rjw | 1f88458 | 2022-01-06 17:20:42 +0800 | [diff] [blame^] | 1 | /* |
| 2 | * Linux NET3: IP/IP protocol decoder modified to support |
| 3 | * virtual tunnel interface |
| 4 | * |
| 5 | * Authors: |
| 6 | * Saurabh Mohan (saurabh.mohan@vyatta.com) 05/07/2012 |
| 7 | * |
| 8 | * This program is free software; you can redistribute it and/or |
| 9 | * modify it under the terms of the GNU General Public License |
| 10 | * as published by the Free Software Foundation; either version |
| 11 | * 2 of the License, or (at your option) any later version. |
| 12 | * |
| 13 | */ |
| 14 | |
| 15 | /* |
| 16 | This version of net/ipv4/ip_vti.c is cloned of net/ipv4/ipip.c |
| 17 | |
| 18 | For comments look at net/ipv4/ip_gre.c --ANK |
| 19 | */ |
| 20 | |
| 21 | |
| 22 | #include <linux/capability.h> |
| 23 | #include <linux/module.h> |
| 24 | #include <linux/types.h> |
| 25 | #include <linux/kernel.h> |
| 26 | #include <linux/uaccess.h> |
| 27 | #include <linux/skbuff.h> |
| 28 | #include <linux/netdevice.h> |
| 29 | #include <linux/in.h> |
| 30 | #include <linux/tcp.h> |
| 31 | #include <linux/udp.h> |
| 32 | #include <linux/if_arp.h> |
| 33 | #include <linux/init.h> |
| 34 | #include <linux/netfilter_ipv4.h> |
| 35 | #include <linux/if_ether.h> |
| 36 | #include <linux/icmpv6.h> |
| 37 | |
| 38 | #include <net/sock.h> |
| 39 | #include <net/ip.h> |
| 40 | #include <net/icmp.h> |
| 41 | #include <net/ip_tunnels.h> |
| 42 | #include <net/inet_ecn.h> |
| 43 | #include <net/xfrm.h> |
| 44 | #include <net/net_namespace.h> |
| 45 | #include <net/netns/generic.h> |
| 46 | |
| 47 | static struct rtnl_link_ops vti_link_ops __read_mostly; |
| 48 | |
| 49 | static unsigned int vti_net_id __read_mostly; |
| 50 | static int vti_tunnel_init(struct net_device *dev); |
| 51 | |
| 52 | static int vti_input(struct sk_buff *skb, int nexthdr, __be32 spi, |
| 53 | int encap_type, bool update_skb_dev) |
| 54 | { |
| 55 | struct ip_tunnel *tunnel; |
| 56 | const struct iphdr *iph = ip_hdr(skb); |
| 57 | struct net *net = dev_net(skb->dev); |
| 58 | struct ip_tunnel_net *itn = net_generic(net, vti_net_id); |
| 59 | |
| 60 | tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY, |
| 61 | iph->saddr, iph->daddr, 0); |
| 62 | if (tunnel) { |
| 63 | if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) |
| 64 | goto drop; |
| 65 | |
| 66 | XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4 = tunnel; |
| 67 | |
| 68 | if (update_skb_dev) |
| 69 | skb->dev = tunnel->dev; |
| 70 | |
| 71 | return xfrm_input(skb, nexthdr, spi, encap_type); |
| 72 | } |
| 73 | |
| 74 | return -EINVAL; |
| 75 | drop: |
| 76 | kfree_skb(skb); |
| 77 | return 0; |
| 78 | } |
| 79 | |
| 80 | static int vti_input_proto(struct sk_buff *skb, int nexthdr, __be32 spi, |
| 81 | int encap_type) |
| 82 | { |
| 83 | return vti_input(skb, nexthdr, spi, encap_type, false); |
| 84 | } |
| 85 | |
| 86 | static int vti_rcv(struct sk_buff *skb, __be32 spi, bool update_skb_dev) |
| 87 | { |
