addrlabel.c 13.8 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31
/*
 * 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>
#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
{
32 33 34
#ifdef CONFIG_NET_NS
	struct net *lbl_net;
#endif
35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
	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;

52 53 54 55 56 57 58 59 60 61
static inline
struct net *ip6addrlbl_net(const struct ip6addrlbl_entry *lbl)
{
#ifdef CONFIG_NET_NS
	return lbl->lbl_net;
#else
	return &init_net;
#endif
}

62 63 64 65 66 67 68 69 70 71 72 73
/*
 * Default policy table (RFC3484 + extensions)
 *
 * 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)
74
 * 2001:10::/28		N/A		7		ORCHID (RFC 4843)
75 76 77 78 79 80
 *
 * Note: 0xffffffff is used if we do not have any policies.
 */

#define IPV6_ADDR_LABEL_DEFAULT	0xffffffffUL

81
static const __net_initdata struct ip6addrlbl_init_table
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
{
	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,
	},{	/* 2002::/16 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x02 }}},
		.prefixlen = 16,
		.label = 2,
	},{	/* 2001::/32 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x01 }}},
		.prefixlen = 32,
		.label = 6,
102 103 104 105
	},{	/* 2001:10::/28 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x01, 0x00, 0x10 }}},
		.prefixlen = 28,
		.label = 7,
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123
	},{	/* ::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)
{
124 125 126
#ifdef CONFIG_NET_NS
	release_net(p->lbl_net);
#endif
127 128 129
	kfree(p);
}

130 131 132 133 134
static void ip6addrlbl_free_rcu(struct rcu_head *h)
{
	ip6addrlbl_free(container_of(h, struct ip6addrlbl_entry, rcu));
}

135 136 137 138 139 140 141 142
static inline int ip6addrlbl_hold(struct ip6addrlbl_entry *p)
{
	return atomic_inc_not_zero(&p->refcnt);
}

static inline void ip6addrlbl_put(struct ip6addrlbl_entry *p)
{
	if (atomic_dec_and_test(&p->refcnt))
143
		call_rcu(&p->rcu, ip6addrlbl_free_rcu);
144 145 146
}

/* Find label */
147 148
static int __ip6addrlbl_match(struct net *net,
			      struct ip6addrlbl_entry *p,
149 150 151
			      const struct in6_addr *addr,
			      int addrtype, int ifindex)
{
152 153
	if (!net_eq(ip6addrlbl_net(p), net))
		return 0;
154 155 156 157 158 159 160 161 162
	if (p->ifindex && p->ifindex != ifindex)
		return 0;
	if (p->addrtype && p->addrtype != addrtype)
		return 0;
	if (!ipv6_prefix_equal(addr, &p->prefix, p->prefixlen))
		return 0;
	return 1;
}

163 164
static struct ip6addrlbl_entry *__ipv6_addr_label(struct net *net,
						  const struct in6_addr *addr,
165 166 167 168 169
						  int type, int ifindex)
{
	struct hlist_node *pos;
	struct ip6addrlbl_entry *p;
	hlist_for_each_entry_rcu(p, pos, &ip6addrlbl_table.head, list) {
170
		if (__ip6addrlbl_match(net, p, addr, type, ifindex))
171 172 173 174 175
			return p;
	}
	return NULL;
}

176 177
u32 ipv6_addr_label(struct net *net,
		    const struct in6_addr *addr, int type, int ifindex)
178 179 180 181 182 183 184
{
	u32 label;
	struct ip6addrlbl_entry *p;

	type &= IPV6_ADDR_MAPPED | IPV6_ADDR_COMPATv4 | IPV6_ADDR_LOOPBACK;

	rcu_read_lock();
185
	p = __ipv6_addr_label(net, addr, type, ifindex);
186 187 188 189
	label = p ? p->label : IPV6_ADDR_LABEL_DEFAULT;
	rcu_read_unlock();

	ADDRLABEL(KERN_DEBUG "%s(addr=" NIP6_FMT ", type=%d, ifindex=%d) => %08x\n",
190
			__func__,
191 192 193 194 195 196 197
			NIP6(*addr), type, ifindex,
			label);

	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)
202 203 204 205 206
{
	struct ip6addrlbl_entry *newp;
	int addrtype;

