addrlabel.c 14 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
/*
 * IPv6 Address Label subsystem
 * for the IPv6 "Default" Source Address Selection
 *
 * Copyright (C)2007 USAGI/WIDE Project
 */
/*
 * Author:
 * 	YOSHIFUJI Hideaki @ USAGI/WIDE Project <yoshfuji@linux-ipv6.org>
 */

#include <linux/kernel.h>
#include <linux/list.h>
#include <linux/rcupdate.h>
#include <linux/in6.h>
16
#include <linux/slab.h>
17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
#include <net/addrconf.h>
#include <linux/if_addrlabel.h>
#include <linux/netlink.h>
#include <linux/rtnetlink.h>

#if 0
#define ADDRLABEL(x...) printk(x)
#else
#define ADDRLABEL(x...) do { ; } while(0)
#endif

/*
 * Policy Table
 */
struct ip6addrlbl_entry
{
33 34 35
#ifdef CONFIG_NET_NS
	struct net *lbl_net;
#endif
36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
	struct in6_addr prefix;
	int prefixlen;
	int ifindex;
	int addrtype;
	u32 label;
	struct hlist_node list;
	atomic_t refcnt;
	struct rcu_head rcu;
};

static struct ip6addrlbl_table
{
	struct hlist_head head;
	spinlock_t lock;
	u32 seq;
} ip6addrlbl_table;

53 54 55
static inline
struct net *ip6addrlbl_net(const struct ip6addrlbl_entry *lbl)
{
Eric Dumazet's avatar
Eric Dumazet committed
56
	return read_pnet(&lbl->lbl_net);
57 58
}

59
/*
60
 * Default policy table (RFC6724 + extensions)
61 62 63 64 65 66 67 68 69 70
 *
 * prefix		addr_type	label
 * -------------------------------------------------------------------------
 * ::1/128		LOOPBACK	0
 * ::/0			N/A		1
 * 2002::/16		N/A		2
 * ::/96		COMPATv4	3
 * ::ffff:0:0/96	V4MAPPED	4
 * fc00::/7		N/A		5		ULA (RFC 4193)
 * 2001::/32		N/A		6		Teredo (RFC 4380)
71
 * 2001:10::/28		N/A		7		ORCHID (RFC 4843)
72 73 74
 * fec0::/10		N/A		11		Site-local
 *							(deprecated by RFC3879)
 * 3ffe::/16		N/A		12		6bone
75 76
 *
 * Note: 0xffffffff is used if we do not have any policies.
77
 * Note: Labels for ULA and 6to4 are different from labels listed in RFC6724.
78 79 80 81
 */

#define IPV6_ADDR_LABEL_DEFAULT	0xffffffffUL

82
static const __net_initconst struct ip6addrlbl_init_table
83 84 85 86 87 88 89 90 91 92 93 94
{
	const struct in6_addr *prefix;
	int prefixlen;
	u32 label;
} ip6addrlbl_init_table[] = {
	{	/* ::/0 */
		.prefix = &in6addr_any,
		.label = 1,
	},{	/* fc00::/7 */
		.prefix = &(struct in6_addr){{{ 0xfc }}},
		.prefixlen = 7,
		.label = 5,
95 96 97 98
	},{	/* fec0::/10 */
		.prefix = &(struct in6_addr){{{ 0xfe, 0xc0 }}},
		.prefixlen = 10,
		.label = 11,
99 100 101 102
	},{	/* 2002::/16 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x02 }}},
		.prefixlen = 16,
		.label = 2,
103 104 105 106
	},{	/* 3ffe::/16 */
		.prefix = &(struct in6_addr){{{ 0x3f, 0xfe }}},
		.prefixlen = 16,
		.label = 12,
107 108 109 110
	},{	/* 2001::/32 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x01 }}},
		.prefixlen = 32,
		.label = 6,
111 112 113 114
	},{	/* 2001:10::/28 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x01, 0x00, 0x10 }}},
		.prefixlen = 28,
		.label = 7,
115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132
	},{	/* ::ffff:0:0 */
		.prefix = &(struct in6_addr){{{ [10] = 0xff, [11] = 0xff }}},
		.prefixlen = 96,
		.label = 4,
	},{	/* ::/96 */
		.prefix = &in6addr_any,
		.prefixlen = 96,
		.label = 3,
	},{	/* ::1/128 */
		.prefix = &in6addr_loopback,
		.prefixlen = 128,
		.label = 0,
	}
};

