1/*
2 * Copyright (c) 2013 Nicira, Inc.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of version 2 of the GNU General Public
6 * License as published by the Free Software Foundation.
7 *
8 * This program is distributed in the hope that it will be useful, but
9 * WITHOUT ANY WARRANTY; without even the implied warranty of
10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
11 * General Public License for more details.
12 *
13 * You should have received a copy of the GNU General Public License
14 * along with this program; if not, write to the Free Software
15 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
16 * 02110-1301, USA
17 */
18
19#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
20
21#include <linux/types.h>
22#include <linux/kernel.h>
23#include <linux/skbuff.h>
24#include <linux/netdevice.h>
25#include <linux/in.h>
26#include <linux/if_arp.h>
27#include <linux/mroute.h>
28#include <linux/init.h>
29#include <linux/in6.h>
30#include <linux/inetdevice.h>
31#include <linux/netfilter_ipv4.h>
32#include <linux/etherdevice.h>
33#include <linux/if_ether.h>
34#include <linux/if_vlan.h>
35#include <linux/static_key.h>
36
37#include <net/ip.h>
38#include <net/icmp.h>
39#include <net/protocol.h>
40#include <net/ip_tunnels.h>
41#include <net/arp.h>
42#include <net/checksum.h>
43#include <net/dsfield.h>
44#include <net/inet_ecn.h>
45#include <net/xfrm.h>
46#include <net/net_namespace.h>
47#include <net/netns/generic.h>
48#include <net/rtnetlink.h>
49#include <net/dst_metadata.h>
50
51int iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
52		  __be32 src, __be32 dst, __u8 proto,
53		  __u8 tos, __u8 ttl, __be16 df, bool xnet)
54{
55	int pkt_len = skb->len - skb_inner_network_offset(skb);
56	struct net *net = dev_net(rt->dst.dev);
57	struct iphdr *iph;
58	int err;
59
60	skb_scrub_packet(skb, xnet);
61
62	skb_clear_hash(skb);
63	skb_dst_set(skb, &rt->dst);
64	memset(IPCB(skb), 0, sizeof(*IPCB(skb)));
65
66	/* Push down and install the IP header. */
67	skb_push(skb, sizeof(struct iphdr));
68	skb_reset_network_header(skb);
69
70	iph = ip_hdr(skb);
71
72	iph->version	=	4;
73	iph->ihl	=	sizeof(struct iphdr) >> 2;
74	iph->frag_off	=	df;
75	iph->protocol	=	proto;
76	iph->tos	=	tos;
77	iph->daddr	=	dst;
78	iph->saddr	=	src;
79	iph->ttl	=	ttl;
80	__ip_select_ident(net, iph, skb_shinfo(skb)->gso_segs ?: 1);
81
82	err = ip_local_out(net, sk, skb);
83	if (unlikely(net_xmit_eval(err)))
84		pkt_len = 0;
85	return pkt_len;
86}
87EXPORT_SYMBOL_GPL(iptunnel_xmit);
88
89int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto)
90{
91	if (unlikely(!pskb_may_pull(skb, hdr_len)))
92		return -ENOMEM;
93
94	skb_pull_rcsum(skb, hdr_len);
95
96	if (inner_proto == htons(ETH_P_TEB)) {
97		struct ethhdr *eh;
98
99		if (unlikely(!pskb_may_pull(skb, ETH_HLEN)))
100			return -ENOMEM;
101
102		eh = (struct ethhdr *)skb->data;
103		if (likely(eth_proto_is_802_3(eh->h_proto)))
104			skb->protocol = eh->h_proto;
105		else
106			skb->protocol = htons(ETH_P_802_2);
107
108	} else {
109		skb->protocol = inner_proto;
110	}
111
112	nf_reset(skb);
113	secpath_reset(skb);
114	skb_clear_hash_if_not_l4(skb);
115	skb_dst_drop(skb);
116	skb->vlan_tci = 0;
117	skb_set_queue_mapping(skb, 0);
