blob: c891235b4966cc4994d81acf39070d27c3060f13 [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +09002 * Linux NET3: IP/IP protocol decoder.
Linus Torvalds1da177e2005-04-16 15:20:36 -07003 *
Linus Torvalds1da177e2005-04-16 15:20:36 -07004 * Authors:
5 * Sam Lantinga (slouken@cs.ucdavis.edu) 02/01/95
6 *
7 * Fixes:
8 * Alan Cox : Merged and made usable non modular (its so tiny its silly as
9 * a module taking up 2 pages).
10 * Alan Cox : Fixed bug with 1.3.18 and IPIP not working (now needs to set skb->h.iph)
11 * to keep ip_forward happy.
12 * Alan Cox : More fixes for 1.3.21, and firewall fix. Maybe this will work soon 8).
13 * Kai Schulte : Fixed #defines for IP_FIREWALL->FIREWALL
14 * David Woodhouse : Perform some basic ICMP handling.
15 * IPIP Routing without decapsulation.
16 * Carlos Picoto : GRE over IP support
17 * Alexey Kuznetsov: Reworked. Really, now it is truncated version of ipv4/ip_gre.c.
18 * I do not want to merge them together.
19 *
20 * This program is free software; you can redistribute it and/or
21 * modify it under the terms of the GNU General Public License
22 * as published by the Free Software Foundation; either version
23 * 2 of the License, or (at your option) any later version.
24 *
25 */
26
27/* tunnel.c: an IP tunnel driver
28
29 The purpose of this driver is to provide an IP tunnel through
30 which you can tunnel network traffic transparently across subnets.
31
32 This was written by looking at Nick Holloway's dummy driver
33 Thanks for the great code!
34
35 -Sam Lantinga (slouken@cs.ucdavis.edu) 02/01/95
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +090036
Linus Torvalds1da177e2005-04-16 15:20:36 -070037 Minor tweaks:
38 Cleaned up the code a little and added some pre-1.3.0 tweaks.
39 dev->hard_header/hard_header_len changed to use no headers.
40 Comments/bracketing tweaked.
41 Made the tunnels use dev->name not tunnel: when error reporting.
42 Added tx_dropped stat
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +090043
Alan Cox113aa832008-10-13 19:01:08 -070044 -Alan Cox (alan@lxorguk.ukuu.org.uk) 21 March 95
Linus Torvalds1da177e2005-04-16 15:20:36 -070045
46 Reworked:
47 Changed to tunnel to destination gateway in addition to the
48 tunnel's pointopoint address
49 Almost completely rewritten
50 Note: There is currently no firewall or ICMP handling done.
51
52 -Sam Lantinga (slouken@cs.ucdavis.edu) 02/13/96
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +090053
Linus Torvalds1da177e2005-04-16 15:20:36 -070054*/
55
56/* Things I wish I had known when writing the tunnel driver:
57
58 When the tunnel_xmit() function is called, the skb contains the
59 packet to be sent (plus a great deal of extra info), and dev
60 contains the tunnel device that _we_ are.
61
62 When we are passed a packet, we are expected to fill in the
63 source address with our source IP address.
64
65 What is the proper way to allocate, copy and free a buffer?
66 After you allocate it, it is a "0 length" chunk of memory
67 starting at zero. If you want to add headers to the buffer
68 later, you'll have to call "skb_reserve(skb, amount)" with
69 the amount of memory you want reserved. Then, you call
70 "skb_put(skb, amount)" with the amount of space you want in
71 the buffer. skb_put() returns a pointer to the top (#0) of
72 that buffer. skb->len is set to the amount of space you have
73 "allocated" with skb_put(). You can then write up to skb->len
74 bytes to that buffer. If you need more, you can call skb_put()
75 again with the additional amount of space you need. You can
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +090076 find out how much more space you can allocate by calling
Linus Torvalds1da177e2005-04-16 15:20:36 -070077 "skb_tailroom(skb)".
78 Now, to add header space, call "skb_push(skb, header_len)".
79 This creates space at the beginning of the buffer and returns
80 a pointer to this new space. If later you need to strip a
81 header from a buffer, call "skb_pull(skb, header_len)".
82 skb_headroom() will return how much space is left at the top
83 of the buffer (before the main data). Remember, this headroom
84 space must be reserved before the skb_put() function is called.
