blob: 6d0fa8fb8af0afd652d99de84faa9ab1eb4e99f4 [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>
Quentin Armitagef5220d62014-07-23 09:58:01 +010061#include <linux/igmp.h>
Eric W. Biederman457c4cb2007-09-12 12:01:34 +020062#include <net/net_namespace.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070063#include <net/dst.h>
64#include <net/sock.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070065#include <linux/ip.h>
66#include <linux/net.h>
67#include <net/ip.h>
68#include <net/icmp.h>
69#include <net/udp.h>
70#include <net/raw.h>
71#include <net/snmp.h>
Arnaldo Carvalho de Meloc752f072005-08-09 20:08:28 -070072#include <net/tcp_states.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070073#include <net/inet_common.h>
74#include <net/checksum.h>
75#include <net/xfrm.h>
76#include <linux/rtnetlink.h>
77#include <linux/proc_fs.h>
78#include <linux/seq_file.h>
79#include <linux/netfilter.h>
80#include <linux/netfilter_ipv4.h>
Eric W. Biederman709b46e2011-01-29 16:15:56 +000081#include <linux/compat.h>
Herbert Xuc008ba5b2014-11-07 21:27:09 +080082#include <linux/uio.h>
83
84struct raw_frag_vec {
Al Virob61e9dc2014-11-24 10:52:29 -050085 struct msghdr *msg;
Herbert Xuc008ba5b2014-11-07 21:27:09 +080086 union {
87 struct icmphdr icmph;
88 char c[1];
89 } hdr;
90 int hlen;
91};
Linus Torvalds1da177e2005-04-16 15:20:36 -070092
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080093static struct raw_hashinfo raw_v4_hashinfo = {
Robert P. J. Day938b93a2008-03-18 00:59:23 -070094 .lock = __RW_LOCK_UNLOCKED(raw_v4_hashinfo.lock),
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -080095};
Linus Torvalds1da177e2005-04-16 15:20:36 -070096
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -070097void raw_hash_sk(struct sock *sk)
Linus Torvalds1da177e2005-04-16 15:20:36 -070098{
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -070099 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
Pavel Emelyanov65b4c502007-11-19 22:37:24 -0800100 struct hlist_head *head;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700101
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000102 head = &h->ht[inet_sk(sk)->inet_num & (RAW_HTABLE_SIZE - 1)];
Pavel Emelyanov65b4c502007-11-19 22:37:24 -0800103
104 write_lock_bh(&h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105 sk_add_node(sk, head);
Pavel Emelyanovc29a0bc2008-03-31 19:41:46 -0700106 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
Pavel Emelyanov65b4c502007-11-19 22:37:24 -0800107 write_unlock_bh(&h->lock);
108}
109EXPORT_SYMBOL_GPL(raw_hash_sk);
110
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700111void raw_unhash_sk(struct sock *sk)
Pavel Emelyanovab707682007-11-19 22:37:58 -0800112{
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700113 struct raw_hashinfo *h = sk->sk_prot->h.raw_hash;
114
Pavel Emelyanovab707682007-11-19 22:37:58 -0800115 write_lock_bh(&h->lock);
116 if (sk_del_node_init(sk))
Pavel Emelyanovc29a0bc2008-03-31 19:41:46 -0700117 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
Pavel Emelyanovab707682007-11-19 22:37:58 -0800118 write_unlock_bh(&h->lock);
119}
120EXPORT_SYMBOL_GPL(raw_unhash_sk);
121
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800122static struct sock *__raw_v4_lookup(struct net *net, struct sock *sk,
123 unsigned short num, __be32 raddr, __be32 laddr, int dif)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700124{
Sasha Levinb67bfe02013-02-27 17:06:00 -0800125 sk_for_each_from(sk) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126 struct inet_sock *inet = inet_sk(sk);
127
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000128 if (net_eq(sock_net(sk), net) && inet->inet_num == num &&
129 !(inet->inet_daddr && inet->inet_daddr != raddr) &&
130 !(inet->inet_rcv_saddr && inet->inet_rcv_saddr != laddr) &&
Linus Torvalds1da177e2005-04-16 15:20:36 -0700131 !(sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif))
132 goto found; /* gotcha */
133 }
134 sk = NULL;
135found:
136 return sk;
137}
138
139/*
140 * 0 - deliver
141 * 1 - block
142 */
Eric Dumazetab43ed82012-09-22 00:08:29 +0000143static int icmp_filter(const struct sock *sk, const struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700144{
Eric Dumazetab43ed82012-09-22 00:08:29 +0000145 struct icmphdr _hdr;
146 const struct icmphdr *hdr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147
Eric Dumazetab43ed82012-09-22 00:08:29 +0000148 hdr = skb_header_pointer(skb, skb_transport_offset(skb),
149 sizeof(_hdr), &_hdr);
150 if (!hdr)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700151 return 1;
152
Eric Dumazetab43ed82012-09-22 00:08:29 +0000153 if (hdr->type < 32) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154 __u32 data = raw_sk(sk)->filter.data;
155
Eric Dumazetab43ed82012-09-22 00:08:29 +0000156 return ((1U << hdr->type) & data) != 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700157 }
158
159 /* Do not block unknown ICMP types */
160 return 0;
161}
162
163/* IP input processing comes here for RAW socket delivery.
164 * Caller owns SKB, so we must make clones.
165 *
166 * RFC 1122: SHOULD pass TOS value up to the transport layer.
167 * -> It does. And not only TOS, but all IP header.
