[linux-2.6-block.git] / net / ipv6 / addrlabel.c

/*
 * 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 <linux/slab.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 {
#ifdef CONFIG_NET_NS
	struct net *lbl_net;
#endif
	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;

static inline
struct net *ip6addrlbl_net(const struct ip6addrlbl_entry *lbl)
{
	return read_pnet(&lbl->lbl_net);
}

/*
 * Default policy table (RFC6724 + 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)
 * 2001:10::/28		N/A		7		ORCHID (RFC 4843)
 * fec0::/10		N/A		11		Site-local
 *							(deprecated by RFC3879)
 * 3ffe::/16		N/A		12		6bone
 *
 * Note: 0xffffffff is used if we do not have any policies.
 * Note: Labels for ULA and 6to4 are different from labels listed in RFC6724.
 */

#define IPV6_ADDR_LABEL_DEFAULT	0xffffffffUL

static const __net_initconst struct ip6addrlbl_init_table
{
	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,
	},{	/* fec0::/10 */
		.prefix = &(struct in6_addr){{{ 0xfe, 0xc0 }}},
		.prefixlen = 10,
		.label = 11,
	},{	/* 2002::/16 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x02 }}},
		.prefixlen = 16,
		.label = 2,
	},{	/* 3ffe::/16 */
		.prefix = &(struct in6_addr){{{ 0x3f, 0xfe }}},
		.prefixlen = 16,
		.label = 12,
	},{	/* 2001::/32 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x01 }}},
		.prefixlen = 32,
		.label = 6,
	},{	/* 2001:10::/28 */
		.prefix = &(struct in6_addr){{{ 0x20, 0x01, 0x00, 0x10 }}},
		.prefixlen = 28,
		.label = 7,
	},{	/* ::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)
{
#ifdef CONFIG_NET_NS
	release_net(p->lbl_net);
#endif
	kfree(p);
}

static void ip6addrlbl_free_rcu(struct rcu_head *h)
{
	ip6addrlbl_free(container_of(h, struct ip6addrlbl_entry, rcu));
}

static bool 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))
		call_rcu(&p->rcu, ip6addrlbl_free_rcu);
}

/* Find label */
static bool __ip6addrlbl_match(struct net *net,
			       const struct ip6addrlbl_entry *p,
			       const struct in6_addr *addr,
			       int addrtype, int ifindex)
{
	if (!net_eq(ip6addrlbl_net(p), net))
		return false;
	if (p->ifindex && p->ifindex != ifindex)
		return false;
	if (p->addrtype && p->addrtype != addrtype)
		return false;
	if (!ipv6_prefix_equal(addr, &p->prefix, p->prefixlen))
		return false;
	return true;
}

static struct ip6addrlbl_entry *__ipv6_addr_label(struct net *net,
						  const struct in6_addr *addr,
						  int type, int ifindex)
{
	struct ip6addrlbl_entry *p;
	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
		if (__ip6addrlbl_match(net, p, addr, type, ifindex))
			return p;
	}
	return NULL;
}

u32 ipv6_addr_label(struct net *net,
		    const struct in6_addr *addr, int type, int ifindex)
{
	u32 label;
	struct ip6addrlbl_entry *p;

	type &= IPV6_ADDR_MAPPED | IPV6_ADDR_COMPATv4 | IPV6_ADDR_LOOPBACK;

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

	ADDRLABEL(KERN_DEBUG "%s(addr=%pI6, type=%d, ifindex=%d) => %08x\n",
		  __func__, addr, type, ifindex, label);

	return label;
}

/* allocate one entry */
static struct ip6addrlbl_entry *ip6addrlbl_alloc(struct net *net,
						 const struct in6_addr *prefix,
						 int prefixlen, int ifindex,
						 u32 label)
{
	struct ip6addrlbl_entry *newp;
	int addrtype;

	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u)\n",
		  __func__, 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);
#ifdef CONFIG_NET_NS
	newp->lbl_net = hold_net(net);
#endif
	atomic_set(&newp->refcnt, 1);
	return newp;
}

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

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

	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;
				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;
		}
		last = p;
	}
	if (last)
		hlist_add_after_rcu(&last->list, &newp->list);
	else
		hlist_add_head_rcu(&newp->list, &ip6addrlbl_table.head);
out:
	if (!ret)
		ip6addrlbl_table.seq++;
	return ret;
}