/* Object management */
static inline void ip6addrlbl_free(struct ip6addrlbl_entry *p)
{
133 134 135
#ifdef CONFIG_NET_NS
	release_net(p->lbl_net);
#endif
136 137 138
	kfree(p);
}

139 140 141 142 143
static void ip6addrlbl_free_rcu(struct rcu_head *h)
{
	ip6addrlbl_free(container_of(h, struct ip6addrlbl_entry, rcu));
}

144
static bool ip6addrlbl_hold(struct ip6addrlbl_entry *p)
145 146 147 148 149 150 151
{
	return atomic_inc_not_zero(&p->refcnt);
}

static inline void ip6addrlbl_put(struct ip6addrlbl_entry *p)
{
	if (atomic_dec_and_test(&p->refcnt))
152
		call_rcu(&p->rcu, ip6addrlbl_free_rcu);
153 154 155
}

/* Find label */
156 157 158 159
static bool __ip6addrlbl_match(struct net *net,
			       const struct ip6addrlbl_entry *p,
			       const struct in6_addr *addr,
			       int addrtype, int ifindex)
160
{
161
	if (!net_eq(ip6addrlbl_net(p), net))
162
		return false;
163
	if (p->ifindex && p->ifindex != ifindex)
164
		return false;
165
	if (p->addrtype && p->addrtype != addrtype)
166
		return false;
167
	if (!ipv6_prefix_equal(addr, &p->prefix, p->prefixlen))
168 169
		return false;
	return true;
170 171
}

172 173
static struct ip6addrlbl_entry *__ipv6_addr_label(struct net *net,
						  const struct in6_addr *addr,
174 175 176
						  int type, int ifindex)
{
	struct ip6addrlbl_entry *p;
177
	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
178
		if (__ip6addrlbl_match(net, p, addr, type, ifindex))
179 180 181 182 183
			return p;
	}
	return NULL;
}

184 185
u32 ipv6_addr_label(struct net *net,
		    const struct in6_addr *addr, int type, int ifindex)
186 187 188 189 190 191 192
{
	u32 label;
	struct ip6addrlbl_entry *p;

	type &= IPV6_ADDR_MAPPED | IPV6_ADDR_COMPATv4 | IPV6_ADDR_LOOPBACK;

	rcu_read_lock();
193
	p = __ipv6_addr_label(net, addr, type, ifindex);
194 195 196
	label = p ? p->label : IPV6_ADDR_LABEL_DEFAULT;
	rcu_read_unlock();

Harvey Harrison's avatar
Harvey Harrison committed
197
	ADDRLABEL(KERN_DEBUG "%s(addr=%pI6, type=%d, ifindex=%d) => %08x\n",
198
		  __func__, addr, type, ifindex, label);
199 200 201 202 203

	return label;
}

/* allocate one entry */
204 205
static struct ip6addrlbl_entry *ip6addrlbl_alloc(struct net *net,
						 const struct in6_addr *prefix,
206 207
						 int prefixlen, int ifindex,
						 u32 label)
208 209 210 211
{
	struct ip6addrlbl_entry *newp;
	int addrtype;

Harvey Harrison's avatar
Harvey Harrison committed
212
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u)\n",
213
		  __func__, prefix, prefixlen, ifindex, (unsigned int)label);
214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243

	addrtype = ipv6_addr_type(prefix) & (IPV6_ADDR_MAPPED | IPV6_ADDR_COMPATv4 | IPV6_ADDR_LOOPBACK);

	switch (addrtype) {
	case IPV6_ADDR_MAPPED:
		if (prefixlen > 96)
			return ERR_PTR(-EINVAL);
		if (prefixlen < 96)
			addrtype = 0;
		break;
	case IPV6_ADDR_COMPATv4:
		if (prefixlen != 96)
			addrtype = 0;
		break;
	case IPV6_ADDR_LOOPBACK:
		if (prefixlen != 128)
			addrtype = 0;
		break;
	}

	newp = kmalloc(sizeof(*newp), GFP_KERNEL);
	if (!newp)
		return ERR_PTR(-ENOMEM);

	ipv6_addr_prefix(&newp->prefix, prefix, prefixlen);
	newp->prefixlen = prefixlen;
	newp->ifindex = ifindex;
	newp->addrtype = addrtype;
	newp->label = label;
	INIT_HLIST_NODE(&newp->list);
244 245 246
#ifdef CONFIG_NET_NS
	newp->lbl_net = hold_net(net);
#endif
247 248 249 250 251
	atomic_set(&newp->refcnt, 1);
	return newp;
}

