ndisc.c 44.3 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 *	Neighbour Discovery for IPv6
3
 *	Linux INET6 implementation
L
Linus Torvalds 已提交
4 5
 *
 *	Authors:
6
 *	Pedro Roque		<roque@di.fc.ul.pt>
L
Linus Torvalds 已提交
7 8 9 10 11 12 13 14 15 16 17
 *	Mike Shaver		<shaver@ingenia.com>
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 */

/*
 *	Changes:
 *
18 19
 *	Pierre Ynard			:	export userland ND options
 *						through netlink (RDNSS support)
L
Linus Torvalds 已提交
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
 *	Lars Fenneberg			:	fixed MTU setting on receipt
 *						of an RA.
 *	Janos Farkas			:	kmalloc failure checks
 *	Alexey Kuznetsov		:	state machine reworked
 *						and moved to net/core.
 *	Pekka Savola			:	RFC2461 validation
 *	YOSHIFUJI Hideaki @USAGI	:	Verify ND options properly
 */

/* Set to 3 to get tracing... */
#define ND_DEBUG 1

#define ND_PRINTK(fmt, args...) do { if (net_ratelimit()) { printk(fmt, ## args); } } while(0)
#define ND_NOPRINTK(x...) do { ; } while(0)
#define ND_PRINTK0 ND_PRINTK
#define ND_PRINTK1 ND_NOPRINTK
#define ND_PRINTK2 ND_NOPRINTK
#define ND_PRINTK3 ND_NOPRINTK
#if ND_DEBUG >= 1
#undef ND_PRINTK1
#define ND_PRINTK1 ND_PRINTK
#endif
#if ND_DEBUG >= 2
#undef ND_PRINTK2
#define ND_PRINTK2 ND_PRINTK
#endif
#if ND_DEBUG >= 3
#undef ND_PRINTK3
#define ND_PRINTK3 ND_PRINTK
#endif

#include <linux/module.h>
#include <linux/errno.h>
#include <linux/types.h>
#include <linux/socket.h>
#include <linux/sockios.h>
#include <linux/sched.h>
#include <linux/net.h>
#include <linux/in6.h>
#include <linux/route.h>
#include <linux/init.h>
#include <linux/rcupdate.h>
#ifdef CONFIG_SYSCTL
#include <linux/sysctl.h>
#endif

66
#include <linux/if_addr.h>
L
Linus Torvalds 已提交
67 68 69 70 71 72 73 74 75 76 77 78 79 80 81
#include <linux/if_arp.h>
#include <linux/ipv6.h>
#include <linux/icmpv6.h>
#include <linux/jhash.h>

#include <net/sock.h>
#include <net/snmp.h>

#include <net/ipv6.h>
#include <net/protocol.h>
#include <net/ndisc.h>
#include <net/ip6_route.h>
#include <net/addrconf.h>
#include <net/icmp.h>

82 83 84
#include <net/netlink.h>
#include <linux/rtnetlink.h>

L
Linus Torvalds 已提交
85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160
#include <net/flow.h>
#include <net/ip6_checksum.h>
#include <linux/proc_fs.h>

#include <linux/netfilter.h>
#include <linux/netfilter_ipv6.h>

static u32 ndisc_hash(const void *pkey, const struct net_device *dev);
static int ndisc_constructor(struct neighbour *neigh);
static void ndisc_solicit(struct neighbour *neigh, struct sk_buff *skb);
static void ndisc_error_report(struct neighbour *neigh, struct sk_buff *skb);
static int pndisc_constructor(struct pneigh_entry *n);
static void pndisc_destructor(struct pneigh_entry *n);
static void pndisc_redo(struct sk_buff *skb);

static struct neigh_ops ndisc_generic_ops = {
	.family =		AF_INET6,
	.solicit =		ndisc_solicit,
	.error_report =		ndisc_error_report,
	.output =		neigh_resolve_output,
	.connected_output =	neigh_connected_output,
	.hh_output =		dev_queue_xmit,
	.queue_xmit =		dev_queue_xmit,
};

static struct neigh_ops ndisc_hh_ops = {
	.family =		AF_INET6,
	.solicit =		ndisc_solicit,
	.error_report =		ndisc_error_report,
	.output =		neigh_resolve_output,
	.connected_output =	neigh_resolve_output,
	.hh_output =		dev_queue_xmit,
	.queue_xmit =		dev_queue_xmit,
};


static struct neigh_ops ndisc_direct_ops = {
	.family =		AF_INET6,
	.output =		dev_queue_xmit,
	.connected_output =	dev_queue_xmit,
	.hh_output =		dev_queue_xmit,
	.queue_xmit =		dev_queue_xmit,
};

struct neigh_table nd_tbl = {
	.family =	AF_INET6,
	.entry_size =	sizeof(struct neighbour) + sizeof(struct in6_addr),
	.key_len =	sizeof(struct in6_addr),
	.hash =		ndisc_hash,
	.constructor =	ndisc_constructor,
	.pconstructor =	pndisc_constructor,
	.pdestructor =	pndisc_destructor,
	.proxy_redo =	pndisc_redo,
	.id =		"ndisc_cache",
	.parms = {
		.tbl =			&nd_tbl,
		.base_reachable_time =	30 * HZ,
		.retrans_time =	 1 * HZ,
		.gc_staletime =	60 * HZ,
		.reachable_time =		30 * HZ,
		.delay_probe_time =	 5 * HZ,
		.queue_len =		 3,
		.ucast_probes =	 3,
		.mcast_probes =	 3,
		.anycast_delay =	 1 * HZ,
		.proxy_delay =		(8 * HZ) / 10,
		.proxy_qlen =		64,
	},
	.gc_interval =	  30 * HZ,
	.gc_thresh1 =	 128,
	.gc_thresh2 =	 512,
	.gc_thresh3 =	1024,
};

/* ND options */
struct ndisc_options {
161 162 163 164 165
	struct nd_opt_hdr *nd_opt_array[__ND_OPT_ARRAY_MAX];
#ifdef CONFIG_IPV6_ROUTE_INFO
	struct nd_opt_hdr *nd_opts_ri;
	struct nd_opt_hdr *nd_opts_ri_end;
#endif
166 167
	struct nd_opt_hdr *nd_useropts;
	struct nd_opt_hdr *nd_useropts_end;
L
Linus Torvalds 已提交
168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231
};

#define nd_opts_src_lladdr	nd_opt_array[ND_OPT_SOURCE_LL_ADDR]
#define nd_opts_tgt_lladdr	nd_opt_array[ND_OPT_TARGET_LL_ADDR]
#define nd_opts_pi		nd_opt_array[ND_OPT_PREFIX_INFO]
#define nd_opts_pi_end		nd_opt_array[__ND_OPT_PREFIX_INFO_END]
#define nd_opts_rh		nd_opt_array[ND_OPT_REDIRECT_HDR]
#define nd_opts_mtu		nd_opt_array[ND_OPT_MTU]

#define NDISC_OPT_SPACE(len) (((len)+2+7)&~7)

/*
 * Return the padding between the option length and the start of the
 * link addr.  Currently only IP-over-InfiniBand needs this, although
 * if RFC 3831 IPv6-over-Fibre Channel is ever implemented it may
 * also need a pad of 2.
 */
static int ndisc_addr_option_pad(unsigned short type)
{
	switch (type) {
	case ARPHRD_INFINIBAND: return 2;
	default:                return 0;
	}
}

static inline int ndisc_opt_addr_space(struct net_device *dev)
{
	return NDISC_OPT_SPACE(dev->addr_len + ndisc_addr_option_pad(dev->type));
}

static u8 *ndisc_fill_addr_option(u8 *opt, int type, void *data, int data_len,
				  unsigned short addr_type)
{
	int space = NDISC_OPT_SPACE(data_len);
	int pad   = ndisc_addr_option_pad(addr_type);

	opt[0] = type;
	opt[1] = space>>3;

	memset(opt + 2, 0, pad);
	opt   += pad;
	space -= pad;

	memcpy(opt+2, data, data_len);
	data_len += 2;
	opt += data_len;
	if ((space -= data_len) > 0)
		memset(opt, 0, space);
	return opt + space;
}

static struct nd_opt_hdr *ndisc_next_option(struct nd_opt_hdr *cur,
					    struct nd_opt_hdr *end)
{
	int type;
	if (!cur || !end || cur >= end)
		return NULL;
	type = cur->nd_opt_type;
	do {
		cur = ((void *)cur) + (cur->nd_opt_len << 3);
	} while(cur < end && cur->nd_opt_type != type);
	return (cur <= end && cur->nd_opt_type == type ? cur : NULL);
}

232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247
static inline int ndisc_is_useropt(struct nd_opt_hdr *opt)
{
	return (opt->nd_opt_type == ND_OPT_RDNSS);
}

static struct nd_opt_hdr *ndisc_next_useropt(struct nd_opt_hdr *cur,
					     struct nd_opt_hdr *end)
{
	if (!cur || !end || cur >= end)
		return NULL;
	do {
		cur = ((void *)cur) + (cur->nd_opt_len << 3);
	} while(cur < end && !ndisc_is_useropt(cur));
	return (cur <= end && ndisc_is_useropt(cur) ? cur : NULL);
}

