1// SPDX-License-Identifier: GPL-2.0-or-later 2/* 3 * NET3 IP device support routines. 4 * 5 * Derived from the IP parts of dev.c 1.0.19 6 * Authors: Ross Biro 7 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> 8 * Mark Evans, <evansmp@uhura.aston.ac.uk> 9 * 10 * Additional Authors: 11 * Alan Cox, <gw4pts@gw4pts.ampr.org> 12 * Alexey Kuznetsov, <kuznet@ms2.inr.ac.ru> 13 * 14 * Changes: 15 * Alexey Kuznetsov: pa_* fields are replaced with ifaddr 16 * lists. 17 * Cyrus Durgin: updated for kmod 18 * Matthias Andree: in devinet_ioctl, compare label and 19 * address (4.4BSD alias style support), 20 * fall back to comparing just the label 21 * if no match found. 22 */ 23 24 25#include <linux/uaccess.h> 26#include <linux/bitops.h> 27#include <linux/capability.h> 28#include <linux/module.h> 29#include <linux/types.h> 30#include <linux/kernel.h> 31#include <linux/sched/signal.h> 32#include <linux/string.h> 33#include <linux/mm.h> 34#include <linux/socket.h> 35#include <linux/sockios.h> 36#include <linux/in.h> 37#include <linux/errno.h> 38#include <linux/interrupt.h> 39#include <linux/if_addr.h> 40#include <linux/if_ether.h> 41#include <linux/inet.h> 42#include <linux/netdevice.h> 43#include <linux/etherdevice.h> 44#include <linux/skbuff.h> 45#include <linux/init.h> 46#include <linux/notifier.h> 47#include <linux/inetdevice.h> 48#include <linux/igmp.h> 49#include <linux/slab.h> 50#include <linux/hash.h> 51#ifdef CONFIG_SYSCTL 52#include <linux/sysctl.h> 53#endif 54#include <linux/kmod.h> 55#include <linux/netconf.h> 56 57#include <net/arp.h> 58#include <net/ip.h> 59#include <net/route.h> 60#include <net/ip_fib.h> 61#include <net/rtnetlink.h> 62#include <net/net_namespace.h> 63#include <net/addrconf.h> 64 65#define IPV6ONLY_FLAGS \ 66 (IFA_F_NODAD | IFA_F_OPTIMISTIC | IFA_F_DADFAILED | \ 67 IFA_F_HOMEADDRESS | IFA_F_TENTATIVE | \ 68 IFA_F_MANAGETEMPADDR | IFA_F_STABLE_PRIVACY) 69 70static struct ipv4_devconf ipv4_devconf = { 71 .data = { 72 [IPV4_DEVCONF_ACCEPT_REDIRECTS - 1] = 1, 73 [IPV4_DEVCONF_SEND_REDIRECTS - 1] = 1, 74 [IPV4_DEVCONF_SECURE_REDIRECTS - 1] = 1, 75 [IPV4_DEVCONF_SHARED_MEDIA - 1] = 1, 76 [IPV4_DEVCONF_IGMPV2_UNSOLICITED_REPORT_INTERVAL - 1] = 10000 /*ms*/, 77 [IPV4_DEVCONF_IGMPV3_UNSOLICITED_REPORT_INTERVAL - 1] = 1000 /*ms*/, 78 }, 79}; 80 81static struct ipv4_devconf ipv4_devconf_dflt = { 82 .data = { 83 [IPV4_DEVCONF_ACCEPT_REDIRECTS - 1] = 1, 84 [IPV4_DEVCONF_SEND_REDIRECTS - 1] = 1, 85 [IPV4_DEVCONF_SECURE_REDIRECTS - 1] = 1, 86 [IPV4_DEVCONF_SHARED_MEDIA - 1] = 1, 87 [IPV4_DEVCONF_ACCEPT_SOURCE_ROUTE - 1] = 1, 88 [IPV4_DEVCONF_IGMPV2_UNSOLICITED_REPORT_INTERVAL - 1] = 10000 /*ms*/, 89 [IPV4_DEVCONF_IGMPV3_UNSOLICITED_REPORT_INTERVAL - 1] = 1000 /*ms*/, 90 }, 91}; 92 93#define IPV4_DEVCONF_DFLT(net, attr) \ 94 IPV4_DEVCONF((*net->ipv4.devconf_dflt), attr) 95 96static const struct nla_policy ifa_ipv4_policy[IFA_MAX+1] = { 97 [IFA_LOCAL] = { .type = NLA_U32 }, 98 [IFA_ADDRESS] = { .type = NLA_U32 }, 99 [IFA_BROADCAST] = { .type = NLA_U32 }, 100 [IFA_LABEL] = { .type = NLA_STRING, .len = IFNAMSIZ - 1 }, 101 [IFA_CACHEINFO] = { .len = sizeof(struct ifa_cacheinfo) }, 102 [IFA_FLAGS] = { .type = NLA_U32 }, 103 [IFA_RT_PRIORITY] = { .type = NLA_U32 }, 104 [IFA_TARGET_NETNSID] = { .type = NLA_S32 }, 105}; 106 107struct inet_fill_args { 108 u32 portid; 109 u32 seq; 110 int event; 111 unsigned int flags; 112 int netnsid; 113 int ifindex; 114}; 115 116#define IN4_ADDR_HSIZE_SHIFT 8 117#define IN4_ADDR_HSIZE (1U << IN4_ADDR_HSIZE_SHIFT) 118 119static struct hlist_head inet_addr_lst[IN4_ADDR_HSIZE]; 120 121static u32 inet_addr_hash(const struct net *net, __be32 addr) 122{ 123 u32 val = (__force u32) addr ^ net_hash_mix(net); 124 125 return hash_32(val, IN4_ADDR_HSIZE_SHIFT); 126} 127 128static void inet_hash_insert(struct net *net, struct in_ifaddr *ifa) 129{ 130 u32 hash = inet_addr_hash(net, ifa->ifa_local); 131 132 ASSERT_RTNL(); 133 hlist_add_head_rcu(&ifa->hash, &inet_addr_lst[hash]); 134} 135 136static void inet_hash_remove(struct in_ifaddr *ifa) 137{ 138 ASSERT_RTNL(); 139 hlist_del_init_rcu(&ifa->hash); 140} 141 142/** 143 * __ip_dev_find - find the first device with a given source address. 144 * @net: the net namespace 145 * @addr: the source address 146 * @devref: if true, take a reference on the found device 147 * 148 * If a caller uses devref=false, it should be protected by RCU, or RTNL 149 */ 150struct net_device *__ip_dev_find(struct net *net, __be32 addr, bool devref) 151{ 152 struct net_device *result = NULL; 153 struct in_ifaddr *ifa; 154 155 rcu_read_lock(); 156 ifa = inet_lookup_ifaddr_rcu(net, addr); 157 if (!ifa) { 158 struct flowi4 fl4 = { .daddr = addr }; 159 struct fib_result res = { 0 }; 160 struct fib_table *local; 161 162 /* Fallback to FIB local table so that communication 163 * over loopback subnets work. 164 */ 165 local = fib_get_table(net, RT_TABLE_LOCAL); 166 if (local && 167 !fib_table_lookup(local, &fl4, &res, FIB_LOOKUP_NOREF) && 168 res.type == RTN_LOCAL) 169 result = FIB_RES_DEV(res); 170 } else { 171 result = ifa->ifa_dev->dev; 172 } 173 if (result && devref) 174 dev_hold(result); 175 rcu_read_unlock(); 176 return result; 177} 178EXPORT_SYMBOL(__ip_dev_find); 179 180/* called under RCU lock */ 181struct in_ifaddr *inet_lookup_ifaddr_rcu(struct net *net, __be32 addr) 182{ 183 u32 hash = inet_addr_hash(net, addr); 184 struct in_ifaddr *ifa; 185 186 hlist_for_each_entry_rcu(ifa, &inet_addr_lst[hash], hash) 187 if (ifa->ifa_local == addr && 188 net_eq(dev_net(ifa->ifa_dev->dev), net)) 189 return ifa; 190 191 return NULL; 192} 193 194static void rtmsg_ifa(int event, struct in_ifaddr *, struct nlmsghdr *, u32); 195 196static BLOCKING_NOTIFIER_HEAD(inetaddr_chain); 197static BLOCKING_NOTIFIER_HEAD(inetaddr_validator_chain); 198static void inet_del_ifa(struct in_device *in_dev, 199 struct in_ifaddr __rcu **ifap, 200 int destroy); 201#ifdef CONFIG_SYSCTL 202static int devinet_sysctl_register(struct in_device *idev); 203static void devinet_sysctl_unregister(struct in_device *idev); 204#else 205static int devinet_sysctl_register(struct in_device *idev) 206{ 207 return 0; 208} 209static void devinet_sysctl_unregister(struct in_device *idev) 210{ 211} 212#endif 213 214/* Locks all the inet devices. */ 215 216static struct in_ifaddr *inet_alloc_ifa(void) 217{ 218 return kzalloc(sizeof(struct in_ifaddr), GFP_KERNEL); 219} 220 221static void inet_rcu_free_ifa(struct rcu_head *head) 222{ 223 struct in_ifaddr *ifa = container_of(head, struct in_ifaddr, rcu_head); 224 if (ifa->ifa_dev) 225 in_dev_put(ifa->ifa_dev); 226 kfree(ifa); 227} 228 229static void inet_free_ifa(struct in_ifaddr *ifa) 230{ 231 call_rcu(&ifa->rcu_head, inet_rcu_free_ifa); 232} 233 234void in_dev_finish_destroy(struct in_device *idev) 235{ 236 struct net_device *dev = idev->dev; 237 238 WARN_ON(idev->ifa_list); 239 WARN_ON(idev->mc_list); 240 kfree(rcu_dereference_protected(idev->mc_hash, 1)); 241#ifdef NET_REFCNT_DEBUG 242 pr_debug("%s: %p=%s\n", __func__, idev, dev ? dev->name : "NIL"); 243#endif 244 dev_put(dev); 245 if (!idev->dead) 246 pr_err("Freeing alive in_device %p\n", idev); 247 else 248 kfree(idev); 249} 250EXPORT_SYMBOL(in_dev_finish_destroy); 251 252static struct in_device *inetdev_init(struct net_device *dev) 253{ 254 struct in_device *in_dev; 255 int err = -ENOMEM; 256 257 ASSERT_RTNL(); 258 259 in_dev = kzalloc(sizeof(*in_dev), GFP_KERNEL); 260 if (!in_dev) 261 goto out; 262 memcpy(&in_dev->cnf, dev_net(dev)->ipv4.devconf_dflt, 263 sizeof(in_dev->cnf)); 264 in_dev->cnf.sysctl = NULL; 265 in_dev->dev = dev; 266 in_dev->arp_parms = neigh_parms_alloc(dev, &arp_tbl); 267 if (!in_dev->arp_parms) 268 goto out_kfree; 269 if (IPV4_DEVCONF(in_dev->cnf, FORWARDING)) 270 dev_disable_lro(dev); 271 /* Reference in_dev->dev */ 272 dev_hold(dev); 273 /* Account for reference dev->ip_ptr (below) */ 274 refcount_set(&in_dev->refcnt, 1); 275 276 err = devinet_sysctl_register(in_dev); 277 if (err) { 278 in_dev->dead = 1; 279 neigh_parms_release(&arp_tbl, in_dev->arp_parms); 280 in_dev_put(in_dev); 281 in_dev = NULL; 282 goto out; 283 } 284 ip_mc_init_dev(in_dev); 285 if (dev->flags & IFF_UP) 286 ip_mc_up(in_dev); 287 288 /* we can receive as soon as ip_ptr is set -- do this last */ 289 rcu_assign_pointer(dev->ip_ptr, in_dev); 290out: 291 return in_dev ?: ERR_PTR(err); 292out_kfree: 293 kfree(in_dev); 294 in_dev = NULL; 295 goto out; 296} 297 298static void in_dev_rcu_put(struct rcu_head *head) 299{ 300 struct in_device *idev = container_of(head, struct in_device, rcu_head); 301 in_dev_put(idev); 302} 303 304static void inetdev_destroy(struct in_device *in_dev) 305{ 306 struct net_device *dev; 307 struct in_ifaddr *ifa; 308 309 ASSERT_RTNL(); 310 311 dev = in_dev->dev; 312 313 in_dev->dead = 1; 314 315 ip_mc_destroy_dev(in_dev); 316 317 while ((ifa = rtnl_dereference(in_dev->ifa_list)) != NULL) { 318 inet_del_ifa(in_dev, &in_dev->ifa_list, 0); 319 inet_free_ifa(ifa); 320 } 321 322 RCU_INIT_POINTER(dev->ip_ptr, NULL); 323 324 devinet_sysctl_unregister(in_dev); 325 neigh_parms_release(&arp_tbl, in_dev->arp_parms); 326 arp_ifdown(dev); 327 328 call_rcu(&in_dev->rcu_head, in_dev_rcu_put); 329} 330 331int inet_addr_onlink(struct in_device *in_dev, __be32 a, __be32 b) 332{ 333 const struct in_ifaddr *ifa; 334 335 rcu_read_lock(); 336 in_dev_for_each_ifa_rcu(ifa, in_dev) { 337 if (inet_ifa_match(a, ifa)) { 338 if (!b || inet_ifa_match(b, ifa)) { 339 rcu_read_unlock(); 340 return 1; 341 } 342 } 343 } 344 rcu_read_unlock(); 345 return 0; 346} 347 348static void __inet_del_ifa(struct in_device *in_dev, 349 struct in_ifaddr __rcu **ifap, 350 int destroy, struct nlmsghdr *nlh, u32 portid) 351{ 352 struct in_ifaddr *promote = NULL; 353 struct in_ifaddr *ifa, *ifa1; 354 struct in_ifaddr __rcu **last_prim; 355 struct in_ifaddr *prev_prom = NULL; 356 int do_promote = IN_DEV_PROMOTE_SECONDARIES(in_dev); 357 358 ASSERT_RTNL(); 359 360 ifa1 = rtnl_dereference(*ifap); 361 last_prim = ifap; 362 if (in_dev->dead) 363 goto no_promotions; 364 365 /* 1. Deleting primary ifaddr forces deletion all secondaries 366 * unless alias promotion is set 367 **/ 368 369 if (!