/* add a label */
252
static int __ip6addrlbl_add(struct ip6addrlbl_entry *newp, int replace)
253
{
254 255
	struct hlist_node *n;
	struct ip6addrlbl_entry *last = NULL, *p = NULL;
256 257
	int ret = 0;

258 259
	ADDRLABEL(KERN_DEBUG "%s(newp=%p, replace=%d)\n", __func__, newp,
		  replace);
260

261 262 263 264 265 266 267
	hlist_for_each_entry_safe(p, n,	&ip6addrlbl_table.head, list) {
		if (p->prefixlen == newp->prefixlen &&
		    net_eq(ip6addrlbl_net(p), ip6addrlbl_net(newp)) &&
		    p->ifindex == newp->ifindex &&
		    ipv6_addr_equal(&p->prefix, &newp->prefix)) {
			if (!replace) {
				ret = -EEXIST;
268 269
				goto out;
			}
270 271 272 273 274 275 276
			hlist_replace_rcu(&p->list, &newp->list);
			ip6addrlbl_put(p);
			goto out;
		} else if ((p->prefixlen == newp->prefixlen && !p->ifindex) ||
			   (p->prefixlen < newp->prefixlen)) {
			hlist_add_before_rcu(&newp->list, &p->list);
			goto out;
277
		}
278
		last = p;
279
	}
280 281 282 283
	if (last)
		hlist_add_after_rcu(&last->list, &newp->list);
	else
		hlist_add_head_rcu(&newp->list, &ip6addrlbl_table.head);
284 285 286 287 288 289 290
out:
	if (!ret)
		ip6addrlbl_table.seq++;
	return ret;
}

/* add a label */
291 292
static int ip6addrlbl_add(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
293
			  int ifindex, u32 label, int replace)
294 295 296 297
{
	struct ip6addrlbl_entry *newp;
	int ret = 0;

Harvey Harrison's avatar
Harvey Harrison committed
298
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u, replace=%d)\n",
299 300
		  __func__, prefix, prefixlen, ifindex, (unsigned int)label,
		  replace);
301

302
	newp = ip6addrlbl_alloc(net, prefix, prefixlen, ifindex, label);
303 304 305 306 307 308 309 310 311 312 313
	if (IS_ERR(newp))
		return PTR_ERR(newp);
	spin_lock(&ip6addrlbl_table.lock);
	ret = __ip6addrlbl_add(newp, replace);
	spin_unlock(&ip6addrlbl_table.lock);
	if (ret)
		ip6addrlbl_free(newp);
	return ret;
}

/* remove a label */
314 315
static int __ip6addrlbl_del(struct net *net,
			    const struct in6_addr *prefix, int prefixlen,
316
			    int ifindex)
317 318
{
	struct ip6addrlbl_entry *p = NULL;
319
	struct hlist_node *n;
320 321
	int ret = -ESRCH;

Harvey Harrison's avatar
Harvey Harrison committed
322
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
323
		  __func__, prefix, prefixlen, ifindex);
324

325
	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
326
		if (p->prefixlen == prefixlen &&
327
		    net_eq(ip6addrlbl_net(p), net) &&
328 329 330 331 332 333 334 335 336 337 338
		    p->ifindex == ifindex &&
		    ipv6_addr_equal(&p->prefix, prefix)) {
			hlist_del_rcu(&p->list);
			ip6addrlbl_put(p);
			ret = 0;
			break;
		}
	}
	return ret;
}

339 340
static int ip6addrlbl_del(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
341
			  int ifindex)
342 343 344 345
{
	struct in6_addr prefix_buf;
	int ret;

Harvey Harrison's avatar
Harvey Harrison committed
346
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
347
		  __func__, prefix, prefixlen, ifindex);
348 349 350

	ipv6_addr_prefix(&prefix_buf, prefix, prefixlen);
	spin_lock(&ip6addrlbl_table.lock);
351
	ret = __ip6addrlbl_del(net, &prefix_buf, prefixlen, ifindex);
352 353 354 355 356
	spin_unlock(&ip6addrlbl_table.lock);
	return ret;
}

/* add default label */
357
static int __net_init ip6addrlbl_net_init(struct net *net)
358 359 360 361
{
	int err = 0;
	int i;