168 */
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000169static int raw_v4_input(struct sk_buff *skb, const struct iphdr *iph, int hash)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700170{
171 struct sock *sk;
172 struct hlist_head *head;
Patrick McHardyd13964f2005-08-09 19:45:02 -0700173 int delivered = 0;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800174 struct net *net;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700175
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800176 read_lock(&raw_v4_hashinfo.lock);
177 head = &raw_v4_hashinfo.ht[hash];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700178 if (hlist_empty(head))
179 goto out;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800180
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900181 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800182 sk = __raw_v4_lookup(net, __sk_head(head), iph->protocol,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700183 iph->saddr, iph->daddr,
184 skb->dev->ifindex);
185
186 while (sk) {
Patrick McHardyd13964f2005-08-09 19:45:02 -0700187 delivered = 1;
Quentin Armitagef5220d62014-07-23 09:58:01 +0100188 if ((iph->protocol != IPPROTO_ICMP || !icmp_filter(sk, skb)) &&
189 ip_mc_sf_allow(sk, iph->daddr, iph->saddr,
190 skb->dev->ifindex)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700191 struct sk_buff *clone = skb_clone(skb, GFP_ATOMIC);
192
193 /* Not releasing hash table! */
194 if (clone)
195 raw_rcv(sk, clone);
196 }
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800197 sk = __raw_v4_lookup(net, sk_next(sk), iph->protocol,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700198 iph->saddr, iph->daddr,
199 skb->dev->ifindex);
200 }
201out:
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800202 read_unlock(&raw_v4_hashinfo.lock);
Patrick McHardyd13964f2005-08-09 19:45:02 -0700203 return delivered;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700204}
205
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800206int raw_local_deliver(struct sk_buff *skb, int protocol)
207{
208 int hash;
209 struct sock *raw_sk;
210
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800211 hash = protocol & (RAW_HTABLE_SIZE - 1);
212 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800213
214 /* If there maybe a raw socket we must check - if not we
215 * don't care less
216 */
217 if (raw_sk && !raw_v4_input(skb, ip_hdr(skb), hash))
218 raw_sk = NULL;
219
220 return raw_sk != NULL;
221
222}
223
224static void raw_err(struct sock *sk, struct sk_buff *skb, u32 info)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225{
226 struct inet_sock *inet = inet_sk(sk);
Arnaldo Carvalho de Melo88c76642007-03-13 14:43:18 -0300227 const int type = icmp_hdr(skb)->type;
228 const int code = icmp_hdr(skb)->code;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700229 int err = 0;
230 int harderr = 0;
231
David S. Miller36393392012-06-14 22:21:46 -0700232 if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED)
233 ipv4_sk_update_pmtu(skb, sk, info);
Duan Jiong8d65b112013-09-20 18:21:25 +0800234 else if (type == ICMP_REDIRECT) {
David S. Miller55be7a92012-07-11 21:27:49 -0700235 ipv4_sk_redirect(skb, sk);
Duan Jiong8d65b112013-09-20 18:21:25 +0800236 return;
237 }
David S. Miller36393392012-06-14 22:21:46 -0700238
Linus Torvalds1da177e2005-04-16 15:20:36 -0700239 /* Report error on raw socket, if:
240 1. User requested ip_recverr.
241 2. Socket is connected (otherwise the error indication
242 is useless without ip_recverr and error is hard.
243 */
244 if (!inet->recverr && sk->sk_state != TCP_ESTABLISHED)
245 return;
246
247 switch (type) {
248 default:
249 case ICMP_TIME_EXCEEDED:
250 err = EHOSTUNREACH;
251 break;
252 case ICMP_SOURCE_QUENCH:
253 return;
254 case ICMP_PARAMETERPROB:
255 err = EPROTO;
256 harderr = 1;
257 break;
258 case ICMP_DEST_UNREACH:
259 err = EHOSTUNREACH;
260 if (code > NR_ICMP_UNREACH)
261 break;
262 err = icmp_err_convert[code].errno;
263 harderr = icmp_err_convert[code].fatal;
264 if (code == ICMP_FRAG_NEEDED) {
265 harderr = inet->pmtudisc != IP_PMTUDISC_DONT;
266 err = EMSGSIZE;
267 }
268 }
269
270 if (inet->recverr) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000271 const struct iphdr *iph = (const struct iphdr *)skb->data;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700272 u8 *payload = skb->data + (iph->ihl << 2);
273
274 if (inet->hdrincl)
275 payload = skb->data;
276 ip_icmp_error(sk, skb, err, 0, info, payload);
277 }
278
279 if (inet->recverr || harderr) {
280 sk->sk_err = err;
281 sk->sk_error_report(sk);
282 }
283}
284
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800285void raw_icmp_error(struct sk_buff *skb, int protocol, u32 info)
286{
287 int hash;
288 struct sock *raw_sk;
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000289 const struct iphdr *iph;
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800290 struct net *net;
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800291
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800292 hash = protocol & (RAW_HTABLE_SIZE - 1);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800293
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800294 read_lock(&raw_v4_hashinfo.lock);
295 raw_sk = sk_head(&raw_v4_hashinfo.ht[hash]);
Ian Morris00db4122015-04-03 09:17:27 +0100296 if (raw_sk) {
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000297 iph = (const struct iphdr *)skb->data;
YOSHIFUJI Hideakic346dca2008-03-25 21:47:49 +0900298 net = dev_net(skb->dev);
Pavel Emelyanovbe185882008-01-14 05:35:31 -0800299
300 while ((raw_sk = __raw_v4_lookup(net, raw_sk, protocol,
301 iph->daddr, iph->saddr,
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800302 skb->dev->ifindex)) != NULL) {
303 raw_err(raw_sk, skb, info);
304 raw_sk = sk_next(raw_sk);
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000305 iph = (const struct iphdr *)skb->data;
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800306 }
307 }
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800308 read_unlock(&raw_v4_hashinfo.lock);
Pavel Emelyanov7bc54c92007-11-19 22:35:07 -0800309}
310
