1 /* SPDX-License-Identifier: GPL-2.0 */
2 #ifndef __NET_IP_TUNNELS_H
3 #define __NET_IP_TUNNELS_H 1
4
5 #include <linux/if_tunnel.h>
6 #include <linux/netdevice.h>
7 #include <linux/skbuff.h>
8 #include <linux/socket.h>
9 #include <linux/types.h>
10 #include <linux/u64_stats_sync.h>
11 #include <linux/bitops.h>
12
13 #include <net/dsfield.h>
14 #include <net/flow.h>
15 #include <net/gro_cells.h>
16 #include <net/inet_dscp.h>
17 #include <net/inet_ecn.h>
18 #include <net/netns/generic.h>
19 #include <net/rtnetlink.h>
20 #include <net/lwtunnel.h>
21 #include <net/dst_cache.h>
22 #include <net/netdev_lock.h>
23
24 #if IS_ENABLED(CONFIG_IPV6)
25 #include <net/ipv6.h>
26 #include <net/ip6_fib.h>
27 #include <net/ip6_route.h>
28 #endif
29
30 /* Recursion limit for tunnel xmit to detect routing loops.
31 * Unlike XMIT_RECURSION_LIMIT (8) used in the no-qdisc path, tunnel
32 * recursion involves route lookups and full IP output, consuming much
33 * more stack per level, so a lower limit is needed.
34 */
35 #define IP_TUNNEL_RECURSION_LIMIT 4
36
37 /* Keep error state on tunnel for 30 sec */
38 #define IPTUNNEL_ERR_TIMEO (30*HZ)
39
40 /* Used to memset ip_tunnel padding. */
41 #define IP_TUNNEL_KEY_SIZE offsetofend(struct ip_tunnel_key, tp_dst)
42
43 /* Used to memset ipv4 address padding. */
44 #define IP_TUNNEL_KEY_IPV4_PAD offsetofend(struct ip_tunnel_key, u.ipv4.dst)
45 #define IP_TUNNEL_KEY_IPV4_PAD_LEN \
46 (sizeof_field(struct ip_tunnel_key, u) - \
47 sizeof_field(struct ip_tunnel_key, u.ipv4))
48
49 #define __ipt_flag_op(op, ...) \
50 op(__VA_ARGS__, __IP_TUNNEL_FLAG_NUM)
51
52 #define IP_TUNNEL_DECLARE_FLAGS(...) \
53 __ipt_flag_op(DECLARE_BITMAP, __VA_ARGS__)
54
55 #define ip_tunnel_flags_zero(...) __ipt_flag_op(bitmap_zero, __VA_ARGS__)
56 #define ip_tunnel_flags_copy(...) __ipt_flag_op(bitmap_copy, __VA_ARGS__)
57 #define ip_tunnel_flags_and(...) __ipt_flag_op(bitmap_and, __VA_ARGS__)
58 #define ip_tunnel_flags_or(...) __ipt_flag_op(bitmap_or, __VA_ARGS__)
59
60 #define ip_tunnel_flags_empty(...) \
61 __ipt_flag_op(bitmap_empty, __VA_ARGS__)
62 #define ip_tunnel_flags_intersect(...) \
63 __ipt_flag_op(bitmap_intersects, __VA_ARGS__)
64 #define ip_tunnel_flags_subset(...) \
65 __ipt_flag_op(bitmap_subset, __VA_ARGS__)
66
67 struct ip_tunnel_key {
68 __be64 tun_id;
69 union {
70 struct {
71 __be32 src;
72 __be32 dst;
73 } ipv4;
74 struct {
75 struct in6_addr src;
76 struct in6_addr dst;
77 } ipv6;
78 } u;
79 IP_TUNNEL_DECLARE_FLAGS(tun_flags);
80 __be32 label; /* Flow Label for IPv6 */
81 u32 nhid;
82 u8 tos; /* TOS for IPv4, TC for IPv6 */
83 u8 ttl; /* TTL for IPv4, HL for IPv6 */
84 __be16 tp_src;
85 __be16 tp_dst;
86 __u8 flow_flags;
87 };
88
89 struct ip_tunnel_encap {
90 u16 type;
91 u16 flags;
92 __be16 sport;
93 __be16 dport;
94 };
95
96 /* Flags for ip_tunnel_info mode. */
97 #define IP_TUNNEL_INFO_TX 0x01 /* represents tx tunnel parameters */
98 #define IP_TUNNEL_INFO_IPV6 0x02 /* key contains IPv6 addresses */
99 #define IP_TUNNEL_INFO_BRIDGE 0x04 /* represents a bridged tunnel id */
100
101 /* Maximum tunnel options length. */
102 #define IP_TUNNEL_OPTS_MAX \
103 GENMASK((sizeof_field(struct ip_tunnel_info, \
