| 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 | |
| 50 | int iptunnel_xmit(struct sock *sk, struct rtable *rt, struct sk_buff *skb, |
| 51 | __be32 src, __be32 dst, __u8 proto, |
| 52 | __u8 tos, __u8 ttl, __be16 df, bool xnet) |
| 53 | { |
| 54 | int pkt_len = skb->len; |
| 55 | struct iphdr *iph; |
| 56 | int err; |
| 57 | |
| 58 | skb_scrub_packet(skb, xnet); |
| 59 | |
| 60 | skb_clear_hash(skb); |
| 61 | skb_dst_set(skb, &rt->dst); |
| 62 | memset(IPCB(skb), 0, sizeof(*IPCB(skb))); |
| 63 | |
| 64 | /* Push down and install the IP header. */ |
| 65 | skb_push(skb, sizeof(struct iphdr)); |
| 66 | skb_reset_network_header(skb); |
| 67 | |
| 68 | iph = ip_hdr(skb); |
| 69 | |
| 70 | iph->version = 4; |
| 71 | iph->ihl = sizeof(struct iphdr) >> 2; |
| 72 | iph->frag_off = df; |
| 73 | iph->protocol = proto; |
| 74 | iph->tos = tos; |
| 75 | iph->daddr = dst; |
| 76 | iph->saddr = src; |
| 77 | iph->ttl = ttl; |
| 78 | __ip_select_ident(dev_net(rt->dst.dev), iph, |
| 79 | skb_shinfo(skb)->gso_segs ?: 1); |
| 80 | |
| 81 | err = ip_local_out_sk(sk, skb); |
| 82 | if (unlikely(net_xmit_eval(err))) |
| 83 | pkt_len = 0; |
| 84 | return pkt_len; |
| 85 | } |
| 86 | EXPORT_SYMBOL_GPL(iptunnel_xmit); |
| 87 | |
| 88 | int iptunnel_pull_header(struct sk_buff *skb, int hdr_len, __be16 inner_proto) |
| 89 | { |
| 90 | if (unlikely(!pskb_may_pull(skb, hdr_len))) |
| 91 | return -ENOMEM; |
| 92 | |
| 93 | skb_pull_rcsum(skb, hdr_len); |
| 94 | |
| 95 | if (inner_proto == htons(ETH_P_TEB)) { |
| 96 | struct ethhdr *eh; |
| 97 | |
| 98 | if (unlikely(!pskb_may_pull(skb, ETH_HLEN))) |
| 99 | return -ENOMEM; |
| 100 | |
| 101 | eh = (struct ethhdr *)skb->data; |
| 102 | if (likely(eth_proto_is_802_3(eh->h_proto))) |
| 103 | skb->protocol = eh->h_proto; |
| 104 | else |
| 105 | skb->protocol = htons(ETH_P_802_2); |
| 106 | |
| 107 | } else { |
| 108 | skb->protocol = inner_proto; |
| 109 | } |
| 110 | |
| 111 | nf_reset(skb); |
| 112 | secpath_reset(skb); |
| 113 | skb_clear_hash_if_not_l4(skb); |
| 114 | skb_dst_drop(skb); |
| 115 | skb->vlan_tci = 0; |
| 116 | skb_set_queue_mapping(skb, 0); |
| 117 | skb->pkt_type = PACKET_HOST; |
| 118 | return 0; |
| 119 | } |
| 120 | EXPORT_SYMBOL_GPL(iptunnel_pull_header); |
| 121 | |
| 122 | struct sk_buff *iptunnel_handle_offloads(struct sk_buff *skb, |
| 123 | bool csum_help, |
| 124 | int gso_type_mask) |
| 125 | { |
| 126 | int err; |
| 127 | |
| 128 | if (likely(!skb->encapsulation)) { |
| 129 | skb_reset_inner_headers(skb); |
| 130 | skb->encapsulation = 1; |
| 131 | } |
| 132 | |
| 133 | if (skb_is_gso(skb)) { |
| 134 | err = skb_unclone(skb, GFP_ATOMIC); |
| 135 | if (unlikely(err)) |
| 136 | goto error; |
| 137 | skb_shinfo(skb)->gso_type |= gso_type_mask; |