Daniel Baluta5e73ea12012-04-15 01:34:41 +0000311static int raw_rcv_skb(struct sock *sk, struct sk_buff *skb)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700312{
313 /* Charge it to the socket. */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900314
Shawn Bohrerfbf88662013-10-07 11:01:40 -0500315 ipv4_pktinfo_prepare(sk, skb);
Eric Dumazetd826eb12011-11-09 07:24:35 +0000316 if (sock_queue_rcv_skb(sk, skb) < 0) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700317 kfree_skb(skb);
318 return NET_RX_DROP;
319 }
320
321 return NET_RX_SUCCESS;
322}
323
324int raw_rcv(struct sock *sk, struct sk_buff *skb)
325{
326 if (!xfrm4_policy_check(sk, XFRM_POLICY_IN, skb)) {
Wang Chen33c732c2007-11-13 20:30:01 -0800327 atomic_inc(&sk->sk_drops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700328 kfree_skb(skb);
329 return NET_RX_DROP;
330 }
Patrick McHardyb59c2702006-01-06 23:06:10 -0800331 nf_reset(skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700332
Arnaldo Carvalho de Melod56f90a2007-04-10 20:50:43 -0700333 skb_push(skb, skb->data - skb_network_header(skb));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700334
335 raw_rcv_skb(sk, skb);
336 return 0;
337}
338
David S. Miller77968b72011-05-08 17:12:19 -0700339static int raw_send_hdrinc(struct sock *sk, struct flowi4 *fl4,
Al Viro7ae9abf2014-11-27 19:30:51 -0500340 struct msghdr *msg, size_t length,
David S. Miller77968b72011-05-08 17:12:19 -0700341 struct rtable **rtp,
342 unsigned int flags)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700343{
344 struct inet_sock *inet = inet_sk(sk);
Pavel Emelyanov0388b002008-07-14 23:00:43 -0700345 struct net *net = sock_net(sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700346 struct iphdr *iph;
347 struct sk_buff *skb;
Herbert Xuf844c742008-01-05 23:14:44 -0800348 unsigned int iphlen;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700349 int err;
Eric Dumazet1789a642010-06-03 22:23:57 +0000350 struct rtable *rt = *rtp;
Herbert Xu66088242011-11-18 02:20:04 +0000351 int hlen, tlen;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700352
Changli Gaod8d1f302010-06-10 23:31:35 -0700353 if (length > rt->dst.dev->mtu) {
David S. Miller77968b72011-05-08 17:12:19 -0700354 ip_local_error(sk, EMSGSIZE, fl4->daddr, inet->inet_dport,
Changli Gaod8d1f302010-06-10 23:31:35 -0700355 rt->dst.dev->mtu);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700356 return -EMSGSIZE;
357 }
358 if (flags&MSG_PROBE)
359 goto out;
360
Herbert Xu66088242011-11-18 02:20:04 +0000361 hlen = LL_RESERVED_SPACE(rt->dst.dev);
362 tlen = rt->dst.dev->needed_tailroom;
Johannes Bergf5184d22008-05-12 20:48:31 -0700363 skb = sock_alloc_send_skb(sk,
Herbert Xu66088242011-11-18 02:20:04 +0000364 length + hlen + tlen + 15,
Johannes Bergf5184d22008-05-12 20:48:31 -0700365 flags & MSG_DONTWAIT, &err);
Ian Morris51456b22015-04-03 09:17:26 +0100366 if (!skb)
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900367 goto error;
Herbert Xu66088242011-11-18 02:20:04 +0000368 skb_reserve(skb, hlen);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700369
370 skb->priority = sk->sk_priority;
Laszlo Attila Toth4a19ec52008-01-30 19:08:16 -0800371 skb->mark = sk->sk_mark;
Changli Gaod8d1f302010-06-10 23:31:35 -0700372 skb_dst_set(skb, &rt->dst);
Eric Dumazet1789a642010-06-03 22:23:57 +0000373 *rtp = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700374
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300375 skb_reset_network_header(skb);
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700376 iph = ip_hdr(skb);
Arnaldo Carvalho de Melo7e28ecc2007-03-10 18:40:59 -0300377 skb_put(skb, length);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700378
379 skb->ip_summed = CHECKSUM_NONE;
380
Willem de Bruijn11878b42014-07-14 17:55:06 -0400381 sock_tx_timestamp(sk, &skb_shinfo(skb)->tx_flags);
382
Arnaldo Carvalho de Melob0e380b2007-04-10 21:21:55 -0700383 skb->transport_header = skb->network_header;
Neil Horman55888df2009-10-28 08:59:47 +0000384 err = -EFAULT;
Al Viro21226ab2014-11-28 15:48:29 -0500385 if (memcpy_from_msg(iph, msg, length))
Neil Horman55888df2009-10-28 08:59:47 +0000386 goto error_free;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700387
Herbert Xuf844c742008-01-05 23:14:44 -0800388 iphlen = iph->ihl * 4;
Neil Horman55888df2009-10-28 08:59:47 +0000389
390 /*
391 * We don't want to modify the ip header, but we do need to
392 * be sure that it won't cause problems later along the network
393 * stack. Specifically we want to make sure that iph->ihl is a
394 * sane value. If ihl points beyond the length of the buffer passed
395 * in, reject the frame as invalid
396 */
397 err = -EINVAL;
398 if (iphlen > length)
399 goto error_free;
400
401 if (iphlen >= sizeof(*iph)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700402 if (!iph->saddr)
David S. Miller77968b72011-05-08 17:12:19 -0700403 iph->saddr = fl4->saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700404 iph->check = 0;
405 iph->tot_len = htons(length);
406 if (!iph->id)
Hannes Frederic Sowab6a77192015-03-25 17:07:44 +0100407 ip_select_ident(net, skb, NULL);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700408
409 iph->check = ip_fast_csum((unsigned char *)iph, iph->ihl);
410 }
David L Stevens96793b42007-09-17 09:57:33 -0700411 if (iph->protocol == IPPROTO_ICMP)
Pavel Emelyanov0388b002008-07-14 23:00:43 -0700412 icmp_out_count(net, ((struct icmphdr *)
David L Stevens96793b42007-09-17 09:57:33 -0700413 skb_transport_header(skb))->type);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700414
Jan Engelhardt9bbc7682010-03-23 04:07:29 +0100415 err = NF_HOOK(NFPROTO_IPV4, NF_INET_LOCAL_OUT, skb, NULL,
Changli Gaod8d1f302010-06-10 23:31:35 -0700416 rt->dst.dev, dst_output);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700417 if (err > 0)
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700418 err = net_xmit_errno(err);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700419 if (err)
420 goto error;
421out:
422 return 0;
423
Neil Horman55888df2009-10-28 08:59:47 +0000424error_free:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700425 kfree_skb(skb);
426error:
