addrlabel.c 13.9 KB
Newer Older
1 2 3 4 5 6 7 8
/*
 * IPv6 Address Label subsystem
 * for the IPv6 "Default" Source Address Selection
 *
 * Copyright (C)2007 USAGI/WIDE Project
 */
/*
 * Author:
9
 *	YOSHIFUJI Hideaki @ USAGI/WIDE Project <yoshfuji@linux-ipv6.org>
10 11 12 13 14 15
 */

#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
#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
25
#define ADDRLABEL(x...) do { ; } while (0)
26 27 28 29 30
#endif

/*
 * Policy Table
 */
31
struct ip6addrlbl_entry {
32
	possible_net_t lbl_net;
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49
	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;

50 51 52
static inline
struct net *ip6addrlbl_net(const struct ip6addrlbl_entry *lbl)
{
Eric Dumazet's avatar
Eric Dumazet committed
53
	return read_pnet(&lbl->lbl_net);
54 55
}

56
/*
57
 * Default policy table (RFC6724 + extensions)
58 59 60 61 62 63 64 65 66 67
 *
 * 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)
68
 * 2001:10::/28		N/A		7		ORCHID (RFC 4843)
69 70 71
 * fec0::/10		N/A		11		Site-local
 *							(deprecated by RFC3879)
 * 3ffe::/16		N/A		12		6bone
72 73
 *
 * Note: 0xffffffff is used if we do not have any policies.
74
 * Note: Labels for ULA and 6to4 are different from labels listed in RFC6724.
75 76 77 78
 */

#define IPV6_ADDR_LABEL_DEFAULT	0xffffffffUL

79
static const __net_initconst struct ip6addrlbl_init_table
80 81 82 83 84 85 86 87
{
	const struct in6_addr *prefix;
	int prefixlen;
	u32 label;
} ip6addrlbl_init_table[] = {
	{	/* ::/0 */
		.prefix = &in6addr_any,
		.label = 1,
88 89
	}, {	/* fc00::/7 */
		.prefix = &(struct in6_addr){ { { 0xfc } } } ,
90 91
		.prefixlen = 7,
		.label = 5,
92 93
	}, {	/* fec0::/10 */
		.prefix = &(struct in6_addr){ { { 0xfe, 0xc0 } } },
94 95
		.prefixlen = 10,
		.label = 11,
96 97
	}, {	/* 2002::/16 */
		.prefix = &(struct in6_addr){ { { 0x20, 0x02 } } },
98 99
		.prefixlen = 16,
		.label = 2,
100 101
	}, {	/* 3ffe::/16 */
		.prefix = &(struct in6_addr){ { { 0x3f, 0xfe } } },
102 103
		.prefixlen = 16,
		.label = 12,
104 105
	}, {	/* 2001::/32 */
		.prefix = &(struct in6_addr){ { { 0x20, 0x01 } } },
106 107
		.prefixlen = 32,
		.label = 6,
108 109
	}, {	/* 2001:10::/28 */
		.prefix = &(struct in6_addr){ { { 0x20, 0x01, 0x00, 0x10 } } },
110 111
		.prefixlen = 28,
		.label = 7,
112 113
	}, {	/* ::ffff:0:0 */
		.prefix = &(struct in6_addr){ { { [10] = 0xff, [11] = 0xff } } },
114 115
		.prefixlen = 96,
		.label = 4,
116
	}, {	/* ::/96 */
117 118 119
		.prefix = &in6addr_any,
		.prefixlen = 96,
		.label = 3,
120
	}, {	/* ::1/128 */
121 122 123 124 125 126 127 128 129 130 131 132
		.prefix = &in6addr_loopback,
		.prefixlen = 128,
		.label = 0,
	}
};

/* Object management */
static inline void ip6addrlbl_free(struct ip6addrlbl_entry *p)
{
	kfree(p);
}