L
Linus Torvalds 已提交
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
static struct ndisc_options *ndisc_parse_options(u8 *opt, int opt_len,
						 struct ndisc_options *ndopts)
{
	struct nd_opt_hdr *nd_opt = (struct nd_opt_hdr *)opt;

	if (!nd_opt || opt_len < 0 || !ndopts)
		return NULL;
	memset(ndopts, 0, sizeof(*ndopts));
	while (opt_len) {
		int l;
		if (opt_len < sizeof(struct nd_opt_hdr))
			return NULL;
		l = nd_opt->nd_opt_len << 3;
		if (opt_len < l || l == 0)
			return NULL;
		switch (nd_opt->nd_opt_type) {
		case ND_OPT_SOURCE_LL_ADDR:
		case ND_OPT_TARGET_LL_ADDR:
		case ND_OPT_MTU:
		case ND_OPT_REDIRECT_HDR:
			if (ndopts->nd_opt_array[nd_opt->nd_opt_type]) {
				ND_PRINTK2(KERN_WARNING
					   "%s(): duplicated ND6 option found: type=%d\n",
271
					   __func__,
L
Linus Torvalds 已提交
272 273 274 275 276 277 278
					   nd_opt->nd_opt_type);
			} else {
				ndopts->nd_opt_array[nd_opt->nd_opt_type] = nd_opt;
			}
			break;
		case ND_OPT_PREFIX_INFO:
			ndopts->nd_opts_pi_end = nd_opt;
S
Stephen Hemminger 已提交
279
			if (!ndopts->nd_opt_array[nd_opt->nd_opt_type])
L
Linus Torvalds 已提交
280 281
				ndopts->nd_opt_array[nd_opt->nd_opt_type] = nd_opt;
			break;
282 283 284 285 286 287 288
#ifdef CONFIG_IPV6_ROUTE_INFO
		case ND_OPT_ROUTE_INFO:
			ndopts->nd_opts_ri_end = nd_opt;
			if (!ndopts->nd_opts_ri)
				ndopts->nd_opts_ri = nd_opt;
			break;
#endif
L
Linus Torvalds 已提交
289
		default:
290 291 292 293 294 295 296 297 298 299 300 301
			if (ndisc_is_useropt(nd_opt)) {
				ndopts->nd_useropts_end = nd_opt;
				if (!ndopts->nd_useropts)
					ndopts->nd_useropts = nd_opt;
			} else {
				/*
				 * Unknown options must be silently ignored,
				 * to accommodate future extension to the
				 * protocol.
				 */
				ND_PRINTK2(KERN_NOTICE
					   "%s(): ignored unsupported option; type=%d, len=%d\n",
302
					   __func__,
303 304
					   nd_opt->nd_opt_type, nd_opt->nd_opt_len);
			}
L
Linus Torvalds 已提交
305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337
		}
		opt_len -= l;
		nd_opt = ((void *)nd_opt) + l;
	}
	return ndopts;
}

static inline u8 *ndisc_opt_addr_data(struct nd_opt_hdr *p,
				      struct net_device *dev)
{
	u8 *lladdr = (u8 *)(p + 1);
	int lladdrlen = p->nd_opt_len << 3;
	int prepad = ndisc_addr_option_pad(dev->type);
	if (lladdrlen != NDISC_OPT_SPACE(dev->addr_len + prepad))
		return NULL;
	return (lladdr + prepad);
}

int ndisc_mc_map(struct in6_addr *addr, char *buf, struct net_device *dev, int dir)
{
	switch (dev->type) {
	case ARPHRD_ETHER:
	case ARPHRD_IEEE802:	/* Not sure. Check it later. --ANK */
	case ARPHRD_FDDI:
		ipv6_eth_mc_map(addr, buf);
		return 0;
	case ARPHRD_IEEE802_TR:
		ipv6_tr_mc_map(addr,buf);
		return 0;
	case ARPHRD_ARCNET:
		ipv6_arcnet_mc_map(addr, buf);
		return 0;
	case ARPHRD_INFINIBAND:
338
		ipv6_ib_mc_map(addr, dev->broadcast, buf);
L
Linus Torvalds 已提交
339 340 341 342 343 344 345 346 347 348
		return 0;
	default:
		if (dir) {
			memcpy(buf, dev->broadcast, dev->addr_len);
			return 0;
		}
	}
	return -EINVAL;
}

349 350
EXPORT_SYMBOL(ndisc_mc_map);

L
Linus Torvalds 已提交
351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383
static u32 ndisc_hash(const void *pkey, const struct net_device *dev)
{
	const u32 *p32 = pkey;
	u32 addr_hash, i;

	addr_hash = 0;
	for (i = 0; i < (sizeof(struct in6_addr) / sizeof(u32)); i++)
		addr_hash ^= *p32++;

	return jhash_2words(addr_hash, dev->ifindex, nd_tbl.hash_rnd);
}

static int ndisc_constructor(struct neighbour *neigh)
{
	struct in6_addr *addr = (struct in6_addr*)&neigh->primary_key;
	struct net_device *dev = neigh->dev;
	struct inet6_dev *in6_dev;
	struct neigh_parms *parms;
	int is_multicast = ipv6_addr_is_multicast(addr);

	rcu_read_lock();
	in6_dev = in6_dev_get(dev);
	if (in6_dev == NULL) {
		rcu_read_unlock();
		return -EINVAL;
	}

	parms = in6_dev->nd_parms;
	__neigh_parms_put(neigh->parms);
	neigh->parms = neigh_parms_clone(parms);
	rcu_read_unlock();

	neigh->type = is_multicast ? RTN_MULTICAST : RTN_UNICAST;
384
	if (!dev->header_ops) {
L
Linus Torvalds 已提交
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
		neigh->nud_state = NUD_NOARP;
		neigh->ops = &ndisc_direct_ops;
		neigh->output = neigh->ops->queue_xmit;
	} else {
		if (is_multicast) {
			neigh->nud_state = NUD_NOARP;
			ndisc_mc_map(addr, neigh->ha, dev, 1);
		} else if (dev->flags&(IFF_NOARP|IFF_LOOPBACK)) {
			neigh->nud_state = NUD_NOARP;
			memcpy(neigh->ha, dev->dev_addr, dev->addr_len);
			if (dev->flags&IFF_LOOPBACK)
				neigh->type = RTN_LOCAL;
		} else if (dev->flags&IFF_POINTOPOINT) {
			neigh->nud_state = NUD_NOARP;
			memcpy(neigh->ha, dev->broadcast, dev->addr_len);
		}
401
		if (dev->header_ops->cache)
L
Linus Torvalds 已提交
402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441
			neigh->ops = &ndisc_hh_ops;
		else
			neigh->ops = &ndisc_generic_ops;
		if (neigh->nud_state&NUD_VALID)
			neigh->output = neigh->ops->connected_output;
		else
			neigh->output = neigh->ops->output;
	}
	in6_dev_put(in6_dev);
	return 0;
}

static int pndisc_constructor(struct pneigh_entry *n)
{
	struct in6_addr *addr = (struct in6_addr*)&n->key;
	struct in6_addr maddr;
	struct net_device *dev = n->dev;

	if (dev == NULL || __in6_dev_get(dev) == NULL)
		return -EINVAL;
	addrconf_addr_solict_mult(addr, &maddr);
	ipv6_dev_mc_inc(dev, &maddr);
	return 0;
}

static void pndisc_destructor(struct pneigh_entry *n)
{
	struct in6_addr *addr = (struct in6_addr*)&n->key;
	struct in6_addr maddr;
	struct net_device *dev = n->dev;

	if (dev == NULL || __in6_dev_get(dev) == NULL)
		return;
	addrconf_addr_solict_mult(addr, &maddr);
	ipv6_dev_mc_dec(dev, &maddr);
}

/*
 *	Send a Neighbour Advertisement
 */
442 443 444 445
static void __ndisc_send(struct net_device *dev,
			 struct neighbour *neigh,
			 struct in6_addr *daddr, struct in6_addr *saddr,
			 struct icmp6hdr *icmp6h, struct in6_addr *target,
446
			 int llinfo)
L
Linus Torvalds 已提交
447 448
{
	struct flowi fl;
449
	struct dst_entry *dst;
450
	struct net *net = dev_net(dev);
451
	struct sock *sk = net->ipv6.ndisc_sk;
452
	struct sk_buff *skb;
453 454 455
	struct icmp6hdr *hdr;
	struct inet6_dev *idev;
	int len;
L
Linus Torvalds 已提交
456
	int err;
457 458 459
	u8 *opt, type;

	type = icmp6h->icmp6_type;
L
Linus Torvalds 已提交
460

461
	icmpv6_flow_init(sk, &fl, type, saddr, daddr, dev->ifindex);
L
Linus Torvalds 已提交
462

463
	dst = icmp6_dst_alloc(dev, neigh, daddr);
L
Linus Torvalds 已提交
464 465 466 467
	if (!dst)
		return;

	err = xfrm_lookup(&dst, &fl, NULL, 0);
468
	if (err < 0)
L
Linus Torvalds 已提交
469 470
		return;

471 472 473 474 475 476
	if (!dev->addr_len)
		llinfo = 0;

	len = sizeof(struct icmp6hdr) + (target ? sizeof(*target) : 0);
	if (llinfo)
		len += ndisc_opt_addr_space(dev);
L
Linus Torvalds 已提交
477