362
	ADDRLABEL(KERN_DEBUG "%s\n", __func__);
363 364

	for (i = 0; i < ARRAY_SIZE(ip6addrlbl_init_table); i++) {
365 366
		int ret = ip6addrlbl_add(net,
					 ip6addrlbl_init_table[i].prefix,
367 368 369 370 371 372 373 374 375 376
					 ip6addrlbl_init_table[i].prefixlen,
					 0,
					 ip6addrlbl_init_table[i].label, 0);
		/* XXX: should we free all rules when we catch an error? */
		if (ret && (!err || err != -ENOMEM))
			err = ret;
	}
	return err;
}

377 378 379
static void __net_exit ip6addrlbl_net_exit(struct net *net)
{
	struct ip6addrlbl_entry *p = NULL;
380
	struct hlist_node *n;
381 382 383

	/* Remove all labels belonging to the exiting net */
	spin_lock(&ip6addrlbl_table.lock);
384
	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
385 386 387 388 389 390 391 392 393 394 395 396 397
		if (net_eq(ip6addrlbl_net(p), net)) {
			hlist_del_rcu(&p->list);
			ip6addrlbl_put(p);
		}
	}
	spin_unlock(&ip6addrlbl_table.lock);
}

static struct pernet_operations ipv6_addr_label_ops = {
	.init = ip6addrlbl_net_init,
	.exit = ip6addrlbl_net_exit,
};

398 399 400 401
int __init ipv6_addr_label_init(void)
{
	spin_lock_init(&ip6addrlbl_table.lock);

402
	return register_pernet_subsys(&ipv6_addr_label_ops);
403 404
}

405 406 407 408 409
void ipv6_addr_label_cleanup(void)
{
	unregister_pernet_subsys(&ipv6_addr_label_ops);
}

410 411 412 413 414
static const struct nla_policy ifal_policy[IFAL_MAX+1] = {
	[IFAL_ADDRESS]		= { .len = sizeof(struct in6_addr), },
	[IFAL_LABEL]		= { .len = sizeof(u32), },
};

415
static int ip6addrlbl_newdel(struct sk_buff *skb, struct nlmsghdr *nlh)
416
{
417
	struct net *net = sock_net(skb->sk);
418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445
	struct ifaddrlblmsg *ifal;
	struct nlattr *tb[IFAL_MAX+1];
	struct in6_addr *pfx;
	u32 label;
	int err = 0;

	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
	if (err < 0)
		return err;

	ifal = nlmsg_data(nlh);

	if (ifal->ifal_family != AF_INET6 ||
	    ifal->ifal_prefixlen > 128)
		return -EINVAL;

	if (!tb[IFAL_ADDRESS])
		return -EINVAL;
	pfx = nla_data(tb[IFAL_ADDRESS]);

	if (!tb[IFAL_LABEL])
		return -EINVAL;
	label = nla_get_u32(tb[IFAL_LABEL]);
	if (label == IPV6_ADDR_LABEL_DEFAULT)
		return -EINVAL;

	switch(nlh->nlmsg_type) {
	case RTM_NEWADDRLABEL:
446 447 448 449
		if (ifal->ifal_index &&
		    !__dev_get_by_index(net, ifal->ifal_index))
			return -EINVAL;

450
		err = ip6addrlbl_add(net, pfx, ifal->ifal_prefixlen,
451 452 453 454
				     ifal->ifal_index, label,
				     nlh->nlmsg_flags & NLM_F_REPLACE);
		break;
	case RTM_DELADDRLABEL:
455
		err = ip6addrlbl_del(net, pfx, ifal->ifal_prefixlen,
456 457 458 459 460 461 462 463
				     ifal->ifal_index);
		break;
	default:
		err = -EOPNOTSUPP;
	}
	return err;
}

464 465
static void ip6addrlbl_putmsg(struct nlmsghdr *nlh,
			      int prefixlen, int ifindex, u32 lseq)
466 467 468 469 470 471 472 473 474 475 476 477
{
	struct ifaddrlblmsg *ifal = nlmsg_data(nlh);
	ifal->ifal_family = AF_INET6;
	ifal->ifal_prefixlen = prefixlen;
	ifal->ifal_flags = 0;
	ifal->ifal_index = ifindex;
	ifal->ifal_seq = lseq;
};

static int ip6addrlbl_fill(struct sk_buff *skb,
			   struct ip6addrlbl_entry *p,
			   u32 lseq,
478
			   u32 portid, u32 seq, int event,
479 480
			   unsigned int flags)
{
481
	struct nlmsghdr *nlh = nlmsg_put(skb, portid, seq, event,
482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498
					 sizeof(struct ifaddrlblmsg), flags);
	if (!nlh)
		return -EMSGSIZE;

