAW_HTABLE_SIZE - 1);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800292
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800293 read_lock(&raw_v4_hashinfo.lock);
294 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
Ian Morris00db4122015-04-03 09:17:27 +0100295 if (raw_sk) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000296 iph = (const struct iphdr *)skb->data;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900297 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800298
299 while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol,
300 iph->daddr, iph->saddr,
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800301 skb->dev->ifindex)) != NULL) {
302 raw_err(raw_sk, skb, info);
303 raw_sk = sk_next(raw_sk);
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000304 iph = (const struct iphdr *)skb->data;
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800305 }
306 }
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800307 read_unlock(&raw_v4_hashinfo.lock);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800308}
309
Daniel Baluta5e73ea12012-04-15 01:34:41 +0000310static int raw_rcv_skb(struct sock *sk, struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311{
312 /* Charge it to the socket. */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900313
Shawn Bohrerfbf88662013-10-07 11:01:40 -0500314 ipv4_pktinfo_prepare(sk, skb);
Eric Dumazetd826eb12011-11-09 07:24:35 +0000315 if (sock_queue_rcv_skb(sk, skb) < 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316 kfree_skb(skb);
317 return NET_RX_DROP;
318 }
319
320 return NET_RX_SUCCESS;
321}
322
323int raw_rcv(struct sock *sk, struct sk_buff *skb)
324{
325 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
Wang Chen33c732c2007-11-13 20:30:01 -0800326 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700327 kfree_skb(skb);
328 return NET_RX_DROP;
329 }
Patrick McHardyb59c2702006-01-06 23:06:10 -0800330 nf_reset(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700331
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700332 skb_push(skb, skb->data - skb_network_header(skb));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700333
334 raw_rcv_skb(sk, skb);
335 return 0;
336}
337
David S. Miller77968b72011-05-08 17:12:19 -0700338static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4,
Al Viro7ae9abf2014-11-27 19:30:51 -0500339 struct msghdr *msg, size_t length,
David S. Miller77968b72011-05-08 17:12:19 -0700340 struct rtable **rtp,
341 unsigned int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700342{
343 struct inet_sock *inet = inet_sk(sk);
Pavel Emelyanov0388b002008-07-14 23:00:43 -0700344 struct net *net = sock_net(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700345 struct iphdr *iph;
346 struct sk_buff *skb;
Herbert Xuf844c742008-01-05 23:14:44 -0800347 unsigned int iphlen;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700348 int err;
Eric Dumazet1789a642010-06-03 22:23:57 +0000349 struct rtable *rt = *rtp;
Herbert Xu66088242011-11-18 02:20:04 +0000350 int hlen, tlen;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351
Changli Gaod8d1f302010-06-10 23:31:35 -0700352 if (length > rt->dst.dev->mtu) {
David S. Miller77968b72011-05-08 17:12:19 -0700353 ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport,
Changli Gaod8d1f302010-06-10 23:31:35 -0700354 rt->dst.dev->mtu);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700355 return -EMSGSIZE;
356 }
357 if (flags&MSG_PROBE)
358 goto out;
359
Herbert Xu66088242011-11-18 02:20:04 +0000360 hlen = LL_RESERVED_SPACE(rt->dst.dev);
361 tlen = rt->dst.dev->needed_tailroom;
Johannes Bergf5184d22008-05-12 20:48:31 -0700362 skb = sock_alloc_send_skb(sk,
Herbert Xu66088242011-11-18 02:20:04 +0000363 length + hlen + tlen + 15,
Johannes Bergf5184d22008-05-12 20:48:31 -0700364 flags & MSG_DONTWAIT, &err);
Ian Morris51456b22015-04-03 09:17:26 +0100365 if (!skb)
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900366 goto error;
Herbert Xu66088242011-11-18 02:20:04 +0000367 skb_reserve(skb, hlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700368
369 skb->priority = sk->sk_priority;
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800370 skb->mark = sk->sk_mark;
Changli Gaod8d1f302010-06-10 23:31:35 -0700371 skb_dst_set(skb, &rt->dst);
Eric Dumazet1789a642010-06-03 22:23:57 +0000372 *rtp = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700373