(ifa1->ifa_flags & IFA_F_SECONDARY)) { 370 struct in_ifaddr __rcu **ifap1 = &ifa1->ifa_next; 371 372 while ((ifa = rtnl_dereference(*ifap1)) != NULL) { 373 if (!(ifa->ifa_flags & IFA_F_SECONDARY) && 374 ifa1->ifa_scope <= ifa->ifa_scope) 375 last_prim = &ifa->ifa_next; 376 377 if (!(ifa->ifa_flags & IFA_F_SECONDARY) || 378 ifa1->ifa_mask != ifa->ifa_mask || 379 !inet_ifa_match(ifa1->ifa_address, ifa)) { 380 ifap1 = &ifa->ifa_next; 381 prev_prom = ifa; 382 continue; 383 } 384 385 if (!do_promote) { 386 inet_hash_remove(ifa); 387 *ifap1 = ifa->ifa_next; 388 389 rtmsg_ifa(RTM_DELADDR, ifa, nlh, portid); 390 blocking_notifier_call_chain(&inetaddr_chain, 391 NETDEV_DOWN, ifa); 392 inet_free_ifa(ifa); 393 } else { 394 promote = ifa; 395 break; 396 } 397 } 398 } 399 400 /* On promotion all secondaries from subnet are changing 401 * the primary IP, we must remove all their routes silently 402 * and later to add them back with new prefsrc. Do this 403 * while all addresses are on the device list. 404 */ 405 for (ifa = promote; ifa; ifa = rtnl_dereference(ifa->ifa_next)) { 406 if (ifa1->ifa_mask == ifa->ifa_mask && 407 inet_ifa_match(ifa1->ifa_address, ifa)) 408 fib_del_ifaddr(ifa, ifa1); 409 } 410 411no_promotions: 412 /* 2. Unlink it */ 413 414 *ifap = ifa1->ifa_next; 415 inet_hash_remove(ifa1); 416 417 /* 3. Announce address deletion */ 418 419 /* Send message first, then call notifier. 420 At first sight, FIB update triggered by notifier 421 will refer to already deleted ifaddr, that could confuse 422 netlink listeners. It is not true: look, gated sees 423 that route deleted and if it still thinks that ifaddr 424 is valid, it will try to restore deleted routes... Grr. 425 So that, this order is correct. 426 */ 427 rtmsg_ifa(RTM_DELADDR, ifa1, nlh, portid); 428 blocking_notifier_call_chain(&inetaddr_chain, NETDEV_DOWN, ifa1); 429 430 if (promote) { 431 struct in_ifaddr *next_sec; 432 433 next_sec = rtnl_dereference(promote->ifa_next); 434 if (prev_prom) { 435 struct in_ifaddr *last_sec; 436 437 rcu_assign_pointer(prev_prom->ifa_next, next_sec); 438 439 last_sec = rtnl_dereference(*last_prim); 440 rcu_assign_pointer(promote->ifa_next, last_sec); 441 rcu_assign_pointer(*last_prim, promote); 442 } 443 444 promote->ifa_flags &= ~IFA_F_SECONDARY; 445 rtmsg_ifa(RTM_NEWADDR, promote, nlh, portid); 446 blocking_notifier_call_chain(&inetaddr_chain, 447 NETDEV_UP, promote); 448 for (ifa = next_sec; ifa; 449 ifa = rtnl_dereference(ifa->ifa_next)) { 450 if (ifa1->ifa_mask != ifa->ifa_mask || 451 !inet_ifa_match(ifa1->ifa_address, ifa)) 452 continue; 453 fib_add_ifaddr(ifa); 454 } 455 456 } 457 if (destroy) 458 inet_free_ifa(ifa1); 459} 460 461static void inet_del_ifa(struct in_device *in_dev, 462 struct in_ifaddr __rcu **ifap, 463 int destroy) 464{ 465 __inet_del_ifa(in_dev, ifap, destroy, NULL, 0); 466} 467 468static void check_lifetime(struct work_struct *work); 469 470static DECLARE_DELAYED_WORK(check_lifetime_work, check_lifetime); 471 472static int __inet_insert_ifa(struct in_ifaddr *ifa, struct nlmsghdr *nlh, 473 u32 portid, struct netlink_ext_ack *extack) 474{ 475 struct in_ifaddr __rcu **last_primary, **ifap; 476 struct in_device *in_dev = ifa->ifa_dev; 477 struct in_validator_info ivi; 478 struct in_ifaddr *ifa1; 479 int ret; 480 481 ASSERT_RTNL(); 482 483 if (!ifa->ifa_local) { 484 inet_free_ifa(ifa); 485 return 0; 486 } 487 488 ifa->ifa_flags &= ~IFA_F_SECONDARY; 489 last_primary = &in_dev->ifa_list; 490 491 /* Don't set IPv6 only flags to IPv4 addresses */ 492 ifa->ifa_flags &= ~IPV6ONLY_FLAGS; 493 494 ifap = &in_dev->ifa_list; 495 ifa1 = rtnl_dereference(*ifap); 496 497 while (ifa1) { 498 if (!(ifa1->ifa_flags & IFA_F_SECONDARY) && 499 ifa->ifa_scope <= ifa1->ifa_scope) 500 last_primary = &ifa1->ifa_next; 501 if (ifa1->ifa_mask == ifa->ifa_mask && 502 inet_ifa_match(ifa1->ifa_address, ifa)) { 503 if (ifa1->ifa_local == ifa->ifa_local) { 504 inet_free_ifa(ifa); 505 return -EEXIST; 506 } 507 if (ifa1->ifa_scope != ifa->ifa_scope) { 508 inet_free_ifa(ifa); 509 return -EINVAL; 510 } 511 ifa->ifa_flags |= IFA_F_SECONDARY; 512 } 513 514 ifap = &ifa1->ifa_next; 515 ifa1 = rtnl_dereference(*ifap); 516 } 517 518 /* Allow any devices that wish to register ifaddr validtors to weigh 519 * in now, before changes are committed. The rntl lock is serializing 520 * access here, so the state should not change between a validator call 521 * and a final notify on commit. This isn't invoked on promotion under 522 * the assumption that validators are checking the address itself, and 523 * not the flags. 524 */ 525 ivi.ivi_addr = ifa->ifa_address; 526 ivi.ivi_dev = ifa->ifa_dev; 527 ivi.extack = extack; 528 ret = blocking_notifier_call_chain(&inetaddr_validator_chain, 529 NETDEV_UP, &ivi); 530 ret = notifier_to_errno(ret); 531 if (ret) { 532 inet_free_ifa(ifa); 533 return ret; 534 } 535 536 if (!(ifa->ifa_flags & IFA_F_SECONDARY)) { 537 prandom_seed((__force u32) ifa->ifa_local); 538 ifap = last_primary; 539 } 540 541 rcu_assign_pointer(ifa->ifa_next, *ifap); 542 rcu_assign_pointer(*ifap, ifa); 543 544 inet_hash_insert(dev_net(in_dev->dev), ifa); 545 546 cancel_delayed_work(&check_lifetime_work); 547 queue_delayed_work(system_power_efficient_wq, &check_lifetime_work, 0); 548 549 /* Send message first, then call notifier. 550 Notifier will trigger FIB update, so that 551 listeners of netlink will know about new ifaddr */ 552 rtmsg_ifa(RTM_NEWADDR, ifa, nlh, portid); 553 blocking_notifier_call_chain(&inetaddr_chain, NETDEV_UP, ifa); 554 555 return 0; 556} 557 558static int inet_insert_ifa(struct in_ifaddr *ifa) 559{ 560 return __inet_insert_ifa(ifa, NULL, 0, NULL); 561} 562 563static int inet_set_ifa(struct net_device *dev, struct in_ifaddr *ifa) 564{ 565 struct in_device *in_dev = __in_dev_get_rtnl(dev); 566 567 ASSERT_RTNL(); 568 569 if (!in_dev) { 570 inet_free_ifa(ifa); 571 return -ENOBUFS; 572 } 573 ipv4_devconf_setall(in_dev); 574 neigh_parms_data_state_setall(in_dev->arp_parms); 575 if (ifa->ifa_dev != in_dev) { 576 WARN_ON(ifa->ifa_dev); 577 in_dev_hold(in_dev); 578 ifa->ifa_dev = in_dev; 579 } 580 if (ipv4_is_loopback(ifa->ifa_local)) 581 ifa->ifa_scope = RT_SCOPE_HOST; 582 return inet_insert_ifa(ifa); 583} 584 585/* Caller must hold RCU or RTNL : 586 * We dont take a reference on found in_device 587 */ 588struct in_device *inetdev_by_index(struct net *net, int ifindex) 589{ 590 struct net_device *dev; 591 struct in_device *in_dev = NULL; 592 593 rcu_read_lock(); 594 dev = dev_get_by_index_rcu(net, ifindex); 595 if (dev) 596 in_dev = rcu_dereference_rtnl(dev->ip_ptr); 597 rcu_read_unlock(); 598 return in_dev; 599} 600EXPORT_SYMBOL(inetdev_by_index); 601 602/* Called only from RTNL semaphored context. No locks. */ 603 604struct in_ifaddr *inet_ifa_byprefix(struct in_device *in_dev, __be32 prefix, 605 __be32 mask) 606{ 607 struct in_ifaddr *ifa; 608 609 ASSERT_RTNL(); 610 611 in_dev_for_each_ifa_rtnl(ifa, in_dev) { 612 if (ifa->ifa_mask == mask && inet_ifa_match(prefix, ifa)) 613 return ifa; 614 } 615 return NULL; 616} 617 618static int ip_mc_autojoin_config(struct net *net, bool join, 619 const struct in_ifaddr *ifa) 620{ 621#if defined(CONFIG_IP_MULTICAST) 622 struct ip_mreqn mreq = { 623 .imr_multiaddr.s_addr = ifa->ifa_address, 624 .imr_ifindex = ifa->ifa_dev->dev->ifindex, 625 }; 626 struct sock *sk = net->ipv4.mc_autojoin_sk; 627 int ret; 628 629 ASSERT_RTNL(); 630 631 lock_sock(sk); 632 if (join) 633 ret = ip_mc_join_group(sk, &mreq); 634 else 635 ret = ip_mc_leave_group(sk, &mreq); 636 release_sock(sk); 637 638 return ret; 639#else 640 return -EOPNOTSUPP; 641#endif 642} 643 644static int inet_rtm_deladdr(struct sk_buff *skb, struct nlmsghdr *nlh, 645 struct netlink_ext_ack *extack) 646{ 647 struct net *net = sock_net(skb->sk); 648 struct in_ifaddr __rcu **ifap; 649 struct nlattr *tb[IFA_MAX+1]; 650 struct in_device *in_dev; 651 struct ifaddrmsg *ifm; 652 struct in_ifaddr *ifa; 653 654 int err = -EINVAL; 655 656 ASSERT_RTNL(); 657 658 err = nlmsg_parse_deprecated(nlh, sizeof(*ifm), tb, IFA_MAX, 659 ifa_ipv4_policy, extack); 660 if (err < 0) 661 goto errout; 662 663 ifm = nlmsg_data(nlh); 664 in_dev = inetdev_by_index(net, ifm->ifa_index); 665 if (!in_dev) { 666 err = -ENODEV; 667 goto errout; 668 } 669 670 for (ifap = &in_dev->ifa_list; (ifa = rtnl_dereference(*ifap)) != NULL; 671 ifap = &ifa->ifa_next) { 672 if (tb[IFA_LOCAL] && 673 ifa->ifa_local != nla_get_in_addr(tb[IFA_LOCAL])) 674 continue; 675 676 if (tb[IFA_LABEL] && nla_strcmp(tb[IFA_LABEL], ifa->ifa_label)) 677 continue; 678 679 if (tb[IFA_ADDRESS] && 680 (ifm->ifa_prefixlen != ifa->ifa_prefixlen || 681 !inet_ifa_match(nla_get_in_addr(tb[IFA_ADDRESS]), ifa))) 682 continue; 683 684 if (ipv4_is_multicast(ifa->ifa_address)) 685 ip_mc_autojoin_config(net, false, ifa); 686 __inet_del_ifa(in_dev, ifap, 1, nlh, NETLINK_CB(skb).portid); 687 return 0; 688 } 689 690 err = -EADDRNOTAVAIL; 691errout: 692 return err; 693} 694 695#define INFINITY_LIFE_TIME 0xFFFFFFFF 696 697static void check_lifetime(struct work_struct *work) 698{ 699 unsigned long now, next, next_sec, next_sched; 700 struct in_ifaddr *ifa; 701 struct hlist_node *n; 702 int i; 703 704 now = jiffies; 705 next = round_jiffies_up(now + ADDR_CHECK_FREQUENCY); 706 707 for (i = 0; i < IN4_ADDR_HSIZE; i++) { 708 bool change_needed = false; 709 710 rcu_read_lock(); 711 hlist_for_each_entry_rcu(ifa, &inet_addr_lst[i], hash) { 712 unsigned long age; 713 714 if (ifa->ifa_flags & IFA_F_PERMANENT) 715 continue; 716 717 /* We try to batch several events at once. */ 718 age = (now - ifa->ifa_tstamp + 719 ADDRCONF_TIMER_FUZZ_MINUS) / HZ; 720 721 if (ifa->ifa_valid_lft != INFINITY_LIFE_TIME && 722 age >= ifa->ifa_valid_lft) { 723 change_needed = true; 724 } else if (ifa->ifa_preferred_lft == 725 INFINITY_LIFE_TIME) { 726 continue; 727 } else if (age >= ifa->ifa_preferred_lft) { 728 if (time_before(ifa->ifa_tstamp + 729 ifa->ifa_valid_lft * HZ, next)) 730 next = ifa->ifa_tstamp + 731 ifa->ifa_valid_lft * HZ; 732 733 if (!