	ip6addrlbl_putmsg(nlh, p->prefixlen, p->ifindex, lseq);

	if (nla_put(skb, IFAL_ADDRESS, 16, &p->prefix) < 0 ||
	    nla_put_u32(skb, IFAL_LABEL, p->label) < 0) {
		nlmsg_cancel(skb, nlh);
		return -EMSGSIZE;
	}

	return nlmsg_end(skb, nlh);
}

static int ip6addrlbl_dump(struct sk_buff *skb, struct netlink_callback *cb)
{
499
	struct net *net = sock_net(skb->sk);
500 501 502 503 504
	struct ip6addrlbl_entry *p;
	int idx = 0, s_idx = cb->args[0];
	int err;

	rcu_read_lock();
505
	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
506 507
		if (idx >= s_idx &&
		    net_eq(ip6addrlbl_net(p), net)) {
508 509
			if ((err = ip6addrlbl_fill(skb, p,
						   ip6addrlbl_table.seq,
510
						   NETLINK_CB(cb->skb).portid,
511 512 513 514 515 516 517 518 519 520 521 522 523 524
						   cb->nlh->nlmsg_seq,
						   RTM_NEWADDRLABEL,
						   NLM_F_MULTI)) <= 0)
				break;
		}
		idx++;
	}
	rcu_read_unlock();
	cb->args[0] = idx;
	return skb->len;
}

static inline int ip6addrlbl_msgsize(void)
{
Eric Dumazet's avatar
Eric Dumazet committed
525
	return NLMSG_ALIGN(sizeof(struct ifaddrlblmsg))
526
		+ nla_total_size(16)	/* IFAL_ADDRESS */
Eric Dumazet's avatar
Eric Dumazet committed
527
		+ nla_total_size(4);	/* IFAL_LABEL */
528 529
}

530
static int ip6addrlbl_get(struct sk_buff *in_skb, struct nlmsghdr* nlh)
531
{
532
	struct net *net = sock_net(in_skb->sk);
533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551
	struct ifaddrlblmsg *ifal;
	struct nlattr *tb[IFAL_MAX+1];
	struct in6_addr *addr;
	u32 lseq;
	int err = 0;
	struct ip6addrlbl_entry *p;
	struct sk_buff *skb;

	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
	if (err < 0)
		return err;

	ifal = nlmsg_data(nlh);

	if (ifal->ifal_family != AF_INET6 ||
	    ifal->ifal_prefixlen != 128)
		return -EINVAL;

	if (ifal->ifal_index &&
552
	    !__dev_get_by_index(net, ifal->ifal_index))
553 554 555 556 557 558 559
		return -EINVAL;

	if (!tb[IFAL_ADDRESS])
		return -EINVAL;
	addr = nla_data(tb[IFAL_ADDRESS]);

	rcu_read_lock();
560
	p = __ipv6_addr_label(net, addr, ipv6_addr_type(addr), ifal->ifal_index);
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576
	if (p && ip6addrlbl_hold(p))
		p = NULL;
	lseq = ip6addrlbl_table.seq;
	rcu_read_unlock();

	if (!p) {
		err = -ESRCH;
		goto out;
	}

	if (!(skb = nlmsg_new(ip6addrlbl_msgsize(), GFP_KERNEL))) {
		ip6addrlbl_put(p);
		return -ENOBUFS;
	}

	err = ip6addrlbl_fill(skb, p, lseq,
577
			      NETLINK_CB(in_skb).portid, nlh->nlmsg_seq,
578 579 580 581 582 583 584 585 586 587
			      RTM_NEWADDRLABEL, 0);

	ip6addrlbl_put(p);

	if (err < 0) {
		WARN_ON(err == -EMSGSIZE);
		kfree_skb(skb);
		goto out;
	}

588
	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
589 590 591 592 593 594
out:
	return err;
}

void __init ipv6_addr_label_rtnl_register(void)
{
595 596 597 598 599 600
	__rtnl_register(PF_INET6, RTM_NEWADDRLABEL, ip6addrlbl_newdel,
			NULL, NULL);
	__rtnl_register(PF_INET6, RTM_DELADDRLABEL, ip6addrlbl_newdel,
			NULL, NULL);
	__rtnl_register(PF_INET6, RTM_GETADDRLABEL, ip6addrlbl_get,
			ip6addrlbl_dump, NULL);
601 602
}