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300374 skb_reset_network_header(skb);
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700375 iph = ip_hdr(skb);
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300376 skb_put(skb, length);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700377
378 skb->ip_summed = CHECKSUM_NONE;
379
Willem de Bruijn11878b42014-07-14 17:55:06 -0400380 sock_tx_timestamp(sk, &skb_shinfo(skb)->tx_flags);
381
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -0700382 skb->transport_header = skb->network_header;
Neil Horman55888df2009-10-28 08:59:47 +0000383 err = -EFAULT;
Al Viro21226ab2014-11-28 15:48:29 -0500384 if (memcpy_from_msg(iph, msg, length))
Neil Horman55888df2009-10-28 08:59:47 +0000385 goto error_free;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700386
Herbert Xuf844c742008-01-05 23:14:44 -0800387 iphlen = iph->ihl * 4;
Neil Horman55888df2009-10-28 08:59:47 +0000388
389 /*
390 * We don't want to modify the ip header, but we do need to
391 * be sure that it won't cause problems later along the network
392 * stack. Specifically we want to make sure that iph->ihl is a
393 * sane value. If ihl points beyond the length of the buffer passed
394 * in, reject the frame as invalid
395 */
396 err = -EINVAL;
397 if (iphlen > length)
398 goto error_free;
399
400 if (iphlen >= sizeof(*iph)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700401 if (!iph->saddr)
David S. Miller77968b72011-05-08 17:12:19 -0700402 iph->saddr = fl4->saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700403 iph->check = 0;
404 iph->tot_len = htons(length);
405 if (!iph->id)
Hannes Frederic Sowab6a77192015-03-25 17:07:44 +0100406 ip_select_ident(net, skb, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700407
408 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
Ben Cartwright-Cox027ac582015-11-14 15:13:58 +0000409 skb->transport_header += iphlen;
410 if (iph->protocol == IPPROTO_ICMP &&
411 length >= iphlen + sizeof(struct icmphdr))
412 icmp_out_count(net, ((struct icmphdr *)
413 skb_transport_header(skb))->type);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414 }
415
Eric W. Biederman29a26a52015-09-15 20:04:16 -0500416 err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT,
417 net, sk, skb, NULL, rt->dst.dev,
Eric W. Biederman13206b62015-10-07 16:48:35 -0500418 dst_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700419 if (err > 0)
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700420 err = net_xmit_errno(err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700421 if (err)
422 goto error;
423out:
424 return 0;
425
Neil Horman55888df2009-10-28 08:59:47 +0000426error_free:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700427 kfree_skb(skb);
428error:
Pavel Emelyanov5e38e272008-07-16 20:19:49 -0700429 IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700430 if (err == -ENOBUFS && !inet->recverr)
431 err = 0;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900432 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700433}
434
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800435static int raw_probe_proto_opt(struct raw_frag_vec *rfv, struct flowi4 *fl4)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700436{
Herbert Xu32b59132014-11-07 21:27:08 +0800437 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700438
Herbert Xu32b59132014-11-07 21:27:08 +0800439 if (fl4->flowi4_proto != IPPROTO_ICMP)
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800440 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700441
Herbert Xu32b59132014-11-07 21:27:08 +0800442 /* We only need the first two bytes. */
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800443 rfv->hlen = 2;
444
Al Virob61e9dc2014-11-24 10:52:29 -0500445 err = memcpy_from_msg(rfv->hdr.c, rfv->msg, rfv->hlen);
Herbert Xu32b59132014-11-07 21:27:08 +0800446 if (err)
447 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700448
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800449 fl4->fl4_icmp_type = rfv->hdr.icmph.type;
450 fl4->fl4_icmp_code = rfv->hdr.icmph.code;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700451
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800452 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700453}
454
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800455static int raw_getfrag(void *from, char *to, int offset, int len, int odd,
456 struct sk_buff *skb)
457{