| 138 | return skb; |
| 139 | } |
| 140 | |
| 141 | /* If packet is not gso and we are resolving any partial checksum, |
| 142 | * clear encapsulation flag. This allows setting CHECKSUM_PARTIAL |
| 143 | * on the outer header without confusing devices that implement |
| 144 | * NETIF_F_IP_CSUM with encapsulation. |
| 145 | */ |
| 146 | if (csum_help) |
| 147 | skb->encapsulation = 0; |
| 148 | |
| 149 | if (skb->ip_summed == CHECKSUM_PARTIAL && csum_help) { |
| 150 | err = skb_checksum_help(skb); |
| 151 | if (unlikely(err)) |
| 152 | goto error; |
| 153 | } else if (skb->ip_summed != CHECKSUM_PARTIAL) |
| 154 | skb->ip_summed = CHECKSUM_NONE; |
| 155 | |
| 156 | return skb; |
| 157 | error: |
| 158 | kfree_skb(skb); |
| 159 | return ERR_PTR(err); |
| 160 | } |
| 161 | EXPORT_SYMBOL_GPL(iptunnel_handle_offloads); |
| 162 | |
| 163 | /* Often modified stats are per cpu, other are shared (netdev->stats) */ |
| 164 | struct rtnl_link_stats64 *ip_tunnel_get_stats64(struct net_device *dev, |
| 165 | struct rtnl_link_stats64 *tot) |
| 166 | { |
| 167 | int i; |
| 168 | |
| 169 | netdev_stats_to_stats64(tot, &dev->stats); |
| 170 | |
| 171 | for_each_possible_cpu(i) { |
| 172 | const struct pcpu_sw_netstats *tstats = |
| 173 | per_cpu_ptr(dev->tstats, i); |
| 174 | u64 rx_packets, rx_bytes, tx_packets, tx_bytes; |
| 175 | unsigned int start; |
| 176 | |
| 177 | do { |
| 178 | start = u64_stats_fetch_begin_irq(&tstats->syncp); |
| 179 | rx_packets = tstats->rx_packets; |
| 180 | tx_packets = tstats->tx_packets; |
| 181 | rx_bytes = tstats->rx_bytes; |
| 182 | tx_bytes = tstats->tx_bytes; |
| 183 | } while (u64_stats_fetch_retry_irq(&tstats->syncp, start)); |
| 184 | |
| 185 | tot->rx_packets += rx_packets; |
| 186 | tot->tx_packets += tx_packets; |
| 187 | tot->rx_bytes += rx_bytes; |
| 188 | tot->tx_bytes += tx_bytes; |
| 189 | } |
| 190 | |
| 191 | return tot; |
| 192 | } |
| 193 | EXPORT_SYMBOL_GPL(ip_tunnel_get_stats64); |
| 194 | |
| 195 | static const struct nla_policy ip_tun_policy[LWTUNNEL_IP_MAX + 1] = { |
| 196 | [LWTUNNEL_IP_ID] = { .type = NLA_U64 }, |
| 197 | [LWTUNNEL_IP_DST] = { .type = NLA_U32 }, |
| 198 | [LWTUNNEL_IP_SRC] = { .type = NLA_U32 }, |
| 199 | [LWTUNNEL_IP_TTL] = { .type = NLA_U8 }, |
| 200 | [LWTUNNEL_IP_TOS] = { .type = NLA_U8 }, |
| 201 | [LWTUNNEL_IP_SPORT] = { .type = NLA_U16 }, |
| 202 | [LWTUNNEL_IP_DPORT] = { .type = NLA_U16 }, |
| 203 | [LWTUNNEL_IP_FLAGS] = { .type = NLA_U16 }, |
| 204 | }; |
| 205 | |
| 206 | static int ip_tun_build_state(struct net_device *dev, struct nlattr *attr, |
| 207 | unsigned int family, const void *cfg, |
| 208 | struct lwtunnel_state **ts) |
| 209 | { |
| 210 | struct ip_tunnel_info *tun_info; |
| 211 | struct lwtunnel_state *new_state; |