| 88 | XFRM_SPI_SKB_CB(skb)->family = AF_INET; |
| 89 | XFRM_SPI_SKB_CB(skb)->daddroff = offsetof(struct iphdr, daddr); |
| 90 | |
| 91 | return vti_input(skb, ip_hdr(skb)->protocol, spi, 0, update_skb_dev); |
| 92 | } |
| 93 | |
| 94 | static int vti_rcv_proto(struct sk_buff *skb) |
| 95 | { |
| 96 | return vti_rcv(skb, 0, false); |
| 97 | } |
| 98 | |
| 99 | static int vti_rcv_tunnel(struct sk_buff *skb) |
| 100 | { |
| 101 | struct ip_tunnel_net *itn = net_generic(dev_net(skb->dev), vti_net_id); |
| 102 | const struct iphdr *iph = ip_hdr(skb); |
| 103 | struct ip_tunnel *tunnel; |
| 104 | |
| 105 | tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY, |
| 106 | iph->saddr, iph->daddr, 0); |
| 107 | if (tunnel) { |
| 108 | struct tnl_ptk_info tpi = { |
| 109 | .proto = htons(ETH_P_IP), |
| 110 | }; |
| 111 | |
| 112 | if (!xfrm4_policy_check(NULL, XFRM_POLICY_IN, skb)) |
| 113 | goto drop; |
| 114 | if (iptunnel_pull_header(skb, 0, tpi.proto, false)) |
| 115 | goto drop; |
| 116 | return ip_tunnel_rcv(tunnel, skb, &tpi, NULL, false); |
| 117 | } |
| 118 | |
| 119 | return -EINVAL; |
| 120 | drop: |
| 121 | kfree_skb(skb); |
| 122 | return 0; |
| 123 | } |
| 124 | |
| 125 | static int vti_rcv_cb(struct sk_buff *skb, int err) |
| 126 | { |
| 127 | unsigned short family; |
| 128 | struct net_device *dev; |
| 129 | struct pcpu_sw_netstats *tstats; |
| 130 | struct xfrm_state *x; |
| 131 | struct xfrm_mode *inner_mode; |
| 132 | struct ip_tunnel *tunnel = XFRM_TUNNEL_SKB_CB(skb)->tunnel.ip4; |
| 133 | u32 orig_mark = skb->mark; |
| 134 | int ret; |
| 135 | |
| 136 | if (!tunnel) |
| 137 | return 1; |
| 138 | |
| 139 | dev = tunnel->dev; |
| 140 | |
| 141 | if (err) { |
| 142 | dev->stats.rx_errors++; |
| 143 | dev->stats.rx_dropped++; |
| 144 | |
| 145 | return 0; |
| 146 | } |
| 147 | |
| 148 | x = xfrm_input_state(skb); |
| 149 | |
| 150 | inner_mode = x->inner_mode; |
| 151 | |
| 152 | if (x->sel.family == AF_UNSPEC) { |
| 153 | inner_mode = xfrm_ip2inner_mode(x, XFRM_MODE_SKB_CB(skb)->protocol); |
| 154 | if (inner_mode == NULL) { |
| 155 | XFRM_INC_STATS(dev_net(skb->dev), |
| 156 | LINUX_MIB_XFRMINSTATEMODEERROR); |
| 157 | return -EINVAL; |
| 158 | } |
| 159 | } |
| 160 | |
| 161 | family = inner_mode->afinfo->family; |
| 162 | |
| 163 | skb->mark = be32_to_cpu(tunnel->parms.i_key); |
| 164 | ret = xfrm_policy_check(NULL, XFRM_POLICY_IN, skb, family); |
| 165 | skb->mark = orig_mark; |
| 166 | |
| 167 | if (!ret) |
| 168 | return -EPERM; |
| 169 | |
| 170 | skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(skb->dev))); |
| 171 | skb->dev = dev; |
| 172 | |
| 173 | tstats = this_cpu_ptr(dev->tstats); |
| 174 | |
| 175 | u64_stats_update_begin(&tstats->syncp); |
| 176 | tstats->rx_packets++; |
| 177 | tstats->rx_bytes += skb->len; |
| 178 | u64_stats_update_end(&tstats->syncp); |
| 179 | |
| 180 | return 0; |
| 181 | } |
| 182 | |
| 183 | static bool vti_state_check(const struct xfrm_state *x, __be32 dst, __be32 src) |
| 184 | { |