458 struct raw_frag_vec *rfv = from;
459
460 if (offset < rfv->hlen) {
461 int copy = min(rfv->hlen - offset, len);
462
463 if (skb->ip_summed == CHECKSUM_PARTIAL)
464 memcpy(to, rfv->hdr.c + offset, copy);
465 else
466 skb->csum = csum_block_add(
467 skb->csum,
468 csum_partial_copy_nocheck(rfv->hdr.c + offset,
469 to, copy, 0),
470 odd);
471
472 odd = 0;
473 offset += copy;
474 to += copy;
475 len -= copy;
476
477 if (!len)
478 return 0;
479 }
480
481 offset -= rfv->hlen;
482
Al Virof69e6d12014-11-24 13:23:40 -0500483 return ip_generic_getfrag(rfv->msg, to, offset, len, odd, skb);
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800484}
485
Ying Xue1b784142015-03-02 15:37:48 +0800486static int raw_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700487{
488 struct inet_sock *inet = inet_sk(sk);
David Ahernbb191c32015-10-05 08:51:27 -0700489 struct net *net = sock_net(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490 struct ipcm_cookie ipc;
491 struct rtable *rt = NULL;
David S. Miller77968b72011-05-08 17:12:19 -0700492 struct flowi4 fl4;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700493 int free = 0;
Al Viro3ca3c682006-09-27 18:28:07 -0700494 __be32 daddr;
Al Viroc1d18f92006-09-27 18:28:28 -0700495 __be32 saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700496 u8 tos;
497 int err;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000498 struct ip_options_data opt_copy;
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800499 struct raw_frag_vec rfv;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700500
501 err = -EMSGSIZE;
Jesper Juhl926d4b82005-06-18 23:00:00 -0700502 if (len > 0xFFFF)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700503 goto out;
504
505 /*
506 * Check the flags.
507 */
508
509 err = -EOPNOTSUPP;
510 if (msg->msg_flags & MSG_OOB) /* Mirror BSD error message */
511 goto out; /* compatibility */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900512
Linus Torvalds1da177e2005-04-16 15:20:36 -0700513 /*
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900514 * Get and verify the address.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700515 */
516
517 if (msg->msg_namelen) {
Steffen Hurrle342dfc32014-01-17 22:53:15 +0100518 DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700519 err = -EINVAL;
520 if (msg->msg_namelen < sizeof(*usin))
521 goto out;
522 if (usin->sin_family != AF_INET) {
Joe Perches058bd4d2012-03-11 18:36:11 +0000523 pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n",
524 __func__, current->comm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700525 err = -EAFNOSUPPORT;
526 if (usin->sin_family)
527 goto out;
528 }
529 daddr = usin->sin_addr.s_addr;
530 /* ANK: I did not forget to get protocol from port field.
531 * I just do not know, who uses this weirdness.
532 * IP_HDRINCL is much more convenient.
533 */
534 } else {
535 err = -EDESTADDRREQ;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900536 if (sk->sk_state != TCP_ESTABLISHED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700537 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000538 daddr = inet->inet_daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700539 }
540
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000541 ipc.addr = inet->inet_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700542 ipc.opt = NULL;
Oliver Hartkopp2244d072010-08-17 08:59:14 +0000543 ipc.tx_flags = 0;
Francesco Fuscoaa661582013-09-24 15:43:09 +0200544 ipc.ttl = 0;
545 ipc.tos = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700546 ipc.oif = sk->sk_bound_dev_if;
547
548 if (msg->msg_controllen) {
David Ahernbb191c32015-10-05 08:51:27 -0700549 err = ip_cmsg_send(net, msg, &ipc, false);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700550 if (err)
551 goto out;
552 if (ipc.opt)
553 free = 1;
554 }
555
556 saddr = ipc.addr;
557 ipc.addr = daddr;
558
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000559 if (!ipc.opt) {
560 struct ip_options_rcu *inet_opt;
561
562 rcu_read_lock();
563 inet_opt = rcu_dereference(inet->inet_opt);
564 if (inet_opt) {
565 memcpy(&opt_copy, inet_opt,
566 sizeof(*inet_opt) + inet_opt->opt.optlen);
567 ipc.opt = &opt_copy.opt;
568 }
569 rcu_read_unlock();
570 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700571
572 if (ipc.opt) {
573 err = -EINVAL;
574 /* Linux does not mangle headers on raw sockets,
575 * so that IP options + IP_HDRINCL is non-sense.