| 212 | struct nlattr *tb[LWTUNNEL_IP_MAX + 1]; |
| 213 | int err; |
| 214 | |
| 215 | err = nla_parse_nested(tb, LWTUNNEL_IP_MAX, attr, ip_tun_policy); |
| 216 | if (err < 0) |
| 217 | return err; |
| 218 | |
| 219 | new_state = lwtunnel_state_alloc(sizeof(*tun_info)); |
| 220 | if (!new_state) |
| 221 | return -ENOMEM; |
| 222 | |
| 223 | new_state->type = LWTUNNEL_ENCAP_IP; |
| 224 | |
| 225 | tun_info = lwt_tun_info(new_state); |
| 226 | |
| 227 | if (tb[LWTUNNEL_IP_ID]) |
| 228 | tun_info->key.tun_id = nla_get_u64(tb[LWTUNNEL_IP_ID]); |
| 229 | |
| 230 | if (tb[LWTUNNEL_IP_DST]) |
| 231 | tun_info->key.u.ipv4.dst = nla_get_be32(tb[LWTUNNEL_IP_DST]); |
| 232 | |
| 233 | if (tb[LWTUNNEL_IP_SRC]) |
| 234 | tun_info->key.u.ipv4.src = nla_get_be32(tb[LWTUNNEL_IP_SRC]); |
| 235 | |
| 236 | if (tb[LWTUNNEL_IP_TTL]) |
| 237 | tun_info->key.ttl = nla_get_u8(tb[LWTUNNEL_IP_TTL]); |
| 238 | |
| 239 | if (tb[LWTUNNEL_IP_TOS]) |
| 240 | tun_info->key.tos = nla_get_u8(tb[LWTUNNEL_IP_TOS]); |
| 241 | |
| 242 | if (tb[LWTUNNEL_IP_SPORT]) |
| 243 | tun_info->key.tp_src = nla_get_be16(tb[LWTUNNEL_IP_SPORT]); |
| 244 | |
| 245 | if (tb[LWTUNNEL_IP_DPORT]) |
| 246 | tun_info->key.tp_dst = nla_get_be16(tb[LWTUNNEL_IP_DPORT]); |
| 247 | |
| 248 | if (tb[LWTUNNEL_IP_FLAGS]) |
| 249 | tun_info->key.tun_flags = nla_get_u16(tb[LWTUNNEL_IP_FLAGS]); |
| 250 | |
| 251 | tun_info->mode = IP_TUNNEL_INFO_TX; |
| 252 | tun_info->options_len = 0; |
| 253 | |
| 254 | *ts = new_state; |
| 255 | |
| 256 | return 0; |
| 257 | } |
| 258 | |
| 259 | static int ip_tun_fill_encap_info(struct sk_buff *skb, |
| 260 | struct lwtunnel_state *lwtstate) |
| 261 | { |
| 262 | struct ip_tunnel_info *tun_info = lwt_tun_info(lwtstate); |
| 263 | |
| 264 | if (nla_put_u64(skb, LWTUNNEL_IP_ID, tun_info->key.tun_id) || |
| 265 | nla_put_be32(skb, LWTUNNEL_IP_DST, tun_info->key.u.ipv4.dst) || |
| 266 | nla_put_be32(skb, LWTUNNEL_IP_SRC, tun_info->key.u.ipv4.src) || |
| 267 | nla_put_u8(skb, LWTUNNEL_IP_TOS, tun_info->key.tos) || |
| 268 | nla_put_u8(skb, LWTUNNEL_IP_TTL, tun_info->key.ttl) || |
| 269 | nla_put_u16(skb, LWTUNNEL_IP_SPORT, tun_info->key.tp_src) || |
| 270 | nla_put_u16(skb, LWTUNNEL_IP_DPORT, tun_info->key.tp_dst) || |
| 271 | nla_put_u16(skb, LWTUNNEL_IP_FLAGS, tun_info->key.tun_flags)) |
| 272 | return -ENOMEM; |
| 273 | |
| 274 | return 0; |
| 275 | } |
| 276 | |
| 277 | static int ip_tun_encap_nlsize(struct lwtunnel_state *lwtstate) |
| 278 | { |
| 279 | return nla_total_size(8) /* LWTUNNEL_IP_ID */ |
| 280 | + nla_total_size(4) /* LWTUNNEL_IP_DST */ |
| 281 | + nla_total_size(4) /* LWTUNNEL_IP_SRC */ |
| 282 | + nla_total_size(1) /* LWTUNNEL_IP_TOS */ |
| 283 | + nla_total_size(1) /* LWTUNNEL_IP_TTL */ |