104 options_len) * BITS_PER_BYTE) - 1, 0)
105
106 #define ip_tunnel_info_opts(info) \
107 _Generic(info, \
108 const struct ip_tunnel_info * : ((const void *)(info)->options),\
109 struct ip_tunnel_info * : ((void *)(info)->options)\
110 )
111
112 struct ip_tunnel_info {
113 struct ip_tunnel_key key;
114 struct ip_tunnel_encap encap;
115 #ifdef CONFIG_DST_CACHE
116 struct dst_cache dst_cache;
117 #endif
118 u8 options_len;
119 u8 mode;
120 u8 options[] __aligned_largest __counted_by(options_len);
121 };
122
123 /* 6rd prefix/relay information */
124 #ifdef CONFIG_IPV6_SIT_6RD
125 struct ip_tunnel_6rd_parm {
126 struct in6_addr prefix;
127 __be32 relay_prefix;
128 u16 prefixlen;
129 u16 relay_prefixlen;
130 };
131 #endif
132
133 struct ip_tunnel_prl_entry {
134 struct ip_tunnel_prl_entry __rcu *next;
135 __be32 addr;
136 u16 flags;
137 struct rcu_head rcu_head;
138 };
139
140 struct metadata_dst;
141
142 /* Kernel-side variant of ip_tunnel_parm */
143 struct ip_tunnel_parm_kern {
144 char name[IFNAMSIZ];
145 IP_TUNNEL_DECLARE_FLAGS(i_flags);
146 IP_TUNNEL_DECLARE_FLAGS(o_flags);
147 __be32 i_key;
148 __be32 o_key;
149 int link;
150 struct iphdr iph;
151 };
152
153 struct ip_tunnel {
154 struct ip_tunnel __rcu *next;
155 struct hlist_node hash_node;
156
157 struct net_device *dev;
158 netdevice_tracker dev_tracker;
159
160 struct net *net; /* netns for packet i/o */
161
162 unsigned long err_time; /* Time when the last ICMP error
163 * arrived */
164 int err_count; /* Number of arrived ICMP errors */
165
166 /* These four fields used only by GRE */
167 u32 i_seqno; /* The last seen seqno */
168 atomic_t o_seqno; /* The last output seqno */
169 int tun_hlen; /* Precalculated header length */
170
171 /* These four fields used only by ERSPAN */
172 u32 index; /* ERSPAN type II index */
173 u8 erspan_ver; /* ERSPAN version */
174 u8 dir; /* ERSPAN direction */
175 u16 hwid; /* ERSPAN hardware ID */
176
177 struct dst_cache dst_cache;
178
179 struct ip_tunnel_parm_kern parms;
180
181 int mlink;
182 int encap_hlen; /* Encap header length (FOU,GUE) */
183 int hlen; /* tun_hlen + encap_hlen */
184 struct ip_tunnel_encap encap;
185
186 /* for SIT */
187 #ifdef CONFIG_IPV6_SIT_6RD
188 struct ip_tunnel_6rd_parm ip6rd;
189 #endif
190 struct ip_tunnel_prl_entry __rcu *prl; /* potential router list */
191 unsigned int prl_count; /* # of entries in PRL */
192 unsigned int ip_tnl_net_id;
193 struct gro_cells gro_cells;
194 __u32 fwmark;
195 bool collect_md;
196 bool ignore_df;
197 };
198
199 struct tnl_ptk_info {
200 IP_TUNNEL_DECLARE_FLAGS(flags);
201 __be16 proto;
202 __be32 key;
203 __be32 seq;
204 int hdr_len;
205 };
206
207 #define PACKET_RCVD 0
208 #define PACKET_REJECT 1
209 #define PACKET_NEXT 2
210
211 #define IP_TNL_HASH_BITS 7
212 #define IP_TNL_HASH_SIZE (1 << IP_TNL_HASH_BITS)
213
214 struct ip_tunnel_net {
215 struct net_device *fb_tunnel_dev;
216 struct rtnl_link_ops *rtnl_link_ops;
217 struct hlist_head tunnels[IP_TNL_HASH_SIZE];
218 struct ip_tunnel __rcu *collect_md_tun;
219 int type;
220 };
221
ip_tunnel_set_options_present(unsigned long * flags)222 static inline void ip_tunnel_set_options_present(unsigned long *flags)