(ifa->ifa_flags & IFA_F_DEPRECATED)) 734 change_needed = true; 735 } else if (time_before(ifa->ifa_tstamp + 736 ifa->ifa_preferred_lft * HZ, 737 next)) { 738 next = ifa->ifa_tstamp + 739 ifa->ifa_preferred_lft * HZ; 740 } 741 } 742 rcu_read_unlock(); 743 if (!change_needed) 744 continue; 745 rtnl_lock(); 746 hlist_for_each_entry_safe(ifa, n, &inet_addr_lst[i], hash) { 747 unsigned long age; 748 749 if (ifa->ifa_flags & IFA_F_PERMANENT) 750 continue; 751 752 /* We try to batch several events at once. */ 753 age = (now - ifa->ifa_tstamp + 754 ADDRCONF_TIMER_FUZZ_MINUS) / HZ; 755 756 if (ifa->ifa_valid_lft != INFINITY_LIFE_TIME && 757 age >= ifa->ifa_valid_lft) { 758 struct in_ifaddr __rcu **ifap; 759 struct in_ifaddr *tmp; 760 761 ifap = &ifa->ifa_dev->ifa_list; 762 tmp = rtnl_dereference(*ifap); 763 while (tmp) { 764 if (tmp == ifa) { 765 inet_del_ifa(ifa->ifa_dev, 766 ifap, 1); 767 break; 768 } 769 ifap = &tmp->ifa_next; 770 tmp = rtnl_dereference(*ifap); 771 } 772 } else if (ifa->ifa_preferred_lft != 773 INFINITY_LIFE_TIME && 774 age >= ifa->ifa_preferred_lft && 775 !(ifa->ifa_flags & IFA_F_DEPRECATED)) { 776 ifa->ifa_flags |= IFA_F_DEPRECATED; 777 rtmsg_ifa(RTM_NEWADDR, ifa, NULL, 0); 778 } 779 } 780 rtnl_unlock(); 781 } 782 783 next_sec = round_jiffies_up(next); 784 next_sched = next; 785 786 /* If rounded timeout is accurate enough, accept it. */ 787 if (time_before(next_sec, next + ADDRCONF_TIMER_FUZZ)) 788 next_sched = next_sec; 789 790 now = jiffies; 791 /* And minimum interval is ADDRCONF_TIMER_FUZZ_MAX. */ 792 if (time_before(next_sched, now + ADDRCONF_TIMER_FUZZ_MAX)) 793 next_sched = now + ADDRCONF_TIMER_FUZZ_MAX; 794 795 queue_delayed_work(system_power_efficient_wq, &check_lifetime_work, 796 next_sched - now); 797} 798 799static void set_ifa_lifetime(struct in_ifaddr *ifa, __u32 valid_lft, 800 __u32 prefered_lft) 801{ 802 unsigned long timeout; 803 804 ifa->ifa_flags &= ~(IFA_F_PERMANENT | IFA_F_DEPRECATED); 805 806 timeout = addrconf_timeout_fixup(valid_lft, HZ); 807 if (addrconf_finite_timeout(timeout)) 808 ifa->ifa_valid_lft = timeout; 809 else 810 ifa->ifa_flags |= IFA_F_PERMANENT; 811 812 timeout = addrconf_timeout_fixup(prefered_lft, HZ); 813 if (addrconf_finite_timeout(timeout)) { 814 if (timeout == 0) 815 ifa->ifa_flags |= IFA_F_DEPRECATED; 816 ifa->ifa_preferred_lft = timeout; 817 } 818 ifa->ifa_tstamp = jiffies; 819 if (!ifa->ifa_cstamp) 820 ifa->ifa_cstamp = ifa->ifa_tstamp; 821} 822 823static struct in_ifaddr *rtm_to_ifaddr(struct net *net, struct nlmsghdr *nlh, 824 __u32 *pvalid_lft, __u32 *pprefered_lft, 825 struct netlink_ext_ack *extack) 826{ 827 struct nlattr *tb[IFA_MAX+1]; 828 struct in_ifaddr *ifa; 829 struct ifaddrmsg *ifm; 830 struct net_device *dev; 831 struct in_device *in_dev; 832 int err; 833 834 err = nlmsg_parse_deprecated(nlh, sizeof(*ifm), tb, IFA_MAX, 835 ifa_ipv4_policy, extack); 836 if (err < 0) 837 goto errout; 838 839 ifm = nlmsg_data(nlh); 840 err = -EINVAL; 841 if (ifm->ifa_prefixlen > 32 || !tb[IFA_LOCAL]) 842 goto errout; 843 844 dev = __dev_get_by_index(net, ifm->ifa_index); 845 err = -ENODEV; 846 if (!dev) 847 goto errout; 848 849 in_dev = __in_dev_get_rtnl(dev); 850 err = -ENOBUFS; 851 if (!in_dev) 852 goto errout; 853 854 ifa = inet_alloc_ifa(); 855 if (!ifa) 856 /* 857 * A potential indev allocation can be left alive, it stays 858 * assigned to its device and is destroy with it. 859 */ 860 goto errout; 861 862 ipv4_devconf_setall(in_dev); 863 neigh_parms_data_state_setall(in_dev->arp_parms); 864 in_dev_hold(in_dev); 865 866 if (!tb[IFA_ADDRESS]) 867 tb[IFA_ADDRESS] = tb[IFA_LOCAL]; 868 869 INIT_HLIST_NODE(&ifa->hash); 870 ifa->ifa_prefixlen = ifm->ifa_prefixlen; 871 ifa->ifa_mask = inet_make_mask(ifm->ifa_prefixlen); 872 ifa->ifa_flags = tb[IFA_FLAGS] ? nla_get_u32(tb[IFA_FLAGS]) : 873 ifm->ifa_flags; 874 ifa->ifa_scope = ifm->ifa_scope; 875 ifa->ifa_dev = in_dev; 876 877 ifa->ifa_local = nla_get_in_addr(tb[IFA_LOCAL]); 878 ifa->ifa_address = nla_get_in_addr(tb[IFA_ADDRESS]); 879 880 if (tb[IFA_BROADCAST]) 881 ifa->ifa_broadcast = nla_get_in_addr(tb[IFA_BROADCAST]); 882 883 if (tb[IFA_LABEL]) 884 nla_strlcpy(ifa->ifa_label, tb[IFA_LABEL], IFNAMSIZ); 885 else 886 memcpy(ifa->ifa_label, dev->name, IFNAMSIZ); 887 888 if (tb[IFA_RT_PRIORITY]) 889 ifa->ifa_rt_priority = nla_get_u32(tb[IFA_RT_PRIORITY]); 890 891 if (tb[IFA_CACHEINFO]) { 892 struct ifa_cacheinfo *ci; 893 894 ci = nla_data(tb[IFA_CACHEINFO]); 895 if (!ci->ifa_valid || ci->ifa_prefered > ci->ifa_valid) { 896 err = -EINVAL; 897 goto errout_free; 898 } 899 *pvalid_lft = ci->ifa_valid; 900 *pprefered_lft = ci->ifa_prefered; 901 } 902 903 return ifa; 904 905errout_free: 906 inet_free_ifa(ifa); 907errout: 908 return ERR_PTR(err); 909} 910 911static struct in_ifaddr *find_matching_ifa(struct in_ifaddr *ifa) 912{ 913 struct in_device *in_dev = ifa->ifa_dev; 914 struct in_ifaddr *ifa1; 915 916 if (!ifa->ifa_local) 917 return NULL; 918 919 in_dev_for_each_ifa_rtnl(ifa1, in_dev) { 920 if (ifa1->ifa_mask == ifa->ifa_mask && 921 inet_ifa_match(ifa1->ifa_address, ifa) && 922 ifa1->ifa_local == ifa->ifa_local) 923 return ifa1; 924 } 925 return NULL; 926} 927 928static int inet_rtm_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh, 929 struct netlink_ext_ack *extack) 930{ 931 struct net *net = sock_net(skb->sk); 932 struct in_ifaddr *ifa; 933 struct in_ifaddr *ifa_existing; 934 __u32 valid_lft = INFINITY_LIFE_TIME; 935 __u32 prefered_lft = INFINITY_LIFE_TIME; 936 937 ASSERT_RTNL(); 938 939 ifa = rtm_to_ifaddr(net, nlh, &valid_lft, &prefered_lft, extack); 940 if (IS_ERR(ifa)) 941 return PTR_ERR(ifa); 942 943 ifa_existing = find_matching_ifa(ifa); 944 if (!ifa_existing) { 945 /* It would be best to check for !NLM_F_CREATE here but 946 * userspace already relies on not having to provide this. 947 */ 948 set_ifa_lifetime(ifa, valid_lft, prefered_lft); 949 if (ifa->ifa_flags & IFA_F_MCAUTOJOIN) { 950 int ret = ip_mc_autojoin_config(net, true, ifa); 951 952 if (ret < 0) { 953 inet_free_ifa(ifa); 954 return ret; 955 } 956 } 957 return __inet_insert_ifa(ifa, nlh, NETLINK_CB(skb).portid, 958 extack); 959 } else { 960 u32 new_metric = ifa->ifa_rt_priority; 961 962 inet_free_ifa(ifa); 963 964 if (nlh->nlmsg_flags & NLM_F_EXCL || 965 !(nlh->nlmsg_flags & NLM_F_REPLACE)) 966 return -EEXIST; 967 ifa = ifa_existing; 968 969 if (ifa->ifa_rt_priority != new_metric) { 970 fib_modify_prefix_metric(ifa, new_metric); 971 ifa->ifa_rt_priority = new_metric; 972 } 973 974 set_ifa_lifetime(ifa, valid_lft, prefered_lft); 975 cancel_delayed_work(&check_lifetime_work); 976 queue_delayed_work(system_power_efficient_wq, 977 &check_lifetime_work, 0); 978 rtmsg_ifa(RTM_NEWADDR, ifa, nlh, NETLINK_CB(skb).portid); 979 } 980 return 0; 981} 982 983/* 984 * Determine a default network mask, based on the IP address. 985 */ 986 987static int inet_abc_len(__be32 addr) 988{ 989 int rc = -1; /* Something else, probably a multicast. */ 990 991 if (ipv4_is_zeronet(addr) || ipv4_is_lbcast(addr)) 992 rc = 0; 993 else { 994 __u32 haddr = ntohl(addr); 995 if (IN_CLASSA(haddr)) 996 rc = 8; 997 else if (IN_CLASSB(haddr)) 998 rc = 16; 999 else if (IN_CLASSC(haddr)) 1000 rc = 24; 1001 else if (IN_CLASSE(haddr)) 1002 rc = 32; 1003 } 1004 1005 return rc; 1006} 1007 1008 1009int devinet_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr) 1010{ 1011 struct sockaddr_in sin_orig; 1012 struct sockaddr_in *sin = (struct sockaddr_in *)&ifr->ifr_addr; 1013 struct in_ifaddr __rcu **ifap = NULL; 1014 struct in_device *in_dev; 1015 struct in_ifaddr *ifa = NULL; 1016 struct net_device *dev; 1017 char *colon; 1018 int ret = -EFAULT; 1019 int tryaddrmatch = 0; 1020 1021 ifr->ifr_name[IFNAMSIZ - 1] = 0; 1022 1023 /* save original address for comparison */ 1024 memcpy(&sin_orig, sin, sizeof(*sin)); 1025 1026 colon = strchr(ifr->ifr_name, ':'); 1027 if (colon) 1028 *colon = 0; 1029 1030 dev_load(net, ifr->ifr_name); 1031 1032 switch (cmd) { 1033 case SIOCGIFADDR: /* Get interface address */ 1034 case SIOCGIFBRDADDR: /* Get the broadcast address */ 1035 case SIOCGIFDSTADDR: /* Get the destination address */ 1036 case SIOCGIFNETMASK: /* Get the netmask for the interface */ 1037 /* Note that these ioctls will not sleep, 1038 so that we do not impose a lock. 1039 One day we will be forced to put shlock here (I mean SMP) 1040 */ 1041 tryaddrmatch = (sin_orig.sin_family == AF_INET); 1042 memset(sin, 0, sizeof(*sin)); 1043 sin->sin_family = AF_INET; 1044 break; 1045 1046 case SIOCSIFFLAGS: 1047 ret = -EPERM; 1048 if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) 1049 goto out; 1050 break; 1051 case SIOCSIFADDR: /* Set interface address (and family) */ 1052 case SIOCSIFBRDADDR: /* Set the broadcast address */ 1053 case SIOCSIFDSTADDR: /* Set the destination address */ 1054 case SIOCSIFNETMASK: /* Set the netmask for the