| 284 | + nla_total_size(2) /* LWTUNNEL_IP_SPORT */ |
| 285 | + nla_total_size(2) /* LWTUNNEL_IP_DPORT */ |
| 286 | + nla_total_size(2); /* LWTUNNEL_IP_FLAGS */ |
| 287 | } |
| 288 | |
| 289 | static int ip_tun_cmp_encap(struct lwtunnel_state *a, struct lwtunnel_state *b) |
| 290 | { |
| 291 | return memcmp(lwt_tun_info(a), lwt_tun_info(b), |
| 292 | sizeof(struct ip_tunnel_info)); |
| 293 | } |
| 294 | |
| 295 | static const struct lwtunnel_encap_ops ip_tun_lwt_ops = { |
| 296 | .build_state = ip_tun_build_state, |
| 297 | .fill_encap = ip_tun_fill_encap_info, |
| 298 | .get_encap_size = ip_tun_encap_nlsize, |
| 299 | .cmp_encap = ip_tun_cmp_encap, |
| 300 | }; |
| 301 | |
| 302 | static const struct nla_policy ip6_tun_policy[LWTUNNEL_IP6_MAX + 1] = { |
| 303 | [LWTUNNEL_IP6_ID] = { .type = NLA_U64 }, |
| 304 | [LWTUNNEL_IP6_DST] = { .len = sizeof(struct in6_addr) }, |
| 305 | [LWTUNNEL_IP6_SRC] = { .len = sizeof(struct in6_addr) }, |
| 306 | [LWTUNNEL_IP6_HOPLIMIT] = { .type = NLA_U8 }, |
| 307 | [LWTUNNEL_IP6_TC] = { .type = NLA_U8 }, |
| 308 | [LWTUNNEL_IP6_SPORT] = { .type = NLA_U16 }, |
| 309 | [LWTUNNEL_IP6_DPORT] = { .type = NLA_U16 }, |
| 310 | [LWTUNNEL_IP6_FLAGS] = { .type = NLA_U16 }, |
| 311 | }; |
| 312 | |
| 313 | static int ip6_tun_build_state(struct net_device *dev, struct nlattr *attr, |
| 314 | unsigned int family, const void *cfg, |
| 315 | struct lwtunnel_state **ts) |
| 316 | { |
| 317 | struct ip_tunnel_info *tun_info; |
| 318 | struct lwtunnel_state *new_state; |
| 319 | struct nlattr *tb[LWTUNNEL_IP6_MAX + 1]; |
| 320 | int err; |
| 321 | |
| 322 | err = nla_parse_nested(tb, LWTUNNEL_IP6_MAX, attr, ip6_tun_policy); |
| 323 | if (err < 0) |
| 324 | return err; |
| 325 | |
| 326 | new_state = lwtunnel_state_alloc(sizeof(*tun_info)); |
| 327 | if (!new_state) |
| 328 | return -ENOMEM; |
| 329 | |
| 330 | new_state->type = LWTUNNEL_ENCAP_IP6; |
| 331 | |
| 332 | tun_info = lwt_tun_info(new_state); |
| 333 | |
| 334 | if (tb[LWTUNNEL_IP6_ID]) |
| 335 | tun_info->key.tun_id = nla_get_u64(tb[LWTUNNEL_IP6_ID]); |
| 336 | |
| 337 | if (tb[LWTUNNEL_IP6_DST]) |
| 338 | tun_info->key.u.ipv6.dst = nla_get_in6_addr(tb[LWTUNNEL_IP6_DST]); |
| 339 | |
| 340 | if (tb[LWTUNNEL_IP6_SRC]) |
| 341 | tun_info->key.u.ipv6.src = nla_get_in6_addr(tb[LWTUNNEL_IP6_SRC]); |
| 342 | |
| 343 | if (tb[LWTUNNEL_IP6_HOPLIMIT]) |
| 344 | tun_info->key.ttl = nla_get_u8(tb[LWTUNNEL_IP6_HOPLIMIT]); |
| 345 | |
| 346 | if (tb[LWTUNNEL_IP6_TC]) |
| 347 | tun_info->key.tos = nla_get_u8(tb[LWTUNNEL_IP6_TC]); |
| 348 | |
| 349 | if (tb[LWTUNNEL_IP6_SPORT]) |
| 350 | tun_info->key.tp_src = nla_get_be16(tb[LWTUNNEL_IP6_SPORT]); |
| 351 | |
| 352 | if (tb[LWTUNNEL_IP6_DPORT]) |
| 353 | tun_info->key.tp_dst = nla_get_be16(tb[LWTUNNEL_IP6_DPORT]); |