576 */
577 if (inet->hdrincl)
578 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000579 if (ipc.opt->opt.srr) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700580 if (!daddr)
581 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000582 daddr = ipc.opt->opt.faddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700583 }
584 }
Francesco Fuscoaa661582013-09-24 15:43:09 +0200585 tos = get_rtconn_flags(&ipc, sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700586 if (msg->msg_flags & MSG_DONTROUTE)
587 tos |= RTO_ONLINK;
588
Joe Perchesf97c1e02007-12-16 13:45:43 -0800589 if (ipv4_is_multicast(daddr)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700590 if (!ipc.oif)
591 ipc.oif = inet->mc_index;
592 if (!saddr)
593 saddr = inet->mc_addr;
Erich E. Hoover76e21052012-02-08 09:11:07 +0000594 } else if (!ipc.oif)
595 ipc.oif = inet->uc_index;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700596
David S. Miller77968b72011-05-08 17:12:19 -0700597 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
598 RT_SCOPE_UNIVERSE,
599 inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
Steffen Klassert0e0d44a2013-08-28 08:04:14 +0200600 inet_sk_flowi_flags(sk) |
Chris Clarkc27c9322013-08-27 12:02:15 -0600601 (inet->hdrincl ? FLOWI_FLAG_KNOWN_NH : 0),
Julian Anastasov47670b72011-08-07 09:16:09 +0000602 daddr, saddr, 0, 0);
David S. Milleref164ae2011-03-31 04:53:51 -0700603
David Ahernbb191c32015-10-05 08:51:27 -0700604 if (!saddr && ipc.oif)
605 l3mdev_get_saddr(net, ipc.oif, &fl4);
606
David S. Miller77968b72011-05-08 17:12:19 -0700607 if (!inet->hdrincl) {
Al Virob61e9dc2014-11-24 10:52:29 -0500608 rfv.msg = msg;
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800609 rfv.hlen = 0;
610
611 err = raw_probe_proto_opt(&rfv, &fl4);
David S. Miller77968b72011-05-08 17:12:19 -0700612 if (err)
David S. Millerb23dd4f2011-03-02 14:31:35 -0800613 goto done;
David S. Miller77968b72011-05-08 17:12:19 -0700614 }
615
616 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
David Ahernbb191c32015-10-05 08:51:27 -0700617 rt = ip_route_output_flow(net, &fl4, sk);
David S. Miller77968b72011-05-08 17:12:19 -0700618 if (IS_ERR(rt)) {
619 err = PTR_ERR(rt);
620 rt = NULL;
621 goto done;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700622 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700623
624 err = -EACCES;
625 if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
626 goto done;
627
628 if (msg->msg_flags & MSG_CONFIRM)
629 goto do_confirm;
630back_from_confirm:
631
632 if (inet->hdrincl)
Al Viro7ae9abf2014-11-27 19:30:51 -0500633 err = raw_send_hdrinc(sk, &fl4, msg, len,
David S. Miller77968b72011-05-08 17:12:19 -0700634 &rt, msg->msg_flags);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900635
Linus Torvalds1da177e2005-04-16 15:20:36 -0700636 else {
Willem de Bruijn11878b42014-07-14 17:55:06 -0400637 sock_tx_timestamp(sk, &ipc.tx_flags);
638
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 if (!ipc.addr)
David S. Miller77968b72011-05-08 17:12:19 -0700640 ipc.addr = fl4.daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700641 lock_sock(sk);
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800642 err = ip_append_data(sk, &fl4, raw_getfrag,
643 &rfv, len, 0,
David S. Miller77968b72011-05-08 17:12:19 -0700644 &ipc, &rt, msg->msg_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700645 if (err)
646 ip_flush_pending_frames(sk);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700647 else if (!(msg->msg_flags & MSG_MORE)) {
David S. Miller77968b72011-05-08 17:12:19 -0700648 err = ip_push_pending_frames(sk, &fl4);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700649 if (err == -ENOBUFS && !inet->recverr)
650 err = 0;
651 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700652 release_sock(sk);
653 }
654done:
655 if (free)
656 kfree(ipc.opt);
657 ip_rt_put(rt);
658
Jesper Juhl5418c6922005-06-18 22:59:45 -0700659out:
660 if (err < 0)
661 return err;
662 return len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700663
664do_confirm:
Changli Gaod8d1f302010-06-10 23:31:35 -0700665 dst_confirm(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700666 if (!(msg->msg_flags & MSG_PROBE) || len)
667 goto back_from_confirm;
668 err = 0;
669 goto done;
670}
671
672static void raw_close(struct sock *sk, long timeout)
673{
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900674 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300675 * Raw sockets may have direct kernel references. Kill them.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700676 */