223 {
224 IP_TUNNEL_DECLARE_FLAGS(present) = { };
225
226 __set_bit(IP_TUNNEL_GENEVE_OPT_BIT, present);
227 __set_bit(IP_TUNNEL_VXLAN_OPT_BIT, present);
228 __set_bit(IP_TUNNEL_ERSPAN_OPT_BIT, present);
229 __set_bit(IP_TUNNEL_GTP_OPT_BIT, present);
230 __set_bit(IP_TUNNEL_PFCP_OPT_BIT, present);
231
232 ip_tunnel_flags_or(flags, flags, present);
233 }
234
ip_tunnel_clear_options_present(unsigned long * flags)235 static inline void ip_tunnel_clear_options_present(unsigned long *flags)
236 {
237 IP_TUNNEL_DECLARE_FLAGS(present) = { };
238
239 __set_bit(IP_TUNNEL_GENEVE_OPT_BIT, present);
240 __set_bit(IP_TUNNEL_VXLAN_OPT_BIT, present);
241 __set_bit(IP_TUNNEL_ERSPAN_OPT_BIT, present);
242 __set_bit(IP_TUNNEL_GTP_OPT_BIT, present);
243 __set_bit(IP_TUNNEL_PFCP_OPT_BIT, present);
244
245 __ipt_flag_op(bitmap_andnot, flags, flags, present);
246 }
247
ip_tunnel_is_options_present(const unsigned long * flags)248 static inline bool ip_tunnel_is_options_present(const unsigned long *flags)
249 {
250 IP_TUNNEL_DECLARE_FLAGS(present) = { };
251
252 __set_bit(IP_TUNNEL_GENEVE_OPT_BIT, present);
253 __set_bit(IP_TUNNEL_VXLAN_OPT_BIT, present);
254 __set_bit(IP_TUNNEL_ERSPAN_OPT_BIT, present);
255 __set_bit(IP_TUNNEL_GTP_OPT_BIT, present);
256 __set_bit(IP_TUNNEL_PFCP_OPT_BIT, present);
257
258 return ip_tunnel_flags_intersect(flags, present);
259 }
260
ip_tunnel_flags_is_be16_compat(const unsigned long * flags)261 static inline bool ip_tunnel_flags_is_be16_compat(const unsigned long *flags)
262 {
263 IP_TUNNEL_DECLARE_FLAGS(supp) = { };
264
265 bitmap_set(supp, 0, BITS_PER_TYPE(__be16));
266 __set_bit(IP_TUNNEL_VTI_BIT, supp);
267
268 return ip_tunnel_flags_subset(flags, supp);
269 }
270
ip_tunnel_flags_from_be16(unsigned long * dst,__be16 flags)271 static inline void ip_tunnel_flags_from_be16(unsigned long *dst, __be16 flags)
272 {
273 ip_tunnel_flags_zero(dst);
274
275 bitmap_write(dst, be16_to_cpu(flags), 0, BITS_PER_TYPE(__be16));
276 __assign_bit(IP_TUNNEL_VTI_BIT, dst, flags & VTI_ISVTI);
277 }
278
ip_tunnel_flags_to_be16(const unsigned long * flags)279 static inline __be16 ip_tunnel_flags_to_be16(const unsigned long *flags)
280 {
281 __be16 ret;
282
283 ret = cpu_to_be16(bitmap_read(flags, 0, BITS_PER_TYPE(__be16)));
284 if (test_bit(IP_TUNNEL_VTI_BIT, flags))
285 ret |= VTI_ISVTI;
286
287 return ret;
288 }
289
ip_tunnel_key_init(struct ip_tunnel_key * key,__be32 saddr,__be32 daddr,u8 tos,u8 ttl,__be32 label,__be16 tp_src,__be16 tp_dst,__be64 tun_id,const unsigned long * tun_flags)290 static inline void ip_tunnel_key_init(struct ip_tunnel_key *key,
291 __be32 saddr, __be32 daddr,
292 u8 tos, u8 ttl, __be32 label,
293 __be16 tp_src, __be16 tp_dst,
294 __be64 tun_id,
295 const unsigned long *tun_flags)
296 {
297 key->tun_id = tun_id;
298 key->u.ipv4.src = saddr;
299 key->u.ipv4.dst = daddr;
300 memset((unsigned char *)key + IP_TUNNEL_KEY_IPV4_PAD,
301 0, IP_TUNNEL_KEY_IPV4_PAD_LEN);
302 key->tos = tos;
303 key->ttl = ttl;
304 key->label = label;
305 ip_tunnel_flags_copy(key->tun_flags, tun_flags);
306
307 /* For the tunnel types on the top of IPsec, the tp_src and tp_dst of
308 * the upper tunnel are used.