Pavel Emelyanov5e38e272008-07-16 20:19:49 -0700427 IP_INC_STATS(net, IPSTATS_MIB_OUTDISCARDS);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700428 if (err == -ENOBUFS && !inet->recverr)
429 err = 0;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900430 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700431}
432
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800433static int raw_probe_proto_opt(struct raw_frag_vec *rfv, struct flowi4 *fl4)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700434{
Herbert Xu32b59132014-11-07 21:27:08 +0800435 int err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700436
Herbert Xu32b59132014-11-07 21:27:08 +0800437 if (fl4->flowi4_proto != IPPROTO_ICMP)
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800438 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700439
Herbert Xu32b59132014-11-07 21:27:08 +0800440 /* We only need the first two bytes. */
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800441 rfv->hlen = 2;
442
Al Virob61e9dc2014-11-24 10:52:29 -0500443 err = memcpy_from_msg(rfv->hdr.c, rfv->msg, rfv->hlen);
Herbert Xu32b59132014-11-07 21:27:08 +0800444 if (err)
445 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700446
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800447 fl4->fl4_icmp_type = rfv->hdr.icmph.type;
448 fl4->fl4_icmp_code = rfv->hdr.icmph.code;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700449
Heiko Carstensa27b58f2006-10-30 15:06:12 -0800450 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700451}
452
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800453static int raw_getfrag(void *from, char *to, int offset, int len, int odd,
454 struct sk_buff *skb)
455{
456 struct raw_frag_vec *rfv = from;
457
458 if (offset < rfv->hlen) {
459 int copy = min(rfv->hlen - offset, len);
460
461 if (skb->ip_summed == CHECKSUM_PARTIAL)
462 memcpy(to, rfv->hdr.c + offset, copy);
463 else
464 skb->csum = csum_block_add(
465 skb->csum,
466 csum_partial_copy_nocheck(rfv->hdr.c + offset,
467 to, copy, 0),
468 odd);
469
470 odd = 0;
471 offset += copy;
472 to += copy;
473 len -= copy;
474
475 if (!len)
476 return 0;
477 }
478
479 offset -= rfv->hlen;
480
Al Virof69e6d12014-11-24 13:23:40 -0500481 return ip_generic_getfrag(rfv->msg, to, offset, len, odd, skb);
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800482}
483
Ying Xue1b784142015-03-02 15:37:48 +0800484static int raw_sendmsg(struct sock *sk, struct msghdr *msg, size_t len)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700485{
486 struct inet_sock *inet = inet_sk(sk);
487 struct ipcm_cookie ipc;
488 struct rtable *rt = NULL;
David S. Miller77968b72011-05-08 17:12:19 -0700489 struct flowi4 fl4;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700490 int free = 0;
Al Viro3ca3c682006-09-27 18:28:07 -0700491 __be32 daddr;
Al Viroc1d18f92006-09-27 18:28:28 -0700492 __be32 saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700493 u8 tos;
494 int err;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000495 struct ip_options_data opt_copy;
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800496 struct raw_frag_vec rfv;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700497
498 err = -EMSGSIZE;
Jesper Juhl926d4b82005-06-18 23:00:00 -0700499 if (len > 0xFFFF)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700500 goto out;
501
502 /*
503 * Check the flags.
504 */
505
506 err = -EOPNOTSUPP;
507 if (msg->msg_flags & MSG_OOB) /* Mirror BSD error message */
508 goto out; /* compatibility */
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900509
Linus Torvalds1da177e2005-04-16 15:20:36 -0700510 /*
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900511 * Get and verify the address.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700512 */
513
514 if (msg->msg_namelen) {
Steffen Hurrle342dfc32014-01-17 22:53:15 +0100515 DECLARE_SOCKADDR(struct sockaddr_in *, usin, msg->msg_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700516 err = -EINVAL;
517 if (msg->msg_namelen < sizeof(*usin))
518 goto out;
519 if (usin->sin_family != AF_INET) {
Joe Perches058bd4d2012-03-11 18:36:11 +0000520 pr_info_once("%s: %s forgot to set AF_INET. Fix it!\n",
521 __func__, current->comm);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700522 err = -EAFNOSUPPORT;
523 if (usin->sin_family)
524 goto out;
525 }
526 daddr = usin->sin_addr.s_addr;
527 /* ANK: I did not forget to get protocol from port field.
528 * I just do not know, who uses this weirdness.
529 * IP_HDRINCL is much more convenient.
530 */
531 } else {
532 err = -EDESTADDRREQ;
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900533 if (sk->sk_state != TCP_ESTABLISHED)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700534 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000535 daddr = inet->inet_daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700536 }
537
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000538 ipc.addr = inet->inet_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700539 ipc.opt = NULL;
Oliver Hartkopp2244d072010-08-17 08:59:14 +0000540 ipc.tx_flags = 0;
Francesco Fuscoaa661582013-09-24 15:43:09 +0200541 ipc.ttl = 0;
542 ipc.tos = -1;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700543 ipc.oif = sk->sk_bound_dev_if;
544
545 if (msg->msg_controllen) {
Hannes Frederic Sowac8e6ad02014-02-18 21:38:08 +0100546 err = ip_cmsg_send(sock_net(sk), msg, &ipc, false);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700547 if (err)
548 goto out;
549 if (ipc.opt)
550 free = 1;
551 }
552
553 saddr = ipc.addr;
554 ipc.addr = daddr;
555
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000556 if (!ipc.opt) {
557 struct ip_options_rcu *inet_opt;
558
559 rcu_read_lock();
560 inet_opt = rcu_dereference(inet->inet_opt);
561 if (inet_opt) {
562 memcpy(&opt_copy, inet_opt,
563 sizeof(*inet_opt) + inet_opt->opt.optlen);
564 ipc.opt = &opt_copy.opt;
565 }
566 rcu_read_unlock();
567 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700568
569 if (ipc.opt) {
570 err = -EINVAL;
571 /* Linux does not mangle headers on raw sockets,
572 * so that IP options + IP_HDRINCL is non-sense.