677 ip_ra_control(sk, 0, NULL);
678
679 sk_common_release(sk);
680}
681
Brian Haley7d06b2e2008-06-14 17:04:49 -0700682static void raw_destroy(struct sock *sk)
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700683{
684 lock_sock(sk);
685 ip_flush_pending_frames(sk);
686 release_sock(sk);
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700687}
688
Linus Torvalds1da177e2005-04-16 15:20:36 -0700689/* This gets rid of all the nasties in af_inet. -DaveM */
690static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
691{
692 struct inet_sock *inet = inet_sk(sk);
693 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
694 int ret = -EINVAL;
695 int chk_addr_ret;
696
697 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
698 goto out;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900699 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700 ret = -EADDRNOTAVAIL;
701 if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
702 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
703 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000704 inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700705 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000706 inet->inet_saddr = 0; /* Use device */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700707 sk_dst_reset(sk);
708 ret = 0;
709out: return ret;
710}
711
712/*
713 * This should be easy, if there is something there
714 * we return it, otherwise we block.
715 */
716
Ying Xue1b784142015-03-02 15:37:48 +0800717static int raw_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
718 int noblock, int flags, int *addr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700719{
720 struct inet_sock *inet = inet_sk(sk);
721 size_t copied = 0;
722 int err = -EOPNOTSUPP;
Steffen Hurrle342dfc32014-01-17 22:53:15 +0100723 DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700724 struct sk_buff *skb;
725
726 if (flags & MSG_OOB)
727 goto out;
728
Linus Torvalds1da177e2005-04-16 15:20:36 -0700729 if (flags & MSG_ERRQUEUE) {
Hannes Frederic Sowa85fbaa72013-11-23 00:46:12 +0100730 err = ip_recv_error(sk, msg, len, addr_len);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700731 goto out;
732 }
733
734 skb = skb_recv_datagram(sk, flags, noblock, &err);
735 if (!skb)
736 goto out;
737
738 copied = skb->len;
739 if (len < copied) {
740 msg->msg_flags |= MSG_TRUNC;
741 copied = len;
742 }
743
David S. Miller51f3d022014-11-05 16:46:40 -0500744 err = skb_copy_datagram_msg(skb, 0, msg, copied);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700745 if (err)
746 goto done;
747
Neil Horman3b885782009-10-12 13:26:31 -0700748 sock_recv_ts_and_drops(msg, sk, skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700749
750 /* Copy the address. */
751 if (sin) {
752 sin->sin_family = AF_INET;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700753 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
Tetsuo Handaf59fc7f2006-07-25 17:05:35 -0700754 sin->sin_port = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700755 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
Hannes Frederic Sowabceaa902013-11-18 04:20:45 +0100756 *addr_len = sizeof(*sin);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700757 }
758 if (inet->cmsg_flags)
759 ip_cmsg_recv(msg, skb);
760 if (flags & MSG_TRUNC)
761 copied = skb->len;
762done:
763 skb_free_datagram(sk, skb);
Jesper Juhl5418c6922005-06-18 22:59:45 -0700764out:
765 if (err)
766 return err;
767 return copied;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700768}
769
770static int raw_init(struct sock *sk)
771{
772 struct raw_sock *rp = raw_sk(sk);
773
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000774 if (inet_sk(sk)->inet_num == IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700775 memset(&rp->filter, 0, sizeof(rp->filter));
776 return 0;
777}
778
779static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
780{
781 if (optlen > sizeof(struct icmp_filter))
782 optlen = sizeof(struct icmp_filter);
783 if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
784 return -EFAULT;
785 return 0;
786}
787
788static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
789{
790 int len, ret = -EFAULT;
791
792 if (get_user(len, optlen))
793 goto out;
794 ret = -EINVAL;
795 if (len < 0)
796 goto out;
797 if (len > sizeof(struct icmp_filter))
798 len = sizeof(struct icmp_filter);
799 ret = -EFAULT;
800 if (put_user(len, optlen) ||
801 copy_to_user(optval, &raw_sk(sk)->filter, len))