478 479 480
	skb = sock_alloc_send_skb(sk,
				  (MAX_HEADER + sizeof(struct ipv6hdr) +
				   len + LL_RESERVED_SPACE(dev)),
L
Linus Torvalds 已提交
481
				  1, &err);
482
	if (!skb) {
L
Linus Torvalds 已提交
483
		ND_PRINTK0(KERN_ERR
484
			   "ICMPv6 ND: %s() failed to allocate an skb.\n",
485
			   __func__);
L
Linus Torvalds 已提交
486 487 488 489 490
		dst_release(dst);
		return;
	}

	skb_reserve(skb, LL_RESERVED_SPACE(dev));
491
	ip6_nd_hdr(sk, skb, dev, saddr, daddr, IPPROTO_ICMPV6, len);
L
Linus Torvalds 已提交
492

493
	skb->transport_header = skb->tail;
494
	skb_put(skb, len);
L
Linus Torvalds 已提交
495

496 497
	hdr = (struct icmp6hdr *)skb_transport_header(skb);
	memcpy(hdr, icmp6h, sizeof(*hdr));
L
Linus Torvalds 已提交
498

499 500 501 502 503
	opt = skb_transport_header(skb) + sizeof(struct icmp6hdr);
	if (target) {
		ipv6_addr_copy((struct in6_addr *)opt, target);
		opt += sizeof(*target);
	}
L
Linus Torvalds 已提交
504

505 506
	if (llinfo)
		ndisc_fill_addr_option(opt, llinfo, dev->dev_addr,
L
Linus Torvalds 已提交
507 508
				       dev->addr_len, dev->type);

509 510 511 512
	hdr->icmp6_cksum = csum_ipv6_magic(saddr, daddr, len,
					   IPPROTO_ICMPV6,
					   csum_partial((__u8 *) hdr,
							len, 0));
L
Linus Torvalds 已提交
513 514

	skb->dst = dst;
515

L
Linus Torvalds 已提交
516
	idev = in6_dev_get(dst->dev);
517
	IP6_INC_STATS(idev, IPSTATS_MIB_OUTREQUESTS);
518

519 520
	err = NF_HOOK(PF_INET6, NF_INET_LOCAL_OUT, skb, NULL, dst->dev,
		      dst_output);
L
Linus Torvalds 已提交
521
	if (!err) {
522
		ICMP6MSGOUT_INC_STATS(idev, type);
L
Linus Torvalds 已提交
523 524 525 526 527
		ICMP6_INC_STATS(idev, ICMP6_MIB_OUTMSGS);
	}

	if (likely(idev != NULL))
		in6_dev_put(idev);
528
}
L
Linus Torvalds 已提交
529

530 531 532 533 534 535 536 537 538 539 540 541
static void ndisc_send_na(struct net_device *dev, struct neighbour *neigh,
		   struct in6_addr *daddr, struct in6_addr *solicited_addr,
		   int router, int solicited, int override, int inc_opt)
{
	struct in6_addr tmpaddr;
	struct inet6_ifaddr *ifp;
	struct in6_addr *src_addr;
	struct icmp6hdr icmp6h = {
		.icmp6_type = NDISC_NEIGHBOUR_ADVERTISEMENT,
	};

	/* for anycast or proxy, solicited_addr != src_addr */
542
	ifp = ipv6_get_ifaddr(dev_net(dev), solicited_addr, dev, 1);
543 544 545 546 547 548
	if (ifp) {
		src_addr = solicited_addr;
		if (ifp->flags & IFA_F_OPTIMISTIC)
			override = 0;
		in6_ifa_put(ifp);
	} else {
549
		if (ipv6_dev_get_saddr(dev, daddr,
550
				       inet6_sk(dev_net(dev)->ipv6.ndisc_sk)->srcprefs,
551
				       &tmpaddr))
552 553 554 555 556 557 558 559 560 561
			return;
		src_addr = &tmpaddr;
	}

	icmp6h.icmp6_router = router;
	icmp6h.icmp6_solicited = solicited;
	icmp6h.icmp6_override = override;

	__ndisc_send(dev, neigh, daddr, src_addr,
		     &icmp6h, solicited_addr,
562
		     inc_opt ? ND_OPT_TARGET_LL_ADDR : 0);
563 564
}

L
Linus Torvalds 已提交
565 566
void ndisc_send_ns(struct net_device *dev, struct neighbour *neigh,
		   struct in6_addr *solicit,
567
		   struct in6_addr *daddr, struct in6_addr *saddr)
L
Linus Torvalds 已提交
568 569
{
	struct in6_addr addr_buf;
570 571 572
	struct icmp6hdr icmp6h = {
		.icmp6_type = NDISC_NEIGHBOUR_SOLICITATION,
	};
L
Linus Torvalds 已提交
573 574

	if (saddr == NULL) {
575 576
		if (ipv6_get_lladdr(dev, &addr_buf,
				   (IFA_F_TENTATIVE|IFA_F_OPTIMISTIC)))
L
Linus Torvalds 已提交
577 578 579 580
			return;
		saddr = &addr_buf;
	}

581 582
	__ndisc_send(dev, neigh, daddr, saddr,
		     &icmp6h, solicit,
583
		     !ipv6_addr_any(saddr) ? ND_OPT_SOURCE_LL_ADDR : 0);
L
Linus Torvalds 已提交
584 585 586 587 588
}

void ndisc_send_rs(struct net_device *dev, struct in6_addr *saddr,
		   struct in6_addr *daddr)
{
589 590 591
	struct icmp6hdr icmp6h = {
		.icmp6_type = NDISC_ROUTER_SOLICITATION,
	};
592 593 594 595 596 597 598 599 600
	int send_sllao = dev->addr_len;

#ifdef CONFIG_IPV6_OPTIMISTIC_DAD
	/*
	 * According to section 2.2 of RFC 4429, we must not
	 * send router solicitations with a sllao from
	 * optimistic addresses, but we may send the solicitation
	 * if we don't include the sllao.  So here we check
	 * if our address is optimistic, and if so, we
J
Joe Perches 已提交
601
	 * suppress the inclusion of the sllao.
602 603
	 */
	if (send_sllao) {
604
		struct inet6_ifaddr *ifp = ipv6_get_ifaddr(dev_net(dev), saddr,
605
							   dev, 1);
606 607
		if (ifp) {
			if (ifp->flags & IFA_F_OPTIMISTIC)  {
608
				send_sllao = 0;
609
			}
610
			in6_ifa_put(ifp);
611 612 613 614 615
		} else {
			send_sllao = 0;
		}
	}
#endif
616 617
	__ndisc_send(dev, NULL, daddr, saddr,
		     &icmp6h, NULL,
618
		     send_sllao ? ND_OPT_SOURCE_LL_ADDR : 0);
L
Linus Torvalds 已提交
619
}
620

L
Linus Torvalds 已提交
621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641

static void ndisc_error_report(struct neighbour *neigh, struct sk_buff *skb)
{
	/*
	 *	"The sender MUST return an ICMP
	 *	 destination unreachable"
	 */
	dst_link_failure(skb);
	kfree_skb(skb);
}

/* Called with locked neigh: either read or both */

static void ndisc_solicit(struct neighbour *neigh, struct sk_buff *skb)
{
	struct in6_addr *saddr = NULL;
	struct in6_addr mcaddr;
	struct net_device *dev = neigh->dev;
	struct in6_addr *target = (struct in6_addr *)&neigh->primary_key;
	int probes = atomic_read(&neigh->probes);

642
	if (skb && ipv6_chk_addr(dev_net(dev), &ipv6_hdr(skb)->saddr, dev, 1))
643
		saddr = &ipv6_hdr(skb)->saddr;
L
Linus Torvalds 已提交
644 645 646 647 648

	if ((probes -= neigh->parms->ucast_probes) < 0) {
		if (!(neigh->nud_state & NUD_VALID)) {
			ND_PRINTK1(KERN_DEBUG
				   "%s(): trying to ucast probe in NUD_INVALID: "
J
Joe Perches 已提交
649
				   NIP6_FMT "\n",
650
				   __func__,
L
Linus Torvalds 已提交
651 652 653 654 655 656 657 658 659 660 661 662 663
				   NIP6(*target));
		}
		ndisc_send_ns(dev, neigh, target, target, saddr);
	} else if ((probes -= neigh->parms->app_probes) < 0) {
#ifdef CONFIG_ARPD
		neigh_app_ns(neigh);
#endif
	} else {
		addrconf_addr_solict_mult(target, &mcaddr);
		ndisc_send_ns(dev, NULL, target, &mcaddr, saddr);
	}
}

664 665
static int pndisc_is_router(const void *pkey,
			    struct net_device *dev)
666 667
{
	struct pneigh_entry *n;
668
	int ret = -1;
669 670

	read_lock_bh(&nd_tbl.lock);
671 672 673
	n = __pneigh_lookup(&nd_tbl, dev_net(dev), pkey, dev);
	if (n)
		ret = !!(n->flags & NTF_ROUTER);
674 675
	read_unlock_bh(&nd_tbl.lock);