573 */
574 if (inet->hdrincl)
575 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000576 if (ipc.opt->opt.srr) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700577 if (!daddr)
578 goto done;
Eric Dumazetf6d8bd02011-04-21 09:45:37 +0000579 daddr = ipc.opt->opt.faddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700580 }
581 }
Francesco Fuscoaa661582013-09-24 15:43:09 +0200582 tos = get_rtconn_flags(&ipc, sk);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700583 if (msg->msg_flags & MSG_DONTROUTE)
584 tos |= RTO_ONLINK;
585
Joe Perchesf97c1e02007-12-16 13:45:43 -0800586 if (ipv4_is_multicast(daddr)) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700587 if (!ipc.oif)
588 ipc.oif = inet->mc_index;
589 if (!saddr)
590 saddr = inet->mc_addr;
Erich E. Hoover76e21052012-02-08 09:11:07 +0000591 } else if (!ipc.oif)
592 ipc.oif = inet->uc_index;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700593
David S. Miller77968b72011-05-08 17:12:19 -0700594 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
595 RT_SCOPE_UNIVERSE,
596 inet->hdrincl ? IPPROTO_RAW : sk->sk_protocol,
Steffen Klassert0e0d44a2013-08-28 08:04:14 +0200597 inet_sk_flowi_flags(sk) |
Chris Clarkc27c9322013-08-27 12:02:15 -0600598 (inet->hdrincl ? FLOWI_FLAG_KNOWN_NH : 0),
Julian Anastasov47670b72011-08-07 09:16:09 +0000599 daddr, saddr, 0, 0);
David S. Milleref164ae2011-03-31 04:53:51 -0700600
David S. Miller77968b72011-05-08 17:12:19 -0700601 if (!inet->hdrincl) {
Al Virob61e9dc2014-11-24 10:52:29 -0500602 rfv.msg = msg;
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800603 rfv.hlen = 0;
604
605 err = raw_probe_proto_opt(&rfv, &fl4);
David S. Miller77968b72011-05-08 17:12:19 -0700606 if (err)
David S. Millerb23dd4f2011-03-02 14:31:35 -0800607 goto done;
David S. Miller77968b72011-05-08 17:12:19 -0700608 }
609
610 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
611 rt = ip_route_output_flow(sock_net(sk), &fl4, sk);
612 if (IS_ERR(rt)) {
613 err = PTR_ERR(rt);
614 rt = NULL;
615 goto done;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700616 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700617
618 err = -EACCES;
619 if (rt->rt_flags & RTCF_BROADCAST && !sock_flag(sk, SOCK_BROADCAST))
620 goto done;
621
622 if (msg->msg_flags & MSG_CONFIRM)
623 goto do_confirm;
624back_from_confirm:
625
626 if (inet->hdrincl)
Al Viro7ae9abf2014-11-27 19:30:51 -0500627 err = raw_send_hdrinc(sk, &fl4, msg, len,
David S. Miller77968b72011-05-08 17:12:19 -0700628 &rt, msg->msg_flags);
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900629
Linus Torvalds1da177e2005-04-16 15:20:36 -0700630 else {
Willem de Bruijn11878b42014-07-14 17:55:06 -0400631 sock_tx_timestamp(sk, &ipc.tx_flags);
632
Linus Torvalds1da177e2005-04-16 15:20:36 -0700633 if (!ipc.addr)
David S. Miller77968b72011-05-08 17:12:19 -0700634 ipc.addr = fl4.daddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700635 lock_sock(sk);
Herbert Xuc008ba5b2014-11-07 21:27:09 +0800636 err = ip_append_data(sk, &fl4, raw_getfrag,
637 &rfv, len, 0,
David S. Miller77968b72011-05-08 17:12:19 -0700638 &ipc, &rt, msg->msg_flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700639 if (err)
640 ip_flush_pending_frames(sk);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700641 else if (!(msg->msg_flags & MSG_MORE)) {
David S. Miller77968b72011-05-08 17:12:19 -0700642 err = ip_push_pending_frames(sk, &fl4);
Eric Dumazet6ce9e7b2009-09-02 18:05:33 -0700643 if (err == -ENOBUFS && !inet->recverr)
644 err = 0;
645 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700646 release_sock(sk);
647 }
648done:
649 if (free)
650 kfree(ipc.opt);
651 ip_rt_put(rt);
652
Jesper Juhl5418c6922005-06-18 22:59:45 -0700653out:
654 if (err < 0)
655 return err;
656 return len;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700657
658do_confirm:
Changli Gaod8d1f302010-06-10 23:31:35 -0700659 dst_confirm(&rt->dst);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660 if (!(msg->msg_flags & MSG_PROBE) || len)
661 goto back_from_confirm;
662 err = 0;
663 goto done;
664}
665
666static void raw_close(struct sock *sk, long timeout)
667{
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +0900668 /*
Lucas De Marchi25985ed2011-03-30 22:57:33 -0300669 * Raw sockets may have direct kernel references. Kill them.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700670 */
671 ip_ra_control(sk, 0, NULL);
672
673 sk_common_release(sk);
674}
675
Brian Haley7d06b2e2008-06-14 17:04:49 -0700676static void raw_destroy(struct sock *sk)
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700677{
678 lock_sock(sk);
679 ip_flush_pending_frames(sk);
680 release_sock(sk);
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700681}
682
Linus Torvalds1da177e2005-04-16 15:20:36 -0700683/* This gets rid of all the nasties in af_inet. -DaveM */
684static int raw_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
685{
686 struct inet_sock *inet = inet_sk(sk);
687 struct sockaddr_in *addr = (struct sockaddr_in *) uaddr;
688 int ret = -EINVAL;
689 int chk_addr_ret;
690
691 if (sk->sk_state != TCP_CLOSE || addr_len < sizeof(struct sockaddr_in))
692 goto out;
YOSHIFUJI Hideaki3b1e0a62008-03-26 02:26:21 +0900693 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700694 ret = -EADDRNOTAVAIL;
695 if (addr->sin_addr.s_addr && chk_addr_ret != RTN_LOCAL &&
696 chk_addr_ret != RTN_MULTICAST && chk_addr_ret != RTN_BROADCAST)
697 goto out;
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000698 inet->inet_rcv_saddr = inet->inet_saddr = addr->sin_addr.s_addr;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700699 if (chk_addr_ret == RTN_MULTICAST || chk_addr_ret == RTN_BROADCAST)
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000700 inet->inet_saddr = 0; /* Use device */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700701 sk_dst_reset(sk);
702 ret = 0;
703out: return ret;
704}
705
706/*
707 * This should be easy, if there is something there
708 * we return it, otherwise we block.