	ADDRLABEL(KERN_DEBUG "%s(prefix=" NIP6_FMT ", prefixlen=%d, ifindex=%d, label=%u)\n",
207
			__func__,
208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
			NIP6(*prefix), prefixlen,
			ifindex,
			(unsigned int)label);

	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);
241 242 243
#ifdef CONFIG_NET_NS
	newp->lbl_net = hold_net(net);
#endif
244 245 246 247 248
	atomic_set(&newp->refcnt, 1);
	return newp;
}

/* add a label */
249
static int __ip6addrlbl_add(struct ip6addrlbl_entry *newp, int replace)
250 251 252 253
{
	int ret = 0;

	ADDRLABEL(KERN_DEBUG "%s(newp=%p, replace=%d)\n",
254
			__func__,
255 256 257 258 259 260 261 262 263 264
			newp, replace);

	if (hlist_empty(&ip6addrlbl_table.head)) {
		hlist_add_head_rcu(&newp->list, &ip6addrlbl_table.head);
	} else {
		struct hlist_node *pos, *n;
		struct ip6addrlbl_entry *p = NULL;
		hlist_for_each_entry_safe(p, pos, n,
					  &ip6addrlbl_table.head, list) {
			if (p->prefixlen == newp->prefixlen &&
265
			    net_eq(ip6addrlbl_net(p), ip6addrlbl_net(newp)) &&
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289
			    p->ifindex == newp->ifindex &&
			    ipv6_addr_equal(&p->prefix, &newp->prefix)) {
				if (!replace) {
					ret = -EEXIST;
					goto out;
				}
				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;
			}
		}
		hlist_add_after_rcu(&p->list, &newp->list);
	}
out:
	if (!ret)
		ip6addrlbl_table.seq++;
	return ret;
}

/* add a label */
290 291
static int ip6addrlbl_add(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
292
			  int ifindex, u32 label, int replace)
293 294 295 296 297
{
	struct ip6addrlbl_entry *newp;
	int ret = 0;

	ADDRLABEL(KERN_DEBUG "%s(prefix=" NIP6_FMT ", prefixlen=%d, ifindex=%d, label=%u, replace=%d)\n",
298
			__func__,
299 300 301 302 303
			NIP6(*prefix), prefixlen,
			ifindex,
			(unsigned int)label,
			replace);

304
	newp = ip6addrlbl_alloc(net, prefix, prefixlen, ifindex, label);
305 306 307 308 309 310 311 312 313 314 315
	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 */
316 317
static int __ip6addrlbl_del(struct net *net,
			    const struct in6_addr *prefix, int prefixlen,
318
			    int ifindex)
319 320 321 322 323 324
{
	struct ip6addrlbl_entry *p = NULL;
	struct hlist_node *pos, *n;
	int ret = -ESRCH;

	ADDRLABEL(KERN_DEBUG "%s(prefix=" NIP6_FMT ", prefixlen=%d, ifindex=%d)\n",
325
			__func__,
326 327 328 329 330
			NIP6(*prefix), prefixlen,
			ifindex);

	hlist_for_each_entry_safe(p, pos, n, &ip6addrlbl_table.head, list) {
		if (p->prefixlen == prefixlen &&
331
		    net_eq(ip6addrlbl_net(p), net) &&
332 333 334 335 336 337 338 339 340 341 342
		    p->ifindex == ifindex &&
		    ipv6_addr_equal(&p->prefix, prefix)) {
			hlist_del_rcu(&p->list);
			ip6addrlbl_put(p);
			ret = 0;
			break;
		}
	}
	return ret;
}

343 344
static int ip6addrlbl_del(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
345
			  int ifindex)
346 347 348 349 350
{
	struct in6_addr prefix_buf;
	int ret;

	ADDRLABEL(KERN_DEBUG "%s(prefix=" NIP6_FMT ", prefixlen=%d, ifindex=%d)\n",
351
			__func__,
352 353 354 355 356
			NIP6(*prefix), prefixlen,
			ifindex);

	ipv6_addr_prefix(&prefix_buf, prefix, prefixlen);
	spin_lock(&ip6addrlbl_table.lock);
357
	ret = __ip6addrlbl_del(net, &prefix_buf, prefixlen, ifindex);
358 359 360 361 362
	spin_unlock(&ip6addrlbl_table.lock);
	return ret;
}

/* add default label */
363
static int __net_init ip6addrlbl_net_init(struct net *net)
364 365 366 367
{
	int err = 0;
	int i;