309 * E.g: GRE over IPSEC, the tp_src and tp_port are zero.
310 */
311 key->tp_src = tp_src;
312 key->tp_dst = tp_dst;
313
314 /* Clear struct padding. */
315 if (sizeof(*key) != IP_TUNNEL_KEY_SIZE)
316 memset((unsigned char *)key + IP_TUNNEL_KEY_SIZE,
317 0, sizeof(*key) - IP_TUNNEL_KEY_SIZE);
318 }
319
320 static inline bool
ip_tunnel_dst_cache_usable(const struct sk_buff * skb,const struct ip_tunnel_info * info)321 ip_tunnel_dst_cache_usable(const struct sk_buff *skb,
322 const struct ip_tunnel_info *info)
323 {
324 if (skb->mark)
325 return false;
326
327 return !info || !test_bit(IP_TUNNEL_NOCACHE_BIT, info->key.tun_flags);
328 }
329
ip_tunnel_info_af(const struct ip_tunnel_info * tun_info)330 static inline unsigned short ip_tunnel_info_af(const struct ip_tunnel_info
331 *tun_info)
332 {
333 return tun_info->mode & IP_TUNNEL_INFO_IPV6 ? AF_INET6 : AF_INET;
334 }
335
key32_to_tunnel_id(__be32 key)336 static inline __be64 key32_to_tunnel_id(__be32 key)
337 {
338 #ifdef __BIG_ENDIAN
339 return (__force __be64)key;
340 #else
341 return (__force __be64)((__force u64)key << 32);
342 #endif
343 }
344
345 /* Returns the least-significant 32 bits of a __be64. */
tunnel_id_to_key32(__be64 tun_id)346 static inline __be32 tunnel_id_to_key32(__be64 tun_id)
347 {
348 #ifdef __BIG_ENDIAN
349 return (__force __be32)tun_id;
350 #else
351 return (__force __be32)((__force u64)tun_id >> 32);
352 #endif
353 }
354
355 #ifdef CONFIG_INET
356
ip_tunnel_init_flow(struct flowi4 * fl4,int proto,__be32 daddr,__be32 saddr,__be32 key,__u8 tos,struct net * net,int oif,__u32 mark,__u32 tun_inner_hash,__u8 flow_flags)357 static inline void ip_tunnel_init_flow(struct flowi4 *fl4,
358 int proto,
359 __be32 daddr, __be32 saddr,
360 __be32 key, __u8 tos,
361 struct net *net, int oif,
362 __u32 mark, __u32 tun_inner_hash,
363 __u8 flow_flags)
364 {
365 memset(fl4, 0, sizeof(*fl4));
366
367 if (oif) {
368 fl4->flowi4_l3mdev = l3mdev_master_upper_ifindex_by_index(net, oif);
369 /* Legacy VRF/l3mdev use case */
370 fl4->flowi4_oif = fl4->flowi4_l3mdev ? 0 : oif;
371 }
372
373 fl4->daddr = daddr;
374 fl4->saddr = saddr;
375 fl4->flowi4_dscp = inet_dsfield_to_dscp(tos);
376 fl4->flowi4_proto = proto;
377 fl4->fl4_gre_key = key;
378 fl4->flowi4_mark = mark;
379 fl4->flowi4_multipath_hash = tun_inner_hash;
380 fl4->flowi4_flags = flow_flags;
381 }
382
383 int __ip_tunnel_init(struct net_device *dev);
384 #define ip_tunnel_init(DEV) \
385 ({ \
386 struct net_device *__dev = (DEV); \
387 int __res = __ip_tunnel_init(__dev); \
388 \
389 if (!__res) \
390 netdev_lockdep_set_classes(__dev);\
391 __res; \
392 })
393
394 void ip_tunnel_uninit(struct net_device *dev);
395 void ip_tunnel_dellink(struct net_device *dev, struct list_head *head);
396 struct net *ip_tunnel_get_link_net(const struct net_device *dev);
397 int ip_tunnel_get_iflink(const struct net_device *dev);
398 int ip_tunnel_init_net(struct net *net, unsigned int ip_tnl_net_id,
399 struct rtnl_link_ops *ops, char *devname);
400 void ip_tunnel_delete_net(struct net *net, unsigned int id,
401 struct rtnl_link_ops *ops,
402 struct list_head *dev_to_kill);
403
404 void ip_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