85 */
86
87/*
88 This version of net/ipv4/ipip.c is cloned of net/ipv4/ip_gre.c
89
90 For comments look at net/ipv4/ip_gre.c --ANK
91 */
92
YOSHIFUJI Hideakie905a9e2007-02-09 23:24:47 +090093
Randy Dunlap4fc268d2006-01-11 12:17:47 -080094#include <linux/capability.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070095#include <linux/module.h>
96#include <linux/types.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -070097#include <linux/kernel.h>
Tejun Heo5a0e3ad2010-03-24 17:04:11 +090098#include <linux/slab.h>
Linus Torvalds7c0f6ba2016-12-24 11:46:01 -080099#include <linux/uaccess.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100#include <linux/skbuff.h>
101#include <linux/netdevice.h>
102#include <linux/in.h>
103#include <linux/tcp.h>
104#include <linux/udp.h>
105#include <linux/if_arp.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700106#include <linux/init.h>
107#include <linux/netfilter_ipv4.h>
Kris Katterjohn46f25df2006-01-05 16:35:42 -0800108#include <linux/if_ether.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700109
110#include <net/sock.h>
111#include <net/ip.h>
112#include <net/icmp.h>
Pravin B Shelarc5441932013-03-25 14:49:35 +0000113#include <net/ip_tunnels.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700114#include <net/inet_ecn.h>
115#include <net/xfrm.h>
Pavel Emelyanov10dc4c72008-04-16 01:03:13 -0700116#include <net/net_namespace.h>
117#include <net/netns/generic.h>
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700118#include <net/dst_metadata.h>
Linus Torvalds1da177e2005-04-16 15:20:36 -0700119
stephen hemmingereccc1bb2012-09-25 11:02:48 +0000120static bool log_ecn_error = true;
121module_param(log_ecn_error, bool, 0644);
122MODULE_PARM_DESC(log_ecn_error, "Log packets received with corrupted ECN");
123
Alexey Dobriyanc7d03a02016-11-17 04:58:21 +0300124static unsigned int ipip_net_id __read_mostly;
Pavel Emelyanov10dc4c72008-04-16 01:03:13 -0700125
Eric Dumazet3c97af92010-09-27 00:35:50 +0000126static int ipip_tunnel_init(struct net_device *dev);
Nicolas Dichtel09746582012-11-09 06:09:59 +0000127static struct rtnl_link_ops ipip_link_ops __read_mostly;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700128
Herbert Xud2acc342006-03-28 01:12:13 -0800129static int ipip_err(struct sk_buff *skb, u32 info)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130{
Xin Longf3594f02017-10-26 19:19:56 +0800131 /* All the routers (except for Linux) return only
132 * 8 bytes of packet payload. It means, that precise relaying of
133 * ICMP in the real Internet is absolutely infeasible.
134 */
Pravin B Shelarfd581562013-03-25 14:49:41 +0000135 struct net *net = dev_net(skb->dev);
136 struct ip_tunnel_net *itn = net_generic(net, ipip_net_id);
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000137 const struct iphdr *iph = (const struct iphdr *)skb->data;
Pravin B Shelarfd581562013-03-25 14:49:41 +0000138 const int type = icmp_hdr(skb)->type;
139 const int code = icmp_hdr(skb)->code;
Xin Longf3594f02017-10-26 19:19:56 +0800140 struct ip_tunnel *t;
141 int err = 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700142
Xin Longf3594f02017-10-26 19:19:56 +0800143 switch (type) {
144 case ICMP_DEST_UNREACH:
145 switch (code) {
146 case ICMP_SR_FAILED:
147 /* Impossible event. */
148 goto out;
149 default:
150 /* All others are translated to HOST_UNREACH.
151 * rfc2003 contains "deep thoughts" about NET_UNREACH,
152 * I believe they are just ether pollution. --ANK
153 */
154 break;
155 }
156 break;
157
158 case ICMP_TIME_EXCEEDED:
159 if (code != ICMP_EXC_TTL)
160 goto out;
161 break;
162
163 case ICMP_REDIRECT:
164 break;
165
166 default:
167 goto out;
168 }
169
Pravin B Shelarfd581562013-03-25 14:49:41 +0000170 t = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY,
171 iph->daddr, iph->saddr, 0);
Xin Longf3594f02017-10-26 19:19:56 +0800172 if (!t) {
173 err = -ENOENT;
David S. Miller36393392012-06-14 22:21:46 -0700174 goto out;
Xin Longf3594f02017-10-26 19:19:56 +0800175 }
David S. Miller36393392012-06-14 22:21:46 -0700176
177 if (type == ICMP_DEST_UNREACH && code == ICMP_FRAG_NEEDED) {
Xin Longf3594f02017-10-26 19:19:56 +0800178 ipv4_update_pmtu(skb, net, info, t->parms.link, 0,
179 iph->protocol, 0);