676
	return ret;
677 678
}

L
Linus Torvalds 已提交
679 680
static void ndisc_recv_ns(struct sk_buff *skb)
{
681
	struct nd_msg *msg = (struct nd_msg *)skb_transport_header(skb);
682 683
	struct in6_addr *saddr = &ipv6_hdr(skb)->saddr;
	struct in6_addr *daddr = &ipv6_hdr(skb)->daddr;
L
Linus Torvalds 已提交
684
	u8 *lladdr = NULL;
685 686
	u32 ndoptlen = skb->tail - (skb->transport_header +
				    offsetof(struct nd_msg, opt));
L
Linus Torvalds 已提交
687 688 689 690 691 692 693
	struct ndisc_options ndopts;
	struct net_device *dev = skb->dev;
	struct inet6_ifaddr *ifp;
	struct inet6_dev *idev = NULL;
	struct neighbour *neigh;
	int dad = ipv6_addr_any(saddr);
	int inc;
694
	int is_router = -1;
L
Linus Torvalds 已提交
695 696

	if (ipv6_addr_is_multicast(&msg->target)) {
697
		ND_PRINTK2(KERN_WARNING
L
Linus Torvalds 已提交
698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
			   "ICMPv6 NS: multicast target address");
		return;
	}

	/*
	 * RFC2461 7.1.1:
	 * DAD has to be destined for solicited node multicast address.
	 */
	if (dad &&
	    !(daddr->s6_addr32[0] == htonl(0xff020000) &&
	      daddr->s6_addr32[1] == htonl(0x00000000) &&
	      daddr->s6_addr32[2] == htonl(0x00000001) &&
	      daddr->s6_addr [12] == 0xff )) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 NS: bad DAD packet (wrong destination)\n");
		return;
	}

	if (!ndisc_parse_options(msg->opt, ndoptlen, &ndopts)) {
717
		ND_PRINTK2(KERN_WARNING
L
Linus Torvalds 已提交
718 719 720 721 722 723 724 725 726 727 728 729 730
			   "ICMPv6 NS: invalid ND options\n");
		return;
	}

	if (ndopts.nd_opts_src_lladdr) {
		lladdr = ndisc_opt_addr_data(ndopts.nd_opts_src_lladdr, dev);
		if (!lladdr) {
			ND_PRINTK2(KERN_WARNING
				   "ICMPv6 NS: invalid link-layer address length\n");
			return;
		}

		/* RFC2461 7.1.1:
731 732
		 *	If the IP source address is the unspecified address,
		 *	there MUST NOT be source link-layer address option
L
Linus Torvalds 已提交
733 734 735
		 *	in the message.
		 */
		if (dad) {
736
			ND_PRINTK2(KERN_WARNING
L
Linus Torvalds 已提交
737 738 739 740 741 742 743
				   "ICMPv6 NS: bad DAD packet (link-layer address option)\n");
			return;
		}
	}

	inc = ipv6_addr_is_multicast(daddr);

744
	ifp = ipv6_get_ifaddr(dev_net(dev), &msg->target, dev, 1);
745
	if (ifp) {
746 747 748 749

		if (ifp->flags & (IFA_F_TENTATIVE|IFA_F_OPTIMISTIC)) {
			if (dad) {
				if (dev->type == ARPHRD_IEEE802_TR) {
750 751
					const unsigned char *sadr;
					sadr = skb_mac_header(skb);
752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768
					if (((sadr[8] ^ dev->dev_addr[0]) & 0x7f) == 0 &&
					    sadr[9] == dev->dev_addr[1] &&
					    sadr[10] == dev->dev_addr[2] &&
					    sadr[11] == dev->dev_addr[3] &&
					    sadr[12] == dev->dev_addr[4] &&
					    sadr[13] == dev->dev_addr[5]) {
						/* looped-back to us */
						goto out;
					}
				}

				/*
				 * We are colliding with another node
				 * who is doing DAD
				 * so fail our DAD process
				 */
				addrconf_dad_failure(ifp);
769
				return;
770 771 772 773 774 775 776 777
			} else {
				/*
				 * This is not a dad solicitation.
				 * If we are an optimistic node,
				 * we should respond.
				 * Otherwise, we should ignore it.
				 */
				if (!(ifp->flags & IFA_F_OPTIMISTIC))
L
Linus Torvalds 已提交
778 779 780 781 782 783 784 785 786 787 788 789
					goto out;
			}
		}

		idev = ifp->idev;
	} else {
		idev = in6_dev_get(dev);
		if (!idev) {
			/* XXX: count this drop? */
			return;
		}

790
		if (ipv6_chk_acast_addr(dev_net(dev), dev, &msg->target) ||
791
		    (idev->cnf.forwarding &&
792
		     (ipv6_devconf.proxy_ndp || idev->cnf.proxy_ndp) &&
793
		     (is_router = pndisc_is_router(&msg->target, dev)) >= 0)) {
794
			if (!(NEIGH_CB(skb)->flags & LOCALLY_ENQUEUED) &&
L
Linus Torvalds 已提交
795 796 797 798 799
			    skb->pkt_type != PACKET_HOST &&
			    inc != 0 &&
			    idev->nd_parms->proxy_delay != 0) {
				/*
				 * for anycast or proxy,
800 801
				 * sender should delay its response
				 * by a random time between 0 and
L
Linus Torvalds 已提交
802 803 804 805 806 807 808 809 810 811 812 813
				 * MAX_ANYCAST_DELAY_TIME seconds.
				 * (RFC2461) -- yoshfuji
				 */
				struct sk_buff *n = skb_clone(skb, GFP_ATOMIC);
				if (n)
					pneigh_enqueue(&nd_tbl, idev->nd_parms, n);
				goto out;
			}
		} else
			goto out;
	}

814 815
	if (is_router < 0)
		is_router = !!idev->cnf.forwarding;
816

L
Linus Torvalds 已提交
817 818 819 820 821
	if (dad) {
		struct in6_addr maddr;

		ipv6_addr_all_nodes(&maddr);
		ndisc_send_na(dev, NULL, &maddr, &msg->target,
822
			      is_router, 0, (ifp != NULL), 1);
L
Linus Torvalds 已提交
823 824 825 826 827 828 829 830
		goto out;
	}

	if (inc)
		NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_mcast);
	else
		NEIGH_CACHE_STAT_INC(&nd_tbl, rcv_probes_ucast);

831
	/*
L
Linus Torvalds 已提交
832 833 834 835 836 837
	 *	update / create cache entry
	 *	for the source address
	 */
	neigh = __neigh_lookup(&nd_tbl, saddr, dev,
			       !inc || lladdr || !dev->addr_len);
	if (neigh)
838
		neigh_update(neigh, lladdr, NUD_STALE,
L
Linus Torvalds 已提交
839 840
			     NEIGH_UPDATE_F_WEAK_OVERRIDE|
			     NEIGH_UPDATE_F_OVERRIDE);
841
	if (neigh || !dev->header_ops) {
L
Linus Torvalds 已提交
842
		ndisc_send_na(dev, neigh, saddr, &msg->target,
843
			      is_router,
L
Linus Torvalds 已提交
844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859
			      1, (ifp != NULL && inc), inc);
		if (neigh)
			neigh_release(neigh);
	}

out:
	if (ifp)
		in6_ifa_put(ifp);
	else
		in6_dev_put(idev);

	return;
}

static void ndisc_recv_na(struct sk_buff *skb)
{
860
	struct nd_msg *msg = (struct nd_msg *)skb_transport_header(skb);
861 862
	struct in6_addr *saddr = &ipv6_hdr(skb)->saddr;
	struct in6_addr *daddr = &ipv6_hdr(skb)->daddr;
L
Linus Torvalds 已提交
863
	u8 *lladdr = NULL;
864 865
	u32 ndoptlen = skb->tail - (skb->transport_header +
				    offsetof(struct nd_msg, opt));
L
Linus Torvalds 已提交
866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888
	struct ndisc_options ndopts;
	struct net_device *dev = skb->dev;
	struct inet6_ifaddr *ifp;
	struct neighbour *neigh;

	if (skb->len < sizeof(struct nd_msg)) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 NA: packet too short\n");
		return;
	}

	if (ipv6_addr_is_multicast(&msg->target)) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 NA: target address is multicast.\n");
		return;
	}

	if (ipv6_addr_is_multicast(daddr) &&
	    msg->icmph.icmp6_solicited) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 NA: solicited NA is multicasted.\n");
		return;
	}
889

L
Linus Torvalds 已提交
890 891 892 893 894 895 896 897 898 899 900 901 902
	if (!ndisc_parse_options(msg->opt, ndoptlen, &ndopts)) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 NS: invalid ND option\n");
		return;
	}
	if (ndopts.nd_opts_tgt_lladdr) {
		lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr, dev);
		if (!lladdr) {
			ND_PRINTK2(KERN_WARNING
				   "ICMPv6 NA: invalid link-layer address length\n");
			return;
		}
	}
903
	ifp = ipv6_get_ifaddr(dev_net(dev), &msg->target, dev, 1);
904
	if (ifp) {
L
Linus Torvalds 已提交
905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927
		if (ifp->flags & IFA_F_TENTATIVE) {
			addrconf_dad_failure(ifp);
			return;
		}
		/* What should we make now? The advertisement
		   is invalid, but ndisc specs say nothing
		   about it. It could be misconfiguration, or
		   an smart proxy agent tries to help us :-)
		 */
		ND_PRINTK1(KERN_WARNING
			   "ICMPv6 NA: someone advertises our address on %s!\n",
			   ifp->idev->dev->name);
		in6_ifa_put(ifp);
		return;
	}
	neigh = neigh_lookup(&nd_tbl, &msg->target, dev);

	if (neigh) {
		u8 old_flags = neigh->flags;

		if (neigh->nud_state & NUD_FAILED)
			goto out;