118	skb->pkt_type = PACKET_HOST;
119	return 0;
120}
121EXPORT_SYMBOL_GPL(iptunnel_pull_header);
122
123struct metadata_dst *iptunnel_metadata_reply(struct metadata_dst *md,
124					     gfp_t flags)
125{
126	struct metadata_dst *res;
127	struct ip_tunnel_info *dst, *src;
128
129	if (!md || md->u.tun_info.mode & IP_TUNNEL_INFO_TX)
130		return NULL;
131
132	res = metadata_dst_alloc(0, flags);
133	if (!res)
134		return NULL;
135
136	dst = &res->u.tun_info;
137	src = &md->u.tun_info;
138	dst->key.tun_id = src->key.tun_id;
139	if (src->mode & IP_TUNNEL_INFO_IPV6)
140		memcpy(&dst->key.u.ipv6.dst, &src->key.u.ipv6.src,
141		       sizeof(struct in6_addr));
142	else
143		dst->key.u.ipv4.dst = src->key.u.ipv4.src;
144	dst->mode = src->mode | IP_TUNNEL_INFO_TX;
145
146	return res;
147}
148EXPORT_SYMBOL_GPL(iptunnel_metadata_reply);
149
150struct sk_buff *iptunnel_handle_offloads(struct sk_buff *skb,
151					 bool csum_help,
152					 int gso_type_mask)
153{
154	int err;
155
156	if (likely(!skb->encapsulation)) {
157		skb_reset_inner_headers(skb);
158		skb->encapsulation = 1;
159	}
160
161	if (skb_is_gso(skb)) {
162		err = skb_unclone(skb, GFP_ATOMIC);
163		if (unlikely(err))
164			goto error;
165		skb_shinfo(skb)->gso_type |= gso_type_mask;
166		return skb;
167	}
168
169	/* If packet is not gso and we are resolving any partial checksum,
170	 * clear encapsulation flag. This allows setting CHECKSUM_PARTIAL
171	 * on the outer header without confusing devices that implement
172	 * NETIF_F_IP_CSUM with encapsulation.
173	 */
174	if (csum_help)
175		skb->encapsulation = 0;
176
177	if (skb->ip_summed == CHECKSUM_PARTIAL && csum_help) {
178		err = skb_checksum_help(skb);
179		if (unlikely(err))
180			goto error;
181	} else if (skb->ip_summed != CHECKSUM_PARTIAL)
182		skb->ip_summed = CHECKSUM_NONE;
183
184	return skb;
185error:
186	kfree_skb(skb);
187	return ERR_PTR(err);
188}
189EXPORT_SYMBOL_GPL(iptunnel_handle_offloads);
190
191/* Often modified stats are per cpu, other are shared (netdev->stats) */
192struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev,
193						struct rtnl_link_stats64 *tot)
194{
195	int i;
196
197	netdev_stats_to_stats64(tot, &dev->stats);
198
199	for_each_possible_cpu(i) {
200		const struct pcpu_sw_netstats *tstats =
201						   per_cpu_ptr(dev->tstats, i);
202		u64 rx_packets, rx_bytes, tx_packets, tx_bytes;
203		unsigned int start;
204
205		do {
206			start = u64_stats_fetch_begin_irq(&tstats->syncp);
207			rx_packets = tstats->rx_packets;
208			tx_packets = tstats->tx_packets;
209			rx_bytes = tstats->rx_bytes;
210			tx_bytes = tstats->tx_bytes;
211		} while (u64_stats_fetch_retry_irq(&tstats->syncp, start));
212
213		tot->rx_packets += rx_packets;
214		tot->tx_packets += tx_packets;
215		tot->rx_bytes   += rx_bytes;
216		tot->tx_bytes   += tx_bytes;
217	}
218
219	return tot;
220}
221EXPORT_SYMBOL_GPL(ip_tunnel_get_stats64);
222
223static const struct nla_policy ip_tun_policy[LWTUNNEL_IP_MAX + 1] = {
224	[LWTUNNEL_IP_ID]	= { .type = NLA_U64 },
225	[LWTUNNEL_IP_DST]	= { .type = NLA_U32 },
226	[LWTUNNEL_IP_SRC]	= { .type = NLA_U32 },
227	[LWTUNNEL_IP_TTL]	= { .type = NLA_U8 },
228	[LWTUNNEL_IP_TOS]	= { .type = NLA_U8 },