| 185 | xfrm_address_t *daddr = (xfrm_address_t *)&dst; |
| 186 | xfrm_address_t *saddr = (xfrm_address_t *)&src; |
| 187 | |
| 188 | /* if there is no transform then this tunnel is not functional. |
| 189 | * Or if the xfrm is not mode tunnel. |
| 190 | */ |
| 191 | if (!x || x->props.mode != XFRM_MODE_TUNNEL || |
| 192 | x->props.family != AF_INET) |
| 193 | return false; |
| 194 | |
| 195 | if (!dst) |
| 196 | return xfrm_addr_equal(saddr, &x->props.saddr, AF_INET); |
| 197 | |
| 198 | if (!xfrm_state_addr_check(x, daddr, saddr, AF_INET)) |
| 199 | return false; |
| 200 | |
| 201 | return true; |
| 202 | } |
| 203 | |
| 204 | static netdev_tx_t vti_xmit(struct sk_buff *skb, struct net_device *dev, |
| 205 | struct flowi *fl) |
| 206 | { |
| 207 | struct ip_tunnel *tunnel = netdev_priv(dev); |
| 208 | struct ip_tunnel_parm *parms = &tunnel->parms; |
| 209 | struct dst_entry *dst = skb_dst(skb); |
| 210 | struct net_device *tdev; /* Device to other host */ |
| 211 | int pkt_len = skb->len; |
| 212 | int err; |
| 213 | int mtu; |
| 214 | |
| 215 | if (!dst) { |
| 216 | switch (skb->protocol) { |
| 217 | case htons(ETH_P_IP): { |
| 218 | struct rtable *rt; |
| 219 | |
| 220 | fl->u.ip4.flowi4_oif = dev->ifindex; |
| 221 | fl->u.ip4.flowi4_flags |= FLOWI_FLAG_ANYSRC; |
| 222 | rt = __ip_route_output_key(dev_net(dev), &fl->u.ip4); |
| 223 | if (IS_ERR(rt)) { |
| 224 | dev->stats.tx_carrier_errors++; |
| 225 | goto tx_error_icmp; |
| 226 | } |
| 227 | dst = &rt->dst; |
| 228 | skb_dst_set(skb, dst); |
| 229 | break; |
| 230 | } |
| 231 | #if IS_ENABLED(CONFIG_IPV6) |
| 232 | case htons(ETH_P_IPV6): |
| 233 | fl->u.ip6.flowi6_oif = dev->ifindex; |
| 234 | fl->u.ip6.flowi6_flags |= FLOWI_FLAG_ANYSRC; |
| 235 | dst = ip6_route_output(dev_net(dev), NULL, &fl->u.ip6); |
| 236 | if (dst->error) { |
| 237 | dst_release(dst); |
| 238 | dst = NULL; |
| 239 | dev->stats.tx_carrier_errors++; |
| 240 | goto tx_error_icmp; |
| 241 | } |
| 242 | skb_dst_set(skb, dst); |
| 243 | break; |
| 244 | #endif |
| 245 | default: |
| 246 | dev->stats.tx_carrier_errors++; |
| 247 | goto tx_error_icmp; |
| 248 | } |
| 249 | } |
| 250 | |
| 251 | dst_hold(dst); |
| 252 | dst = xfrm_lookup(tunnel->net, dst, fl, NULL, 0); |
| 253 | if (IS_ERR(dst)) { |
| 254 | dev->stats.tx_carrier_errors++; |
| 255 | goto tx_error_icmp; |
| 256 | } |
| 257 | |
| 258 | if (!vti_state_check(dst->xfrm, parms->iph.daddr, parms->iph.saddr)) { |
| 259 | dev->stats.tx_carrier_errors++; |
| 260 | dst_release(dst); |
| 261 | goto tx_error_icmp; |
| 262 | } |
| 263 | |
| 264 | tdev = dst->dev; |
| 265 | |
| 266 | if (tdev == dev) { |
| 267 | dst_release(dst); |
| 268 | dev->stats.collisions++; |
| 269 | goto tx_error; |
| 270 | } |
| 271 | |
| 272 | if (tunnel->err_count > 0) { |
| 273 | if (time_before(jiffies, |
| 274 | tunnel->err_time + IPTUNNEL_ERR_TIMEO)) { |
| 275 | tunnel->err_count--; |
| 276 | dst_link_failure(skb); |
| 277 | } else |
| 278 | tunnel->err_count = 0; |
| 279 | } |
| 280 | |