interface */ 1055 ret = -EPERM; 1056 if (!ns_capable(net->user_ns, CAP_NET_ADMIN)) 1057 goto out; 1058 ret = -EINVAL; 1059 if (sin->sin_family != AF_INET) 1060 goto out; 1061 break; 1062 default: 1063 ret = -EINVAL; 1064 goto out; 1065 } 1066 1067 rtnl_lock(); 1068 1069 ret = -ENODEV; 1070 dev = __dev_get_by_name(net, ifr->ifr_name); 1071 if (!dev) 1072 goto done; 1073 1074 if (colon) 1075 *colon = ':'; 1076 1077 in_dev = __in_dev_get_rtnl(dev); 1078 if (in_dev) { 1079 if (tryaddrmatch) { 1080 /* Matthias Andree */ 1081 /* compare label and address (4.4BSD style) */ 1082 /* note: we only do this for a limited set of ioctls 1083 and only if the original address family was AF_INET. 1084 This is checked above. */ 1085 1086 for (ifap = &in_dev->ifa_list; 1087 (ifa = rtnl_dereference(*ifap)) != NULL; 1088 ifap = &ifa->ifa_next) { 1089 if (!strcmp(ifr->ifr_name, ifa->ifa_label) && 1090 sin_orig.sin_addr.s_addr == 1091 ifa->ifa_local) { 1092 break; /* found */ 1093 } 1094 } 1095 } 1096 /* we didn't get a match, maybe the application is 1097 4.3BSD-style and passed in junk so we fall back to 1098 comparing just the label */ 1099 if (!ifa) { 1100 for (ifap = &in_dev->ifa_list; 1101 (ifa = rtnl_dereference(*ifap)) != NULL; 1102 ifap = &ifa->ifa_next) 1103 if (!strcmp(ifr->ifr_name, ifa->ifa_label)) 1104 break; 1105 } 1106 } 1107 1108 ret = -EADDRNOTAVAIL; 1109 if (!ifa && cmd != SIOCSIFADDR && cmd != SIOCSIFFLAGS) 1110 goto done; 1111 1112 switch (cmd) { 1113 case SIOCGIFADDR: /* Get interface address */ 1114 ret = 0; 1115 sin->sin_addr.s_addr = ifa->ifa_local; 1116 break; 1117 1118 case SIOCGIFBRDADDR: /* Get the broadcast address */ 1119 ret = 0; 1120 sin->sin_addr.s_addr = ifa->ifa_broadcast; 1121 break; 1122 1123 case SIOCGIFDSTADDR: /* Get the destination address */ 1124 ret = 0; 1125 sin->sin_addr.s_addr = ifa->ifa_address; 1126 break; 1127 1128 case SIOCGIFNETMASK: /* Get the netmask for the interface */ 1129 ret = 0; 1130 sin->sin_addr.s_addr = ifa->ifa_mask; 1131 break; 1132 1133 case SIOCSIFFLAGS: 1134 if (colon) { 1135 ret = -EADDRNOTAVAIL; 1136 if (!ifa) 1137 break; 1138 ret = 0; 1139 if (!(ifr->ifr_flags & IFF_UP)) 1140 inet_del_ifa(in_dev, ifap, 1); 1141 break; 1142 } 1143 ret = dev_change_flags(dev, ifr->ifr_flags, NULL); 1144 break; 1145 1146 case SIOCSIFADDR: /* Set interface address (and family) */ 1147 ret = -EINVAL; 1148 if (inet_abc_len(sin->sin_addr.s_addr) < 0) 1149 break; 1150 1151 if (!ifa) { 1152 ret = -ENOBUFS; 1153 ifa = inet_alloc_ifa(); 1154 if (!ifa) 1155 break; 1156 INIT_HLIST_NODE(&ifa->hash); 1157 if (colon) 1158 memcpy(ifa->ifa_label, ifr->ifr_name, IFNAMSIZ); 1159 else 1160 memcpy(ifa->ifa_label, dev->name, IFNAMSIZ); 1161 } else { 1162 ret = 0; 1163 if (ifa->ifa_local == sin->sin_addr.s_addr) 1164 break; 1165 inet_del_ifa(in_dev, ifap, 0); 1166 ifa->ifa_broadcast = 0; 1167 ifa->ifa_scope = 0; 1168 } 1169 1170 ifa->ifa_address = ifa->ifa_local = sin->sin_addr.s_addr; 1171 1172 if (!(dev->flags & IFF_POINTOPOINT)) { 1173 ifa->ifa_prefixlen = inet_abc_len(ifa->ifa_address); 1174 ifa->ifa_mask = inet_make_mask(ifa->ifa_prefixlen); 1175 if ((dev->flags & IFF_BROADCAST) && 1176 ifa->ifa_prefixlen < 31) 1177 ifa->ifa_broadcast = ifa->ifa_address | 1178 ~ifa->ifa_mask; 1179 } else { 1180 ifa->ifa_prefixlen = 32; 1181 ifa->ifa_mask = inet_make_mask(32); 1182 } 1183 set_ifa_lifetime(ifa, INFINITY_LIFE_TIME, INFINITY_LIFE_TIME); 1184 ret = inet_set_ifa(dev, ifa); 1185 break; 1186 1187 case SIOCSIFBRDADDR: /* Set the broadcast address */ 1188 ret = 0; 1189 if (ifa->ifa_broadcast != sin->sin_addr.s_addr) { 1190 inet_del_ifa(in_dev, ifap, 0); 1191 ifa->ifa_broadcast = sin->sin_addr.s_addr; 1192 inet_insert_ifa(ifa); 1193 } 1194 break; 1195 1196 case SIOCSIFDSTADDR: /* Set the destination address */ 1197 ret = 0; 1198 if (ifa->ifa_address == sin->sin_addr.s_addr) 1199 break; 1200 ret = -EINVAL; 1201 if (inet_abc_len(sin->sin_addr.s_addr) < 0) 1202 break; 1203 ret = 0; 1204 inet_del_ifa(in_dev, ifap, 0); 1205 ifa->ifa_address = sin->sin_addr.s_addr; 1206 inet_insert_ifa(ifa); 1207 break; 1208 1209 case SIOCSIFNETMASK: /* Set the netmask for the interface */ 1210 1211 /* 1212 * The mask we set must be legal. 1213 */ 1214 ret = -EINVAL; 1215 if (bad_mask(sin->sin_addr.s_addr, 0)) 1216 break; 1217 ret = 0; 1218 if (ifa->ifa_mask != sin->sin_addr.s_addr) { 1219 __be32 old_mask = ifa->ifa_mask; 1220 inet_del_ifa(in_dev, ifap, 0); 1221 ifa->ifa_mask = sin->sin_addr.s_addr; 1222 ifa->ifa_prefixlen = inet_mask_len(ifa->ifa_mask); 1223 1224 /* See if current broadcast address matches 1225 * with current netmask, then recalculate 1226 * the broadcast address. Otherwise it's a 1227 * funny address, so don't touch it since 1228 * the user seems to know what (s)he's doing... 1229 */ 1230 if ((dev->flags & IFF_BROADCAST) && 1231 (ifa->ifa_prefixlen < 31) && 1232 (ifa->ifa_broadcast == 1233 (ifa->ifa_local|~old_mask))) { 1234 ifa->ifa_broadcast = (ifa->ifa_local | 1235 ~sin->sin_addr.s_addr); 1236 } 1237 inet_insert_ifa(ifa); 1238 } 1239 break; 1240 } 1241done: 1242 rtnl_unlock(); 1243out: 1244 return ret; 1245} 1246 1247int inet_gifconf(struct net_device *dev, char __user *buf, int len, int size) 1248{ 1249 struct in_device *in_dev = __in_dev_get_rtnl(dev); 1250 const struct in_ifaddr *ifa; 1251 struct ifreq ifr; 1252 int done = 0; 1253 1254 if (WARN_ON(size > sizeof(struct ifreq))) 1255 goto out; 1256 1257 if (!in_dev) 1258 goto out; 1259 1260 in_dev_for_each_ifa_rtnl(ifa, in_dev) { 1261 if (!buf) { 1262 done += size; 1263 continue; 1264 } 1265 if (len < size) 1266 break; 1267 memset(&ifr, 0, sizeof(struct ifreq)); 1268 strcpy(ifr.ifr_name, ifa->ifa_label); 1269 1270 (*(struct sockaddr_in *)&ifr.ifr_addr).sin_family = AF_INET; 1271 (*(struct sockaddr_in *)&ifr.ifr_addr).sin_addr.s_addr = 1272 ifa->ifa_local; 1273 1274 if (copy_to_user(buf + done, &ifr, size)) { 1275 done = -EFAULT; 1276 break; 1277 } 1278 len -= size; 1279 done += size; 1280 } 1281out: 1282 return done; 1283} 1284 1285static __be32 in_dev_select_addr(const struct in_device *in_dev, 1286 int scope) 1287{ 1288 const struct in_ifaddr *ifa; 1289 1290 in_dev_for_each_ifa_rcu(ifa, in_dev) { 1291 if (ifa->ifa_flags & IFA_F_SECONDARY) 1292 continue; 1293 if (ifa->ifa_scope != RT_SCOPE_LINK && 1294 ifa->ifa_scope <= scope) 1295 return ifa->ifa_local; 1296 } 1297 1298 return 0; 1299} 1300 1301__be32 inet_select_addr(const struct net_device *dev, __be32 dst, int scope) 1302{ 1303 const struct in_ifaddr *ifa; 1304 __be32 addr = 0; 1305 unsigned char localnet_scope = RT_SCOPE_HOST; 1306 struct in_device *in_dev; 1307 struct net *net = dev_net(dev); 1308 int master_idx; 1309 1310 rcu_read_lock(); 1311 in_dev = __in_dev_get_rcu(dev); 1312 if (!in_dev) 1313 goto no_in_dev; 1314 1315 if (unlikely(IN_DEV_ROUTE_LOCALNET(in_dev))) 1316 localnet_scope = RT_SCOPE_LINK; 1317 1318 in_dev_for_each_ifa_rcu(ifa, in_dev) { 1319 if (ifa->ifa_flags & IFA_F_SECONDARY) 1320 continue; 1321 if (min(ifa->ifa_scope, localnet_scope) > scope) 1322 continue; 1323 if (!dst || inet_ifa_match(dst, ifa)) { 1324 addr = ifa->ifa_local; 1325 break; 1326 } 1327 if (!addr) 1328 addr = ifa->ifa_local; 1329 } 1330 1331 if (addr) 1332 goto out_unlock; 1333no_in_dev: 1334 master_idx = l3mdev_master_ifindex_rcu(dev); 1335 1336 /* For VRFs, the VRF device takes the place of the loopback device, 1337 * with addresses on it being preferred. Note in such cases the 1338 * loopback device will be among the devices that fail the master_idx 1339 * equality check in the loop below. 1340 */ 1341 if (master_idx && 1342 (dev = dev_get_by_index_rcu(net, master_idx)) && 1343 (in_dev = __in_dev_get_rcu(dev))) { 1344 addr = in_dev_select_addr(in_dev, scope); 1345 if (addr) 1346 goto out_unlock; 1347 } 1348 1349 /* Not loopback addresses on loopback should be preferred 1350 in this case. It is important that lo is the first interface 1351 in dev_base list. 1352 */ 1353 for_each_netdev_rcu(net, dev) { 1354 if (l3mdev_master_ifindex_rcu(dev) != master_idx) 1355 continue; 1356 1357 in_dev = __in_dev_get_rcu(dev); 1358 if (!in_dev) 1359 continue; 1360 1361 addr = in_dev_select_addr(in_dev, scope); 1362 if (addr) 1363 goto out_unlock; 1364 } 1365out_unlock: 1366 rcu_read_unlock(); 1367 return addr; 1368} 1369EXPORT_SYMBOL(inet_select_addr); 1370 1371static __be32 confirm_addr_indev(struct in_device *in_dev, __be32 dst, 1372 __be32 local, int scope) 1373{ 1374 unsigned char localnet_scope = RT_SCOPE_HOST; 1375 const struct in_ifaddr *ifa; 1376 __be32 addr = 0; 1377 int same = 0; 1378 1379 if (unlikely(IN_DEV_ROUTE_LOCALNET(in_dev))) 1380 localnet_scope = RT_SCOPE_LINK; 1381 1382 in_dev_for_each_ifa_rcu(ifa, in_dev) { 1383 unsigned char min_scope = min(ifa->ifa_scope, localnet_scope); 1384 1385 if (!addr && 1386 (local == ifa->ifa_local || !local) && 1387 min_scope <= scope) { 1388 addr = ifa->ifa_local; 1389 if (same) 1390 break; 1391 } 1392 if (!same) { 1393 same = (!local || inet_ifa_match(local, ifa)) && 1394 (!dst || inet_ifa_match(dst, ifa)); 1395 if (same && addr) { 1396 if (local || !dst) 1397 break; 1398 /* Is the selected addr into dst subnet? */ 1399 if (inet_ifa_match(addr, ifa)) 1400 break; 1401 /* No, then can we use new local src? */ 1402 if (min_scope <= scope) { 1403 addr = ifa->ifa_local; 1404 break; 1405 } 1406 /* search for large dst subnet for addr */ 1407 same = 0; 1408 } 1409 } 1410 } 1411 1412 return same ? addr : 0; 1413} 1414 1415/* 1416 * Confirm that local IP address exists using wildcards: 1417 * - net: netns to check, cannot be NULL 1418 * - in_dev: only on this interface, NULL=any interface 1419 * - dst: only in the same subnet as dst, 0=any