405 const struct iphdr *tnl_params, const u8 protocol);
406 void ip_md_tunnel_xmit(struct sk_buff *skb, struct net_device *dev,
407 const u8 proto, int tunnel_hlen);
408 int ip_tunnel_ctl(struct net_device *dev, struct ip_tunnel_parm_kern *p,
409 int cmd);
410 bool ip_tunnel_parm_from_user(struct ip_tunnel_parm_kern *kp,
411 const void __user *data);
412 bool ip_tunnel_parm_to_user(void __user *data, struct ip_tunnel_parm_kern *kp);
413 int ip_tunnel_siocdevprivate(struct net_device *dev, struct ifreq *ifr,
414 void __user *data, int cmd);
415 int __ip_tunnel_change_mtu(struct net_device *dev, int new_mtu, bool strict);
416 int ip_tunnel_change_mtu(struct net_device *dev, int new_mtu);
417
418 struct ip_tunnel *ip_tunnel_lookup(struct ip_tunnel_net *itn,
419 int link, const unsigned long *flags,
420 __be32 remote, __be32 local,
421 __be32 key);
422
423 void ip_tunnel_md_udp_encap(struct sk_buff *skb, struct ip_tunnel_info *info);
424 int ip_tunnel_rcv(struct ip_tunnel *tunnel, struct sk_buff *skb,
425 const struct tnl_ptk_info *tpi, struct metadata_dst *tun_dst,
426 bool log_ecn_error);
427 int ip_tunnel_changelink(struct net_device *dev, struct nlattr *tb[],
428 struct ip_tunnel_parm_kern *p, __u32 fwmark);
429 int ip_tunnel_newlink(struct net *net, struct net_device *dev,
430 struct nlattr *tb[], struct ip_tunnel_parm_kern *p,
431 __u32 fwmark);
432 void ip_tunnel_setup(struct net_device *dev, unsigned int net_id);
433
434 bool ip_tunnel_netlink_encap_parms(struct nlattr *data[],
435 struct ip_tunnel_encap *encap);
436
437 void ip_tunnel_netlink_parms(struct nlattr *data[],
438 struct ip_tunnel_parm_kern *parms);
439
440 extern const struct header_ops ip_tunnel_header_ops;
441 __be16 ip_tunnel_parse_protocol(const struct sk_buff *skb);
442
443 struct ip_tunnel_encap_ops {
444 size_t (*encap_hlen)(struct ip_tunnel_encap *e);
445 int (*build_header)(struct sk_buff *skb, struct ip_tunnel_encap *e,
446 u8 *protocol, struct flowi4 *fl4);
447 int (*err_handler)(struct sk_buff *skb, u32 info);
448 };
449
450 #define MAX_IPTUN_ENCAP_OPS 8
451
452 extern const struct ip_tunnel_encap_ops __rcu *
453 iptun_encaps[MAX_IPTUN_ENCAP_OPS];
454
455 int ip_tunnel_encap_add_ops(const struct ip_tunnel_encap_ops *op,
456 unsigned int num);
457 int ip_tunnel_encap_del_ops(const struct ip_tunnel_encap_ops *op,
458 unsigned int num);
459
460 int ip_tunnel_encap_setup(struct ip_tunnel *t,
461 struct ip_tunnel_encap *ipencap);
462
463 static inline enum skb_drop_reason
pskb_inet_may_pull_reason(struct sk_buff * skb)464 pskb_inet_may_pull_reason(struct sk_buff *skb)
465 {
466 int nhlen;
467
468 switch (skb->protocol) {
469 #if IS_ENABLED(CONFIG_IPV6)
470 case htons(ETH_P_IPV6):
471 nhlen = sizeof(struct ipv6hdr);
472 break;
473 #endif
474 case htons(ETH_P_IP):
475 nhlen = sizeof(struct iphdr);
476 break;
477 default:
478 nhlen = 0;
479 }
480
481 return pskb_network_may_pull_reason(skb, nhlen);
482 }
483
pskb_inet_may_pull(struct sk_buff * skb)484 static inline bool pskb_inet_may_pull(struct sk_buff *skb)
485 {
486 return pskb_inet_may_pull_reason(skb) == SKB_NOT_DROPPED_YET;
487 }
488
489 /* Variant of pskb_inet_may_pull().