/* add a label */
static int ip6addrlbl_add(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
			  int ifindex, u32 label, int replace)
{
	struct ip6addrlbl_entry *newp;
	int ret = 0;

	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u, replace=%d)\n",
		  __func__, prefix, prefixlen, ifindex, (unsigned int)label,
		  replace);

	newp = ip6addrlbl_alloc(net, prefix, prefixlen, ifindex, label);
	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 */
static int __ip6addrlbl_del(struct net *net,
			    const struct in6_addr *prefix, int prefixlen,
			    int ifindex)
{
	struct ip6addrlbl_entry *p = NULL;
	struct hlist_node *n;
	int ret = -ESRCH;

	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
		  __func__, prefix, prefixlen, ifindex);

	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
		if (p->prefixlen == prefixlen &&
		    net_eq(ip6addrlbl_net(p), net) &&
		    p->ifindex == ifindex &&
		    ipv6_addr_equal(&p->prefix, prefix)) {
			hlist_del_rcu(&p->list);
			ip6addrlbl_put(p);
			ret = 0;
			break;
		}
	}
	return ret;
}

static int ip6addrlbl_del(struct net *net,
			  const struct in6_addr *prefix, int prefixlen,
			  int ifindex)
{
	struct in6_addr prefix_buf;
	int ret;

	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
		  __func__, prefix, prefixlen, ifindex);

	ipv6_addr_prefix(&prefix_buf, prefix, prefixlen);
	spin_lock(&ip6addrlbl_table.lock);
	ret = __ip6addrlbl_del(net, &prefix_buf, prefixlen, ifindex);
	spin_unlock(&ip6addrlbl_table.lock);
	return ret;
}

/* add default label */
static int __net_init ip6addrlbl_net_init(struct net *net)
{
	int err = 0;
	int i;

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

	for (i = 0; i < ARRAY_SIZE(ip6addrlbl_init_table); i++) {
		int ret = ip6addrlbl_add(net,
					 ip6addrlbl_init_table[i].prefix,
					 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;
}

static void __net_exit ip6addrlbl_net_exit(struct net *net)
{
	struct ip6addrlbl_entry *p = NULL;
	struct hlist_node *n;

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

int __init ipv6_addr_label_init(void)
{
	spin_lock_init(&ip6addrlbl_table.lock);

	return register_pernet_subsys(&ipv6_addr_label_ops);
}

void ipv6_addr_label_cleanup(void)
{
	unregister_pernet_subsys(&ipv6_addr_label_ops);
}

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)
{
	struct net *net = sock_net(skb->sk);
	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:
		if (ifal->ifal_index &&
		    !__dev_get_by_index(net, ifal->ifal_index))
			return -EINVAL;

		err = ip6addrlbl_add(net, pfx, ifal->ifal_prefixlen,
				     ifal->ifal_index, label,
				     nlh->nlmsg_flags & NLM_F_REPLACE);
		break;
	case RTM_DELADDRLABEL:
		err = ip6addrlbl_del(net, pfx, ifal->ifal_prefixlen,
				     ifal->ifal_index);
		break;
	default:
		err = -EOPNOTSUPP;
	}
	return err;
}

static 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 portid, u32 seq, int event,
			   unsigned int flags)
{
	struct nlmsghdr *nlh = nlmsg_put(skb, portid, 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)
{
	struct net *net = sock_net(skb->sk);
	struct ip6addrlbl_entry *p;
	int idx = 0, s_idx = cb->args[0];
	int err;

	rcu_read_lock();
	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
		if (idx >= s_idx &&
		    net_eq(ip6addrlbl_net(p), net)) {
			if ((err = ip6addrlbl_fill(skb, p,
						   ip6addrlbl_table.seq,
						   NETLINK_CB(cb->skb).portid,
						   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)
{
	struct net *net = sock_net(in_skb->sk);
	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 &&
	    !__dev_get_by_index(net, ifal->ifal_index))
		return -EINVAL;

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

	rcu_read_lock();
	p = __ipv6_addr_label(net, addr, ipv6_addr_type(addr), ifal->ifal_index);
	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).portid, nlh->nlmsg_seq,
			      RTM_NEWADDRLABEL, 0);

	ip6addrlbl_put(p);