229	[LWTUNNEL_IP_FLAGS]	= { .type = NLA_U16 },
230};
231
232static int ip_tun_build_state(struct net_device *dev, struct nlattr *attr,
233			      unsigned int family, const void *cfg,
234			      struct lwtunnel_state **ts)
235{
236	struct ip_tunnel_info *tun_info;
237	struct lwtunnel_state *new_state;
238	struct nlattr *tb[LWTUNNEL_IP_MAX + 1];
239	int err;
240
241	err = nla_parse_nested(tb, LWTUNNEL_IP_MAX, attr, ip_tun_policy);
242	if (err < 0)
243		return err;
244
245	new_state = lwtunnel_state_alloc(sizeof(*tun_info));
246	if (!new_state)
247		return -ENOMEM;
248
249	new_state->type = LWTUNNEL_ENCAP_IP;
250
251	tun_info = lwt_tun_info(new_state);
252
253	if (tb[LWTUNNEL_IP_ID])
254		tun_info->key.tun_id = nla_get_u64(tb[LWTUNNEL_IP_ID]);
255
256	if (tb[LWTUNNEL_IP_DST])
257		tun_info->key.u.ipv4.dst = nla_get_be32(tb[LWTUNNEL_IP_DST]);
258
259	if (tb[LWTUNNEL_IP_SRC])
260		tun_info->key.u.ipv4.src = nla_get_be32(tb[LWTUNNEL_IP_SRC]);
261
262	if (tb[LWTUNNEL_IP_TTL])
263		tun_info->key.ttl = nla_get_u8(tb[LWTUNNEL_IP_TTL]);
264
265	if (tb[LWTUNNEL_IP_TOS])
266		tun_info->key.tos = nla_get_u8(tb[LWTUNNEL_IP_TOS]);
267
268	if (tb[LWTUNNEL_IP_FLAGS])
269		tun_info->key.tun_flags = nla_get_u16(tb[LWTUNNEL_IP_FLAGS]);
270
271	tun_info->mode = IP_TUNNEL_INFO_TX;
272	tun_info->options_len = 0;
273
274	*ts = new_state;
275
276	return 0;
277}
278
279static int ip_tun_fill_encap_info(struct sk_buff *skb,
280				  struct lwtunnel_state *lwtstate)
281{
282	struct ip_tunnel_info *tun_info = lwt_tun_info(lwtstate);
283
284	if (nla_put_u64(skb, LWTUNNEL_IP_ID, tun_info->key.tun_id) ||
285	    nla_put_be32(skb, LWTUNNEL_IP_DST, tun_info->key.u.ipv4.dst) ||
286	    nla_put_be32(skb, LWTUNNEL_IP_SRC, tun_info->key.u.ipv4.src) ||
287	    nla_put_u8(skb, LWTUNNEL_IP_TOS, tun_info->key.tos) ||
288	    nla_put_u8(skb, LWTUNNEL_IP_TTL, tun_info->key.ttl) ||
289	    nla_put_u16(skb, LWTUNNEL_IP_FLAGS, tun_info->key.tun_flags))
290		return -ENOMEM;
291
292	return 0;
293}
294
295static int ip_tun_encap_nlsize(struct lwtunnel_state *lwtstate)
296{
297	return nla_total_size(8)	/* LWTUNNEL_IP_ID */
298		+ nla_total_size(4)	/* LWTUNNEL_IP_DST */
299		+ nla_total_size(4)	/* LWTUNNEL_IP_SRC */
300		+ nla_total_size(1)	/* LWTUNNEL_IP_TOS */
301		+ nla_total_size(1)	/* LWTUNNEL_IP_TTL */
302		+ nla_total_size(2);	/* LWTUNNEL_IP_FLAGS */
303}
304
305static int ip_tun_cmp_encap(struct lwtunnel_state *a, struct lwtunnel_state *b)
306{
307	return memcmp(lwt_tun_info(a), lwt_tun_info(b),
308		      sizeof(struct ip_tunnel_info));
309}
310
311static const struct lwtunnel_encap_ops ip_tun_lwt_ops = {
312	.build_state = ip_tun_build_state,
313	.fill_encap = ip_tun_fill_encap_info,
314	.get_encap_size = ip_tun_encap_nlsize,
315	.cmp_encap = ip_tun_cmp_encap,
316};
317
318static const struct nla_policy ip6_tun_policy[LWTUNNEL_IP6_MAX + 1] = {
319	[LWTUNNEL_IP6_ID]		= { .type = NLA_U64 },
320	[LWTUNNEL_IP6_DST]		= { .len = sizeof(struct in6_addr) },
321	[LWTUNNEL_IP6_SRC]		= { .len = sizeof(struct in6_addr) },
322	[LWTUNNEL_IP6_HOPLIMIT]		= { .type = NLA_U8 },
323	[LWTUNNEL_IP6_TC]		= { .type = NLA_U8 },
324	[LWTUNNEL_IP6_FLAGS]		= { .type = NLA_U16 },
325};
326