| 281 | mtu = dst_mtu(dst); |
| 282 | if (skb->len > mtu) { |
| 283 | skb_dst_update_pmtu_no_confirm(skb, mtu); |
| 284 | if (skb->protocol == htons(ETH_P_IP)) { |
| 285 | icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, |
| 286 | htonl(mtu)); |
| 287 | } else { |
| 288 | if (mtu < IPV6_MIN_MTU) |
| 289 | mtu = IPV6_MIN_MTU; |
| 290 | |
| 291 | icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu); |
| 292 | } |
| 293 | |
| 294 | dst_release(dst); |
| 295 | goto tx_error; |
| 296 | } |
| 297 | |
| 298 | skb_scrub_packet(skb, !net_eq(tunnel->net, dev_net(dev))); |
| 299 | skb_dst_set(skb, dst); |
| 300 | skb->dev = skb_dst(skb)->dev; |
| 301 | |
| 302 | err = dst_output(tunnel->net, skb->sk, skb); |
| 303 | if (net_xmit_eval(err) == 0) |
| 304 | err = pkt_len; |
| 305 | iptunnel_xmit_stats(dev, err); |
| 306 | return NETDEV_TX_OK; |
| 307 | |
| 308 | tx_error_icmp: |
| 309 | dst_link_failure(skb); |
| 310 | tx_error: |
| 311 | dev->stats.tx_errors++; |
| 312 | kfree_skb(skb); |
| 313 | return NETDEV_TX_OK; |
| 314 | } |
| 315 | |
| 316 | /* This function assumes it is being called from dev_queue_xmit() |
| 317 | * and that skb is filled properly by that function. |
| 318 | */ |
| 319 | static netdev_tx_t vti_tunnel_xmit(struct sk_buff *skb, struct net_device *dev) |
| 320 | { |
| 321 | struct ip_tunnel *tunnel = netdev_priv(dev); |
| 322 | struct flowi fl; |
| 323 | |
| 324 | memset(&fl, 0, sizeof(fl)); |
| 325 | |
| 326 | switch (skb->protocol) { |
| 327 | case htons(ETH_P_IP): |
| 328 | xfrm_decode_session(skb, &fl, AF_INET); |
| 329 | memset(IPCB(skb), 0, sizeof(*IPCB(skb))); |
| 330 | break; |
| 331 | case htons(ETH_P_IPV6): |
| 332 | xfrm_decode_session(skb, &fl, AF_INET6); |
| 333 | memset(IP6CB(skb), 0, sizeof(*IP6CB(skb))); |
| 334 | break; |
| 335 | default: |
| 336 | dev->stats.tx_errors++; |
| 337 | dev_kfree_skb(skb); |
| 338 | return NETDEV_TX_OK; |
| 339 | } |
| 340 | |
| 341 | /* override mark with tunnel output key */ |
| 342 | fl.flowi_mark = be32_to_cpu(tunnel->parms.o_key); |
| 343 | |
| 344 | return vti_xmit(skb, dev, &fl); |
| 345 | } |
| 346 | |
| 347 | static int vti4_err(struct sk_buff *skb, u32 info) |
| 348 | { |
| 349 | __be32 spi; |
| 350 | __u32 mark; |
| 351 | struct xfrm_state *x; |
| 352 | struct ip_tunnel *tunnel; |
| 353 | struct ip_esp_hdr *esph; |
| 354 | struct ip_auth_hdr *ah ; |
| 355 | struct ip_comp_hdr *ipch; |
| 356 | struct net *net = dev_net(skb->dev); |
| 357 | const struct iphdr *iph = (const struct iphdr *)skb->data; |
| 358 | int protocol = iph->protocol; |
| 359 | struct ip_tunnel_net *itn = net_generic(net, vti_net_id); |
| 360 | |
| 361 | tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, TUNNEL_NO_KEY, |
| 362 | iph->daddr, iph->saddr, 0); |
| 363 | if (!tunnel) |
| 364 | return -1; |
| 365 | |
| 366 | mark = be32_to_cpu(tunnel->parms.o_key); |
| 367 | |
| 368 | switch (protocol) { |
| 369 | case IPPROTO_ESP: |
| 370 | esph = (struct ip_esp_hdr *)(skb->data+(iph->ihl<<2)); |