802 goto out;
803 ret = 0;
804out: return ret;
805}
806
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800807static int do_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700808 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700809{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700810 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000811 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700812 return -EOPNOTSUPP;
813 else
814 return raw_seticmpfilter(sk, optval, optlen);
815 }
816 return -ENOPROTOOPT;
817}
818
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800819static int raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700820 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700821{
822 if (level != SOL_RAW)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800823 return ip_setsockopt(sk, level, optname, optval, optlen);
824 return do_raw_setsockopt(sk, level, optname, optval, optlen);
825}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700826
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800827#ifdef CONFIG_COMPAT
828static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700829 char __user *optval, unsigned int optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800830{
831 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800832 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800833 return do_raw_setsockopt(sk, level, optname, optval, optlen);
834}
835#endif
836
837static int do_raw_getsockopt(struct sock *sk, int level, int optname,
838 char __user *optval, int __user *optlen)
839{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700840 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000841 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700842 return -EOPNOTSUPP;
843 else
844 return raw_geticmpfilter(sk, optval, optlen);
845 }
846 return -ENOPROTOOPT;
847}
848
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800849static int raw_getsockopt(struct sock *sk, int level, int optname,
850 char __user *optval, int __user *optlen)
851{
852 if (level != SOL_RAW)
853 return ip_getsockopt(sk, level, optname, optval, optlen);
854 return do_raw_getsockopt(sk, level, optname, optval, optlen);
855}
856
857#ifdef CONFIG_COMPAT
858static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800859 char __user *optval, int __user *optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800860{
861 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800862 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800863 return do_raw_getsockopt(sk, level, optname, optval, optlen);
864}
865#endif
866
Linus Torvalds1da177e2005-04-16 15:20:36 -0700867static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
868{
869 switch (cmd) {
Joe Perches4500ebf2011-07-01 09:43:07 +0000870 case SIOCOUTQ: {
871 int amount = sk_wmem_alloc_get(sk);
Eric Dumazet31e6d362009-06-17 19:05:41 -0700872
Joe Perches4500ebf2011-07-01 09:43:07 +0000873 return put_user(amount, (int __user *)arg);
874 }
875 case SIOCINQ: {
876 struct sk_buff *skb;
877 int amount = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700878
Joe Perches4500ebf2011-07-01 09:43:07 +0000879 spin_lock_bh(&sk->sk_receive_queue.lock);
880 skb = skb_peek(&sk->sk_receive_queue);
Ian Morris00db4122015-04-03 09:17:27 +0100881 if (skb)
Joe Perches4500ebf2011-07-01 09:43:07 +0000882 amount = skb->len;
883 spin_unlock_bh(&sk->sk_receive_queue.lock);
884 return put_user(amount, (int __user *)arg);
885 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700886
Joe Perches4500ebf2011-07-01 09:43:07 +0000887 default:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700888#ifdef CONFIG_IP_MROUTE
Joe Perches4500ebf2011-07-01 09:43:07 +0000889 return ipmr_ioctl(sk, cmd, (void __user *)arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700890#else
Joe Perches4500ebf2011-07-01 09:43:07 +0000891 return -ENOIOCTLCMD;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700892#endif
893 }
894}
895
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000896#ifdef CONFIG_COMPAT
897static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg)
898{
899 switch (cmd) {
900 case SIOCOUTQ:
901 case SIOCINQ:
902 return -ENOIOCTLCMD;
903 default:
904#ifdef CONFIG_IP_MROUTE
905 return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg));
906#else
907 return -ENOIOCTLCMD;
908#endif
909 }
910}
911#endif
912
Linus Torvalds1da177e2005-04-16 15:20:36 -0700913struct proto raw_prot = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800914 .name = "RAW",