David S. Miller36393392012-06-14 22:21:46 -0700180 goto out;
181 }
182
David S. Miller55be7a92012-07-11 21:27:49 -0700183 if (type == ICMP_REDIRECT) {
Xin Longf3594f02017-10-26 19:19:56 +0800184 ipv4_redirect(skb, net, t->parms.link, 0, iph->protocol, 0);
David S. Miller55be7a92012-07-11 21:27:49 -0700185 goto out;
186 }
187
Xin Longf3594f02017-10-26 19:19:56 +0800188 if (t->parms.iph.daddr == 0) {
189 err = -ENOENT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700190 goto out;
Xin Longf3594f02017-10-26 19:19:56 +0800191 }
Herbert Xud2acc342006-03-28 01:12:13 -0800192
Linus Torvalds1da177e2005-04-16 15:20:36 -0700193 if (t->parms.iph.ttl == 0 && type == ICMP_TIME_EXCEEDED)
194 goto out;
195
Wei Yongjun26d94b42009-02-24 23:36:47 -0800196 if (time_before(jiffies, t->err_time + IPTUNNEL_ERR_TIMEO))
Linus Torvalds1da177e2005-04-16 15:20:36 -0700197 t->err_count++;
198 else
199 t->err_count = 1;
200 t->err_time = jiffies;
stephen hemmingerb0558ef2012-09-24 18:12:25 +0000201
Pravin B Shelarfd581562013-03-25 14:49:41 +0000202out:
Herbert Xud2acc342006-03-28 01:12:13 -0800203 return err;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700204}
205
Simon Horman1b69e7e2016-07-07 07:56:14 +0200206static const struct tnl_ptk_info ipip_tpi = {
Pravin B Shelarfd581562013-03-25 14:49:41 +0000207 /* no tunnel info required for ipip. */
208 .proto = htons(ETH_P_IP),
209};
210
Simon Horman1b69e7e2016-07-07 07:56:14 +0200211#if IS_ENABLED(CONFIG_MPLS)
212static const struct tnl_ptk_info mplsip_tpi = {
213 /* no tunnel info required for mplsip. */
214 .proto = htons(ETH_P_MPLS_UC),
215};
216#endif
217
218static int ipip_tunnel_rcv(struct sk_buff *skb, u8 ipproto)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700219{
Pravin B Shelarfd581562013-03-25 14:49:41 +0000220 struct net *net = dev_net(skb->dev);
221 struct ip_tunnel_net *itn = net_generic(net, ipip_net_id);
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700222 struct metadata_dst *tun_dst = NULL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700223 struct ip_tunnel *tunnel;
Pravin B Shelar3d7b46c2013-06-17 17:50:02 -0700224 const struct iphdr *iph;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700225
Pravin B Shelar3d7b46c2013-06-17 17:50:02 -0700226 iph = ip_hdr(skb);
Pravin B Shelarfd581562013-03-25 14:49:41 +0000227 tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY,
228 iph->saddr, iph->daddr, 0);
229 if (tunnel) {
Simon Horman1b69e7e2016-07-07 07:56:14 +0200230 const struct tnl_ptk_info *tpi;
231
232 if (tunnel->parms.iph.protocol != ipproto &&
233 tunnel->parms.iph.protocol != 0)
234 goto drop;
235
stephen hemmingereccc1bb2012-09-25 11:02:48 +0000236 if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb))
237 goto drop;
Simon Horman1b69e7e2016-07-07 07:56:14 +0200238#if IS_ENABLED(CONFIG_MPLS)
239 if (ipproto == IPPROTO_MPLS)
240 tpi = &mplsip_tpi;
241 else
242#endif
243 tpi = &ipip_tpi;
244 if (iptunnel_pull_header(skb, 0, tpi->proto, false))
Li Hongjun737e8282013-08-28 11:54:50 +0200245 goto drop;
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700246 if (tunnel->collect_md) {
247 tun_dst = ip_tun_rx_dst(skb, 0, 0, 0);
248 if (!tun_dst)
249 return 0;
250 }
251 return ip_tunnel_rcv(tunnel, skb, tpi, tun_dst, log_ecn_error);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700252 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700253
Linus Torvalds1da177e2005-04-16 15:20:36 -0700254 return -1;
stephen hemmingereccc1bb2012-09-25 11:02:48 +0000255
256drop:
257 kfree_skb(skb);
258 return 0;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700259}
260
Simon Horman1b69e7e2016-07-07 07:56:14 +0200261static int ipip_rcv(struct sk_buff *skb)
262{
263 return ipip_tunnel_rcv(skb, IPPROTO_IPIP);
264}
265
266#if IS_ENABLED(CONFIG_MPLS)
267static int mplsip_rcv(struct sk_buff *skb)
268{
269 return ipip_tunnel_rcv(skb, IPPROTO_MPLS);
270}
271#endif
272
Linus Torvalds1da177e2005-04-16 15:20:36 -0700273/*
274 * This function assumes it is being called from dev_queue_xmit()
275 * and that skb is filled properly by that function.
276 */
Simon Horman1b69e7e2016-07-07 07:56:14 +0200277static netdev_tx_t ipip_tunnel_xmit(struct sk_buff *skb,
278 struct net_device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700279{