dst 1420 * - local: address, 0=autoselect the local address 1421 * - scope: maximum allowed scope value for the local address 1422 */ 1423__be32 inet_confirm_addr(struct net *net, struct in_device *in_dev, 1424 __be32 dst, __be32 local, int scope) 1425{ 1426 __be32 addr = 0; 1427 struct net_device *dev; 1428 1429 if (in_dev) 1430 return confirm_addr_indev(in_dev, dst, local, scope); 1431 1432 rcu_read_lock(); 1433 for_each_netdev_rcu(net, dev) { 1434 in_dev = __in_dev_get_rcu(dev); 1435 if (in_dev) { 1436 addr = confirm_addr_indev(in_dev, dst, local, scope); 1437 if (addr) 1438 break; 1439 } 1440 } 1441 rcu_read_unlock(); 1442 1443 return addr; 1444} 1445EXPORT_SYMBOL(inet_confirm_addr); 1446 1447/* 1448 * Device notifier 1449 */ 1450 1451int register_inetaddr_notifier(struct notifier_block *nb) 1452{ 1453 return blocking_notifier_chain_register(&inetaddr_chain, nb); 1454} 1455EXPORT_SYMBOL(register_inetaddr_notifier); 1456 1457int unregister_inetaddr_notifier(struct notifier_block *nb) 1458{ 1459 return blocking_notifier_chain_unregister(&inetaddr_chain, nb); 1460} 1461EXPORT_SYMBOL(unregister_inetaddr_notifier); 1462 1463int register_inetaddr_validator_notifier(struct notifier_block *nb) 1464{ 1465 return blocking_notifier_chain_register(&inetaddr_validator_chain, nb); 1466} 1467EXPORT_SYMBOL(register_inetaddr_validator_notifier); 1468 1469int unregister_inetaddr_validator_notifier(struct notifier_block *nb) 1470{ 1471 return blocking_notifier_chain_unregister(&inetaddr_validator_chain, 1472 nb); 1473} 1474EXPORT_SYMBOL(unregister_inetaddr_validator_notifier); 1475 1476/* Rename ifa_labels for a device name change. Make some effort to preserve 1477 * existing alias numbering and to create unique labels if possible. 1478*/ 1479static void inetdev_changename(struct net_device *dev, struct in_device *in_dev) 1480{ 1481 struct in_ifaddr *ifa; 1482 int named = 0; 1483 1484 in_dev_for_each_ifa_rtnl(ifa, in_dev) { 1485 char old[IFNAMSIZ], *dot; 1486 1487 memcpy(old, ifa->ifa_label, IFNAMSIZ); 1488 memcpy(ifa->ifa_label, dev->name, IFNAMSIZ); 1489 if (named++ == 0) 1490 goto skip; 1491 dot = strchr(old, ':'); 1492 if (!dot) { 1493 sprintf(old, ":%d", named); 1494 dot = old; 1495 } 1496 if (strlen(dot) + strlen(dev->name) < IFNAMSIZ) 1497 strcat(ifa->ifa_label, dot); 1498 else 1499 strcpy(ifa->ifa_label + (IFNAMSIZ - strlen(dot) - 1), dot); 1500skip: 1501 rtmsg_ifa(RTM_NEWADDR, ifa, NULL, 0); 1502 } 1503} 1504 1505static void inetdev_send_gratuitous_arp(struct net_device *dev, 1506 struct in_device *in_dev) 1507 1508{ 1509 const struct in_ifaddr *ifa; 1510 1511 in_dev_for_each_ifa_rtnl(ifa, in_dev) { 1512 arp_send(ARPOP_REQUEST, ETH_P_ARP, 1513 ifa->ifa_local, dev, 1514 ifa->ifa_local, NULL, 1515 dev->dev_addr, NULL); 1516 } 1517} 1518 1519/* Called only under RTNL semaphore */ 1520 1521static int inetdev_event(struct notifier_block *this, unsigned long event, 1522 void *ptr) 1523{ 1524 struct net_device *dev = netdev_notifier_info_to_dev(ptr); 1525 struct in_device *in_dev = __in_dev_get_rtnl(dev); 1526 1527 ASSERT_RTNL(); 1528 1529 if (!in_dev) { 1530 if (event == NETDEV_REGISTER) { 1531 in_dev = inetdev_init(dev); 1532 if (IS_ERR(in_dev)) 1533 return notifier_from_errno(PTR_ERR(in_dev)); 1534 if (dev->flags & IFF_LOOPBACK) { 1535 IN_DEV_CONF_SET(in_dev, NOXFRM, 1); 1536 IN_DEV_CONF_SET(in_dev, NOPOLICY, 1); 1537 } 1538 } else if (event == NETDEV_CHANGEMTU) { 1539 /* Re-enabling IP */ 1540 if (inetdev_valid_mtu(dev->mtu)) 1541 in_dev = inetdev_init(dev); 1542 } 1543 goto out; 1544 } 1545 1546 switch (event) { 1547 case NETDEV_REGISTER: 1548 pr_debug("%s: bug\n", __func__); 1549 RCU_INIT_POINTER(dev->ip_ptr, NULL); 1550 break; 1551 case NETDEV_UP: 1552 if (!inetdev_valid_mtu(dev->mtu)) 1553 break; 1554 if (dev->flags & IFF_LOOPBACK) { 1555 struct in_ifaddr *ifa = inet_alloc_ifa(); 1556 1557 if (ifa) { 1558 INIT_HLIST_NODE(&ifa->hash); 1559 ifa->ifa_local = 1560 ifa->ifa_address = htonl(INADDR_LOOPBACK); 1561 ifa->ifa_prefixlen = 8; 1562 ifa->ifa_mask = inet_make_mask(8); 1563 in_dev_hold(in_dev); 1564 ifa->ifa_dev = in_dev; 1565 ifa->ifa_scope = RT_SCOPE_HOST; 1566 memcpy(ifa->ifa_label, dev->name, IFNAMSIZ); 1567 set_ifa_lifetime(ifa, INFINITY_LIFE_TIME, 1568 INFINITY_LIFE_TIME); 1569 ipv4_devconf_setall(in_dev); 1570 neigh_parms_data_state_setall(in_dev->arp_parms); 1571 inet_insert_ifa(ifa); 1572 } 1573 } 1574 ip_mc_up(in_dev); 1575 fallthrough; 1576 case NETDEV_CHANGEADDR: 1577 if (!IN_DEV_ARP_NOTIFY(in_dev)) 1578 break; 1579 fallthrough; 1580 case NETDEV_NOTIFY_PEERS: 1581 /* Send gratuitous ARP to notify of link change */ 1582 inetdev_send_gratuitous_arp(dev, in_dev); 1583 break; 1584 case NETDEV_DOWN: 1585 ip_mc_down(in_dev); 1586 break; 1587 case NETDEV_PRE_TYPE_CHANGE: 1588 ip_mc_unmap(in_dev); 1589 break; 1590 case NETDEV_POST_TYPE_CHANGE: 1591 ip_mc_remap(in_dev); 1592 break; 1593 case NETDEV_CHANGEMTU: 1594 if (inetdev_valid_mtu(dev->mtu)) 1595 break; 1596 /* disable IP when MTU is not enough */ 1597 fallthrough; 1598 case NETDEV_UNREGISTER: 1599 inetdev_destroy(in_dev); 1600 break; 1601 case NETDEV_CHANGENAME: 1602 /* Do not notify about label change, this event is 1603 * not interesting to applications using netlink. 1604 */ 1605 inetdev_changename(dev, in_dev); 1606 1607 devinet_sysctl_unregister(in_dev); 1608 devinet_sysctl_register(in_dev); 1609 break; 1610 } 1611out: 1612 return NOTIFY_DONE; 1613} 1614 1615static struct notifier_block ip_netdev_notifier = { 1616 .notifier_call = inetdev_event, 1617}; 1618 1619static size_t inet_nlmsg_size(void) 1620{ 1621 return NLMSG_ALIGN(sizeof(struct ifaddrmsg)) 1622 + nla_total_size(4) /* IFA_ADDRESS */ 1623 + nla_total_size(4) /* IFA_LOCAL */ 1624 + nla_total_size(4) /* IFA_BROADCAST */ 1625 + nla_total_size(IFNAMSIZ) /* IFA_LABEL */ 1626 + nla_total_size(4) /* IFA_FLAGS */ 1627 + nla_total_size(4) /* IFA_RT_PRIORITY */ 1628 + nla_total_size(sizeof(struct ifa_cacheinfo)); /* IFA_CACHEINFO */ 1629} 1630 1631static inline u32 cstamp_delta(unsigned long cstamp) 1632{ 1633 return (cstamp - INITIAL_JIFFIES) * 100UL / HZ; 1634} 1635 1636static int put_cacheinfo(struct sk_buff *skb, unsigned long cstamp, 1637 unsigned long tstamp, u32 preferred, u32 valid) 1638{ 1639 struct ifa_cacheinfo ci; 1640 1641 ci.cstamp = cstamp_delta(cstamp); 1642 ci.tstamp = cstamp_delta(tstamp); 1643 ci.ifa_prefered = preferred; 1644 ci.ifa_valid = valid; 1645 1646 return nla_put(skb, IFA_CACHEINFO, sizeof(ci), &ci); 1647} 1648 1649static int inet_fill_ifaddr(struct sk_buff *skb, struct in_ifaddr *ifa, 1650 struct inet_fill_args *args) 1651{ 1652 struct ifaddrmsg *ifm; 1653 struct nlmsghdr *nlh; 1654 u32 preferred, valid; 1655 1656 nlh = nlmsg_put(skb, args->portid, args->seq, args->event, sizeof(*ifm), 1657 args->flags); 1658 if (!nlh) 1659 return -EMSGSIZE; 1660 1661 ifm = nlmsg_data(nlh); 1662 ifm->ifa_family = AF_INET; 1663 ifm->ifa_prefixlen = ifa->ifa_prefixlen; 1664 ifm->ifa_flags = ifa->ifa_flags; 1665 ifm->ifa_scope = ifa->ifa_scope; 1666 ifm->ifa_index = ifa->ifa_dev->dev->ifindex; 1667 1668 if (args->netnsid >= 0 && 1669 nla_put_s32(skb, IFA_TARGET_NETNSID, args->netnsid)) 1670 goto nla_put_failure; 1671 1672 if (!(ifm->ifa_flags & IFA_F_PERMANENT)) { 1673 preferred = ifa->ifa_preferred_lft; 1674 valid = ifa->ifa_valid_lft; 1675 if (preferred != INFINITY_LIFE_TIME) { 1676 long tval = (jiffies - ifa->ifa_tstamp) / HZ; 1677 1678 if (preferred > tval) 1679 preferred -= tval; 1680 else 1681 preferred = 0; 1682 if (valid != INFINITY_LIFE_TIME) { 1683 if (valid > tval) 1684 valid -= tval; 1685 else 1686 valid = 0; 1687 } 1688 } 1689 } else { 1690 preferred = INFINITY_LIFE_TIME; 1691 valid = INFINITY_LIFE_TIME; 1692 } 1693 if ((ifa->ifa_address && 1694 nla_put_in_addr(skb, IFA_ADDRESS, ifa->ifa_address)) || 1695 (ifa->ifa_local && 1696 nla_put_in_addr(skb, IFA_LOCAL, ifa->ifa_local)) || 1697 (ifa->ifa_broadcast && 1698 nla_put_in_addr(skb, IFA_BROADCAST, ifa->ifa_broadcast)) || 1699 (ifa->ifa_label[0] && 1700 nla_put_string(skb, IFA_LABEL, ifa->ifa_label)) || 1701 nla_put_u32(skb, IFA_FLAGS, ifa->ifa_flags) || 1702 (ifa->ifa_rt_priority && 1703 nla_put_u32(skb, IFA_RT_PRIORITY, ifa->ifa_rt_priority)) || 1704 put_cacheinfo(skb, ifa->ifa_cstamp, ifa->ifa_tstamp, 1705 preferred, valid)) 1706 goto nla_put_failure; 1707 1708 nlmsg_end(skb, nlh); 1709 return 0; 1710 1711nla_put_failure: 1712 nlmsg_cancel(skb, nlh); 1713 return -EMSGSIZE; 1714} 1715 1716static int inet_valid_dump_ifaddr_req(const struct nlmsghdr *nlh, 1717 struct inet_fill_args *fillargs, 1718 struct net **tgt_net, struct sock *sk, 1719 struct netlink_callback *cb) 1720{ 1721 struct netlink_ext_ack *extack = cb->extack; 1722 struct nlattr *tb[IFA_MAX+1]; 1723 struct ifaddrmsg *ifm; 1724 int err, i; 1725 1726 if (nlh->nlmsg_len < nlmsg_msg_size(sizeof(*ifm))) { 1727 NL_SET_ERR_MSG(extack, "ipv4: Invalid header for address dump request"); 1728 return -EINVAL; 1729 } 1730 1731 ifm = nlmsg_data(nlh); 1732 if (ifm->ifa_prefixlen || ifm->ifa_flags || ifm->ifa_scope) { 1733 NL_SET_ERR_MSG(extack, "ipv4: Invalid values in header for address dump request"); 1734 return -EINVAL; 1735 } 1736 1737 fillargs->ifindex = ifm->ifa_index; 1738 if (fillargs->ifindex) { 1739 cb->answer_flags |= NLM_F_DUMP_FILTERED; 1740 fillargs->flags |= NLM_F_DUMP_FILTERED; 1741 } 1742 1743 err = nlmsg_parse_deprecated_strict(nlh, sizeof(*ifm), tb, IFA_MAX, 1744 ifa_ipv4_policy, extack); 1745 if (err < 0) 1746 return err; 1747 1748 for (i = 0; i <= IFA_MAX; ++i) { 1749 if (!tb[i]) 1750 continue; 1751 1752 if (i == IFA_TARGET_NETNSID) { 1753 struct net *net; 1754 1755 fillargs->netnsid = nla_get_s32(tb[i]); 1756 