133 134 135 136 137
static void ip6addrlbl_free_rcu(struct rcu_head *h)
{
	ip6addrlbl_free(container_of(h, struct ip6addrlbl_entry, rcu));
}

138
static bool ip6addrlbl_hold(struct ip6addrlbl_entry *p)
139 140 141 142 143 144 145
{
	return atomic_inc_not_zero(&p->refcnt);
}

static inline void ip6addrlbl_put(struct ip6addrlbl_entry *p)
{
	if (atomic_dec_and_test(&p->refcnt))
146
		call_rcu(&p->rcu, ip6addrlbl_free_rcu);
147 148 149
}

/* Find label */
150 151 152 153
static bool __ip6addrlbl_match(struct net *net,
			       const struct ip6addrlbl_entry *p,
			       const struct in6_addr *addr,
			       int addrtype, int ifindex)
154
{
155
	if (!net_eq(ip6addrlbl_net(p), net))
156
		return false;
157
	if (p->ifindex && p->ifindex != ifindex)
158
		return false;
159
	if (p->addrtype && p->addrtype != addrtype)
160
		return false;
161
	if (!ipv6_prefix_equal(addr, &p->prefix, p->prefixlen))
162 163
		return false;
	return true;
164 165
}

166 167
static struct ip6addrlbl_entry *__ipv6_addr_label(struct net *net,
						  const struct in6_addr *addr,
168 169 170
						  int type, int ifindex)
{
	struct ip6addrlbl_entry *p;
171
	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
172
		if (__ip6addrlbl_match(net, p, addr, type, ifindex))
173 174 175 176 177
			return p;
	}
	return NULL;
}

178 179
u32 ipv6_addr_label(struct net *net,
		    const struct in6_addr *addr, int type, int ifindex)
180 181 182 183 184 185 186
{
	u32 label;
	struct ip6addrlbl_entry *p;

	type &= IPV6_ADDR_MAPPED | IPV6_ADDR_COMPATv4 | IPV6_ADDR_LOOPBACK;

	rcu_read_lock();
187
	p = __ipv6_addr_label(net, addr, type, ifindex);
188 189 190
	label = p ? p->label : IPV6_ADDR_LABEL_DEFAULT;
	rcu_read_unlock();

Harvey Harrison's avatar
Harvey Harrison committed
191
	ADDRLABEL(KERN_DEBUG "%s(addr=%pI6, type=%d, ifindex=%d) => %08x\n",
192
		  __func__, addr, type, ifindex, label);
193 194 195 196 197

	return label;
}

/* allocate one entry */
198 199
static struct ip6addrlbl_entry *ip6addrlbl_alloc(struct net *net,
						 const struct in6_addr *prefix,
200 201
						 int prefixlen, int ifindex,
						 u32 label)
202 203 204 205
{
	struct ip6addrlbl_entry *newp;
	int addrtype;

Harvey Harrison's avatar
Harvey Harrison committed
206
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u)\n",
207
		  __func__, prefix, prefixlen, ifindex, (unsigned int)label);
208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237

	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);
238
	write_pnet(&newp->lbl_net, net);
239 240 241 242 243
	atomic_set(&newp->refcnt, 1);
	return newp;
}

/* add a label */
244
static int __ip6addrlbl_add(struct ip6addrlbl_entry *newp, int replace)
245
{
246 247
	struct hlist_node *n;
	struct ip6addrlbl_entry *last = NULL, *p = NULL;
248 249
	int ret = 0;

250 251
	ADDRLABEL(KERN_DEBUG "%s(newp=%p, replace=%d)\n", __func__, newp,
		  replace);
252

253 254 255 256 257 258 259
	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;
260 261
				goto out;
			}
262 263 264 265 266 267 268
			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;
269
		}
270
		last = p;
271
	}
272
	if (last)
273
		hlist_add_behind_rcu(&newp->list, &last->list);
274 275
	else
		hlist_add_head_rcu(&newp->list, &ip6addrlbl_table.head);
276 277 278 279 280 281 282
out:
	if (!ret)
		ip6addrlbl_table.seq++;
	return ret;
}