Patrick McHardy2941a482006-01-08 22:05:26 -0800280 struct ip_tunnel *tunnel = netdev_priv(dev);
Eric Dumazetb71d1d42011-04-22 04:53:02 +0000281 const struct iphdr *tiph = &tunnel->parms.iph;
Simon Horman1b69e7e2016-07-07 07:56:14 +0200282 u8 ipproto;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700283
Simon Horman1b69e7e2016-07-07 07:56:14 +0200284 switch (skb->protocol) {
285 case htons(ETH_P_IP):
286 ipproto = IPPROTO_IPIP;
287 break;
288#if IS_ENABLED(CONFIG_MPLS)
289 case htons(ETH_P_MPLS_UC):
290 ipproto = IPPROTO_MPLS;
291 break;
292#endif
293 default:
294 goto tx_error;
295 }
296
297 if (tiph->protocol != ipproto && tiph->protocol != 0)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700298 goto tx_error;
Eric Dumazetcef401d2013-01-25 20:34:37 +0000299
Tom Herbert7e133182016-05-18 09:06:10 -0700300 if (iptunnel_handle_offloads(skb, SKB_GSO_IPXIP4))
Alexander Duyckaed069d2016-04-14 15:33:37 -0400301 goto tx_error;
Pravin B Shelar4f3ed922013-03-08 15:12:52 +0000302
Simon Horman1b69e7e2016-07-07 07:56:14 +0200303 skb_set_inner_ipproto(skb, ipproto);
Tom Herbert077c5a02014-09-29 20:22:31 -0700304
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700305 if (tunnel->collect_md)
306 ip_md_tunnel_xmit(skb, dev, ipproto);
307 else
308 ip_tunnel_xmit(skb, dev, tiph, ipproto);
Patrick McHardy6ed10652009-06-23 06:03:08 +0000309 return NETDEV_TX_OK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700310
Linus Torvalds1da177e2005-04-16 15:20:36 -0700311tx_error:
Eric Dumazet3acfa1e2014-01-18 18:27:49 -0800312 kfree_skb(skb);
Alexander Duyckaed069d2016-04-14 15:33:37 -0400313
Eric Dumazetcb32f512013-10-19 11:42:57 -0700314 dev->stats.tx_errors++;
Patrick McHardy6ed10652009-06-23 06:03:08 +0000315 return NETDEV_TX_OK;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700316}
317
Simon Horman1b69e7e2016-07-07 07:56:14 +0200318static bool ipip_tunnel_ioctl_verify_protocol(u8 ipproto)
319{
320 switch (ipproto) {
321 case 0:
322 case IPPROTO_IPIP:
323#if IS_ENABLED(CONFIG_MPLS)
324 case IPPROTO_MPLS:
325#endif
326 return true;
327 }
328
329 return false;
330}
331
Linus Torvalds1da177e2005-04-16 15:20:36 -0700332static int
Pravin B Shelarfd581562013-03-25 14:49:41 +0000333ipip_tunnel_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700334{
335 int err = 0;
336 struct ip_tunnel_parm p;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700337
Pravin B Shelarfd581562013-03-25 14:49:41 +0000338 if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
339 return -EFAULT;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700340
Cong Wang3b7b5142013-07-02 14:49:34 +0800341 if (cmd == SIOCADDTUNNEL || cmd == SIOCCHGTUNNEL) {
Simon Horman1b69e7e2016-07-07 07:56:14 +0200342 if (p.iph.version != 4 ||
343 !ipip_tunnel_ioctl_verify_protocol(p.iph.protocol) ||
Cong Wang3b7b5142013-07-02 14:49:34 +0800344 p.iph.ihl != 5 || (p.iph.frag_off&htons(~IP_DF)))
345 return -EINVAL;
346 }
347
Eric Dumazet252a8fb2015-05-15 08:58:45 -0700348 p.i_key = p.o_key = 0;
349 p.i_flags = p.o_flags = 0;
Pravin B Shelarfd581562013-03-25 14:49:41 +0000350 err = ip_tunnel_ioctl(dev, &p, cmd);
351 if (err)
352 return err;
353
354 if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p)))
355 return -EFAULT;
356
Linus Torvalds1da177e2005-04-16 15:20:36 -0700357 return 0;
358}
359
Stephen Hemminger23a12b12008-11-20 20:33:21 -0800360static const struct net_device_ops ipip_netdev_ops = {
Pravin B Shelarfd581562013-03-25 14:49:41 +0000361 .ndo_init = ipip_tunnel_init,
362 .ndo_uninit = ip_tunnel_uninit,
Stephen Hemminger23a12b12008-11-20 20:33:21 -0800363 .ndo_start_xmit = ipip_tunnel_xmit,
364 .ndo_do_ioctl = ipip_tunnel_ioctl,
Pravin B Shelarfd581562013-03-25 14:49:41 +0000365 .ndo_change_mtu = ip_tunnel_change_mtu,
366 .ndo_get_stats64 = ip_tunnel_get_stats64,
Nicolas Dichtel1e995842015-04-02 17:07:02 +0200367 .ndo_get_iflink = ip_tunnel_get_iflink,
Stephen Hemminger23a12b12008-11-20 20:33:21 -0800368};
369
Eric Dumazetc3b89fb2012-11-08 09:59:52 +0000370#define IPIP_FEATURES (NETIF_F_SG | \
371 NETIF_F_FRAGLIST | \
372 NETIF_F_HIGHDMA | \
Eric Dumazetcb32f512013-10-19 11:42:57 -0700373 NETIF_F_GSO_SOFTWARE | \
Eric Dumazetc3b89fb2012-11-08 09:59:52 +0000374 NETIF_F_HW_CSUM)
375