928 929 930 931 932 933
		/*
		 * Don't update the neighbor cache entry on a proxy NA from
		 * ourselves because either the proxied node is off link or it
		 * has already sent a NA to us.
		 */
		if (lladdr && !memcmp(lladdr, dev->dev_addr, dev->addr_len) &&
934
		    ipv6_devconf.forwarding && ipv6_devconf.proxy_ndp &&
935
		    pneigh_lookup(&nd_tbl, dev_net(dev), &msg->target, dev, 0)) {
936
			/* XXX: idev->cnf.prixy_ndp */
937
			goto out;
938
		}
939

L
Linus Torvalds 已提交
940 941 942 943 944 945 946 947 948 949 950 951 952 953
		neigh_update(neigh, lladdr,
			     msg->icmph.icmp6_solicited ? NUD_REACHABLE : NUD_STALE,
			     NEIGH_UPDATE_F_WEAK_OVERRIDE|
			     (msg->icmph.icmp6_override ? NEIGH_UPDATE_F_OVERRIDE : 0)|
			     NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
			     (msg->icmph.icmp6_router ? NEIGH_UPDATE_F_ISROUTER : 0));

		if ((old_flags & ~neigh->flags) & NTF_ROUTER) {
			/*
			 * Change: router to host
			 */
			struct rt6_info *rt;
			rt = rt6_get_dflt_router(saddr, dev);
			if (rt)
954
				ip6_del_rt(rt);
L
Linus Torvalds 已提交
955 956 957 958 959 960 961 962 963
		}

out:
		neigh_release(neigh);
	}
}

static void ndisc_recv_rs(struct sk_buff *skb)
{
964
	struct rs_msg *rs_msg = (struct rs_msg *)skb_transport_header(skb);
L
Linus Torvalds 已提交
965 966 967
	unsigned long ndoptlen = skb->len - sizeof(*rs_msg);
	struct neighbour *neigh;
	struct inet6_dev *idev;
968
	struct in6_addr *saddr = &ipv6_hdr(skb)->saddr;
L
Linus Torvalds 已提交
969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018
	struct ndisc_options ndopts;
	u8 *lladdr = NULL;

	if (skb->len < sizeof(*rs_msg))
		return;

	idev = in6_dev_get(skb->dev);
	if (!idev) {
		if (net_ratelimit())
			ND_PRINTK1("ICMP6 RS: can't find in6 device\n");
		return;
	}

	/* Don't accept RS if we're not in router mode */
	if (!idev->cnf.forwarding)
		goto out;

	/*
	 * Don't update NCE if src = ::;
	 * this implies that the source node has no ip address assigned yet.
	 */
	if (ipv6_addr_any(saddr))
		goto out;

	/* Parse ND options */
	if (!ndisc_parse_options(rs_msg->opt, ndoptlen, &ndopts)) {
		if (net_ratelimit())
			ND_PRINTK2("ICMP6 NS: invalid ND option, ignored\n");
		goto out;
	}

	if (ndopts.nd_opts_src_lladdr) {
		lladdr = ndisc_opt_addr_data(ndopts.nd_opts_src_lladdr,
					     skb->dev);
		if (!lladdr)
			goto out;
	}

	neigh = __neigh_lookup(&nd_tbl, saddr, skb->dev, 1);
	if (neigh) {
		neigh_update(neigh, lladdr, NUD_STALE,
			     NEIGH_UPDATE_F_WEAK_OVERRIDE|
			     NEIGH_UPDATE_F_OVERRIDE|
			     NEIGH_UPDATE_F_OVERRIDE_ISROUTER);
		neigh_release(neigh);
	}
out:
	in6_dev_put(idev);
}

1019 1020 1021 1022 1023 1024
static void ndisc_ra_useropt(struct sk_buff *ra, struct nd_opt_hdr *opt)
{
	struct icmp6hdr *icmp6h = (struct icmp6hdr *)skb_transport_header(ra);
	struct sk_buff *skb;
	struct nlmsghdr *nlh;
	struct nduseroptmsg *ndmsg;
1025
	struct net *net = dev_net(ra->dev);
1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
	int err;
	int base_size = NLMSG_ALIGN(sizeof(struct nduseroptmsg)
				    + (opt->nd_opt_len << 3));
	size_t msg_size = base_size + nla_total_size(sizeof(struct in6_addr));

	skb = nlmsg_new(msg_size, GFP_ATOMIC);
	if (skb == NULL) {
		err = -ENOBUFS;
		goto errout;
	}

	nlh = nlmsg_put(skb, 0, 0, RTM_NEWNDUSEROPT, base_size, 0);
	if (nlh == NULL) {
		goto nla_put_failure;
	}

	ndmsg = nlmsg_data(nlh);
	ndmsg->nduseropt_family = AF_INET6;
1044
	ndmsg->nduseropt_ifindex = ra->dev->ifindex;
1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
	ndmsg->nduseropt_icmp_type = icmp6h->icmp6_type;
	ndmsg->nduseropt_icmp_code = icmp6h->icmp6_code;
	ndmsg->nduseropt_opts_len = opt->nd_opt_len << 3;

	memcpy(ndmsg + 1, opt, opt->nd_opt_len << 3);

	NLA_PUT(skb, NDUSEROPT_SRCADDR, sizeof(struct in6_addr),
		&ipv6_hdr(ra)->saddr);
	nlmsg_end(skb, nlh);

1055
	err = rtnl_notify(skb, net, 0, RTNLGRP_ND_USEROPT, NULL,
1056
			  GFP_ATOMIC);
1057 1058 1059 1060 1061 1062 1063 1064 1065
	if (err < 0)
		goto errout;

	return;

nla_put_failure:
	nlmsg_free(skb);
	err = -EMSGSIZE;
errout:
1066
	rtnl_set_sk_err(net, RTNLGRP_ND_USEROPT, err);
1067 1068
}

L
Linus Torvalds 已提交
1069 1070
static void ndisc_router_discovery(struct sk_buff *skb)
{
1071
	struct ra_msg *ra_msg = (struct ra_msg *)skb_transport_header(skb);
L
Linus Torvalds 已提交
1072 1073
	struct neighbour *neigh = NULL;
	struct inet6_dev *in6_dev;
1074
	struct rt6_info *rt = NULL;
L
Linus Torvalds 已提交
1075 1076 1077
	int lifetime;
	struct ndisc_options ndopts;
	int optlen;
1078
	unsigned int pref = 0;
L
Linus Torvalds 已提交
1079 1080 1081

	__u8 * opt = (__u8 *)(ra_msg + 1);

1082
	optlen = (skb->tail - skb->transport_header) - sizeof(struct ra_msg);
L
Linus Torvalds 已提交
1083

1084
	if (!(ipv6_addr_type(&ipv6_hdr(skb)->saddr) & IPV6_ADDR_LINKLOCAL)) {
L
Linus Torvalds 已提交
1085 1086 1087 1088 1089
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 RA: source address is not link-local.\n");
		return;
	}
	if (optlen < 0) {
1090
		ND_PRINTK2(KERN_WARNING
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136
			   "ICMPv6 RA: packet too short\n");
		return;
	}

	/*
	 *	set the RA_RECV flag in the interface
	 */

	in6_dev = in6_dev_get(skb->dev);
	if (in6_dev == NULL) {
		ND_PRINTK0(KERN_ERR
			   "ICMPv6 RA: can't find inet6 device for %s.\n",
			   skb->dev->name);
		return;
	}
	if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_ra) {
		in6_dev_put(in6_dev);
		return;
	}

	if (!ndisc_parse_options(opt, optlen, &ndopts)) {
		in6_dev_put(in6_dev);
		ND_PRINTK2(KERN_WARNING
			   "ICMP6 RA: invalid ND options\n");
		return;
	}

	if (in6_dev->if_flags & IF_RS_SENT) {
		/*
		 *	flag that an RA was received after an RS was sent
		 *	out on this interface.
		 */
		in6_dev->if_flags |= IF_RA_RCVD;
	}

	/*
	 * Remember the managed/otherconf flags from most recently
	 * received RA message (RFC 2462) -- yoshfuji
	 */
	in6_dev->if_flags = (in6_dev->if_flags & ~(IF_RA_MANAGED |
				IF_RA_OTHERCONF)) |
				(ra_msg->icmph.icmp6_addrconf_managed ?
					IF_RA_MANAGED : 0) |
				(ra_msg->icmph.icmp6_addrconf_other ?
					IF_RA_OTHERCONF : 0);

1137 1138 1139
	if (!in6_dev->cnf.accept_ra_defrtr)
		goto skip_defrtr;

L
Linus Torvalds 已提交
1140 1141
	lifetime = ntohs(ra_msg->icmph.icmp6_rt_lifetime);

1142 1143 1144
#ifdef CONFIG_IPV6_ROUTER_PREF
	pref = ra_msg->icmph.icmp6_router_pref;
	/* 10b is handled as if it were 00b (medium) */
1145
	if (pref == ICMPV6_ROUTER_PREF_INVALID ||
1146
	    !in6_dev->cnf.accept_ra_rtr_pref)
1147 1148 1149
		pref = ICMPV6_ROUTER_PREF_MEDIUM;
#endif