| 371 | spi = esph->spi; |
| 372 | break; |
| 373 | case IPPROTO_AH: |
| 374 | ah = (struct ip_auth_hdr *)(skb->data+(iph->ihl<<2)); |
| 375 | spi = ah->spi; |
| 376 | break; |
| 377 | case IPPROTO_COMP: |
| 378 | ipch = (struct ip_comp_hdr *)(skb->data+(iph->ihl<<2)); |
| 379 | spi = htonl(ntohs(ipch->cpi)); |
| 380 | break; |
| 381 | default: |
| 382 | return 0; |
| 383 | } |
| 384 | |
| 385 | switch (icmp_hdr(skb)->type) { |
| 386 | case ICMP_DEST_UNREACH: |
| 387 | if (icmp_hdr(skb)->code != ICMP_FRAG_NEEDED) |
| 388 | return 0; |
| 389 | case ICMP_REDIRECT: |
| 390 | break; |
| 391 | default: |
| 392 | return 0; |
| 393 | } |
| 394 | |
| 395 | x = xfrm_state_lookup(net, mark, (const xfrm_address_t *)&iph->daddr, |
| 396 | spi, protocol, AF_INET); |
| 397 | if (!x) |
| 398 | return 0; |
| 399 | |
| 400 | if (icmp_hdr(skb)->type == ICMP_DEST_UNREACH) |
| 401 | ipv4_update_pmtu(skb, net, info, 0, 0, protocol, 0); |
| 402 | else |
| 403 | ipv4_redirect(skb, net, 0, 0, protocol, 0); |
| 404 | xfrm_state_put(x); |
| 405 | |
| 406 | return 0; |
| 407 | } |
| 408 | |
| 409 | static int |
| 410 | vti_tunnel_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd) |
| 411 | { |
| 412 | int err = 0; |
| 413 | struct ip_tunnel_parm p; |
| 414 | |
| 415 | if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) |
| 416 | return -EFAULT; |
| 417 | |
| 418 | if (cmd == SIOCADDTUNNEL || cmd == SIOCCHGTUNNEL) { |
| 419 | if (p.iph.version != 4 || p.iph.protocol != IPPROTO_IPIP || |
| 420 | p.iph.ihl != 5) |
| 421 | return -EINVAL; |
| 422 | } |
| 423 | |
| 424 | if (!(p.i_flags & GRE_KEY)) |
| 425 | p.i_key = 0; |
| 426 | if (!(p.o_flags & GRE_KEY)) |
| 427 | p.o_key = 0; |
| 428 | |
| 429 | p.i_flags = VTI_ISVTI; |
| 430 | |
| 431 | err = ip_tunnel_ioctl(dev, &p, cmd); |
| 432 | if (err) |
| 433 | return err; |
| 434 | |
| 435 | if (cmd != SIOCDELTUNNEL) { |
| 436 | p.i_flags |= GRE_KEY; |
| 437 | p.o_flags |= GRE_KEY; |
| 438 | } |
| 439 | |
| 440 | if (copy_to_user(ifr->ifr_ifru.ifru_data, &p, sizeof(p))) |
| 441 | return -EFAULT; |
| 442 | return 0; |
| 443 | } |
| 444 | |
| 445 | static const struct net_device_ops vti_netdev_ops = { |
| 446 | .ndo_init = vti_tunnel_init, |
| 447 | .ndo_uninit = ip_tunnel_uninit, |
| 448 | .ndo_start_xmit = vti_tunnel_xmit, |
| 449 | .ndo_do_ioctl = vti_tunnel_ioctl, |
| 450 | .ndo_change_mtu = ip_tunnel_change_mtu, |
| 451 | .ndo_get_stats64 = ip_tunnel_get_stats64, |
| 452 | .ndo_get_iflink = ip_tunnel_get_iflink, |
| 453 | }; |
| 454 | |
| 455 | static void vti_tunnel_setup(struct net_device *dev) |
| 456 | { |
| 457 | dev->netdev_ops = &vti_netdev_ops; |
| 458 | dev->type = ARPHRD_TUNNEL; |
| 459 | ip_tunnel_setup(dev, vti_net_id); |
| 460 | } |
| 461 | |
| 462 | static int vti_tunnel_init(struct net_device *dev) |
| 463 | { |
| 464 | struct ip_tunnel *tunnel = netdev_priv(dev); |
| 465 | struct iphdr *iph = &tunnel->parms.iph; |
| 466 | |
| 467 | memcpy(dev->dev_addr, &iph->saddr, 4); |