709 */
710
Ying Xue1b784142015-03-02 15:37:48 +0800711static int raw_recvmsg(struct sock *sk, struct msghdr *msg, size_t len,
712 int noblock, int flags, int *addr_len)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700713{
714 struct inet_sock *inet = inet_sk(sk);
715 size_t copied = 0;
716 int err = -EOPNOTSUPP;
Steffen Hurrle342dfc32014-01-17 22:53:15 +0100717 DECLARE_SOCKADDR(struct sockaddr_in *, sin, msg->msg_name);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700718 struct sk_buff *skb;
719
720 if (flags & MSG_OOB)
721 goto out;
722
Linus Torvalds1da177e2005-04-16 15:20:36 -0700723 if (flags & MSG_ERRQUEUE) {
Hannes Frederic Sowa85fbaa72013-11-23 00:46:12 +0100724 err = ip_recv_error(sk, msg, len, addr_len);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700725 goto out;
726 }
727
728 skb = skb_recv_datagram(sk, flags, noblock, &err);
729 if (!skb)
730 goto out;
731
732 copied = skb->len;
733 if (len < copied) {
734 msg->msg_flags |= MSG_TRUNC;
735 copied = len;
736 }
737
David S. Miller51f3d022014-11-05 16:46:40 -0500738 err = skb_copy_datagram_msg(skb, 0, msg, copied);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700739 if (err)
740 goto done;
741
Neil Horman3b885782009-10-12 13:26:31 -0700742 sock_recv_ts_and_drops(msg, sk, skb);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700743
744 /* Copy the address. */
745 if (sin) {
746 sin->sin_family = AF_INET;
Arnaldo Carvalho de Meloeddc9ec2007-04-20 22:47:35 -0700747 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
Tetsuo Handaf59fc7f2006-07-25 17:05:35 -0700748 sin->sin_port = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700749 memset(&sin->sin_zero, 0, sizeof(sin->sin_zero));
Hannes Frederic Sowabceaa902013-11-18 04:20:45 +0100750 *addr_len = sizeof(*sin);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700751 }
752 if (inet->cmsg_flags)
753 ip_cmsg_recv(msg, skb);
754 if (flags & MSG_TRUNC)
755 copied = skb->len;
756done:
757 skb_free_datagram(sk, skb);
Jesper Juhl5418c6922005-06-18 22:59:45 -0700758out:
759 if (err)
760 return err;
761 return copied;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700762}
763
764static int raw_init(struct sock *sk)
765{
766 struct raw_sock *rp = raw_sk(sk);
767
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000768 if (inet_sk(sk)->inet_num == IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700769 memset(&rp->filter, 0, sizeof(rp->filter));
770 return 0;
771}
772
773static int raw_seticmpfilter(struct sock *sk, char __user *optval, int optlen)
774{
775 if (optlen > sizeof(struct icmp_filter))
776 optlen = sizeof(struct icmp_filter);
777 if (copy_from_user(&raw_sk(sk)->filter, optval, optlen))
778 return -EFAULT;
779 return 0;
780}
781
782static int raw_geticmpfilter(struct sock *sk, char __user *optval, int __user *optlen)
783{
784 int len, ret = -EFAULT;
785
786 if (get_user(len, optlen))
787 goto out;
788 ret = -EINVAL;
789 if (len < 0)
790 goto out;
791 if (len > sizeof(struct icmp_filter))
792 len = sizeof(struct icmp_filter);
793 ret = -EFAULT;
794 if (put_user(len, optlen) ||
795 copy_to_user(optval, &raw_sk(sk)->filter, len))
796 goto out;
797 ret = 0;
798out: return ret;
799}
800
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800801static int do_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700802 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700803{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700804 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000805 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700806 return -EOPNOTSUPP;
807 else
808 return raw_seticmpfilter(sk, optval, optlen);
809 }
810 return -ENOPROTOOPT;
811}
812
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800813static int raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700814 char __user *optval, unsigned int optlen)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700815{
816 if (level != SOL_RAW)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800817 return ip_setsockopt(sk, level, optname, optval, optlen);
818 return do_raw_setsockopt(sk, level, optname, optval, optlen);
819}
Linus Torvalds1da177e2005-04-16 15:20:36 -0700820
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800821#ifdef CONFIG_COMPAT
822static int compat_raw_setsockopt(struct sock *sk, int level, int optname,
David S. Millerb7058842009-09-30 16:12:20 -0700823 char __user *optval, unsigned int optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800824{
825 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800826 return compat_ip_setsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800827 return do_raw_setsockopt(sk, level, optname, optval, optlen);
828}
829#endif
830
831static int do_raw_getsockopt(struct sock *sk, int level, int optname,
832 char __user *optval, int __user *optlen)
833{
Linus Torvalds1da177e2005-04-16 15:20:36 -0700834 if (optname == ICMP_FILTER) {
Eric Dumazetc720c7e2009-10-15 06:30:45 +0000835 if (inet_sk(sk)->inet_num != IPPROTO_ICMP)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700836 return -EOPNOTSUPP;
837 else
838 return raw_geticmpfilter(sk, optval, optlen);
839 }
840 return -ENOPROTOOPT;
841}
842
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800843static int raw_getsockopt(struct sock *sk, int level, int optname,
844 char __user *optval, int __user *optlen)
845{
846 if (level != SOL_RAW)
847 return ip_getsockopt(sk, level, optname, optval, optlen);
848 return do_raw_getsockopt(sk, level, optname, optval, optlen);
849}
850
851#ifdef CONFIG_COMPAT
852static int compat_raw_getsockopt(struct sock *sk, int level, int optname,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800853 char __user *optval, int __user *optlen)
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800854{
855 if (level != SOL_RAW)