Linus Torvalds1da177e2005-04-16 15:20:36 -0700376static void ipip_tunnel_setup(struct net_device *dev)
377{
Stephen Hemminger23a12b12008-11-20 20:33:21 -0800378 dev->netdev_ops = &ipip_netdev_ops;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700379
380 dev->type = ARPHRD_TUNNEL;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700381 dev->flags = IFF_NOARP;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382 dev->addr_len = 4;
Eric Dumazet153f0942010-09-28 00:17:17 +0000383 dev->features |= NETIF_F_LLTX;
Eric Dumazet02875872014-10-05 18:38:35 -0700384 netif_keep_dst(dev);
Eric Dumazetc3b89fb2012-11-08 09:59:52 +0000385
386 dev->features |= IPIP_FEATURES;
387 dev->hw_features |= IPIP_FEATURES;
Pravin B Shelarfd581562013-03-25 14:49:41 +0000388 ip_tunnel_setup(dev, ipip_net_id);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389}
390
Eric Dumazet3c97af92010-09-27 00:35:50 +0000391static int ipip_tunnel_init(struct net_device *dev)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700392{
Stephen Hemminger23a12b12008-11-20 20:33:21 -0800393 struct ip_tunnel *tunnel = netdev_priv(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700394
Linus Torvalds1da177e2005-04-16 15:20:36 -0700395 memcpy(dev->dev_addr, &tunnel->parms.iph.saddr, 4);
396 memcpy(dev->broadcast, &tunnel->parms.iph.daddr, 4);
397
Tom Herbert473ab822014-09-17 12:26:00 -0700398 tunnel->tun_hlen = 0;
399 tunnel->hlen = tunnel->tun_hlen + tunnel->encap_hlen;
Pravin B Shelarfd581562013-03-25 14:49:41 +0000400 return ip_tunnel_init(dev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700401}
402
Matthias Schiffera8b8a8892017-06-25 23:56:01 +0200403static int ipip_tunnel_validate(struct nlattr *tb[], struct nlattr *data[],
404 struct netlink_ext_ack *extack)
Simon Horman1b69e7e2016-07-07 07:56:14 +0200405{
406 u8 proto;
407
408 if (!data || !data[IFLA_IPTUN_PROTO])
409 return 0;
410
411 proto = nla_get_u8(data[IFLA_IPTUN_PROTO]);
412 if (proto != IPPROTO_IPIP && proto != IPPROTO_MPLS && proto != 0)
413 return -EINVAL;
414
415 return 0;
416}
417
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000418static void ipip_netlink_parms(struct nlattr *data[],
Craig Gallek9830ad42017-04-19 12:30:54 -0400419 struct ip_tunnel_parm *parms, bool *collect_md,
420 __u32 *fwmark)
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000421{
422 memset(parms, 0, sizeof(*parms));
423
424 parms->iph.version = 4;
425 parms->iph.protocol = IPPROTO_IPIP;
426 parms->iph.ihl = 5;
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700427 *collect_md = false;
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000428
429 if (!data)
430 return;
431
432 if (data[IFLA_IPTUN_LINK])
433 parms->link = nla_get_u32(data[IFLA_IPTUN_LINK]);
434
435 if (data[IFLA_IPTUN_LOCAL])
Jiri Benc67b61f62015-03-29 16:59:26 +0200436 parms->iph.saddr = nla_get_in_addr(data[IFLA_IPTUN_LOCAL]);
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000437
438 if (data[IFLA_IPTUN_REMOTE])
Jiri Benc67b61f62015-03-29 16:59:26 +0200439 parms->iph.daddr = nla_get_in_addr(data[IFLA_IPTUN_REMOTE]);
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000440
441 if (data[IFLA_IPTUN_TTL]) {
442 parms->iph.ttl = nla_get_u8(data[IFLA_IPTUN_TTL]);
443 if (parms->iph.ttl)
444 parms->iph.frag_off = htons(IP_DF);
445 }
446
447 if (data[IFLA_IPTUN_TOS])
448 parms->iph.tos = nla_get_u8(data[IFLA_IPTUN_TOS]);
449
Simon Horman1b69e7e2016-07-07 07:56:14 +0200450 if (data[IFLA_IPTUN_PROTO])
451 parms->iph.protocol = nla_get_u8(data[IFLA_IPTUN_PROTO]);
452
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000453 if (!data[IFLA_IPTUN_PMTUDISC] || nla_get_u8(data[IFLA_IPTUN_PMTUDISC]))
454 parms->iph.frag_off = htons(IP_DF);
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700455
456 if (data[IFLA_IPTUN_COLLECT_METADATA])
457 *collect_md = true;
Craig Gallek9830ad42017-04-19 12:30:54 -0400458
459 if (data[IFLA_IPTUN_FWMARK])
460 *fwmark = nla_get_u32(data[IFLA_IPTUN_FWMARK]);
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000461}
462
Tom Herbert473ab822014-09-17 12:26:00 -0700463/* This function returns true when ENCAP attributes are present in the nl msg */
464static bool ipip_netlink_encap_parms(struct nlattr *data[],
465 struct ip_tunnel_encap *ipencap)
466{
467 bool ret = false;
468