| 468 | memcpy(dev->broadcast, &iph->daddr, 4); |
| 469 | |
| 470 | dev->mtu = ETH_DATA_LEN; |
| 471 | dev->flags = IFF_NOARP; |
| 472 | dev->addr_len = 4; |
| 473 | dev->features |= NETIF_F_LLTX; |
| 474 | netif_keep_dst(dev); |
| 475 | |
| 476 | return ip_tunnel_init(dev); |
| 477 | } |
| 478 | |
| 479 | static void __net_init vti_fb_tunnel_init(struct net_device *dev) |
| 480 | { |
| 481 | struct ip_tunnel *tunnel = netdev_priv(dev); |
| 482 | struct iphdr *iph = &tunnel->parms.iph; |
| 483 | |
| 484 | iph->version = 4; |
| 485 | iph->protocol = IPPROTO_IPIP; |
| 486 | iph->ihl = 5; |
| 487 | } |
| 488 | |
| 489 | static struct xfrm4_protocol vti_esp4_protocol __read_mostly = { |
| 490 | .handler = vti_rcv_proto, |
| 491 | .input_handler = vti_input_proto, |
| 492 | .cb_handler = vti_rcv_cb, |
| 493 | .err_handler = vti4_err, |
| 494 | .priority = 100, |
| 495 | }; |
| 496 | |
| 497 | static struct xfrm4_protocol vti_ah4_protocol __read_mostly = { |
| 498 | .handler = vti_rcv_proto, |
| 499 | .input_handler = vti_input_proto, |
| 500 | .cb_handler = vti_rcv_cb, |
| 501 | .err_handler = vti4_err, |
| 502 | .priority = 100, |
| 503 | }; |
| 504 | |
| 505 | static struct xfrm4_protocol vti_ipcomp4_protocol __read_mostly = { |
| 506 | .handler = vti_rcv_proto, |
| 507 | .input_handler = vti_input_proto, |
| 508 | .cb_handler = vti_rcv_cb, |
| 509 | .err_handler = vti4_err, |
| 510 | .priority = 100, |
| 511 | }; |
| 512 | |
| 513 | static struct xfrm_tunnel ipip_handler __read_mostly = { |
| 514 | .handler = vti_rcv_tunnel, |
| 515 | .err_handler = vti4_err, |
| 516 | .priority = 0, |
| 517 | }; |
| 518 | |
| 519 | static int __net_init vti_init_net(struct net *net) |
| 520 | { |
| 521 | int err; |
| 522 | struct ip_tunnel_net *itn; |
| 523 | |
| 524 | err = ip_tunnel_init_net(net, vti_net_id, &vti_link_ops, "ip_vti0"); |
| 525 | if (err) |
| 526 | return err; |
| 527 | itn = net_generic(net, vti_net_id); |
| 528 | vti_fb_tunnel_init(itn->fb_tunnel_dev); |
| 529 | return 0; |
| 530 | } |
| 531 | |
| 532 | static void __net_exit vti_exit_net(struct net *net) |
| 533 | { |
| 534 | struct ip_tunnel_net *itn = net_generic(net, vti_net_id); |
| 535 | ip_tunnel_delete_net(itn, &vti_link_ops); |
| 536 | } |
| 537 | |
| 538 | static struct pernet_operations vti_net_ops = { |
| 539 | .init = vti_init_net, |
| 540 | .exit = vti_exit_net, |
| 541 | .id = &vti_net_id, |
| 542 | .size = sizeof(struct ip_tunnel_net), |
| 543 | }; |
| 544 | |
| 545 | static int vti_tunnel_validate(struct nlattr *tb[], struct nlattr *data[], |
| 546 | struct netlink_ext_ack *extack) |
| 547 | { |
| 548 | return 0; |
| 549 | } |
| 550 | |
| 551 | static void vti_netlink_parms(struct nlattr *data[], |
| 552 | struct ip_tunnel_parm *parms, |
| 553 | __u32 *fwmark) |
| 554 | { |
| 555 | memset(parms, 0, sizeof(*parms)); |
| 556 | |
| 557 | parms->iph.protocol = IPPROTO_IPIP; |
| 558 | |
| 559 | if (!data) |
| 560 | return; |
| 561 | |
| 562 | parms->i_flags = VTI_ISVTI; |
| 563 | |