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800856 return compat_ip_getsockopt(sk, level, optname, optval, optlen);
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800857 return do_raw_getsockopt(sk, level, optname, optval, optlen);
858}
859#endif
860
Linus Torvalds1da177e2005-04-16 15:20:36 -0700861static int raw_ioctl(struct sock *sk, int cmd, unsigned long arg)
862{
863 switch (cmd) {
Joe Perches4500ebf2011-07-01 09:43:07 +0000864 case SIOCOUTQ: {
865 int amount = sk_wmem_alloc_get(sk);
Eric Dumazet31e6d362009-06-17 19:05:41 -0700866
Joe Perches4500ebf2011-07-01 09:43:07 +0000867 return put_user(amount, (int __user *)arg);
868 }
869 case SIOCINQ: {
870 struct sk_buff *skb;
871 int amount = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700872
Joe Perches4500ebf2011-07-01 09:43:07 +0000873 spin_lock_bh(&sk->sk_receive_queue.lock);
874 skb = skb_peek(&sk->sk_receive_queue);
Ian Morris00db4122015-04-03 09:17:27 +0100875 if (skb)
Joe Perches4500ebf2011-07-01 09:43:07 +0000876 amount = skb->len;
877 spin_unlock_bh(&sk->sk_receive_queue.lock);
878 return put_user(amount, (int __user *)arg);
879 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700880
Joe Perches4500ebf2011-07-01 09:43:07 +0000881 default:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700882#ifdef CONFIG_IP_MROUTE
Joe Perches4500ebf2011-07-01 09:43:07 +0000883 return ipmr_ioctl(sk, cmd, (void __user *)arg);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700884#else
Joe Perches4500ebf2011-07-01 09:43:07 +0000885 return -ENOIOCTLCMD;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700886#endif
887 }
888}
889
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000890#ifdef CONFIG_COMPAT
891static int compat_raw_ioctl(struct sock *sk, unsigned int cmd, unsigned long arg)
892{
893 switch (cmd) {
894 case SIOCOUTQ:
895 case SIOCINQ:
896 return -ENOIOCTLCMD;
897 default:
898#ifdef CONFIG_IP_MROUTE
899 return ipmr_compat_ioctl(sk, cmd, compat_ptr(arg));
900#else
901 return -ENOIOCTLCMD;
902#endif
903 }
904}
905#endif
906
Linus Torvalds1da177e2005-04-16 15:20:36 -0700907struct proto raw_prot = {
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800908 .name = "RAW",
909 .owner = THIS_MODULE,
910 .close = raw_close,
Denis V. Lunev22dd4852008-06-04 15:16:12 -0700911 .destroy = raw_destroy,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800912 .connect = ip4_datagram_connect,
913 .disconnect = udp_disconnect,
914 .ioctl = raw_ioctl,
915 .init = raw_init,
916 .setsockopt = raw_setsockopt,
917 .getsockopt = raw_getsockopt,
918 .sendmsg = raw_sendmsg,
919 .recvmsg = raw_recvmsg,
920 .bind = raw_bind,
921 .backlog_rcv = raw_rcv_skb,
Steffen Klassert8141ed92013-01-21 02:00:03 +0000922 .release_cb = ip4_datagram_release_cb,
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700923 .hash = raw_hash_sk,
924 .unhash = raw_unhash_sk,
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800925 .obj_size = sizeof(struct raw_sock),
Pavel Emelyanovfc8717b2008-03-22 16:56:51 -0700926 .h.raw_hash = &raw_v4_hashinfo,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800927#ifdef CONFIG_COMPAT
Arnaldo Carvalho de Melo543d9cf2006-03-20 22:48:35 -0800928 .compat_setsockopt = compat_raw_setsockopt,
929 .compat_getsockopt = compat_raw_getsockopt,
Eric W. Biederman709b46e2011-01-29 16:15:56 +0000930 .compat_ioctl = compat_raw_ioctl,
Dmitry Mishin3fdadf72006-03-20 22:45:21 -0800931#endif
Linus Torvalds1da177e2005-04-16 15:20:36 -0700932};
933
934#ifdef CONFIG_PROC_FS
Linus Torvalds1da177e2005-04-16 15:20:36 -0700935static struct sock *raw_get_first(struct seq_file *seq)
936{
937 struct sock *sk;
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800938 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700939
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800940 for (state->bucket = 0; state->bucket < RAW_HTABLE_SIZE;
941 ++state->bucket) {
Sasha Levinb67bfe02013-02-27 17:06:00 -0800942 sk_for_each(sk, &state->h->ht[state->bucket])
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900943 if (sock_net(sk) == seq_file_net(seq))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700944 goto found;
945 }
946 sk = NULL;
947found:
948 return sk;
949}
950
951static struct sock *raw_get_next(struct seq_file *seq, struct sock *sk)
952{
Jianjun Kong5a5f3a82008-11-03 00:24:34 -0800953 struct raw_iter_state *state = raw_seq_private(seq);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700954
955 do {
956 sk = sk_next(sk);
957try_again:
958 ;
YOSHIFUJI Hideaki12188542008-03-26 02:36:06 +0900959 } while (sk && sock_net(sk) != seq_file_net(seq));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700960
Pavel Emelyanovb673e4d2007-11-19 22:36:45 -0800961 if (!sk && ++state->bucket < RAW_HTABLE_SIZE) {
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800962 sk = sk_head(&state->h->ht[state->bucket]);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700963 goto try_again;
964 }
965 return sk;
966}
967
968static struct sock *raw_get_idx(struct seq_file *seq, loff_t pos)
969{
970 struct sock *sk = raw_get_first(seq);
971
972 if (sk)
973 while (pos && (sk = raw_get_next(seq, sk)) != NULL)
974 --pos;
975 return pos ? NULL : sk;
976}
977
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800978void *raw_seq_start(struct seq_file *seq, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700979{
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800980 struct raw_iter_state *state = raw_seq_private(seq);
981
982 read_lock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700983 return *pos ? raw_get_idx(seq, *pos - 1) : SEQ_START_TOKEN;
984}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800985EXPORT_SYMBOL_GPL(raw_seq_start);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700986
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800987void *raw_seq_next(struct seq_file *seq, void *v, loff_t *pos)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700988{