915 .owner = THIS_MODULE,
916 .close = raw_close,
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700917 .destroy = raw_destroy,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800918 .connect = ip4_datagram_connect,
919 .disconnect = udp_disconnect,
920 .ioctl = raw_ioctl,
921 .init = raw_init,
922 .setsockopt = raw_setsockopt,
923 .getsockopt = raw_getsockopt,
924 .sendmsg = raw_sendmsg,
925 .recvmsg = raw_recvmsg,
926 .bind = raw_bind,
927 .backlog_rcv = raw_rcv_skb,
Steffen Klassert8141ed92013-01-21 02:00:03 +0000928 .release_cb = ip4_datagram_release_cb,
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700929 .hash = raw_hash_sk,
930 .unhash = raw_unhash_sk,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800931 .obj_size = sizeof(struct raw_sock),
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700932 .h.raw_hash = &raw_v4_hashinfo,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800933#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800934 .compat_setsockopt = compat_raw_setsockopt,
935 .compat_getsockopt = compat_raw_getsockopt,
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000936 .compat_ioctl = compat_raw_ioctl,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800937#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700938};
939
940#ifdef CONFIG_PROC_FS
Linus Torvalds1da177e2005-04-16 15:20:36 -0700941static struct sock *raw_get_first(struct seq_file *seq)
942{
943 struct sock *sk;
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800944 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700945
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800946 for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
947 ++state->bucket) {
Sasha Levinb67bfe02013-02-27 17:06:00 -0800948 sk_for_each(sk, &state->h->ht[state->bucket])
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900949 if (sock_net(sk) == seq_file_net(seq))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700950 goto found;
951 }
952 sk = NULL;
953found:
954 return sk;
955}
956
957static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
958{
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800959 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700960
961 do {
962 sk = sk_next(sk);
963try_again:
964 ;
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900965 } while (sk && sock_net(sk) != seq_file_net(seq));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800967 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800968 sk = sk_head(&state->h->ht[state->bucket]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700969 goto try_again;
970 }
971 return sk;
972}
973
974static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
975{
976 struct sock *sk = raw_get_first(seq);
977
978 if (sk)
979 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
980 --pos;
981 return pos ? NULL : sk;
982}
983
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800984void *raw_seq_start(struct seq_file *seq, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700985{
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800986 struct raw_iter_state *state = raw_seq_private(seq);
987
988 read_lock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700989 return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
990}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800991EXPORT_SYMBOL_GPL(raw_seq_start);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700992
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800993void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700994{
995 struct sock *sk;
996
997 if (v == SEQ_START_TOKEN)
998 sk = raw_get_first(seq);
999 else
1000 sk = raw_get_next(seq, v);
1001 ++*pos;
1002 return sk;
1003}
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001004EXPORT_SYMBOL_GPL(raw_seq_next);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001005
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001006void raw_seq_stop(struct seq_file *seq, void *v)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001007{
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001008 struct raw_iter_state *state = raw_seq_private(seq);
1009
1010 read_unlock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001011}
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001012EXPORT_SYMBOL_GPL(raw_seq_stop);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001013
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001014static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001015{