| 564 | if (data[IFLA_VTI_LINK]) |
| 565 | parms->link = nla_get_u32(data[IFLA_VTI_LINK]); |
| 566 | |
| 567 | if (data[IFLA_VTI_IKEY]) |
| 568 | parms->i_key = nla_get_be32(data[IFLA_VTI_IKEY]); |
| 569 | |
| 570 | if (data[IFLA_VTI_OKEY]) |
| 571 | parms->o_key = nla_get_be32(data[IFLA_VTI_OKEY]); |
| 572 | |
| 573 | if (data[IFLA_VTI_LOCAL]) |
| 574 | parms->iph.saddr = nla_get_in_addr(data[IFLA_VTI_LOCAL]); |
| 575 | |
| 576 | if (data[IFLA_VTI_REMOTE]) |
| 577 | parms->iph.daddr = nla_get_in_addr(data[IFLA_VTI_REMOTE]); |
| 578 | |
| 579 | if (data[IFLA_VTI_FWMARK]) |
| 580 | *fwmark = nla_get_u32(data[IFLA_VTI_FWMARK]); |
| 581 | } |
| 582 | |
| 583 | static int vti_newlink(struct net *src_net, struct net_device *dev, |
| 584 | struct nlattr *tb[], struct nlattr *data[], |
| 585 | struct netlink_ext_ack *extack) |
| 586 | { |
| 587 | struct ip_tunnel_parm parms; |
| 588 | __u32 fwmark = 0; |
| 589 | |
| 590 | vti_netlink_parms(data, &parms, &fwmark); |
| 591 | return ip_tunnel_newlink(dev, tb, &parms, fwmark); |
| 592 | } |
| 593 | |
| 594 | static int vti_changelink(struct net_device *dev, struct nlattr *tb[], |
| 595 | struct nlattr *data[], |
| 596 | struct netlink_ext_ack *extack) |
| 597 | { |
| 598 | struct ip_tunnel *t = netdev_priv(dev); |
| 599 | __u32 fwmark = t->fwmark; |
| 600 | struct ip_tunnel_parm p; |
| 601 | |
| 602 | vti_netlink_parms(data, &p, &fwmark); |
| 603 | return ip_tunnel_changelink(dev, tb, &p, fwmark); |
| 604 | } |
| 605 | |
| 606 | static size_t vti_get_size(const struct net_device *dev) |
| 607 | { |
| 608 | return |
| 609 | /* IFLA_VTI_LINK */ |
| 610 | nla_total_size(4) + |
| 611 | /* IFLA_VTI_IKEY */ |
| 612 | nla_total_size(4) + |
| 613 | /* IFLA_VTI_OKEY */ |
| 614 | nla_total_size(4) + |
| 615 | /* IFLA_VTI_LOCAL */ |
| 616 | nla_total_size(4) + |
| 617 | /* IFLA_VTI_REMOTE */ |
| 618 | nla_total_size(4) + |
| 619 | /* IFLA_VTI_FWMARK */ |
| 620 | nla_total_size(4) + |
| 621 | 0; |
| 622 | } |
| 623 | |
| 624 | static int vti_fill_info(struct sk_buff *skb, const struct net_device *dev) |
| 625 | { |
| 626 | struct ip_tunnel *t = netdev_priv(dev); |
| 627 | struct ip_tunnel_parm *p = &t->parms; |
| 628 | |
| 629 | if (nla_put_u32(skb, IFLA_VTI_LINK, p->link) || |
| 630 | nla_put_be32(skb, IFLA_VTI_IKEY, p->i_key) || |
| 631 | nla_put_be32(skb, IFLA_VTI_OKEY, p->o_key) || |
| 632 | nla_put_in_addr(skb, IFLA_VTI_LOCAL, p->iph.saddr) || |
| 633 | nla_put_in_addr(skb, IFLA_VTI_REMOTE, p->iph.daddr) || |
| 634 | nla_put_u32(skb, IFLA_VTI_FWMARK, t->fwmark)) |
| 635 | return -EMSGSIZE; |
| 636 | |
| 637 | return 0; |
| 638 | } |
| 639 | |
| 640 | static const struct nla_policy vti_policy[IFLA_VTI_MAX + 1] = { |
| 641 | [IFLA_VTI_LINK] = { .type = NLA_U32 }, |
| 642 | [IFLA_VTI_IKEY] = { .type = NLA_U32 }, |
| 643 | [IFLA_VTI_OKEY] = { .type = NLA_U32 }, |
| 644 | [IFLA_VTI_LOCAL] = { .len = FIELD_SIZEOF(struct iphdr, saddr) }, |
| 645 | [IFLA_VTI_REMOTE] = { .len = FIELD_SIZEOF(struct iphdr, daddr) }, |