989 struct sock *sk;
990
991 if (v == SEQ_START_TOKEN)
992 sk = raw_get_first(seq);
993 else
994 sk = raw_get_next(seq, v);
995 ++*pos;
996 return sk;
997}
Pavel Emelyanov42a73802007-11-19 22:38:33 -0800998EXPORT_SYMBOL_GPL(raw_seq_next);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700999
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001000void raw_seq_stop(struct seq_file *seq, void *v)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001001{
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001002 struct raw_iter_state *state = raw_seq_private(seq);
1003
1004 read_unlock(&state->h->lock);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001005}
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001006EXPORT_SYMBOL_GPL(raw_seq_stop);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001007
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001008static void raw_sock_seq_show(struct seq_file *seq, struct sock *sp, int i)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001009{
1010 struct inet_sock *inet = inet_sk(sp);
Eric Dumazetc720c7e2009-10-15 06:30:45 +00001011 __be32 dest = inet->inet_daddr,
1012 src = inet->inet_rcv_saddr;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001013 __u16 destp = 0,
Eric Dumazetc720c7e2009-10-15 06:30:45 +00001014 srcp = inet->inet_num;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001015
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001016 seq_printf(seq, "%4d: %08X:%04X %08X:%04X"
Francesco Fuscod14c5ab2013-08-15 13:42:14 +02001017 " %02X %08X:%08X %02X:%08lX %08X %5u %8d %lu %d %pK %d\n",
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09001018 i, src, srcp, dest, destp, sp->sk_state,
Eric Dumazet31e6d362009-06-17 19:05:41 -07001019 sk_wmem_alloc_get(sp),
1020 sk_rmem_alloc_get(sp),
Eric W. Biedermana7cb5a42012-05-24 01:10:10 -06001021 0, 0L, 0,
1022 from_kuid_munged(seq_user_ns(seq), sock_i_uid(sp)),
1023 0, sock_i_ino(sp),
Wang Chen33c732c2007-11-13 20:30:01 -08001024 atomic_read(&sp->sk_refcnt), sp, atomic_read(&sp->sk_drops));
Linus Torvalds1da177e2005-04-16 15:20:36 -07001025}
1026
1027static int raw_seq_show(struct seq_file *seq, void *v)
1028{
Linus Torvalds1da177e2005-04-16 15:20:36 -07001029 if (v == SEQ_START_TOKEN)
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001030 seq_printf(seq, " sl local_address rem_address st tx_queue "
1031 "rx_queue tr tm->when retrnsmt uid timeout "
Eric Dumazetcb61cb92008-06-17 21:04:56 -07001032 "inode ref pointer drops\n");
Denis V. Lunev8cd850e2008-01-31 03:46:43 -08001033 else
1034 raw_sock_seq_show(seq, v, raw_seq_private(seq)->bucket);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001035 return 0;
1036}
1037
Stephen Hemmingerf6908082007-03-12 14:34:29 -07001038static const struct seq_operations raw_seq_ops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001039 .start = raw_seq_start,
1040 .next = raw_seq_next,
1041 .stop = raw_seq_stop,
1042 .show = raw_seq_show,
1043};
1044
Denis V. Lunev3046d762008-01-31 03:48:55 -08001045int raw_seq_open(struct inode *ino, struct file *file,
1046 struct raw_hashinfo *h, const struct seq_operations *ops)
Linus Torvalds1da177e2005-04-16 15:20:36 -07001047{
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001048 int err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001049 struct raw_iter_state *i;
1050
Denis V. Lunev3046d762008-01-31 03:48:55 -08001051 err = seq_open_net(ino, file, ops, sizeof(struct raw_iter_state));
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001052 if (err < 0)
1053 return err;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001054
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001055 i = raw_seq_private((struct seq_file *)file->private_data);
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001056 i->h = h;
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001057 return 0;
1058}
1059EXPORT_SYMBOL_GPL(raw_seq_open);
1060
1061static int raw_v4_seq_open(struct inode *inode, struct file *file)
1062{
Denis V. Lunev3046d762008-01-31 03:48:55 -08001063 return raw_seq_open(inode, file, &raw_v4_hashinfo, &raw_seq_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001064}
1065
Arjan van de Ven9a321442007-02-12 00:55:35 -08001066static const struct file_operations raw_seq_fops = {
Linus Torvalds1da177e2005-04-16 15:20:36 -07001067 .owner = THIS_MODULE,
Pavel Emelyanov42a73802007-11-19 22:38:33 -08001068 .open = raw_v4_seq_open,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001069 .read = seq_read,
1070 .llseek = seq_lseek,
Pavel Emelyanovf51d5992008-01-14 05:35:57 -08001071 .release = seq_release_net,
Linus Torvalds1da177e2005-04-16 15:20:36 -07001072};
1073
Pavel Emelyanova308da12008-01-14 05:36:50 -08001074static __net_init int raw_init_net(struct net *net)
1075{
Gao fengd4beaa62013-02-18 01:34:54 +00001076 if (!proc_create("raw", S_IRUGO, net->proc_net, &raw_seq_fops))
Pavel Emelyanova308da12008-01-14 05:36:50 -08001077 return -ENOMEM;
1078
1079 return 0;
1080}
1081
1082static __net_exit void raw_exit_net(struct net *net)
1083{
Gao fengece31ff2013-02-18 01:34:56 +00001084 remove_proc_entry("raw", net->proc_net);
Pavel Emelyanova308da12008-01-14 05:36:50 -08001085}
1086
1087static __net_initdata struct pernet_operations raw_net_ops = {
1088 .init = raw_init_net,
1089 .exit = raw_exit_net,
1090};
1091
Linus Torvalds1da177e2005-04-16 15:20:36 -07001092int __init raw_proc_init(void)
1093{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001094 return register_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001095}
1096
1097void __init raw_proc_exit(void)
1098{
Pavel Emelyanova308da12008-01-14 05:36:50 -08001099 unregister_pernet_subsys(&raw_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001100}
1101#endif /* CONFIG_PROC_FS */