1757 net = rtnl_get_net_ns_capable(sk, fillargs->netnsid); 1758 if (IS_ERR(net)) { 1759 fillargs->netnsid = -1; 1760 NL_SET_ERR_MSG(extack, "ipv4: Invalid target network namespace id"); 1761 return PTR_ERR(net); 1762 } 1763 *tgt_net = net; 1764 } else { 1765 NL_SET_ERR_MSG(extack, "ipv4: Unsupported attribute in dump request"); 1766 return -EINVAL; 1767 } 1768 } 1769 1770 return 0; 1771} 1772 1773static int in_dev_dump_addr(struct in_device *in_dev, struct sk_buff *skb, 1774 struct netlink_callback *cb, int s_ip_idx, 1775 struct inet_fill_args *fillargs) 1776{ 1777 struct in_ifaddr *ifa; 1778 int ip_idx = 0; 1779 int err; 1780 1781 in_dev_for_each_ifa_rtnl(ifa, in_dev) { 1782 if (ip_idx < s_ip_idx) { 1783 ip_idx++; 1784 continue; 1785 } 1786 err = inet_fill_ifaddr(skb, ifa, fillargs); 1787 if (err < 0) 1788 goto done; 1789 1790 nl_dump_check_consistent(cb, nlmsg_hdr(skb)); 1791 ip_idx++; 1792 } 1793 err = 0; 1794 1795done: 1796 cb->args[2] = ip_idx; 1797 1798 return err; 1799} 1800 1801static int inet_dump_ifaddr(struct sk_buff *skb, struct netlink_callback *cb) 1802{ 1803 const struct nlmsghdr *nlh = cb->nlh; 1804 struct inet_fill_args fillargs = { 1805 .portid = NETLINK_CB(cb->skb).portid, 1806 .seq = nlh->nlmsg_seq, 1807 .event = RTM_NEWADDR, 1808 .flags = NLM_F_MULTI, 1809 .netnsid = -1, 1810 }; 1811 struct net *net = sock_net(skb->sk); 1812 struct net *tgt_net = net; 1813 int h, s_h; 1814 int idx, s_idx; 1815 int s_ip_idx; 1816 struct net_device *dev; 1817 struct in_device *in_dev; 1818 struct hlist_head *head; 1819 int err = 0; 1820 1821 s_h = cb->args[0]; 1822 s_idx = idx = cb->args[1]; 1823 s_ip_idx = cb->args[2]; 1824 1825 if (cb->strict_check) { 1826 err = inet_valid_dump_ifaddr_req(nlh, &fillargs, &tgt_net, 1827 skb->sk, cb); 1828 if (err < 0) 1829 goto put_tgt_net; 1830 1831 err = 0; 1832 if (fillargs.ifindex) { 1833 dev = __dev_get_by_index(tgt_net, fillargs.ifindex); 1834 if (!dev) { 1835 err = -ENODEV; 1836 goto put_tgt_net; 1837 } 1838 1839 in_dev = __in_dev_get_rtnl(dev); 1840 if (in_dev) { 1841 err = in_dev_dump_addr(in_dev, skb, cb, s_ip_idx, 1842 &fillargs); 1843 } 1844 goto put_tgt_net; 1845 } 1846 } 1847 1848 for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) { 1849 idx = 0; 1850 head = &tgt_net->dev_index_head[h]; 1851 rcu_read_lock(); 1852 cb->seq = atomic_read(&tgt_net->ipv4.dev_addr_genid) ^ 1853 tgt_net->dev_base_seq; 1854 hlist_for_each_entry_rcu(dev, head, index_hlist) { 1855 if (idx < s_idx) 1856 goto cont; 1857 if (h > s_h || idx > s_idx) 1858 s_ip_idx = 0; 1859 in_dev = __in_dev_get_rcu(dev); 1860 if (!in_dev) 1861 goto cont; 1862 1863 err = in_dev_dump_addr(in_dev, skb, cb, s_ip_idx, 1864 &fillargs); 1865 if (err < 0) { 1866 rcu_read_unlock(); 1867 goto done; 1868 } 1869cont: 1870 idx++; 1871 } 1872 rcu_read_unlock(); 1873 } 1874 1875done: 1876 cb->args[0] = h; 1877 cb->args[1] = idx; 1878put_tgt_net: 1879 if (fillargs.netnsid >= 0) 1880 put_net(tgt_net); 1881 1882 return skb->len ? : err; 1883} 1884 1885static void rtmsg_ifa(int event, struct in_ifaddr *ifa, struct nlmsghdr *nlh, 1886 u32 portid) 1887{ 1888 struct inet_fill_args fillargs = { 1889 .portid = portid, 1890 .seq = nlh ? nlh->nlmsg_seq : 0, 1891 .event = event, 1892 .flags = 0, 1893 .netnsid = -1, 1894 }; 1895 struct sk_buff *skb; 1896 int err = -ENOBUFS; 1897 struct net *net; 1898 1899 net = dev_net(ifa->ifa_dev->dev); 1900 skb = nlmsg_new(inet_nlmsg_size(), GFP_KERNEL); 1901 if (!skb) 1902 goto errout; 1903 1904 err = inet_fill_ifaddr(skb, ifa, &fillargs); 1905 if (err < 0) { 1906 /* -EMSGSIZE implies BUG in inet_nlmsg_size() */ 1907 WARN_ON(err == -EMSGSIZE); 1908 kfree_skb(skb); 1909 goto errout; 1910 } 1911 rtnl_notify(skb, net, portid, RTNLGRP_IPV4_IFADDR, nlh, GFP_KERNEL); 1912 return; 1913errout: 1914 if (err < 0) 1915 rtnl_set_sk_err(net, RTNLGRP_IPV4_IFADDR, err); 1916} 1917 1918static size_t inet_get_link_af_size(const struct net_device *dev, 1919 u32 ext_filter_mask) 1920{ 1921 struct in_device *in_dev = rcu_dereference_rtnl(dev->ip_ptr); 1922 1923 if (!in_dev) 1924 return 0; 1925 1926 return nla_total_size(IPV4_DEVCONF_MAX * 4); /* IFLA_INET_CONF */ 1927} 1928 1929static int inet_fill_link_af(struct sk_buff *skb, const struct net_device *dev, 1930 u32 ext_filter_mask) 1931{ 1932 struct in_device *in_dev = rcu_dereference_rtnl(dev->ip_ptr); 1933 struct nlattr *nla; 1934 int i; 1935 1936 if (!in_dev) 1937 return -ENODATA; 1938 1939 nla = nla_reserve(skb, IFLA_INET_CONF, IPV4_DEVCONF_MAX * 4); 1940 if (!nla) 1941 return -EMSGSIZE; 1942 1943 for (i = 0; i < IPV4_DEVCONF_MAX; i++) 1944 ((u32 *) nla_data(nla))[i] = in_dev->cnf.data[i]; 1945 1946 return 0; 1947} 1948 1949static const struct nla_policy inet_af_policy[IFLA_INET_MAX+1] = { 1950 [IFLA_INET_CONF] = { .type = NLA_NESTED }, 1951}; 1952 1953static int inet_validate_link_af(const struct net_device *dev, 1954 const struct nlattr *nla) 1955{ 1956 struct nlattr *a, *tb[IFLA_INET_MAX+1]; 1957 int err, rem; 1958 1959 if (dev && !__in_dev_get_rcu(dev)) 1960 return -EAFNOSUPPORT; 1961 1962 err = nla_parse_nested_deprecated(tb, IFLA_INET_MAX, nla, 1963 inet_af_policy, NULL); 1964 if (err < 0) 1965 return err; 1966 1967 if (tb[IFLA_INET_CONF]) { 1968 nla_for_each_nested(a, tb[IFLA_INET_CONF], rem) { 1969 int cfgid = nla_type(a); 1970 1971 if (nla_len(a) < 4) 1972 return -EINVAL; 1973 1974 if (cfgid <= 0 || cfgid > IPV4_DEVCONF_MAX) 1975 return -EINVAL; 1976 } 1977 } 1978 1979 return 0; 1980} 1981 1982static int inet_set_link_af(struct net_device *dev, const struct nlattr *nla) 1983{ 1984 struct in_device *in_dev = __in_dev_get_rcu(dev); 1985 struct nlattr *a, *tb[IFLA_INET_MAX+1]; 1986 int rem; 1987 1988 if (!in_dev) 1989 return -EAFNOSUPPORT; 1990 1991 if (nla_parse_nested_deprecated(tb, IFLA_INET_MAX, nla, NULL, NULL) < 0) 1992 return -EINVAL; 1993 1994 if (tb[IFLA_INET_CONF]) { 1995 nla_for_each_nested(a, tb[IFLA_INET_CONF], rem) 1996 ipv4_devconf_set(in_dev, nla_type(a), nla_get_u32(a)); 1997 } 1998 1999 return 0; 2000} 2001 2002static int inet_netconf_msgsize_devconf(int type) 2003{ 2004 int size = NLMSG_ALIGN(sizeof(struct netconfmsg)) 2005 + nla_total_size(4); /* NETCONFA_IFINDEX */ 2006 bool all = false; 2007 2008 if (type == NETCONFA_ALL) 2009 all = true; 2010 2011 if (all || type == NETCONFA_FORWARDING) 2012 size += nla_total_size(4); 2013 if (all || type == NETCONFA_RP_FILTER) 2014 size += nla_total_size(4); 2015 if (all || type == NETCONFA_MC_FORWARDING) 2016 size += nla_total_size(4); 2017 if (all || type == NETCONFA_BC_FORWARDING) 2018 size += nla_total_size(4); 2019 if (all || type == NETCONFA_PROXY_NEIGH) 2020 size += nla_total_size(4); 2021 if (all || type == NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN) 2022 size += nla_total_size(4); 2023 2024 return size; 2025} 2026 2027static int inet_netconf_fill_devconf(struct sk_buff *skb, int ifindex, 2028 struct ipv4_devconf *devconf, u32 portid, 2029 u32 seq, int event, unsigned int flags, 2030 int type) 2031{ 2032 struct nlmsghdr *nlh; 2033 struct netconfmsg *ncm; 2034 bool all = false; 2035 2036 nlh = nlmsg_put(skb, portid, seq, event, sizeof(struct netconfmsg), 2037 flags); 2038 if (!nlh) 2039 return -EMSGSIZE; 2040 2041 if (type == NETCONFA_ALL) 2042 all = true; 2043 2044 ncm = nlmsg_data(nlh); 2045 ncm->ncm_family = AF_INET; 2046 2047 if (nla_put_s32(skb, NETCONFA_IFINDEX, ifindex) < 0) 2048 goto nla_put_failure; 2049 2050 if (!devconf) 2051 goto out; 2052 2053 if ((all || type == NETCONFA_FORWARDING) && 2054 nla_put_s32(skb, NETCONFA_FORWARDING, 2055 IPV4_DEVCONF(*devconf, FORWARDING)) < 0) 2056 goto nla_put_failure; 2057 if ((all || type == NETCONFA_RP_FILTER) && 2058 nla_put_s32(skb, NETCONFA_RP_FILTER, 2059 IPV4_DEVCONF(*devconf, RP_FILTER)) < 0) 2060 goto nla_put_failure; 2061 if ((all || type == NETCONFA_MC_FORWARDING) && 2062 nla_put_s32(skb, NETCONFA_MC_FORWARDING, 2063 IPV4_DEVCONF(*devconf, MC_FORWARDING)) < 0) 2064 goto nla_put_failure; 2065 if ((all || type == NETCONFA_BC_FORWARDING) && 2066 nla_put_s32(skb, NETCONFA_BC_FORWARDING, 2067 IPV4_DEVCONF(*devconf, BC_FORWARDING)) < 0) 2068 goto nla_put_failure; 2069 if ((all || type == NETCONFA_PROXY_NEIGH) && 2070 nla_put_s32(skb, NETCONFA_PROXY_NEIGH, 2071 IPV4_DEVCONF(*devconf, PROXY_ARP)) < 0) 2072 goto nla_put_failure; 2073 if ((all || type == NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN) && 2074 nla_put_s32(skb, NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN, 2075 IPV4_DEVCONF(*devconf, IGNORE_ROUTES_WITH_LINKDOWN)) < 0) 2076 goto nla_put_failure; 2077 2078out: 2079 nlmsg_end(skb, nlh); 2080 return 0; 2081 2082nla_put_failure: 2083 nlmsg_cancel(skb, nlh); 2084 return -EMSGSIZE; 2085} 2086 2087void inet_netconf_notify_devconf(struct net *net, int event, int type, 2088 int ifindex, struct ipv4_devconf *devconf) 2089{ 2090 struct sk_buff *skb; 2091 int err = -ENOBUFS; 2092 2093 skb = nlmsg_new(inet_netconf_msgsize_devconf(type), GFP_KERNEL); 2094 if (!skb) 2095 goto errout; 2096 2097 err = inet_netconf_fill_devconf(skb, ifindex, devconf, 0, 0, 2098 event, 0, type); 2099 if (err < 0) { 2100 /* -EMSGSIZE implies BUG in inet_netconf_msgsize_devconf() */ 2101 WARN_ON(err == -EMSGSIZE); 2102 kfree_skb(skb); 2103 goto errout; 2104 } 2105 rtnl_notify(skb, net, 0, RTNLGRP_IPV4_NETCONF, NULL, GFP_KERNEL); 2106 return; 2107errout: 2108 if (err < 0) 2109 rtnl_set_sk_err(net, RTNLGRP_IPV4_NETCONF, err); 2110} 2111 2112static const struct nla_policy devconf_ipv4_policy[NETCONFA_MAX+1] = { 2113 [NETCONFA_IFINDEX] = { .len = sizeof(int) }, 2114 [NETCONFA_FORWARDING] = { .len = sizeof(int) }, 2115 [NETCONFA_RP_FILTER] = { .len = sizeof(int) }, 2116 [NETCONFA_PROXY_NEIGH] = { .len = sizeof(int) }, 2117 [NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN] = { .len = sizeof(int) }, 2118}; 2119 2120static int inet_netconf_valid_get_req(struct sk_buff *skb, 2121 const struct nlmsghdr *nlh, 2122 struct nlattr **tb, 2123 struct netlink_ext_ack *extack) 2124{ 2125 int i, err; 2126 2127 if (nlh->nlmsg_len < nlmsg_msg_size(sizeof(struct netconfmsg))) { 2128 NL_SET_ERR_MSG(extack, "ipv4: Invalid header for netconf get request"); 2129 return -EINVAL; 2130 } 2131 2132 if (!netlink_strict_get_check(skb)) 2133 return nlmsg_parse_deprecated(nlh, sizeof(struct netconfmsg), 2134 tb, NETCONFA_MAX, 2135 devconf_ipv4_policy, extack); 2136 2137 err = nlmsg_parse_deprecated_strict(nlh, sizeof(struct netconfmsg), 2138 tb, NETCONFA_MAX, 2139 devconf_ipv4_policy, extack); 2140 if (err) 2141 return err; 2142 2143 for (i = 0; i <= NETCONFA_MAX; i++) { 2144 if (!tb[i]) 2145 continue; 2146 2147 switch (i) { 2148 case NETCONFA_IFINDEX: 2149 break; 2150 default: 2151 NL_SET_ERR_MSG(extack, "ipv4: Unsupported attribute in netconf get request"); 2152 return -EINVAL; 2153 } 2154 } 2155 2156 return 0; 2157} 2158 2159static int inet_netconf_get_devconf(struct sk_buff *in_skb, 2160 struct nlmsghdr *nlh, 2161 struct netlink_ext_ack *extack) 2162{ 2163 struct net *net = sock_net(in_skb->sk); 2164 struct nlattr *tb[NETCONFA_MAX+1]; 2165 struct sk_buff *skb; 2166 struct ipv4_devconf *devconf; 2167 struct in_device *in_dev; 2168 struct net_device *dev; 2169 int ifindex; 2170 int err; 2171 2172 err = inet_netconf_valid_get_req(in_skb, nlh, tb, extack); 2173 if (err) 2174 goto errout; 2175 2176 err = -EINVAL; 2177 if (!tb[NETCONFA_IFINDEX]) 2178 goto errout; 2179 2180 ifindex = nla_get_s32(tb[NETCONFA_IFINDEX]); 2181 switch (ifindex) { 2182 case NETCONFA_IFINDEX_ALL: 2183 devconf = net->ipv4.devconf_all; 2184 break; 2185 case NETCONFA_IFINDEX_DEFAULT: 2186 devconf = net->ipv4.devconf_dflt; 2187 break; 2188 default: 2189 dev = __dev_get_by_index(net, ifindex); 2190 if (!dev) 2191 goto errout; 2192 in_dev = __in_dev_get_rtnl(dev); 2193 if (!in_dev) 2194 goto errout; 2195 devconf = &in_dev->cnf; 2196 break; 2197 } 2198 2199 err = -ENOBUFS; 2200 skb = nlmsg_new(inet_netconf_msgsize_devconf(NETCONFA_ALL), GFP_KERNEL); 2201 if (!skb) 2202 goto errout; 2203 2204 err = inet_netconf_fill_devconf(skb, ifindex, devconf, 2205 NETLINK_CB(in_skb).portid, 2206 nlh->nlmsg_seq, RTM_NEWNETCONF, 0, 2207 NETCONFA_ALL); 2208 if (err < 0) { 2209 /* -EMSGSIZE implies BUG in inet_netconf_msgsize_devconf() */ 2210 WARN_ON(err == -EMSGSIZE); 2211 kfree_skb(skb); 2212 goto errout; 2213 } 2214 err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid); 2215errout: 2216 return err; 2217} 2218 2219static int inet_netconf_dump_devconf(struct sk_buff *skb, 2220 struct netlink_callback *cb) 2221{ 2222 const struct nlmsghdr *nlh = cb->nlh; 2223 struct net *net = sock_net(skb->sk); 2224 int h, s_h; 2225 int idx, s_idx; 2226 struct net_device *dev; 2227 struct in_device *in_dev; 2228 struct hlist_head *head; 2229 2230 if (cb->strict_check) { 2231 struct netlink_ext_ack *extack = cb->extack; 2232 struct netconfmsg *ncm; 2233 2234 if (nlh->nlmsg_len < nlmsg_msg_size(sizeof(*ncm))) { 2235 NL_SET_ERR_MSG(extack, "ipv4: Invalid header for netconf dump request"); 2236 return -EINVAL; 2237 } 2238 2239 if (nlmsg_attrlen(nlh, sizeof(*ncm))) { 2240 NL_SET_ERR_MSG(extack, "ipv4: Invalid data after header in netconf dump request"); 2241 return -EINVAL; 2242 } 2243 } 2244 2245 s_h = cb->args[0]; 2246 s_idx = idx = cb->args[1]; 2247 2248 for (h = s_h; h < NETDEV_HASHENTRIES; h++, s_idx = 0) { 2249 idx = 0; 2250 head = &net->dev_index_head[h]; 2251 rcu_read_lock(); 2252 cb->seq = atomic_read(&net->ipv4.dev_addr_genid) ^ 2253 net->dev_base_seq; 2254 hlist_for_each_entry_rcu(dev, head, index_hlist) { 2255 if (idx < s_idx) 2256 goto cont; 2257 in_dev = __in_dev_get_rcu(dev); 2258 if (!in_dev) 2259 goto cont; 2260 2261 if (inet_netconf_fill_devconf(skb, dev->ifindex, 2262 &in_dev->cnf, 2263 NETLINK_CB(cb->skb).portid, 2264 nlh->nlmsg_seq, 2265 RTM_NEWNETCONF, 2266 NLM_F_MULTI, 2267 NETCONFA_ALL) < 0) { 2268 rcu_read_unlock(); 2269 goto done; 2270 } 2271 nl_dump_check_consistent(cb, nlmsg_hdr(skb)); 2272cont: 2273 idx++; 2274 } 2275 rcu_read_unlock(); 2276 } 2277 if (h == NETDEV_HASHENTRIES) { 2278 if (inet_netconf_fill_devconf(skb, NETCONFA_IFINDEX_ALL, 2279 net->ipv4.devconf_all, 2280 NETLINK_CB(cb->skb).portid, 2281 nlh->nlmsg_seq, 2282 RTM_NEWNETCONF, NLM_F_MULTI, 2283 NETCONFA_ALL) < 0) 2284 goto done; 2285 else 2286 h++; 2287 } 2288 if (h == NETDEV_HASHENTRIES + 1) { 2289 if (inet_netconf_fill_devconf(skb, NETCONFA_IFINDEX_DEFAULT, 2290 net->ipv4.devconf_dflt, 2291 NETLINK_CB(cb->skb).portid, 2292 nlh->nlmsg_seq, 2293 RTM_NEWNETCONF, NLM_F_MULTI, 2294 NETCONFA_ALL) < 0) 2295 goto done; 2296 else 2297 h++; 2298 } 2299done: 2300 cb->args[0] = h; 2301 cb->args[1] = idx; 2302 2303 return skb->len; 2304} 2305 2306#ifdef CONFIG_SYSCTL 2307 2308static void devinet_copy_dflt_conf(struct net *net, int i) 2309{ 2310 struct net_device *dev; 2311 2312 rcu_read_lock(); 2313 for_each_netdev_rcu(net, dev) { 2314 struct in_device *in_dev; 2315 2316 in_dev = __in_dev_get_rcu(dev); 2317 if (in_dev && !test_bit(i, in_dev->cnf.state)) 2318 in_dev->cnf.data[i] = net->ipv4.devconf_dflt->data[i]; 2319 } 2320 rcu_read_unlock(); 2321} 2322 2323/* called with RTNL locked */ 2324static void inet_forward_change(struct net *net) 2325{ 2326 struct net_device *dev; 2327 int on = IPV4_DEVCONF_ALL(net, FORWARDING); 2328 2329 IPV4_DEVCONF_ALL(net, ACCEPT_REDIRECTS) = !on; 2330 IPV4_DEVCONF_DFLT(net, FORWARDING) = on; 2331 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2332 NETCONFA_FORWARDING, 2333 NETCONFA_IFINDEX_ALL, 2334 net->ipv4.devconf_all); 2335 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2336 NETCONFA_FORWARDING, 2337 NETCONFA_IFINDEX_DEFAULT, 2338 net->ipv4.devconf_dflt); 2339 2340 for_each_netdev(net, dev) { 2341 struct in_device *in_dev; 2342 2343 if (on) 2344 dev_disable_lro(dev); 2345 2346 in_dev = __in_dev_get_rtnl(dev); 2347 if (in_dev) { 2348 IN_DEV_CONF_SET(in_dev, FORWARDING, on); 2349 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2350 NETCONFA_FORWARDING, 2351 dev->ifindex, &in_dev->cnf); 2352 } 2353 } 2354} 2355 2356static int devinet_conf_ifindex(struct net *net, struct ipv4_devconf *cnf) 2357{ 2358 if (cnf == net->ipv4.devconf_dflt) 2359 return NETCONFA_IFINDEX_DEFAULT; 2360 else if (cnf == net->ipv4.devconf_all) 2361 return NETCONFA_IFINDEX_ALL; 2362 else { 2363 struct in_device *idev 2364 = container_of(cnf, struct in_device, cnf); 2365 return idev->dev->ifindex; 2366 } 2367} 2368 2369static int devinet_conf_proc(struct ctl_table *ctl, int write, 2370 void *buffer, size_t *lenp, loff_t *ppos) 2371{ 2372 int old_value = *(int *)ctl->data; 2373 int ret = proc_dointvec(ctl, write, buffer, lenp, ppos); 2374 int new_value = *(int *)ctl->data; 2375 2376 if (write) { 2377 struct ipv4_devconf *cnf = ctl->extra1; 2378 struct net *net = ctl->extra2; 2379 int i = (int *)ctl->data - cnf->data; 2380 int ifindex; 2381 2382 set_bit(i, cnf->state); 2383 2384 if (cnf == net->ipv4.devconf_dflt) 2385 devinet_copy_dflt_conf(net, i); 2386 if (i == IPV4_DEVCONF_ACCEPT_LOCAL - 1 || 2387 i == IPV4_DEVCONF_ROUTE_LOCALNET - 1) 2388 if ((new_value == 0) && (old_value != 0)) 2389 rt_cache_flush(net); 2390 2391 if (i == IPV4_DEVCONF_BC_FORWARDING - 1 && 2392 new_value != old_value) 2393 rt_cache_flush(net); 2394 2395 if (i == IPV4_DEVCONF_RP_FILTER - 1 && 2396 new_value != old_value) { 2397 ifindex = devinet_conf_ifindex(net, cnf); 2398 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2399 NETCONFA_RP_FILTER, 2400 ifindex, cnf); 2401 } 2402 if (i == IPV4_DEVCONF_PROXY_ARP - 1 && 2403 new_value != old_value) { 2404 ifindex = devinet_conf_ifindex(net, cnf); 2405 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2406 NETCONFA_PROXY_NEIGH, 2407 ifindex, cnf); 2408 } 2409 if (i == IPV4_DEVCONF_IGNORE_ROUTES_WITH_LINKDOWN - 1 && 2410 new_value != old_value) { 2411 ifindex = devinet_conf_ifindex(net, cnf); 2412 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2413 NETCONFA_IGNORE_ROUTES_WITH_LINKDOWN, 2414 ifindex, cnf); 2415 } 2416 } 2417 2418 return ret; 2419} 2420 2421static int devinet_sysctl_forward(struct ctl_table *ctl, int write, 2422 void *buffer, size_t *lenp, loff_t *ppos) 2423{ 2424 int *valp = ctl->data; 2425 int val = *valp; 2426 loff_t pos = *ppos; 2427 int ret = proc_dointvec(ctl, write, buffer, lenp, ppos); 2428 2429 if (write && *valp != val) { 2430 struct net *net = ctl->extra2; 2431 2432 if (valp != &IPV4_DEVCONF_DFLT(net, FORWARDING)) { 2433 if (!rtnl_trylock()) { 2434 /* Restore the original values before restarting */ 2435 *valp = val; 2436 *ppos = pos; 2437 return restart_syscall(); 2438 } 2439 if (valp == &IPV4_DEVCONF_ALL(net, FORWARDING)) { 2440 inet_forward_change(net); 2441 } else { 2442 struct ipv4_devconf *cnf = ctl->extra1; 2443 struct in_device *idev = 2444 container_of(cnf, struct in_device, cnf); 2445 if (*valp) 2446 dev_disable_lro(idev->dev); 2447 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2448 NETCONFA_FORWARDING, 2449 idev->dev->ifindex, 2450 cnf); 2451 } 2452 rtnl_unlock(); 2453 rt_cache_flush(net); 2454 } else 2455 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, 2456 NETCONFA_FORWARDING, 2457 NETCONFA_IFINDEX_DEFAULT, 2458 net->ipv4.devconf_dflt); 2459 } 2460 2461 return ret; 2462} 2463 2464static int ipv4_doint_and_flush(struct ctl_table *ctl, int write, 2465 void *buffer, size_t *lenp, loff_t *ppos) 2466{ 2467 int *valp = ctl->data; 2468 int val = *valp; 2469 int ret = proc_dointvec(ctl, write, buffer, lenp, ppos); 2470 struct net *net = ctl->extra2; 2471 2472 if (write && *valp != val) 2473 rt_cache_flush(net); 2474 2475 return ret; 2476} 2477 2478#define DEVINET_SYSCTL_ENTRY(attr, name, mval, proc) \ 2479 { \ 2480 .procname = name, \ 2481 .data = ipv4_devconf.data + \ 2482 IPV4_DEVCONF_ ## attr - 1, \ 2483 .maxlen = sizeof(int), \ 2484 .mode = mval, \ 2485 .proc_handler = proc, \ 2486 .extra1 = &ipv4_devconf, \ 2487 } 2488 2489#define DEVINET_SYSCTL_RW_ENTRY(attr, name) \ 2490 DEVINET_SYSCTL_ENTRY(attr, name, 0644, devinet_conf_proc) 2491 2492#define DEVINET_SYSCTL_RO_ENTRY(attr, name) \ 2493 DEVINET_SYSCTL_ENTRY(attr, name, 0444, devinet_conf_proc) 2494 2495#define DEVINET_SYSCTL_COMPLEX_ENTRY(attr, name, proc) \ 2496 DEVINET_SYSCTL_ENTRY(attr, name, 0644, proc) 2497 2498#define DEVINET_SYSCTL_FLUSHING_ENTRY(attr, name) \ 2499 DEVINET_SYSCTL_COMPLEX_ENTRY(attr, name, ipv4_doint_and_flush) 2500 2501static struct devinet_sysctl_table { 2502 struct ctl_table_header *sysctl_header; 2503 struct ctl_table devinet_vars[__IPV4_DEVCONF_MAX]; 2504} devinet_sysctl = { 2505 .devinet_vars = { 2506 DEVINET_SYSCTL_COMPLEX_ENTRY(FORWARDING, "forwarding", 2507 devinet_sysctl_forward), 2508 DEVINET_SYSCTL_RO_ENTRY(MC_FORWARDING, "mc_forwarding"), 2509 DEVINET_SYSCTL_RW_ENTRY(BC_FORWARDING, "bc_forwarding"), 2510 2511 DEVINET_SYSCTL_RW_ENTRY(ACCEPT_REDIRECTS, "accept_redirects"), 2512 DEVINET_SYSCTL_RW_ENTRY(SECURE_REDIRECTS, "secure_redirects"), 2513 DEVINET_SYSCTL_RW_ENTRY(SHARED_MEDIA, "shared_media"), 2514 DEVINET_SYSCTL_RW_ENTRY(RP_FILTER, "rp_filter"), 2515 DEVINET_SYSCTL_RW_ENTRY(SEND_REDIRECTS, "send_redirects"), 2516 DEVINET_SYSCTL_RW_ENTRY(ACCEPT_SOURCE_ROUTE, 2517 "accept_source_route"), 2518 DEVINET_SYSCTL_RW_ENTRY(ACCEPT_LOCAL, "accept_local"), 2519 DEVINET_SYSCTL_RW_ENTRY(SRC_VMARK, "src_valid_mark"), 2520 DEVINET_SYSCTL_RW_ENTRY(PROXY_ARP, "proxy_arp"), 2521 DEVINET_SYSCTL_RW_ENTRY(MEDIUM_ID, "medium_id"), 2522 DEVINET_SYSCTL_RW_ENTRY(BOOTP_RELAY, "bootp_relay"), 2523 DEVINET_SYSCTL_RW_ENTRY(LOG_MARTIANS, "log_martians"), 2524 DEVINET_SYSCTL_RW_ENTRY(TAG, "tag"), 2525 DEVINET_SYSCTL_RW_ENTRY(ARPFILTER, "arp_filter"), 2526 DEVINET_SYSCTL_RW_ENTRY(ARP_ANNOUNCE, "arp_announce"), 2527 DEVINET_SYSCTL_RW_ENTRY(ARP_IGNORE, "arp_ignore"), 2528 DEVINET_SYSCTL_RW_ENTRY(ARP_ACCEPT, "arp_accept"), 2529 DEVINET_SYSCTL_RW_ENTRY(ARP_NOTIFY, "arp_notify"), 2530 DEVINET_SYSCTL_RW_ENTRY(PROXY_ARP_PVLAN, "proxy_arp_pvlan"), 2531 DEVINET_SYSCTL_RW_ENTRY(FORCE_IGMP_VERSION, 2532 "force_igmp_version"), 2533 DEVINET_SYSCTL_RW_ENTRY(IGMPV2_UNSOLICITED_REPORT_INTERVAL, 2534 "igmpv2_unsolicited_report_interval"), 2535 DEVINET_SYSCTL_RW_ENTRY(IGMPV3_UNSOLICITED_REPORT_INTERVAL, 2536 "igmpv3_unsolicited_report_interval"), 2537 DEVINET_SYSCTL_RW_ENTRY(IGNORE_ROUTES_WITH_LINKDOWN, 2538 "ignore_routes_with_linkdown"), 2539 DEVINET_SYSCTL_RW_ENTRY(DROP_GRATUITOUS_ARP, 2540 "drop_gratuitous_arp"), 2541 2542 DEVINET_SYSCTL_FLUSHING_ENTRY(NOXFRM, "disable_xfrm"), 2543 DEVINET_SYSCTL_FLUSHING_ENTRY(NOPOLICY, "disable_policy"), 2544 DEVINET_SYSCTL_FLUSHING_ENTRY(PROMOTE_SECONDARIES, 2545 "promote_secondaries"), 2546 DEVINET_SYSCTL_FLUSHING_ENTRY(ROUTE_LOCALNET, 2547 "route_localnet"), 2548 DEVINET_SYSCTL_FLUSHING_ENTRY(DROP_UNICAST_IN_L2_MULTICAST, 2549 "drop_unicast_in_l2_multicast"), 2550 }, 2551}; 2552 2553static int __devinet_sysctl_register(struct net *net, char *dev_name, 2554 int ifindex, struct ipv4_devconf *p) 2555{ 2556 int i; 2557 struct devinet_sysctl_table *t; 2558 char path[sizeof("net/ipv4/conf/") + IFNAMSIZ]; 2559 2560 t = kmemdup(&devinet_sysctl, sizeof(*t), GFP_KERNEL); 2561 if (!t) 2562 goto out; 2563 2564 for (i = 0; i < ARRAY_SIZE(t->devinet_vars) - 1; i++) { 2565 t->devinet_vars[i].data += (char *)p - (char *)&ipv4_devconf; 2566 t->devinet_vars[i].extra1 = p; 2567 t->devinet_vars[i].extra2 = net; 2568 } 2569 2570 snprintf(path, sizeof(path), "net/ipv4/conf/%s", dev_name); 2571 2572 t->sysctl_header = register_net_sysctl(net, path, t->devinet_vars); 2573 if (!t->sysctl_header) 2574 goto free; 2575 2576 p->sysctl = t; 2577 2578 inet_netconf_notify_devconf(net, RTM_NEWNETCONF, NETCONFA_ALL, 2579 ifindex, p); 2580 return 0; 2581 2582free: 2583 kfree(t); 2584out: 2585 return -ENOMEM; 2586} 2587 2588static void __devinet_sysctl_unregister(struct net *net, 2589 struct ipv4_devconf *cnf, int ifindex) 2590{ 2591 struct devinet_sysctl_table *t = cnf->sysctl; 2592 2593 if (t) { 2594 cnf->sysctl = NULL; 2595 unregister_net_sysctl_table(t->sysctl_header); 2596 kfree(t); 2597 } 2598 2599 inet_netconf_notify_devconf(net, RTM_DELNETCONF, 0, ifindex, NULL); 2600} 2601 2602static int devinet_sysctl_register(struct in_device *idev) 2603{ 2604 int err; 2605 2606 if (!sysctl_dev_name_is_allowed(idev->dev->name)) 2607 return -EINVAL; 2608 2609 err = neigh_sysctl_register(idev->dev, idev->arp_parms, NULL); 2610 if (err) 2611 return err; 2612 err = __devinet_sysctl_register(dev_net(idev->dev), idev->dev->name, 2613 idev->dev->ifindex, &idev->cnf); 2614 if (err) 2615 neigh_sysctl_unregister(idev->arp_parms); 2616 return err; 2617} 2618 2619static void devinet_sysctl_unregister(struct in_device *idev) 2620{ 2621 struct net *net = dev_net(idev->dev); 2622 2623 __devinet_sysctl_unregister(net, &idev->cnf, idev->dev->ifindex); 2624 neigh_sysctl_unregister(idev->arp_parms); 2625} 2626 2627static struct ctl_table ctl_forward_entry[] = { 2628 { 2629 .procname = "ip_forward", 2630 .data = &ipv4_devconf.data[ 2631 IPV4_DEVCONF_FORWARDING - 1], 2632 .maxlen = sizeof(int), 2633 .mode = 0644, 2634 .proc_handler = devinet_sysctl_forward, 2635 .extra1 = &ipv4_devconf, 2636 .extra2 = &init_net, 2637 }, 2638 { }, 2639}; 2640#endif 2641 2642static __net_init int devinet_init_net(struct net *net) 2643{ 2644 int err; 2645 struct ipv4_devconf *all, *dflt; 2646#ifdef CONFIG_SYSCTL 2647 struct ctl_table *tbl; 2648 struct ctl_table_header *forw_hdr; 2649#endif 2650 2651 err = -ENOMEM; 2652 all = kmemdup(&ipv4_devconf, sizeof(ipv4_devconf), GFP_KERNEL); 2653 if (!all) 2654 goto err_alloc_all; 2655 2656 dflt = kmemdup(&ipv4_devconf_dflt, sizeof(ipv4_devconf_dflt), GFP_KERNEL); 2657 if (!dflt) 2658 goto err_alloc_dflt; 2659 2660#ifdef CONFIG_SYSCTL 2661 tbl = kmemdup(ctl_forward_entry, sizeof(ctl_forward_entry), GFP_KERNEL); 2662 if (!tbl) 2663 goto err_alloc_ctl; 2664 2665 tbl[0].data = &all->data[IPV4_DEVCONF_FORWARDING - 1]; 2666 tbl[0].extra1 = all; 2667 tbl[0].extra2 = net; 2668#endif 2669 2670 if (!net_eq(net, &init_net)) { 2671 switch (net_inherit_devconf()) { 2672 case 3: 2673 /* copy from the current netns */ 2674 memcpy(all, current->nsproxy->net_ns->ipv4.devconf_all, 2675 sizeof(ipv4_devconf)); 2676 memcpy(dflt, 2677 current->nsproxy->net_ns->ipv4.devconf_dflt, 2678 sizeof(ipv4_devconf_dflt)); 2679 break; 2680 case 0: 2681 case 1: 2682 /* copy from init_net */ 2683 memcpy(all, init_net.ipv4.devconf_all, 2684 sizeof(ipv4_devconf)); 2685 memcpy(dflt, init_net.ipv4.devconf_dflt, 2686 sizeof(ipv4_devconf_dflt)); 2687 break; 2688 case 2: 2689 /* use compiled values */ 2690 break; 2691 } 2692 } 2693 2694#ifdef CONFIG_SYSCTL 2695 err = __devinet_sysctl_register(net, "all", NETCONFA_IFINDEX_ALL, all); 2696 if (err < 0) 2697 goto err_reg_all; 2698 2699 err = __devinet_sysctl_register(net, "default", 2700 NETCONFA_IFINDEX_DEFAULT, dflt); 2701 if (err < 0) 2702 goto err_reg_dflt; 2703 2704 err = -ENOMEM; 2705 forw_hdr = register_net_sysctl(net, "net/ipv4", tbl); 2706 if (!forw_hdr) 2707 goto err_reg_ctl; 2708 net->ipv4.forw_hdr = forw_hdr; 2709#endif 2710 2711 net->ipv4.devconf_all = all; 2712 net->ipv4.devconf_dflt = dflt; 2713 return 0; 2714 2715#ifdef CONFIG_SYSCTL 2716err_reg_ctl: 2717 __devinet_sysctl_unregister(net, dflt, NETCONFA_IFINDEX_DEFAULT); 2718err_reg_dflt: 2719 __devinet_sysctl_unregister(net, all, NETCONFA_IFINDEX_ALL); 2720err_reg_all: 2721 kfree(tbl); 2722err_alloc_ctl: 2723#endif 2724 kfree(dflt); 2725err_alloc_dflt: 2726 kfree(all); 2727err_alloc_all: 2728 return err; 2729} 2730 2731static __net_exit void devinet_exit_net(struct net *net) 2732{ 2733#ifdef CONFIG_SYSCTL 2734 struct ctl_table *tbl; 2735 2736 tbl = net->ipv4.forw_hdr->ctl_table_arg; 2737 unregister_net_sysctl_table(net->ipv4.forw_hdr); 2738 __devinet_sysctl_unregister(net, net->ipv4.devconf_dflt, 2739 NETCONFA_IFINDEX_DEFAULT); 2740 __devinet_sysctl_unregister(net, net->ipv4.devconf_all, 2741 NETCONFA_IFINDEX_ALL); 2742 kfree(tbl); 2743#endif 2744 kfree(net->ipv4.devconf_dflt); 2745 kfree(net->ipv4.devconf_all); 2746} 2747 2748static __net_initdata struct pernet_operations devinet_ops = { 2749 .init = devinet_init_net, 2750 .exit = devinet_exit_net, 2751}; 2752 2753static struct rtnl_af_ops inet_af_ops __read_mostly = { 2754 .family = AF_INET, 2755 .fill_link_af = inet_fill_link_af, 2756 .get_link_af_size = inet_get_link_af_size, 2757 .validate_link_af = inet_validate_link_af, 2758 .set_link_af = inet_set_link_af, 2759}; 2760 2761void __init devinet_init(void) 2762{ 2763 int i; 2764 2765 for (i = 0; i < IN4_ADDR_HSIZE; i++) 2766 INIT_HLIST_HEAD(&inet_addr_lst[i]); 2767 2768 register_pernet_subsys(&devinet_ops); 2769 register_netdevice_notifier(&ip_netdev_notifier); 2770 2771 queue_delayed_work(system_power_efficient_wq, &check_lifetime_work, 0); 2772 2773 rtnl_af_register(&inet_af_ops); 2774 2775 rtnl_register(PF_INET, RTM_NEWADDR, inet_rtm_newaddr, NULL, 0); 2776 rtnl_register(PF_INET, RTM_DELADDR, inet_rtm_deladdr, NULL, 0); 2777 rtnl_register(PF_INET, RTM_GETADDR, NULL, inet_dump_ifaddr, 0); 2778 rtnl_register(PF_INET, RTM_GETNETCONF, inet_netconf_get_devconf, 2779 inet_netconf_dump_devconf, 0); 2780} 2781