490 */
491 static inline enum skb_drop_reason
skb_vlan_inet_prepare(struct sk_buff * skb,bool inner_proto_inherit)492 skb_vlan_inet_prepare(struct sk_buff *skb, bool inner_proto_inherit)
493 {
494 int nhlen = 0, maclen = inner_proto_inherit ? 0 : ETH_HLEN;
495 __be16 type = skb->protocol;
496 enum skb_drop_reason reason;
497
498 /* Essentially this is skb_protocol(skb, true)
499 * And we get MAC len.
500 */
501 if (eth_type_vlan(type))
502 type = __vlan_get_protocol(skb, type, &maclen);
503
504 switch (type) {
505 #if IS_ENABLED(CONFIG_IPV6)
506 case htons(ETH_P_IPV6):
507 nhlen = sizeof(struct ipv6hdr);
508 break;
509 #endif
510 case htons(ETH_P_IP):
511 nhlen = sizeof(struct iphdr);
512 break;
513 }
514 /* For ETH_P_IPV6/ETH_P_IP we make sure to pull
515 * a base network header in skb->head.
516 */
517 reason = pskb_may_pull_reason(skb, maclen + nhlen);
518 if (reason)
519 return reason;
520
521 skb_set_network_header(skb, maclen);
522
523 return SKB_NOT_DROPPED_YET;
524 }
525
ip_encap_hlen(struct ip_tunnel_encap * e)526 static inline int ip_encap_hlen(struct ip_tunnel_encap *e)
527 {
528 const struct ip_tunnel_encap_ops *ops;
529 int hlen = -EINVAL;
530
531 if (e->type == TUNNEL_ENCAP_NONE)
532 return 0;
533
534 if (e->type >= MAX_IPTUN_ENCAP_OPS)
535 return -EINVAL;
536
537 rcu_read_lock();
538 ops = rcu_dereference(iptun_encaps[e->type]);
539 if (likely(ops && ops->encap_hlen))
540 hlen = ops->encap_hlen(e);
541 rcu_read_unlock();
542
543 return hlen;
544 }
545
ip_tunnel_encap(struct sk_buff * skb,struct ip_tunnel_encap * e,u8 * protocol,struct flowi4 * fl4)546 static inline int ip_tunnel_encap(struct sk_buff *skb,
547 struct ip_tunnel_encap *e,
548 u8 *protocol, struct flowi4 *fl4)
549 {
550 const struct ip_tunnel_encap_ops *ops;
551 int ret = -EINVAL;
552
553 if (e->type == TUNNEL_ENCAP_NONE)
554 return 0;
555
556 if (e->type >= MAX_IPTUN_ENCAP_OPS)
557 return -EINVAL;
558
559 rcu_read_lock();
560 ops = rcu_dereference(iptun_encaps[e->type]);
561 if (likely(ops && ops->build_header))
562 ret = ops->build_header(skb, e, protocol, fl4);
563 rcu_read_unlock();
564
565 return ret;
566 }
567
568 /* Extract dsfield from inner protocol */
ip_tunnel_get_dsfield(const struct iphdr * iph,const struct sk_buff * skb)569 static inline u8 ip_tunnel_get_dsfield(const struct iphdr *iph,
570 const struct sk_buff *skb)
571 {
572 __be16 payload_protocol = skb_protocol(skb, true);
573