1150
	rt = rt6_get_dflt_router(&ipv6_hdr(skb)->saddr, skb->dev);
L
Linus Torvalds 已提交
1151 1152 1153 1154 1155 1156

	if (rt)
		neigh = rt->rt6i_nexthop;

	if (rt && lifetime == 0) {
		neigh_clone(neigh);
1157
		ip6_del_rt(rt);
L
Linus Torvalds 已提交
1158 1159 1160 1161 1162 1163 1164
		rt = NULL;
	}

	if (rt == NULL && lifetime) {
		ND_PRINTK3(KERN_DEBUG
			   "ICMPv6 RA: adding default router.\n");

1165
		rt = rt6_add_dflt_router(&ipv6_hdr(skb)->saddr, skb->dev, pref);
L
Linus Torvalds 已提交
1166 1167 1168
		if (rt == NULL) {
			ND_PRINTK0(KERN_ERR
				   "ICMPv6 RA: %s() failed to add default route.\n",
1169
				   __func__);
L
Linus Torvalds 已提交
1170 1171 1172 1173 1174 1175 1176 1177
			in6_dev_put(in6_dev);
			return;
		}

		neigh = rt->rt6i_nexthop;
		if (neigh == NULL) {
			ND_PRINTK0(KERN_ERR
				   "ICMPv6 RA: %s() got default router without neighbour.\n",
1178
				   __func__);
L
Linus Torvalds 已提交
1179 1180 1181 1182 1183
			dst_release(&rt->u.dst);
			in6_dev_put(in6_dev);
			return;
		}
		neigh->flags |= NTF_ROUTER;
1184 1185
	} else if (rt) {
		rt->rt6i_flags |= (rt->rt6i_flags & ~RTF_PREF_MASK) | RTF_PREF(pref);
L
Linus Torvalds 已提交
1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
	}

	if (rt)
		rt->rt6i_expires = jiffies + (HZ * lifetime);

	if (ra_msg->icmph.icmp6_hop_limit) {
		in6_dev->cnf.hop_limit = ra_msg->icmph.icmp6_hop_limit;
		if (rt)
			rt->u.dst.metrics[RTAX_HOPLIMIT-1] = ra_msg->icmph.icmp6_hop_limit;
	}

1197 1198
skip_defrtr:

L
Linus Torvalds 已提交
1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236
	/*
	 *	Update Reachable Time and Retrans Timer
	 */

	if (in6_dev->nd_parms) {
		unsigned long rtime = ntohl(ra_msg->retrans_timer);

		if (rtime && rtime/1000 < MAX_SCHEDULE_TIMEOUT/HZ) {
			rtime = (rtime*HZ)/1000;
			if (rtime < HZ/10)
				rtime = HZ/10;
			in6_dev->nd_parms->retrans_time = rtime;
			in6_dev->tstamp = jiffies;
			inet6_ifinfo_notify(RTM_NEWLINK, in6_dev);
		}

		rtime = ntohl(ra_msg->reachable_time);
		if (rtime && rtime/1000 < MAX_SCHEDULE_TIMEOUT/(3*HZ)) {
			rtime = (rtime*HZ)/1000;

			if (rtime < HZ/10)
				rtime = HZ/10;

			if (rtime != in6_dev->nd_parms->base_reachable_time) {
				in6_dev->nd_parms->base_reachable_time = rtime;
				in6_dev->nd_parms->gc_staletime = 3 * rtime;
				in6_dev->nd_parms->reachable_time = neigh_rand_reach_time(rtime);
				in6_dev->tstamp = jiffies;
				inet6_ifinfo_notify(RTM_NEWLINK, in6_dev);
			}
		}
	}

	/*
	 *	Process options.
	 */

	if (!neigh)
1237
		neigh = __neigh_lookup(&nd_tbl, &ipv6_hdr(skb)->saddr,
L
Linus Torvalds 已提交
1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256
				       skb->dev, 1);
	if (neigh) {
		u8 *lladdr = NULL;
		if (ndopts.nd_opts_src_lladdr) {
			lladdr = ndisc_opt_addr_data(ndopts.nd_opts_src_lladdr,
						     skb->dev);
			if (!lladdr) {
				ND_PRINTK2(KERN_WARNING
					   "ICMPv6 RA: invalid link-layer address length\n");
				goto out;
			}
		}
		neigh_update(neigh, lladdr, NUD_STALE,
			     NEIGH_UPDATE_F_WEAK_OVERRIDE|
			     NEIGH_UPDATE_F_OVERRIDE|
			     NEIGH_UPDATE_F_OVERRIDE_ISROUTER|
			     NEIGH_UPDATE_F_ISROUTER);
	}

1257
#ifdef CONFIG_IPV6_ROUTE_INFO
1258
	if (in6_dev->cnf.accept_ra_rtr_pref && ndopts.nd_opts_ri) {
1259 1260 1261 1262
		struct nd_opt_hdr *p;
		for (p = ndopts.nd_opts_ri;
		     p;
		     p = ndisc_next_option(p, ndopts.nd_opts_ri_end)) {
1263 1264
			if (((struct route_info *)p)->prefix_len > in6_dev->cnf.accept_ra_rt_info_max_plen)
				continue;
1265
			rt6_route_rcv(skb->dev, (u8*)p, (p->nd_opt_len) << 3,
1266
				      &ipv6_hdr(skb)->saddr);
1267 1268 1269 1270
		}
	}
#endif

1271
	if (in6_dev->cnf.accept_ra_pinfo && ndopts.nd_opts_pi) {
L
Linus Torvalds 已提交
1272 1273 1274 1275 1276 1277 1278 1279 1280
		struct nd_opt_hdr *p;
		for (p = ndopts.nd_opts_pi;
		     p;
		     p = ndisc_next_option(p, ndopts.nd_opts_pi_end)) {
			addrconf_prefix_rcv(skb->dev, (u8*)p, (p->nd_opt_len) << 3);
		}
	}

	if (ndopts.nd_opts_mtu) {
A
Al Viro 已提交
1281
		__be32 n;
L
Linus Torvalds 已提交
1282 1283
		u32 mtu;

A
Al Viro 已提交
1284 1285
		memcpy(&n, ((u8*)(ndopts.nd_opts_mtu+1))+2, sizeof(mtu));
		mtu = ntohl(n);
L
Linus Torvalds 已提交
1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299

		if (mtu < IPV6_MIN_MTU || mtu > skb->dev->mtu) {
			ND_PRINTK2(KERN_WARNING
				   "ICMPv6 RA: invalid mtu: %d\n",
				   mtu);
		} else if (in6_dev->cnf.mtu6 != mtu) {
			in6_dev->cnf.mtu6 = mtu;

			if (rt)
				rt->u.dst.metrics[RTAX_MTU-1] = mtu;

			rt6_mtu_change(skb->dev, mtu);
		}
	}
1300

1301
	if (ndopts.nd_useropts) {
1302 1303 1304 1305 1306
		struct nd_opt_hdr *p;
		for (p = ndopts.nd_useropts;
		     p;
		     p = ndisc_next_useropt(p, ndopts.nd_useropts_end)) {
			ndisc_ra_useropt(skb, p);
1307 1308 1309
		}
	}

L
Linus Torvalds 已提交
1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333
	if (ndopts.nd_opts_tgt_lladdr || ndopts.nd_opts_rh) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 RA: invalid RA options");
	}
out:
	if (rt)
		dst_release(&rt->u.dst);
	else if (neigh)
		neigh_release(neigh);
	in6_dev_put(in6_dev);
}

static void ndisc_redirect_rcv(struct sk_buff *skb)
{
	struct inet6_dev *in6_dev;
	struct icmp6hdr *icmph;
	struct in6_addr *dest;
	struct in6_addr *target;	/* new first hop to destination */
	struct neighbour *neigh;
	int on_link = 0;
	struct ndisc_options ndopts;
	int optlen;
	u8 *lladdr = NULL;

1334
	if (!(ipv6_addr_type(&ipv6_hdr(skb)->saddr) & IPV6_ADDR_LINKLOCAL)) {
L
Linus Torvalds 已提交
1335 1336 1337 1338 1339
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 Redirect: source address is not link-local.\n");
		return;
	}

1340
	optlen = skb->tail - skb->transport_header;
L
Linus Torvalds 已提交
1341 1342 1343 1344 1345 1346 1347 1348
	optlen -= sizeof(struct icmp6hdr) + 2 * sizeof(struct in6_addr);

	if (optlen < 0) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 Redirect: packet too short\n");
		return;
	}

1349
	icmph = icmp6_hdr(skb);
L
Linus Torvalds 已提交
1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360
	target = (struct in6_addr *) (icmph + 1);
	dest = target + 1;

	if (ipv6_addr_is_multicast(dest)) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 Redirect: destination address is multicast.\n");
		return;
	}

	if (ipv6_addr_equal(dest, target)) {
		on_link = 1;
1361 1362
	} else if (ipv6_addr_type(target) !=
		   (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
1363
		ND_PRINTK2(KERN_WARNING
1364
			   "ICMPv6 Redirect: target address is not link-local unicast.\n");
L
Linus Torvalds 已提交
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
		return;
	}

	in6_dev = in6_dev_get(skb->dev);
	if (!in6_dev)
		return;
	if (in6_dev->cnf.forwarding || !in6_dev->cnf.accept_redirects) {
		in6_dev_put(in6_dev);
		return;
	}