1016 struct inet_sock *inet = inet_sk(sp);
Eric Dumazetc720c7e2009-10-15 06:30:45 +00001017 __be32 dest = inet->inet_daddr,
1018 src = inet->inet_rcv_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001019 __u16 destp = 0,
Eric Dumazetc720c7e2009-10-15 06:30:45 +00001020 srcp = inet->inet_num;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001021
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001022 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
Francesco Fuscod14c5ab2013-08-15 13:42:14 +02001023 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d\n",
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001024 i, src, srcp, dest, destp, sp->sk_state,
Eric Dumazet31e6d362009-06-17 19:05:41 -07001025 sk_wmem_alloc_get(sp),
1026 sk_rmem_alloc_get(sp),
Eric W. Biedermana7cb5a42012-05-24 01:10:10 -06001027 0, 0L, 0,
1028 from_kuid_munged(seq_user_ns(seq), sock_i_uid(sp)),
1029 0, sock_i_ino(sp),
Wang Chen33c732c2007-11-13 20:30:01 -08001030 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001031}
1032
1033static int raw_seq_show(struct seq_file *seq, void *v)
1034{
Linus Torvalds1da177e2005-04-16 15:20:36 -07001035 if (v == SEQ_START_TOKEN)
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001036 seq_printf(seq, " sl local_address rem_address st tx_queue "
1037 "rx_queue tr tm->when retrnsmt uid timeout "
Eric Dumazetcb61cb92008-06-17 21:04:56 -07001038 "inode ref pointer drops\n");
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001039 else
1040 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001041 return 0;
1042}
1043
Stephen Hemmingerf6908082007-03-12 14:34:29 -07001044static const struct seq_operations raw_seq_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001045 .start = raw_seq_start,
1046 .next = raw_seq_next,
1047 .stop = raw_seq_stop,
1048 .show = raw_seq_show,
1049};
1050
Denis V. Lunev3046d762008-01-31 03:48:55 -08001051int raw_seq_open(struct inode *ino, struct file *file,
1052 struct raw_hashinfo *h, const struct seq_operations *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001053{
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001054 int err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001055 struct raw_iter_state *i;
1056
Denis V. Lunev3046d762008-01-31 03:48:55 -08001057 err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001058 if (err < 0)
1059 return err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001060
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001061 i = raw_seq_private((struct seq_file *)file->private_data);
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001062 i->h = h;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001063 return 0;
1064}
1065EXPORT_SYMBOL_GPL(raw_seq_open);
1066
1067static int raw_v4_seq_open(struct inode *inode, struct file *file)
1068{
Denis V. Lunev3046d762008-01-31 03:48:55 -08001069 return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001070}
1071
Arjan van de Ven9a321442007-02-12 00:55:35 -08001072static const struct file_operations raw_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001073 .owner = THIS_MODULE,
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001074 .open = raw_v4_seq_open,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001075 .read = seq_read,
1076 .llseek = seq_lseek,
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001077 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001078};
1079
Pavel Emelyanova308da12008-01-14 05:36:50 -08001080static __net_init int raw_init_net(struct net *net)
1081{
Gao fengd4beaa62013-02-18 01:34:54 +00001082 if (!proc_create("raw", S_IRUGO, net->proc_net, &raw_seq_fops))
Pavel Emelyanova308da12008-01-14 05:36:50 -08001083 return -ENOMEM;
1084
1085 return 0;
1086}
1087
1088static __net_exit void raw_exit_net(struct net *net)
1089{
Gao fengece31ff2013-02-18 01:34:56 +00001090 remove_proc_entry("raw", net->proc_net);
Pavel Emelyanova308da12008-01-14 05:36:50 -08001091}
1092
1093static __net_initdata struct pernet_operations raw_net_ops = {
1094 .init = raw_init_net,
1095 .exit = raw_exit_net,
1096};
1097
Linus Torvalds1da177e2005-04-16 15:20:36 -07001098int __init raw_proc_init(void)
1099{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001100 return register_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001101}
1102
1103void __init raw_proc_exit(void)
1104{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001105 unregister_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001106}
1107#endif /* CONFIG_PROC_FS */