574 if (payload_protocol == htons(ETH_P_IP))
575 return iph->tos;
576 else if (payload_protocol == htons(ETH_P_IPV6))
577 return ipv6_get_dsfield((const struct ipv6hdr *)iph);
578 else
579 return 0;
580 }
581
ip_tunnel_get_flowlabel(const struct iphdr * iph,const struct sk_buff * skb)582 static inline __be32 ip_tunnel_get_flowlabel(const struct iphdr *iph,
583 const struct sk_buff *skb)
584 {
585 __be16 payload_protocol = skb_protocol(skb, true);
586
587 if (payload_protocol == htons(ETH_P_IPV6))
588 return ip6_flowlabel((const struct ipv6hdr *)iph);
589 else
590 return 0;
591 }
592
ip_tunnel_get_ttl(const struct iphdr * iph,const struct sk_buff * skb)593 static inline u8 ip_tunnel_get_ttl(const struct iphdr *iph,
594 const struct sk_buff *skb)
595 {
596 __be16 payload_protocol = skb_protocol(skb, true);
597
598 if (payload_protocol == htons(ETH_P_IP))
599 return iph->ttl;
600 else if (payload_protocol == htons(ETH_P_IPV6))
601 return ((const struct ipv6hdr *)iph)->hop_limit;
602 else
603 return 0;
604 }
605
606 /* Propagate ECN bits out */
ip_tunnel_ecn_encap(u8 tos,const struct iphdr * iph,const struct sk_buff * skb)607 static inline u8 ip_tunnel_ecn_encap(u8 tos, const struct iphdr *iph,
608 const struct sk_buff *skb)
609 {
610 u8 inner = ip_tunnel_get_dsfield(iph, skb);
611
612 return INET_ECN_encapsulate(tos, inner);
613 }
614
615 int __iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
616 __be16 inner_proto, bool raw_proto, bool xnet);
617
iptunnel_pull_header(struct sk_buff * skb,int hdr_len,__be16 inner_proto,bool xnet)618 static inline int iptunnel_pull_header(struct sk_buff *skb, int hdr_len,
619 __be16 inner_proto, bool xnet)
620 {
621 return __iptunnel_pull_header(skb, hdr_len, inner_proto, false, xnet);
622 }
623
624 void iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb,
625 __be32 src, __be32 dst, u8 proto,
626 u8 tos, u8 ttl, __be16 df, bool xnet, u16 ipcb_flags);
627 struct metadata_dst *iptunnel_metadata_reply(struct metadata_dst *md,
628 gfp_t flags);
629 int skb_tunnel_check_pmtu(struct sk_buff *skb, struct dst_entry *encap_dst,
630 int headroom, bool reply);
631
ip_tunnel_adj_headroom(struct net_device * dev,unsigned int headroom)632 static inline void ip_tunnel_adj_headroom(struct net_device *dev,
633 unsigned int headroom)
634 {
635 /* we must cap headroom to some upperlimit, else pskb_expand_head
636 * will overflow header offsets in skb_headers_offset_update().