368
	ADDRLABEL(KERN_DEBUG "%s()\n", __func__);
369 370

	for (i = 0; i < ARRAY_SIZE(ip6addrlbl_init_table); i++) {
371 372
		int ret = ip6addrlbl_add(net,
					 ip6addrlbl_init_table[i].prefix,
373 374 375 376 377 378 379 380 381 382
					 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;
}

383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403
static void __net_exit ip6addrlbl_net_exit(struct net *net)
{
	struct ip6addrlbl_entry *p = NULL;
	struct hlist_node *pos, *n;

	/* Remove all labels belonging to the exiting net */
	spin_lock(&ip6addrlbl_table.lock);
	hlist_for_each_entry_safe(p, pos, n, &ip6addrlbl_table.head, list) {
		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,
};

404 405 406 407
int __init ipv6_addr_label_init(void)
{
	spin_lock_init(&ip6addrlbl_table.lock);

408
	return register_pernet_subsys(&ipv6_addr_label_ops);
409 410 411 412 413 414 415 416 417 418
}

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

static int ip6addrlbl_newdel(struct sk_buff *skb, struct nlmsghdr *nlh,
			     void *arg)
{
419
	struct net *net = sock_net(skb->sk);
420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436
	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 (ifal->ifal_index &&
437
	    !__dev_get_by_index(net, ifal->ifal_index))
438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454
		return -EINVAL;

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

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

	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:
455
		err = ip6addrlbl_add(net, pfx, ifal->ifal_prefixlen,
456 457 458 459
				     ifal->ifal_index, label,
				     nlh->nlmsg_flags & NLM_F_REPLACE);
		break;
	case RTM_DELADDRLABEL:
460
		err = ip6addrlbl_del(net, pfx, ifal->ifal_prefixlen,
461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503
				     ifal->ifal_index);
		break;
	default:
		err = -EOPNOTSUPP;
	}
	return err;
}

static inline void ip6addrlbl_putmsg(struct nlmsghdr *nlh,
				     int prefixlen, int ifindex, u32 lseq)
{
	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,
			   u32 pid, u32 seq, int event,
			   unsigned int flags)
{
	struct nlmsghdr *nlh = nlmsg_put(skb, pid, seq, event,
					 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)
{
504
	struct net *net = sock_net(skb->sk);
505 506 507 508 509 510 511
	struct ip6addrlbl_entry *p;
	struct hlist_node *pos;
	int idx = 0, s_idx = cb->args[0];
	int err;

	rcu_read_lock();
	hlist_for_each_entry_rcu(p, pos, &ip6addrlbl_table.head, list) {
512 513
		if (idx >= s_idx &&
		    net_eq(ip6addrlbl_net(p), net)) {
514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
			if ((err = ip6addrlbl_fill(skb, p,
						   ip6addrlbl_table.seq,
						   NETLINK_CB(cb->skb).pid,
						   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)
{
	return (NLMSG_ALIGN(sizeof(struct ifaddrlblmsg))
		+ nla_total_size(16)	/* IFAL_ADDRESS */
		+ nla_total_size(4)	/* IFAL_LABEL */
	);
}

static int ip6addrlbl_get(struct sk_buff *in_skb, struct nlmsghdr* nlh,
			  void *arg)
{
540
	struct net *net = sock_net(in_skb->sk);
541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559
	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 &&
560
	    !__dev_get_by_index(net, ifal->ifal_index))
561 562 563 564 565 566 567 568 569 570
		return -EINVAL;

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

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

	rcu_read_lock();
571
	p = __ipv6_addr_label(net, addr, ipv6_addr_type(addr), ifal->ifal_index);
572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598
	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,
			      NETLINK_CB(in_skb).pid, nlh->nlmsg_seq,
			      RTM_NEWADDRLABEL, 0);

	ip6addrlbl_put(p);

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

599
	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid);
600 601 602 603 604 605 606 607 608 609 610
out:
	return err;
}

void __init ipv6_addr_label_rtnl_register(void)
{
	__rtnl_register(PF_INET6, RTM_NEWADDRLABEL, ip6addrlbl_newdel, NULL);
	__rtnl_register(PF_INET6, RTM_DELADDRLABEL, ip6addrlbl_newdel, NULL);
	__rtnl_register(PF_INET6, RTM_GETADDRLABEL, ip6addrlbl_get, ip6addrlbl_dump);
}