327static int ip6_tun_build_state(struct net_device *dev, struct nlattr *attr,
328			       unsigned int family, const void *cfg,
329			       struct lwtunnel_state **ts)
330{
331	struct ip_tunnel_info *tun_info;
332	struct lwtunnel_state *new_state;
333	struct nlattr *tb[LWTUNNEL_IP6_MAX + 1];
334	int err;
335
336	err = nla_parse_nested(tb, LWTUNNEL_IP6_MAX, attr, ip6_tun_policy);
337	if (err < 0)
338		return err;
339
340	new_state = lwtunnel_state_alloc(sizeof(*tun_info));
341	if (!new_state)
342		return -ENOMEM;
343
344	new_state->type = LWTUNNEL_ENCAP_IP6;
345
346	tun_info = lwt_tun_info(new_state);
347
348	if (tb[LWTUNNEL_IP6_ID])
349		tun_info->key.tun_id = nla_get_u64(tb[LWTUNNEL_IP6_ID]);
350
351	if (tb[LWTUNNEL_IP6_DST])
352		tun_info->key.u.ipv6.dst = nla_get_in6_addr(tb[LWTUNNEL_IP6_DST]);
353
354	if (tb[LWTUNNEL_IP6_SRC])
355		tun_info->key.u.ipv6.src = nla_get_in6_addr(tb[LWTUNNEL_IP6_SRC]);
356
357	if (tb[LWTUNNEL_IP6_HOPLIMIT])
358		tun_info->key.ttl = nla_get_u8(tb[LWTUNNEL_IP6_HOPLIMIT]);
359
360	if (tb[LWTUNNEL_IP6_TC])
361		tun_info->key.tos = nla_get_u8(tb[LWTUNNEL_IP6_TC]);
362
363	if (tb[LWTUNNEL_IP6_FLAGS])
364		tun_info->key.tun_flags = nla_get_u16(tb[LWTUNNEL_IP6_FLAGS]);
365
366	tun_info->mode = IP_TUNNEL_INFO_TX | IP_TUNNEL_INFO_IPV6;
367	tun_info->options_len = 0;
368
369	*ts = new_state;
370
371	return 0;
372}
373
374static int ip6_tun_fill_encap_info(struct sk_buff *skb,
375				   struct lwtunnel_state *lwtstate)
376{
377	struct ip_tunnel_info *tun_info = lwt_tun_info(lwtstate);
378
379	if (nla_put_u64(skb, LWTUNNEL_IP6_ID, tun_info->key.tun_id) ||
380	    nla_put_in6_addr(skb, LWTUNNEL_IP6_DST, &tun_info->key.u.ipv6.dst) ||
381	    nla_put_in6_addr(skb, LWTUNNEL_IP6_SRC, &tun_info->key.u.ipv6.src) ||
382	    nla_put_u8(skb, LWTUNNEL_IP6_HOPLIMIT, tun_info->key.tos) ||
383	    nla_put_u8(skb, LWTUNNEL_IP6_TC, tun_info->key.ttl) ||
384	    nla_put_u16(skb, LWTUNNEL_IP6_FLAGS, tun_info->key.tun_flags))
385		return -ENOMEM;
386
387	return 0;
388}
389
390static int ip6_tun_encap_nlsize(struct lwtunnel_state *lwtstate)
391{
392	return nla_total_size(8)	/* LWTUNNEL_IP6_ID */
393		+ nla_total_size(16)	/* LWTUNNEL_IP6_DST */
394		+ nla_total_size(16)	/* LWTUNNEL_IP6_SRC */
395		+ nla_total_size(1)	/* LWTUNNEL_IP6_HOPLIMIT */
396		+ nla_total_size(1)	/* LWTUNNEL_IP6_TC */
397		+ nla_total_size(2);	/* LWTUNNEL_IP6_FLAGS */
398}
399
400static const struct lwtunnel_encap_ops ip6_tun_lwt_ops = {
401	.build_state = ip6_tun_build_state,
402	.fill_encap = ip6_tun_fill_encap_info,
403	.get_encap_size = ip6_tun_encap_nlsize,
404	.cmp_encap = ip_tun_cmp_encap,
405};
406
407void __init ip_tunnel_core_init(void)
408{
409	lwtunnel_encap_add_ops(&ip_tun_lwt_ops, LWTUNNEL_ENCAP_IP);
410	lwtunnel_encap_add_ops(&ip6_tun_lwt_ops, LWTUNNEL_ENCAP_IP6);
411}
412
413struct static_key ip_tunnel_metadata_cnt = STATIC_KEY_INIT_FALSE;
414EXPORT_SYMBOL(ip_tunnel_metadata_cnt);
415
416void ip_tunnel_need_metadata(void)
417{
418	static_key_slow_inc(&ip_tunnel_metadata_cnt);
419}
420EXPORT_SYMBOL_GPL(ip_tunnel_need_metadata);
421
422void ip_tunnel_unneed_metadata(void)
423{
424	static_key_slow_dec(&ip_tunnel_metadata_cnt);
425}
426EXPORT_SYMBOL_GPL(ip_tunnel_unneed_metadata);
427