637 */
638 const unsigned int max_allowed = 512;
639
640 if (headroom > max_allowed)
641 headroom = max_allowed;
642
643 if (headroom > READ_ONCE(dev->needed_headroom))
644 WRITE_ONCE(dev->needed_headroom, headroom);
645 }
646
647 int iptunnel_handle_offloads(struct sk_buff *skb, int gso_type_mask);
648
iptunnel_pull_offloads(struct sk_buff * skb)649 static inline int iptunnel_pull_offloads(struct sk_buff *skb)
650 {
651 if (skb_is_gso(skb)) {
652 int err;
653
654 err = skb_unclone(skb, GFP_ATOMIC);
655 if (unlikely(err))
656 return err;
657 skb_shinfo(skb)->gso_type &= ~(NETIF_F_GSO_ENCAP_ALL >>
658 NETIF_F_GSO_SHIFT);
659 }
660
661 skb->encapsulation = 0;
662 return 0;
663 }
664
iptunnel_xmit_stats(struct net_device * dev,int pkt_len)665 static inline void iptunnel_xmit_stats(struct net_device *dev, int pkt_len)
666 {
667 if (pkt_len > 0) {
668 if (dev->pcpu_stat_type == NETDEV_PCPU_STAT_DSTATS) {
669 struct pcpu_dstats *dstats = get_cpu_ptr(dev->dstats);
670
671 u64_stats_update_begin(&dstats->syncp);
672 u64_stats_add(&dstats->tx_bytes, pkt_len);
673 u64_stats_inc(&dstats->tx_packets);
674 u64_stats_update_end(&dstats->syncp);
675 put_cpu_ptr(dstats);
676 return;
677 }
678 if (dev->pcpu_stat_type == NETDEV_PCPU_STAT_TSTATS) {
679 struct pcpu_sw_netstats *tstats = get_cpu_ptr(dev->tstats);
680
681 u64_stats_update_begin(&tstats->syncp);
682 u64_stats_add(&tstats->tx_bytes, pkt_len);
683 u64_stats_inc(&tstats->tx_packets);
684 u64_stats_update_end(&tstats->syncp);
685 put_cpu_ptr(tstats);
686 return;
687 }
688 pr_err_once("iptunnel_xmit_stats pcpu_stat_type=%d\n",
689 dev->pcpu_stat_type);
690 WARN_ON_ONCE(1);
691 return;
692 }
693
694 if (pkt_len < 0) {
695 DEV_STATS_INC(dev, tx_errors);
696 DEV_STATS_INC(dev, tx_aborted_errors);
697 } else {
698 DEV_STATS_INC(dev, tx_dropped);
699 }
700 }
701
ip_tunnel_info_opts_get(void * to,const struct ip_tunnel_info * info)702 static inline void ip_tunnel_info_opts_get(void *to,
703 const struct ip_tunnel_info *info)
704 {
705 memcpy(to, ip_tunnel_info_opts(info), info->options_len);
706 }
707
ip_tunnel_info_opts_set(struct ip_tunnel_info * info,const void * from,int len,const unsigned long * flags)708 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
709 const void *from, int len,
710 const unsigned long *flags)
711 {
712 info->options_len = len;
713 if (len > 0) {
714 memcpy(ip_tunnel_info_opts(info), from, len);
715 ip_tunnel_flags_or(info->key.tun_flags, info->key.tun_flags,
716 flags);
717 }
718 }
719
lwt_tun_info(struct lwtunnel_state * lwtstate)720 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
721 {
722 return (struct ip_tunnel_info *)lwtstate->data;
723 }
724
725 DECLARE_STATIC_KEY_FALSE(ip_tunnel_metadata_cnt);
726
727 /* Returns > 0 if metadata should be collected */
ip_tunnel_collect_metadata(void)728 static inline int ip_tunnel_collect_metadata(void)
729 {
730 return static_branch_unlikely(&ip_tunnel_metadata_cnt);
731 }
732
733 void __init ip_tunnel_core_init(void);
734
735 void ip_tunnel_need_metadata(void);
736 void ip_tunnel_unneed_metadata(void);
737
738 #else /* CONFIG_INET */
739
lwt_tun_info(struct lwtunnel_state * lwtstate)740 static inline struct ip_tunnel_info *lwt_tun_info(struct lwtunnel_state *lwtstate)
741 {
742 return NULL;
743 }
744
ip_tunnel_need_metadata(void)745 static inline void ip_tunnel_need_metadata(void)
746 {
747 }
748
ip_tunnel_unneed_metadata(void)749 static inline void ip_tunnel_unneed_metadata(void)
750 {
751 }
752
ip_tunnel_info_opts_get(void * to,const struct ip_tunnel_info * info)753 static inline void ip_tunnel_info_opts_get(void *to,
754 const struct ip_tunnel_info *info)
755 {
756 }
757
ip_tunnel_info_opts_set(struct ip_tunnel_info * info,const void * from,int len,const unsigned long * flags)758 static inline void ip_tunnel_info_opts_set(struct ip_tunnel_info *info,
759 const void *from, int len,
760 const unsigned long *flags)
761 {
762 info->options_len = 0;
763 }
764
765 #endif /* CONFIG_INET */
766
767 #endif /* __NET_IP_TUNNELS_H */
768