469 memset(ipencap, 0, sizeof(*ipencap));
470
471 if (!data)
472 return ret;
473
474 if (data[IFLA_IPTUN_ENCAP_TYPE]) {
475 ret = true;
476 ipencap->type = nla_get_u16(data[IFLA_IPTUN_ENCAP_TYPE]);
477 }
478
479 if (data[IFLA_IPTUN_ENCAP_FLAGS]) {
480 ret = true;
481 ipencap->flags = nla_get_u16(data[IFLA_IPTUN_ENCAP_FLAGS]);
482 }
483
484 if (data[IFLA_IPTUN_ENCAP_SPORT]) {
485 ret = true;
Sabrina Dubroca3e97fa72015-02-06 17:22:22 +0100486 ipencap->sport = nla_get_be16(data[IFLA_IPTUN_ENCAP_SPORT]);
Tom Herbert473ab822014-09-17 12:26:00 -0700487 }
488
489 if (data[IFLA_IPTUN_ENCAP_DPORT]) {
490 ret = true;
Sabrina Dubroca3e97fa72015-02-06 17:22:22 +0100491 ipencap->dport = nla_get_be16(data[IFLA_IPTUN_ENCAP_DPORT]);
Tom Herbert473ab822014-09-17 12:26:00 -0700492 }
493
494 return ret;
495}
496
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000497static int ipip_newlink(struct net *src_net, struct net_device *dev,
Matthias Schiffer7a3f4a12017-06-25 23:55:59 +0200498 struct nlattr *tb[], struct nlattr *data[],
499 struct netlink_ext_ack *extack)
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000500{
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700501 struct ip_tunnel *t = netdev_priv(dev);
Pravin B Shelarfd581562013-03-25 14:49:41 +0000502 struct ip_tunnel_parm p;
Tom Herbert473ab822014-09-17 12:26:00 -0700503 struct ip_tunnel_encap ipencap;
Craig Gallek9830ad42017-04-19 12:30:54 -0400504 __u32 fwmark = 0;
Tom Herbert473ab822014-09-17 12:26:00 -0700505
506 if (ipip_netlink_encap_parms(data, &ipencap)) {
Tom Herbert473ab822014-09-17 12:26:00 -0700507 int err = ip_tunnel_encap_setup(t, &ipencap);
508
509 if (err < 0)
510 return err;
511 }
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000512
Craig Gallek9830ad42017-04-19 12:30:54 -0400513 ipip_netlink_parms(data, &p, &t->collect_md, &fwmark);
514 return ip_tunnel_newlink(dev, tb, &p, fwmark);
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000515}
516
517static int ipip_changelink(struct net_device *dev, struct nlattr *tb[],
Matthias Schifferad744b22017-06-25 23:56:00 +0200518 struct nlattr *data[],
519 struct netlink_ext_ack *extack)
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000520{
Craig Gallek9830ad42017-04-19 12:30:54 -0400521 struct ip_tunnel *t = netdev_priv(dev);
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000522 struct ip_tunnel_parm p;
Tom Herbert473ab822014-09-17 12:26:00 -0700523 struct ip_tunnel_encap ipencap;
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700524 bool collect_md;
Craig Gallek9830ad42017-04-19 12:30:54 -0400525 __u32 fwmark = t->fwmark;
Tom Herbert473ab822014-09-17 12:26:00 -0700526
527 if (ipip_netlink_encap_parms(data, &ipencap)) {
Tom Herbert473ab822014-09-17 12:26:00 -0700528 int err = ip_tunnel_encap_setup(t, &ipencap);
529
530 if (err < 0)
531 return err;
532 }
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000533
Craig Gallek9830ad42017-04-19 12:30:54 -0400534 ipip_netlink_parms(data, &p, &collect_md, &fwmark);
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700535 if (collect_md)
536 return -EINVAL;
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000537
538 if (((dev->flags & IFF_POINTOPOINT) && !p.iph.daddr) ||
539 (!(dev->flags & IFF_POINTOPOINT) && p.iph.daddr))
540 return -EINVAL;
541
Craig Gallek9830ad42017-04-19 12:30:54 -0400542 return ip_tunnel_changelink(dev, tb, &p, fwmark);
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000543}
544
Nicolas Dichtel09746582012-11-09 06:09:59 +0000545static size_t ipip_get_size(const struct net_device *dev)
546{
547 return
548 /* IFLA_IPTUN_LINK */
549 nla_total_size(4) +
550 /* IFLA_IPTUN_LOCAL */
551 nla_total_size(4) +
552 /* IFLA_IPTUN_REMOTE */
553 nla_total_size(4) +
554 /* IFLA_IPTUN_TTL */
555 nla_total_size(1) +
556 /* IFLA_IPTUN_TOS */
557 nla_total_size(1) +
Simon Horman1b69e7e2016-07-07 07:56:14 +0200558 /* IFLA_IPTUN_PROTO */
559 nla_total_size(1) +
Nicolas Dichtelbefe2aa2012-11-14 05:14:02 +0000560 /* IFLA_IPTUN_PMTUDISC */
561 nla_total_size(1) +
Tom Herbert473ab822014-09-17 12:26:00 -0700562 /* IFLA_IPTUN_ENCAP_TYPE */
563 nla_total_size(2) +
564 /* IFLA_IPTUN_ENCAP_FLAGS */
565 nla_total_size(2) +
566 /* IFLA_IPTUN_ENCAP_SPORT */
567 nla_total_size(2) +
568 /* IFLA_IPTUN_ENCAP_DPORT */