| 354 | |
| 355 | if (tb[LWTUNNEL_IP6_FLAGS]) |
| 356 | tun_info->key.tun_flags = nla_get_u16(tb[LWTUNNEL_IP6_FLAGS]); |
| 357 | |
| 358 | tun_info->mode = IP_TUNNEL_INFO_TX | IP_TUNNEL_INFO_IPV6; |
| 359 | tun_info->options_len = 0; |
| 360 | |
| 361 | *ts = new_state; |
| 362 | |
| 363 | return 0; |
| 364 | } |
| 365 | |
| 366 | static int ip6_tun_fill_encap_info(struct sk_buff *skb, |
| 367 | struct lwtunnel_state *lwtstate) |
| 368 | { |
| 369 | struct ip_tunnel_info *tun_info = lwt_tun_info(lwtstate); |
| 370 | |
| 371 | if (nla_put_u64(skb, LWTUNNEL_IP6_ID, tun_info->key.tun_id) || |
| 372 | nla_put_in6_addr(skb, LWTUNNEL_IP6_DST, &tun_info->key.u.ipv6.dst) || |
| 373 | nla_put_in6_addr(skb, LWTUNNEL_IP6_SRC, &tun_info->key.u.ipv6.src) || |
| 374 | nla_put_u8(skb, LWTUNNEL_IP6_HOPLIMIT, tun_info->key.tos) || |
| 375 | nla_put_u8(skb, LWTUNNEL_IP6_TC, tun_info->key.ttl) || |
| 376 | nla_put_u16(skb, LWTUNNEL_IP6_SPORT, tun_info->key.tp_src) || |
| 377 | nla_put_u16(skb, LWTUNNEL_IP6_DPORT, tun_info->key.tp_dst) || |
| 378 | nla_put_u16(skb, LWTUNNEL_IP6_FLAGS, tun_info->key.tun_flags)) |
| 379 | return -ENOMEM; |
| 380 | |
| 381 | return 0; |
| 382 | } |
| 383 | |
| 384 | static int ip6_tun_encap_nlsize(struct lwtunnel_state *lwtstate) |
| 385 | { |
| 386 | return nla_total_size(8) /* LWTUNNEL_IP6_ID */ |
| 387 | + nla_total_size(16) /* LWTUNNEL_IP6_DST */ |
| 388 | + nla_total_size(16) /* LWTUNNEL_IP6_SRC */ |
| 389 | + nla_total_size(1) /* LWTUNNEL_IP6_HOPLIMIT */ |
| 390 | + nla_total_size(1) /* LWTUNNEL_IP6_TC */ |
| 391 | + nla_total_size(2) /* LWTUNNEL_IP6_SPORT */ |
| 392 | + nla_total_size(2) /* LWTUNNEL_IP6_DPORT */ |
| 393 | + nla_total_size(2); /* LWTUNNEL_IP6_FLAGS */ |
| 394 | } |
| 395 | |
| 396 | static const struct lwtunnel_encap_ops ip6_tun_lwt_ops = { |
| 397 | .build_state = ip6_tun_build_state, |
| 398 | .fill_encap = ip6_tun_fill_encap_info, |
| 399 | .get_encap_size = ip6_tun_encap_nlsize, |
| 400 | .cmp_encap = ip_tun_cmp_encap, |
| 401 | }; |
| 402 | |
| 403 | void __init ip_tunnel_core_init(void) |
| 404 | { |
| 405 | lwtunnel_encap_add_ops(&ip_tun_lwt_ops, LWTUNNEL_ENCAP_IP); |
| 406 | lwtunnel_encap_add_ops(&ip6_tun_lwt_ops, LWTUNNEL_ENCAP_IP6); |
| 407 | } |
| 408 | |
| 409 | struct static_key ip_tunnel_metadata_cnt = STATIC_KEY_INIT_FALSE; |
| 410 | EXPORT_SYMBOL(ip_tunnel_metadata_cnt); |
| 411 | |
| 412 | void ip_tunnel_need_metadata(void) |
| 413 | { |
| 414 | static_key_slow_inc(&ip_tunnel_metadata_cnt); |
| 415 | } |
| 416 | EXPORT_SYMBOL_GPL(ip_tunnel_need_metadata); |
| 417 | |
| 418 | void ip_tunnel_unneed_metadata(void) |
| 419 | { |
| 420 | static_key_slow_dec(&ip_tunnel_metadata_cnt); |
| 421 | } |
| 422 | EXPORT_SYMBOL_GPL(ip_tunnel_unneed_metadata); |