1376
	/* RFC2461 8.1:
L
Linus Torvalds 已提交
1377 1378 1379
	 *	The IP source address of the Redirect MUST be the same as the current
	 *	first-hop router for the specified ICMP Destination Address.
	 */
1380

L
Linus Torvalds 已提交
1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399
	if (!ndisc_parse_options((u8*)(dest + 1), optlen, &ndopts)) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 Redirect: invalid ND options\n");
		in6_dev_put(in6_dev);
		return;
	}
	if (ndopts.nd_opts_tgt_lladdr) {
		lladdr = ndisc_opt_addr_data(ndopts.nd_opts_tgt_lladdr,
					     skb->dev);
		if (!lladdr) {
			ND_PRINTK2(KERN_WARNING
				   "ICMPv6 Redirect: invalid link-layer address length\n");
			in6_dev_put(in6_dev);
			return;
		}
	}

	neigh = __neigh_lookup(&nd_tbl, target, skb->dev, 1);
	if (neigh) {
1400 1401
		rt6_redirect(dest, &ipv6_hdr(skb)->daddr,
			     &ipv6_hdr(skb)->saddr, neigh, lladdr,
L
Linus Torvalds 已提交
1402 1403 1404 1405 1406 1407 1408 1409 1410
			     on_link);
		neigh_release(neigh);
	}
	in6_dev_put(in6_dev);
}

void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
			 struct in6_addr *target)
{
1411
	struct net_device *dev = skb->dev;
1412
	struct net *net = dev_net(dev);
1413
	struct sock *sk = net->ipv6.ndisc_sk;
L
Linus Torvalds 已提交
1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427
	int len = sizeof(struct icmp6hdr) + 2 * sizeof(struct in6_addr);
	struct sk_buff *buff;
	struct icmp6hdr *icmph;
	struct in6_addr saddr_buf;
	struct in6_addr *addrp;
	struct rt6_info *rt;
	struct dst_entry *dst;
	struct inet6_dev *idev;
	struct flowi fl;
	u8 *opt;
	int rd_len;
	int err;
	u8 ha_buf[MAX_ADDR_LEN], *ha = NULL;

1428
	if (ipv6_get_lladdr(dev, &saddr_buf, IFA_F_TENTATIVE)) {
L
Linus Torvalds 已提交
1429 1430 1431
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 Redirect: no link-local address on %s\n",
			   dev->name);
1432 1433
		return;
	}
L
Linus Torvalds 已提交
1434

1435
	if (!ipv6_addr_equal(&ipv6_hdr(skb)->daddr, target) &&
1436
	    ipv6_addr_type(target) != (IPV6_ADDR_UNICAST|IPV6_ADDR_LINKLOCAL)) {
1437
		ND_PRINTK2(KERN_WARNING
1438
			"ICMPv6 Redirect: target address is not link-local unicast.\n");
1439 1440 1441
		return;
	}

1442
	icmpv6_flow_init(sk, &fl, NDISC_REDIRECT,
1443
			 &saddr_buf, &ipv6_hdr(skb)->saddr, dev->ifindex);
L
Linus Torvalds 已提交
1444

1445
	dst = ip6_route_output(net, NULL, &fl);
L
Linus Torvalds 已提交
1446 1447 1448 1449
	if (dst == NULL)
		return;

	err = xfrm_lookup(&dst, &fl, NULL, 0);
1450
	if (err)
L
Linus Torvalds 已提交
1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481
		return;

	rt = (struct rt6_info *) dst;

	if (rt->rt6i_flags & RTF_GATEWAY) {
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 Redirect: destination is not a neighbour.\n");
		dst_release(dst);
		return;
	}
	if (!xrlim_allow(dst, 1*HZ)) {
		dst_release(dst);
		return;
	}

	if (dev->addr_len) {
		read_lock_bh(&neigh->lock);
		if (neigh->nud_state & NUD_VALID) {
			memcpy(ha_buf, neigh->ha, dev->addr_len);
			read_unlock_bh(&neigh->lock);
			ha = ha_buf;
			len += ndisc_opt_addr_space(dev);
		} else
			read_unlock_bh(&neigh->lock);
	}

	rd_len = min_t(unsigned int,
		     IPV6_MIN_MTU-sizeof(struct ipv6hdr)-len, skb->len + 8);
	rd_len &= ~0x7;
	len += rd_len;

1482 1483 1484
	buff = sock_alloc_send_skb(sk,
				   (MAX_HEADER + sizeof(struct ipv6hdr) +
				    len + LL_RESERVED_SPACE(dev)),
L
Linus Torvalds 已提交
1485 1486 1487 1488
				   1, &err);
	if (buff == NULL) {
		ND_PRINTK0(KERN_ERR
			   "ICMPv6 Redirect: %s() failed to allocate an skb.\n",
1489
			   __func__);
L
Linus Torvalds 已提交
1490 1491 1492 1493 1494
		dst_release(dst);
		return;
	}

	skb_reserve(buff, LL_RESERVED_SPACE(dev));
1495
	ip6_nd_hdr(sk, buff, dev, &saddr_buf, &ipv6_hdr(skb)->saddr,
L
Linus Torvalds 已提交
1496 1497
		   IPPROTO_ICMPV6, len);

1498
	skb_set_transport_header(buff, skb_tail_pointer(buff) - buff->data);
1499 1500
	skb_put(buff, len);
	icmph = icmp6_hdr(buff);
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511

	memset(icmph, 0, sizeof(struct icmp6hdr));
	icmph->icmp6_type = NDISC_REDIRECT;

	/*
	 *	copy target and destination addresses
	 */

	addrp = (struct in6_addr *)(icmph + 1);
	ipv6_addr_copy(addrp, target);
	addrp++;
1512
	ipv6_addr_copy(addrp, &ipv6_hdr(skb)->daddr);
L
Linus Torvalds 已提交
1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527

	opt = (u8*) (addrp + 1);

	/*
	 *	include target_address option
	 */

	if (ha)
		opt = ndisc_fill_addr_option(opt, ND_OPT_TARGET_LL_ADDR, ha,
					     dev->addr_len, dev->type);

	/*
	 *	build redirect option and copy skb over to the new packet.
	 */

1528
	memset(opt, 0, 8);
L
Linus Torvalds 已提交
1529 1530 1531 1532
	*(opt++) = ND_OPT_REDIRECT_HDR;
	*(opt++) = (rd_len >> 3);
	opt += 6;

1533
	memcpy(opt, ipv6_hdr(skb), rd_len - 8);
L
Linus Torvalds 已提交
1534

1535
	icmph->icmp6_cksum = csum_ipv6_magic(&saddr_buf, &ipv6_hdr(skb)->saddr,
L
Linus Torvalds 已提交
1536 1537 1538 1539 1540
					     len, IPPROTO_ICMPV6,
					     csum_partial((u8 *) icmph, len, 0));

	buff->dst = dst;
	idev = in6_dev_get(dst->dev);
1541
	IP6_INC_STATS(idev, IPSTATS_MIB_OUTREQUESTS);
1542 1543
	err = NF_HOOK(PF_INET6, NF_INET_LOCAL_OUT, buff, NULL, dst->dev,
		      dst_output);
L
Linus Torvalds 已提交
1544
	if (!err) {
1545
		ICMP6MSGOUT_INC_STATS(idev, NDISC_REDIRECT);
L
Linus Torvalds 已提交
1546 1547 1548 1549 1550 1551 1552 1553 1554
		ICMP6_INC_STATS(idev, ICMP6_MIB_OUTMSGS);
	}

	if (likely(idev != NULL))
		in6_dev_put(idev);
}

static void pndisc_redo(struct sk_buff *skb)
{
1555
	ndisc_recv_ns(skb);
L
Linus Torvalds 已提交
1556 1557 1558 1559 1560 1561 1562 1563 1564 1565
	kfree_skb(skb);
}

int ndisc_rcv(struct sk_buff *skb)
{
	struct nd_msg *msg;

	if (!pskb_may_pull(skb, skb->len))
		return 0;

1566
	msg = (struct nd_msg *)skb_transport_header(skb);
L
Linus Torvalds 已提交
1567

1568
	__skb_push(skb, skb->data - skb_transport_header(skb));
L
Linus Torvalds 已提交
1569

1570
	if (ipv6_hdr(skb)->hop_limit != 255) {
L
Linus Torvalds 已提交
1571 1572
		ND_PRINTK2(KERN_WARNING
			   "ICMPv6 NDISC: invalid hop-limit: %d\n",
1573
			   ipv6_hdr(skb)->hop_limit);
L
Linus Torvalds 已提交
1574 1575 1576 1577
		return 0;
	}

	if (msg->icmph.icmp6_code != 0) {
1578
		ND_PRINTK2(KERN_WARNING
L
Linus Torvalds 已提交
1579 1580 1581 1582 1583
			   "ICMPv6 NDISC: invalid ICMPv6 code: %d\n",
			   msg->icmph.icmp6_code);
		return 0;
	}

1584 1585
	memset(NEIGH_CB(skb), 0, sizeof(struct neighbour_cb));