569 nla_total_size(2) +
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700570 /* IFLA_IPTUN_COLLECT_METADATA */
571 nla_total_size(0) +
Craig Gallek9830ad42017-04-19 12:30:54 -0400572 /* IFLA_IPTUN_FWMARK */
573 nla_total_size(4) +
Nicolas Dichtel09746582012-11-09 06:09:59 +0000574 0;
575}
576
577static int ipip_fill_info(struct sk_buff *skb, const struct net_device *dev)
578{
579 struct ip_tunnel *tunnel = netdev_priv(dev);
580 struct ip_tunnel_parm *parm = &tunnel->parms;
581
582 if (nla_put_u32(skb, IFLA_IPTUN_LINK, parm->link) ||
Jiri Benc930345e2015-03-29 16:59:25 +0200583 nla_put_in_addr(skb, IFLA_IPTUN_LOCAL, parm->iph.saddr) ||
584 nla_put_in_addr(skb, IFLA_IPTUN_REMOTE, parm->iph.daddr) ||
Nicolas Dichtel09746582012-11-09 06:09:59 +0000585 nla_put_u8(skb, IFLA_IPTUN_TTL, parm->iph.ttl) ||
Nicolas Dichtelbefe2aa2012-11-14 05:14:02 +0000586 nla_put_u8(skb, IFLA_IPTUN_TOS, parm->iph.tos) ||
Simon Horman1b69e7e2016-07-07 07:56:14 +0200587 nla_put_u8(skb, IFLA_IPTUN_PROTO, parm->iph.protocol) ||
Nicolas Dichtelbefe2aa2012-11-14 05:14:02 +0000588 nla_put_u8(skb, IFLA_IPTUN_PMTUDISC,
Craig Gallek9830ad42017-04-19 12:30:54 -0400589 !!(parm->iph.frag_off & htons(IP_DF))) ||
590 nla_put_u32(skb, IFLA_IPTUN_FWMARK, tunnel->fwmark))
Nicolas Dichtel09746582012-11-09 06:09:59 +0000591 goto nla_put_failure;
Tom Herbert473ab822014-09-17 12:26:00 -0700592
593 if (nla_put_u16(skb, IFLA_IPTUN_ENCAP_TYPE,
594 tunnel->encap.type) ||
Sabrina Dubroca3e97fa72015-02-06 17:22:22 +0100595 nla_put_be16(skb, IFLA_IPTUN_ENCAP_SPORT,
596 tunnel->encap.sport) ||
597 nla_put_be16(skb, IFLA_IPTUN_ENCAP_DPORT,
598 tunnel->encap.dport) ||
Tom Herbert473ab822014-09-17 12:26:00 -0700599 nla_put_u16(skb, IFLA_IPTUN_ENCAP_FLAGS,
Tom Herberte1b2cb62014-11-05 16:49:38 -0800600 tunnel->encap.flags))
Tom Herbert473ab822014-09-17 12:26:00 -0700601 goto nla_put_failure;
602
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700603 if (tunnel->collect_md)
604 if (nla_put_flag(skb, IFLA_IPTUN_COLLECT_METADATA))
605 goto nla_put_failure;
Nicolas Dichtel09746582012-11-09 06:09:59 +0000606 return 0;
607
608nla_put_failure:
609 return -EMSGSIZE;
610}
611
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000612static const struct nla_policy ipip_policy[IFLA_IPTUN_MAX + 1] = {
613 [IFLA_IPTUN_LINK] = { .type = NLA_U32 },
614 [IFLA_IPTUN_LOCAL] = { .type = NLA_U32 },
615 [IFLA_IPTUN_REMOTE] = { .type = NLA_U32 },
616 [IFLA_IPTUN_TTL] = { .type = NLA_U8 },
617 [IFLA_IPTUN_TOS] = { .type = NLA_U8 },
Simon Horman1b69e7e2016-07-07 07:56:14 +0200618 [IFLA_IPTUN_PROTO] = { .type = NLA_U8 },
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000619 [IFLA_IPTUN_PMTUDISC] = { .type = NLA_U8 },
Tom Herbert473ab822014-09-17 12:26:00 -0700620 [IFLA_IPTUN_ENCAP_TYPE] = { .type = NLA_U16 },
621 [IFLA_IPTUN_ENCAP_FLAGS] = { .type = NLA_U16 },
622 [IFLA_IPTUN_ENCAP_SPORT] = { .type = NLA_U16 },
623 [IFLA_IPTUN_ENCAP_DPORT] = { .type = NLA_U16 },
Alexei Starovoitovcfc73812016-09-15 13:00:29 -0700624 [IFLA_IPTUN_COLLECT_METADATA] = { .type = NLA_FLAG },
Craig Gallek9830ad42017-04-19 12:30:54 -0400625 [IFLA_IPTUN_FWMARK] = { .type = NLA_U32 },
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000626};
627
Nicolas Dichtel09746582012-11-09 06:09:59 +0000628static struct rtnl_link_ops ipip_link_ops __read_mostly = {
629 .kind = "ipip",
630 .maxtype = IFLA_IPTUN_MAX,
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000631 .policy = ipip_policy,
Nicolas Dichtel09746582012-11-09 06:09:59 +0000632 .priv_size = sizeof(struct ip_tunnel),
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000633 .setup = ipip_tunnel_setup,
Simon Horman1b69e7e2016-07-07 07:56:14 +0200634 .validate = ipip_tunnel_validate,
Nicolas Dichtelbe42da02012-11-14 05:14:03 +0000635 .newlink = ipip_newlink,
636 .changelink = ipip_changelink,
Pravin B Shelarfd581562013-03-25 14:49:41 +0000637 .dellink = ip_tunnel_dellink,
Nicolas Dichtel09746582012-11-09 06:09:59 +0000638 .get_size = ipip_get_size,
639 .fill_info = ipip_fill_info,
Nicolas Dichtel1728d4f2015-01-15 15:11:17 +0100640 .get_link_net = ip_tunnel_get_link_net,
Nicolas Dichtel09746582012-11-09 06:09:59 +0000641};
642