/* add a label */
283 284
static int ip6addrlbl_add(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
285
			  int ifindex, u32 label, int replace)
286 287 288 289
{
	struct ip6addrlbl_entry *newp;
	int ret = 0;

Harvey Harrison's avatar
Harvey Harrison committed
290
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u, replace=%d)\n",
291 292
		  __func__, prefix, prefixlen, ifindex, (unsigned int)label,
		  replace);
293

294
	newp = ip6addrlbl_alloc(net, prefix, prefixlen, ifindex, label);
295 296 297 298 299 300 301 302 303 304 305
	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 */
306 307
static int __ip6addrlbl_del(struct net *net,
			    const struct in6_addr *prefix, int prefixlen,
308
			    int ifindex)
309 310
{
	struct ip6addrlbl_entry *p = NULL;
311
	struct hlist_node *n;
312 313
	int ret = -ESRCH;

Harvey Harrison's avatar
Harvey Harrison committed
314
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
315
		  __func__, prefix, prefixlen, ifindex);
316

317
	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
318
		if (p->prefixlen == prefixlen &&
319
		    net_eq(ip6addrlbl_net(p), net) &&
320 321 322 323 324 325 326 327 328 329 330
		    p->ifindex == ifindex &&
		    ipv6_addr_equal(&p->prefix, prefix)) {
			hlist_del_rcu(&p->list);
			ip6addrlbl_put(p);
			ret = 0;
			break;
		}
	}
	return ret;
}

331 332
static int ip6addrlbl_del(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
333
			  int ifindex)
334 335 336 337
{
	struct in6_addr prefix_buf;
	int ret;

Harvey Harrison's avatar
Harvey Harrison committed
338
	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
339
		  __func__, prefix, prefixlen, ifindex);
340 341 342

	ipv6_addr_prefix(&prefix_buf, prefix, prefixlen);
	spin_lock(&ip6addrlbl_table.lock);
343
	ret = __ip6addrlbl_del(net, &prefix_buf, prefixlen, ifindex);
344 345 346 347 348
	spin_unlock(&ip6addrlbl_table.lock);
	return ret;
}

/* add default label */
349
static int __net_init ip6addrlbl_net_init(struct net *net)
350 351 352 353
{
	int err = 0;
	int i;

354
	ADDRLABEL(KERN_DEBUG "%s\n", __func__);
355 356

	for (i = 0; i < ARRAY_SIZE(ip6addrlbl_init_table); i++) {
357 358
		int ret = ip6addrlbl_add(net,
					 ip6addrlbl_init_table[i].prefix,
359 360 361 362 363 364 365 366 367 368
					 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;
}

369 370 371
static void __net_exit ip6addrlbl_net_exit(struct net *net)
{
	struct ip6addrlbl_entry *p = NULL;
372
	struct hlist_node *n;
373 374 375

	/* Remove all labels belonging to the exiting net */
	spin_lock(&ip6addrlbl_table.lock);
376
	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
377 378 379 380 381 382 383 384 385 386 387 388 389
		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,
};

390 391 392 393
int __init ipv6_addr_label_init(void)
{
	spin_lock_init(&ip6addrlbl_table.lock);

394
	return register_pernet_subsys(&ipv6_addr_label_ops);
395 396
}

397 398 399 400 401
void ipv6_addr_label_cleanup(void)
{
	unregister_pernet_subsys(&ipv6_addr_label_ops);
}

402 403 404 405 406
static const struct nla_policy ifal_policy[IFAL_MAX+1] = {
	[IFAL_ADDRESS]		= { .len = sizeof(struct in6_addr), },
	[IFAL_LABEL]		= { .len = sizeof(u32), },
};

407
static int ip6addrlbl_newdel(struct sk_buff *skb, struct nlmsghdr *nlh)
408
{
409
	struct net *net = sock_net(skb->sk);
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
	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;