L
Linus Torvalds 已提交
1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605
	switch (msg->icmph.icmp6_type) {
	case NDISC_NEIGHBOUR_SOLICITATION:
		ndisc_recv_ns(skb);
		break;

	case NDISC_NEIGHBOUR_ADVERTISEMENT:
		ndisc_recv_na(skb);
		break;

	case NDISC_ROUTER_SOLICITATION:
		ndisc_recv_rs(skb);
		break;

	case NDISC_ROUTER_ADVERTISEMENT:
		ndisc_router_discovery(skb);
		break;

	case NDISC_REDIRECT:
		ndisc_redirect_rcv(skb);
		break;
1606
	}
L
Linus Torvalds 已提交
1607 1608 1609 1610 1611 1612 1613

	return 0;
}

static int ndisc_netdev_event(struct notifier_block *this, unsigned long event, void *ptr)
{
	struct net_device *dev = ptr;
1614
	struct net *net = dev_net(dev);
L
Linus Torvalds 已提交
1615 1616 1617 1618

	switch (event) {
	case NETDEV_CHANGEADDR:
		neigh_changeaddr(&nd_tbl, dev);
1619
		fib6_run_gc(~0UL, net);
L
Linus Torvalds 已提交
1620 1621 1622
		break;
	case NETDEV_DOWN:
		neigh_ifdown(&nd_tbl, dev);
1623
		fib6_run_gc(~0UL, net);
L
Linus Torvalds 已提交
1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661
		break;
	default:
		break;
	}

	return NOTIFY_DONE;
}

static struct notifier_block ndisc_netdev_notifier = {
	.notifier_call = ndisc_netdev_event,
};

#ifdef CONFIG_SYSCTL
static void ndisc_warn_deprecated_sysctl(struct ctl_table *ctl,
					 const char *func, const char *dev_name)
{
	static char warncomm[TASK_COMM_LEN];
	static int warned;
	if (strcmp(warncomm, current->comm) && warned < 5) {
		strcpy(warncomm, current->comm);
		printk(KERN_WARNING
			"process `%s' is using deprecated sysctl (%s) "
			"net.ipv6.neigh.%s.%s; "
			"Use net.ipv6.neigh.%s.%s_ms "
			"instead.\n",
			warncomm, func,
			dev_name, ctl->procname,
			dev_name, ctl->procname);
		warned++;
	}
}

int ndisc_ifinfo_sysctl_change(struct ctl_table *ctl, int write, struct file * filp, void __user *buffer, size_t *lenp, loff_t *ppos)
{
	struct net_device *dev = ctl->extra1;
	struct inet6_dev *idev;
	int ret;

1662 1663
	if ((strcmp(ctl->procname, "retrans_time") == 0) ||
	    (strcmp(ctl->procname, "base_reachable_time") == 0))
L
Linus Torvalds 已提交
1664 1665
		ndisc_warn_deprecated_sysctl(ctl, "syscall", dev ? dev->name : "default");

1666
	if (strcmp(ctl->procname, "retrans_time") == 0)
L
Linus Torvalds 已提交
1667
		ret = proc_dointvec(ctl, write, filp, buffer, lenp, ppos);
1668 1669

	else if (strcmp(ctl->procname, "base_reachable_time") == 0)
L
Linus Torvalds 已提交
1670 1671
		ret = proc_dointvec_jiffies(ctl, write,
					    filp, buffer, lenp, ppos);
1672 1673

	else if ((strcmp(ctl->procname, "retrans_time_ms") == 0) ||
1674
		 (strcmp(ctl->procname, "base_reachable_time_ms") == 0))
L
Linus Torvalds 已提交
1675 1676
		ret = proc_dointvec_ms_jiffies(ctl, write,
					       filp, buffer, lenp, ppos);
1677
	else
L
Linus Torvalds 已提交
1678 1679 1680
		ret = -1;

	if (write && ret == 0 && dev && (idev = in6_dev_get(dev)) != NULL) {
1681
		if (ctl->data == &idev->nd_parms->base_reachable_time)
L
Linus Torvalds 已提交
1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692
			idev->nd_parms->reachable_time = neigh_rand_reach_time(idev->nd_parms->base_reachable_time);
		idev->tstamp = jiffies;
		inet6_ifinfo_notify(RTM_NEWLINK, idev);
		in6_dev_put(idev);
	}
	return ret;
}

static int ndisc_ifinfo_sysctl_strategy(ctl_table *ctl, int __user *name,
					int nlen, void __user *oldval,
					size_t __user *oldlenp,
1693
					void __user *newval, size_t newlen)
L
Linus Torvalds 已提交
1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705
{
	struct net_device *dev = ctl->extra1;
	struct inet6_dev *idev;
	int ret;

	if (ctl->ctl_name == NET_NEIGH_RETRANS_TIME ||
	    ctl->ctl_name == NET_NEIGH_REACHABLE_TIME)
		ndisc_warn_deprecated_sysctl(ctl, "procfs", dev ? dev->name : "default");

	switch (ctl->ctl_name) {
	case NET_NEIGH_REACHABLE_TIME:
		ret = sysctl_jiffies(ctl, name, nlen,
1706
				     oldval, oldlenp, newval, newlen);
L
Linus Torvalds 已提交
1707 1708 1709 1710
		break;
	case NET_NEIGH_RETRANS_TIME_MS:
	case NET_NEIGH_REACHABLE_TIME_MS:
		 ret = sysctl_ms_jiffies(ctl, name, nlen,
1711
					 oldval, oldlenp, newval, newlen);
L
Linus Torvalds 已提交
1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731
		 break;
	default:
		ret = 0;
	}

	if (newval && newlen && ret > 0 &&
	    dev && (idev = in6_dev_get(dev)) != NULL) {
		if (ctl->ctl_name == NET_NEIGH_REACHABLE_TIME ||
		    ctl->ctl_name == NET_NEIGH_REACHABLE_TIME_MS)
			idev->nd_parms->reachable_time = neigh_rand_reach_time(idev->nd_parms->base_reachable_time);
		idev->tstamp = jiffies;
		inet6_ifinfo_notify(RTM_NEWLINK, idev);
		in6_dev_put(idev);
	}

	return ret;
}

#endif

1732
static int ndisc_net_init(struct net *net)
L
Linus Torvalds 已提交
1733
{
1734
	struct socket *sock;
L
Linus Torvalds 已提交
1735 1736
	struct ipv6_pinfo *np;
	struct sock *sk;
1737
	int err;
L
Linus Torvalds 已提交
1738

1739
	err = sock_create_kern(PF_INET6, SOCK_RAW, IPPROTO_ICMPV6, &sock);
L
Linus Torvalds 已提交
1740 1741
	if (err < 0) {
		ND_PRINTK0(KERN_ERR
1742
			   "ICMPv6 NDISC: Failed to initialize the control socket (err %d).\n",
L
Linus Torvalds 已提交
1743 1744 1745 1746
			   err);
		return err;
	}

1747 1748 1749
	net->ipv6.ndisc_sk = sk = sock->sk;
	sk_change_net(sk, net);

L
Linus Torvalds 已提交
1750 1751 1752 1753 1754 1755 1756
	np = inet6_sk(sk);
	sk->sk_allocation = GFP_ATOMIC;
	np->hop_limit = 255;
	/* Do not loopback ndisc messages */
	np->mc_loop = 0;
	sk->sk_prot->unhash(sk);

1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776
	return 0;
}

static void ndisc_net_exit(struct net *net)
{
	sk_release_kernel(net->ipv6.ndisc_sk);
}

static struct pernet_operations ndisc_net_ops = {
	.init = ndisc_net_init,
	.exit = ndisc_net_exit,
};

int __init ndisc_init(void)
{
	int err;

	err = register_pernet_subsys(&ndisc_net_ops);
	if (err)
		return err;
1777 1778 1779
	/*
	 * Initialize the neighbour table
	 */
L
Linus Torvalds 已提交
1780 1781 1782
	neigh_table_init(&nd_tbl);

#ifdef CONFIG_SYSCTL
1783 1784 1785 1786 1787 1788
	err = neigh_sysctl_register(NULL, &nd_tbl.parms, NET_IPV6,
				    NET_IPV6_NEIGH, "ipv6",
				    &ndisc_ifinfo_sysctl_change,
				    &ndisc_ifinfo_sysctl_strategy);
	if (err)
		goto out_unregister_pernet;
L
Linus Torvalds 已提交
1789
#endif
1790 1791 1792 1793 1794
	err = register_netdevice_notifier(&ndisc_netdev_notifier);
	if (err)
		goto out_unregister_sysctl;
out:
	return err;
L
Linus Torvalds 已提交
1795

1796 1797 1798 1799 1800 1801 1802
out_unregister_sysctl:
#ifdef CONFIG_SYSCTL
	neigh_sysctl_unregister(&nd_tbl.parms);
out_unregister_pernet:
#endif
	unregister_pernet_subsys(&ndisc_net_ops);
	goto out;
L
Linus Torvalds 已提交
1803 1804 1805 1806
}

void ndisc_cleanup(void)
{
1807
	unregister_netdevice_notifier(&ndisc_netdev_notifier);
L
Linus Torvalds 已提交
1808 1809 1810 1811
#ifdef CONFIG_SYSCTL
	neigh_sysctl_unregister(&nd_tbl.parms);
#endif
	neigh_table_clear(&nd_tbl);
1812
	unregister_pernet_subsys(&ndisc_net_ops);
L
Linus Torvalds 已提交
1813
}