Eric Dumazet6dcd8142010-08-30 07:04:14 +0000643static struct xfrm_tunnel ipip_handler __read_mostly = {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700644 .handler = ipip_rcv,
645 .err_handler = ipip_err,
Herbert Xud2acc342006-03-28 01:12:13 -0800646 .priority = 1,
Linus Torvalds1da177e2005-04-16 15:20:36 -0700647};
648
Simon Horman1b69e7e2016-07-07 07:56:14 +0200649#if IS_ENABLED(CONFIG_MPLS)
650static struct xfrm_tunnel mplsip_handler __read_mostly = {
651 .handler = mplsip_rcv,
652 .err_handler = ipip_err,
653 .priority = 1,
654};
655#endif
656
Alexey Dobriyan2c8c1e72010-01-17 03:35:32 +0000657static int __net_init ipip_init_net(struct net *net)
Pavel Emelyanov10dc4c72008-04-16 01:03:13 -0700658{
Pravin B Shelarfd581562013-03-25 14:49:41 +0000659 return ip_tunnel_init_net(net, ipip_net_id, &ipip_link_ops, "tunl0");
Pavel Emelyanov10dc4c72008-04-16 01:03:13 -0700660}
661
Eric Dumazet64bc1782017-09-19 16:27:09 -0700662static void __net_exit ipip_exit_batch_net(struct list_head *list_net)
Pavel Emelyanov10dc4c72008-04-16 01:03:13 -0700663{
Eric Dumazet64bc1782017-09-19 16:27:09 -0700664 ip_tunnel_delete_nets(list_net, ipip_net_id, &ipip_link_ops);
Pavel Emelyanov10dc4c72008-04-16 01:03:13 -0700665}
666
667static struct pernet_operations ipip_net_ops = {
668 .init = ipip_init_net,
Eric Dumazet64bc1782017-09-19 16:27:09 -0700669 .exit_batch = ipip_exit_batch_net,
Eric W. Biederman86de8a62009-11-29 15:46:14 +0000670 .id = &ipip_net_id,
Pravin B Shelarfd581562013-03-25 14:49:41 +0000671 .size = sizeof(struct ip_tunnel_net),
Pavel Emelyanov10dc4c72008-04-16 01:03:13 -0700672};
673
Linus Torvalds1da177e2005-04-16 15:20:36 -0700674static int __init ipip_init(void)
675{
676 int err;
677
Simon Horman1b69e7e2016-07-07 07:56:14 +0200678 pr_info("ipip: IPv4 and MPLS over IPv4 tunneling driver\n");
Linus Torvalds1da177e2005-04-16 15:20:36 -0700679
Eric W. Biederman86de8a62009-11-29 15:46:14 +0000680 err = register_pernet_device(&ipip_net_ops);
Alexey Dobriyand5aa4072010-02-16 09:05:04 +0000681 if (err < 0)
682 return err;
683 err = xfrm4_tunnel_register(&ipip_handler, AF_INET);
684 if (err < 0) {
Joe Perches058bd4d2012-03-11 18:36:11 +0000685 pr_info("%s: can't register tunnel\n", __func__);
Simon Horman1b69e7e2016-07-07 07:56:14 +0200686 goto xfrm_tunnel_ipip_failed;
Alexey Dobriyand5aa4072010-02-16 09:05:04 +0000687 }
Simon Horman1b69e7e2016-07-07 07:56:14 +0200688#if IS_ENABLED(CONFIG_MPLS)
689 err = xfrm4_tunnel_register(&mplsip_handler, AF_MPLS);
690 if (err < 0) {
691 pr_info("%s: can't register tunnel\n", __func__);
692 goto xfrm_tunnel_mplsip_failed;
693 }
694#endif
Nicolas Dichtel09746582012-11-09 06:09:59 +0000695 err = rtnl_link_register(&ipip_link_ops);
696 if (err < 0)
697 goto rtnl_link_failed;
698
699out:
Linus Torvalds1da177e2005-04-16 15:20:36 -0700700 return err;
Nicolas Dichtel09746582012-11-09 06:09:59 +0000701
702rtnl_link_failed:
Simon Horman1b69e7e2016-07-07 07:56:14 +0200703#if IS_ENABLED(CONFIG_MPLS)
704 xfrm4_tunnel_deregister(&mplsip_handler, AF_INET);
705xfrm_tunnel_mplsip_failed:
706
707#endif
Nicolas Dichtel09746582012-11-09 06:09:59 +0000708 xfrm4_tunnel_deregister(&ipip_handler, AF_INET);
Simon Horman1b69e7e2016-07-07 07:56:14 +0200709xfrm_tunnel_ipip_failed:
Nicolas Dichtel09746582012-11-09 06:09:59 +0000710 unregister_pernet_device(&ipip_net_ops);
711 goto out;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700712}
713
714static void __exit ipip_fini(void)
715{
Nicolas Dichtel09746582012-11-09 06:09:59 +0000716 rtnl_link_unregister(&ipip_link_ops);
Kazunori MIYAZAWAc0d56402007-02-13 12:54:47 -0800717 if (xfrm4_tunnel_deregister(&ipip_handler, AF_INET))
Joe Perches058bd4d2012-03-11 18:36:11 +0000718 pr_info("%s: can't deregister tunnel\n", __func__);
Simon Horman1b69e7e2016-07-07 07:56:14 +0200719#if IS_ENABLED(CONFIG_MPLS)
720 if (xfrm4_tunnel_deregister(&mplsip_handler, AF_MPLS))
721 pr_info("%s: can't deregister tunnel\n", __func__);
722#endif
Eric W. Biederman86de8a62009-11-29 15:46:14 +0000723 unregister_pernet_device(&ipip_net_ops);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700724}
725
726module_init(ipip_init);
727module_exit(ipip_fini);
728MODULE_LICENSE("GPL");
Tom Gundersenf98f89a2014-05-15 23:21:30 +0200729MODULE_ALIAS_RTNL_LINK("ipip");
Vasiliy Kulikov8909c9a2011-03-02 00:33:13 +0300730MODULE_ALIAS_NETDEV("tunl0");