| 646 | [IFLA_VTI_FWMARK] = { .type = NLA_U32 }, |
| 647 | }; |
| 648 | |
| 649 | static struct rtnl_link_ops vti_link_ops __read_mostly = { |
| 650 | .kind = "vti", |
| 651 | .maxtype = IFLA_VTI_MAX, |
| 652 | .policy = vti_policy, |
| 653 | .priv_size = sizeof(struct ip_tunnel), |
| 654 | .setup = vti_tunnel_setup, |
| 655 | .validate = vti_tunnel_validate, |
| 656 | .newlink = vti_newlink, |
| 657 | .changelink = vti_changelink, |
| 658 | .dellink = ip_tunnel_dellink, |
| 659 | .get_size = vti_get_size, |
| 660 | .fill_info = vti_fill_info, |
| 661 | .get_link_net = ip_tunnel_get_link_net, |
| 662 | }; |
| 663 | |
| 664 | static int __init vti_init(void) |
| 665 | { |
| 666 | const char *msg; |
| 667 | int err; |
| 668 | |
| 669 | pr_info("IPv4 over IPsec tunneling driver\n"); |
| 670 | |
| 671 | msg = "tunnel device"; |
| 672 | err = register_pernet_device(&vti_net_ops); |
| 673 | if (err < 0) |
| 674 | goto pernet_dev_failed; |
| 675 | |
| 676 | msg = "tunnel protocols"; |
| 677 | err = xfrm4_protocol_register(&vti_esp4_protocol, IPPROTO_ESP); |
| 678 | if (err < 0) |
| 679 | goto xfrm_proto_esp_failed; |
| 680 | err = xfrm4_protocol_register(&vti_ah4_protocol, IPPROTO_AH); |
| 681 | if (err < 0) |
| 682 | goto xfrm_proto_ah_failed; |
| 683 | err = xfrm4_protocol_register(&vti_ipcomp4_protocol, IPPROTO_COMP); |
| 684 | if (err < 0) |
| 685 | goto xfrm_proto_comp_failed; |
| 686 | |
| 687 | msg = "ipip tunnel"; |
| 688 | err = xfrm4_tunnel_register(&ipip_handler, AF_INET); |
| 689 | if (err < 0) |
| 690 | goto xfrm_tunnel_failed; |
| 691 | |
| 692 | msg = "netlink interface"; |
| 693 | err = rtnl_link_register(&vti_link_ops); |
| 694 | if (err < 0) |
| 695 | goto rtnl_link_failed; |
| 696 | |
| 697 | return err; |
| 698 | |
| 699 | rtnl_link_failed: |
| 700 | xfrm4_tunnel_deregister(&ipip_handler, AF_INET); |
| 701 | xfrm_tunnel_failed: |
| 702 | xfrm4_protocol_deregister(&vti_ipcomp4_protocol, IPPROTO_COMP); |
| 703 | xfrm_proto_comp_failed: |
| 704 | xfrm4_protocol_deregister(&vti_ah4_protocol, IPPROTO_AH); |
| 705 | xfrm_proto_ah_failed: |
| 706 | xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP); |
| 707 | xfrm_proto_esp_failed: |
| 708 | unregister_pernet_device(&vti_net_ops); |
| 709 | pernet_dev_failed: |
| 710 | pr_err("vti init: failed to register %s\n", msg); |
| 711 | return err; |
| 712 | } |
| 713 | |
| 714 | static void __exit vti_fini(void) |
| 715 | { |
| 716 | rtnl_link_unregister(&vti_link_ops); |
| 717 | xfrm4_tunnel_deregister(&ipip_handler, AF_INET); |
| 718 | xfrm4_protocol_deregister(&vti_ipcomp4_protocol, IPPROTO_COMP); |
| 719 | xfrm4_protocol_deregister(&vti_ah4_protocol, IPPROTO_AH); |
| 720 | xfrm4_protocol_deregister(&vti_esp4_protocol, IPPROTO_ESP); |
| 721 | unregister_pernet_device(&vti_net_ops); |
| 722 | } |
| 723 | |
| 724 | module_init(vti_init); |
| 725 | module_exit(vti_fini); |
| 726 | MODULE_LICENSE("GPL"); |
| 727 | MODULE_ALIAS_RTNL_LINK("vti"); |
| 728 | MODULE_ALIAS_NETDEV("ip_vti0"); |