436
	switch (nlh->nlmsg_type) {
437
	case RTM_NEWADDRLABEL:
438 439 440 441
		if (ifal->ifal_index &&
		    !__dev_get_by_index(net, ifal->ifal_index))
			return -EINVAL;

442
		err = ip6addrlbl_add(net, pfx, ifal->ifal_prefixlen,
443 444 445 446
				     ifal->ifal_index, label,
				     nlh->nlmsg_flags & NLM_F_REPLACE);
		break;
	case RTM_DELADDRLABEL:
447
		err = ip6addrlbl_del(net, pfx, ifal->ifal_prefixlen,
448 449 450 451 452 453 454 455
				     ifal->ifal_index);
		break;
	default:
		err = -EOPNOTSUPP;
	}
	return err;
}

456 457
static void ip6addrlbl_putmsg(struct nlmsghdr *nlh,
			      int prefixlen, int ifindex, u32 lseq)
458 459 460 461 462 463 464 465 466 467 468 469
{
	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,
470
			   u32 portid, u32 seq, int event,
471 472
			   unsigned int flags)
{
473
	struct nlmsghdr *nlh = nlmsg_put(skb, portid, seq, event,
474 475 476 477 478 479 480 481 482 483 484 485
					 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;
	}

486 487
	nlmsg_end(skb, nlh);
	return 0;
488 489 490 491
}

static int ip6addrlbl_dump(struct sk_buff *skb, struct netlink_callback *cb)
{
492
	struct net *net = sock_net(skb->sk);
493 494 495 496 497
	struct ip6addrlbl_entry *p;
	int idx = 0, s_idx = cb->args[0];
	int err;

	rcu_read_lock();
498
	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
499 500
		if (idx >= s_idx &&
		    net_eq(ip6addrlbl_net(p), net)) {
501 502 503 504 505 506
			err = ip6addrlbl_fill(skb, p,
					      ip6addrlbl_table.seq,
					      NETLINK_CB(cb->skb).portid,
					      cb->nlh->nlmsg_seq,
					      RTM_NEWADDRLABEL,
					      NLM_F_MULTI);
507
			if (err < 0)
508 509 510 511 512 513 514 515 516 517 518
				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
519
	return NLMSG_ALIGN(sizeof(struct ifaddrlblmsg))
520
		+ nla_total_size(16)	/* IFAL_ADDRESS */
Eric Dumazet's avatar
Eric Dumazet committed
521
		+ nla_total_size(4);	/* IFAL_LABEL */
522 523
}

524
static int ip6addrlbl_get(struct sk_buff *in_skb, struct nlmsghdr *nlh)
525
{
526
	struct net *net = sock_net(in_skb->sk);
527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545
	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 &&
546
	    !__dev_get_by_index(net, ifal->ifal_index))
547 548 549 550 551 552 553
		return -EINVAL;

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

	rcu_read_lock();
554
	p = __ipv6_addr_label(net, addr, ipv6_addr_type(addr), ifal->ifal_index);
555 556 557 558 559 560 561 562 563 564
	if (p && ip6addrlbl_hold(p))
		p = NULL;
	lseq = ip6addrlbl_table.seq;
	rcu_read_unlock();

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

565 566
	skb = nlmsg_new(ip6addrlbl_msgsize(), GFP_KERNEL);
	if (!skb) {
567 568 569 570 571
		ip6addrlbl_put(p);
		return -ENOBUFS;
	}

	err = ip6addrlbl_fill(skb, p, lseq,
572
			      NETLINK_CB(in_skb).portid, nlh->nlmsg_seq,
573 574 575 576 577 578 579 580 581 582
			      RTM_NEWADDRLABEL, 0);

	ip6addrlbl_put(p);

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

583
	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
584 585 586 587 588 589
out:
	return err;
}

void __init ipv6_addr_label_rtnl_register(void)
{
590 591 592 593 594 595
	__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);
596 597
}