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

	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
out:
	return err;
}

void __init ipv6_addr_label_rtnl_register(void)
{
	__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);
}
Commit	Line	Data
2a8cc6c8 YH	1	/*
	2	* IPv6 Address Label subsystem
	3	* for the IPv6 "Default" Source Address Selection
	4	*
	5	* Copyright (C)2007 USAGI/WIDE Project
	6	*/
	7	/*
	8	* Author:
	9	* YOSHIFUJI Hideaki @ USAGI/WIDE Project <yoshfuji@linux-ipv6.org>
	10	*/
	11
	12	#include <linux/kernel.h>
	13	#include <linux/list.h>
	14	#include <linux/rcupdate.h>
	15	#include <linux/in6.h>
5a0e3ad6	16	#include <linux/slab.h>
2a8cc6c8 YH	17	#include <net/addrconf.h>
	18	#include <linux/if_addrlabel.h>
	19	#include <linux/netlink.h>
	20	#include <linux/rtnetlink.h>
	21
	22	#if 0
	23	#define ADDRLABEL(x...) printk(x)
	24	#else
	25	#define ADDRLABEL(x...) do { ; } while(0)
	26	#endif
	27
	28	/*
	29	* Policy Table
	30	*/
22b285d6	31	struct ip6addrlbl_entry {
3de23255 BT	32	#ifdef CONFIG_NET_NS
	33	struct net *lbl_net;
	34	#endif
2a8cc6c8 YH	35	struct in6_addr prefix;
	36	int prefixlen;
	37	int ifindex;
	38	int addrtype;
	39	u32 label;
	40	struct hlist_node list;
	41	atomic_t refcnt;
	42	struct rcu_head rcu;
	43	};
	44
	45	static struct ip6addrlbl_table
	46	{
	47	struct hlist_head head;
	48	spinlock_t lock;
	49	u32 seq;
	50	} ip6addrlbl_table;
	51
3de23255 BT	52	static inline
	53	struct net ip6addrlbl_net(const struct ip6addrlbl_entry lbl)
	54	{
c2d9ba9b	55	return read_pnet(&lbl->lbl_net);
3de23255 BT	56	}
3de23255 BT	57
2a8cc6c8	58	/*
417962a0	59	* Default policy table (RFC6724 + extensions)
2a8cc6c8 YH	60	*
	61	* prefix addr_type label
	62	* -------------------------------------------------------------------------
	63	* ::1/128 LOOPBACK 0
	64	* ::/0 N/A 1
	65	* 2002::/16 N/A 2
	66	* ::/96 COMPATv4 3
	67	* ::ffff:0:0/96 V4MAPPED 4
	68	* fc00::/7 N/A 5 ULA (RFC 4193)
	69	* 2001::/32 N/A 6 Teredo (RFC 4380)
5fe47b8a	70	* 2001:10::/28 N/A 7 ORCHID (RFC 4843)
417962a0 YH	71	* fec0::/10 N/A 11 Site-local
	72	* (deprecated by RFC3879)
	73	* 3ffe::/16 N/A 12 6bone
2a8cc6c8 YH	74	*
2a8cc6c8 YH	75	* Note: 0xffffffff is used if we do not have any policies.
417962a0	76	* Note: Labels for ULA and 6to4 are different from labels listed in RFC6724.
2a8cc6c8 YH	77	*/
	78
	79	#define IPV6_ADDR_LABEL_DEFAULT 0xffffffffUL
	80
04a6f82c	81	static const __net_initconst struct ip6addrlbl_init_table
2a8cc6c8 YH	82	{
	83	const struct in6_addr *prefix;
	84	int prefixlen;
	85	u32 label;
	86	} ip6addrlbl_init_table[] = {
	87	{ /* ::/0 */
	88	.prefix = &in6addr_any,
	89	.label = 1,
	90	},{ /* fc00::/7 */
	91	.prefix = &(struct in6_addr){{{ 0xfc }}},
	92	.prefixlen = 7,
	93	.label = 5,
417962a0 YH	94	},{ /* fec0::/10 */
	95	.prefix = &(struct in6_addr){{{ 0xfe, 0xc0 }}},
	96	.prefixlen = 10,
	97	.label = 11,
2a8cc6c8 YH	98	},{ /* 2002::/16 */
	99	.prefix = &(struct in6_addr){{{ 0x20, 0x02 }}},
	100	.prefixlen = 16,
	101	.label = 2,
417962a0 YH	102	},{ /* 3ffe::/16 */
	103	.prefix = &(struct in6_addr){{{ 0x3f, 0xfe }}},
	104	.prefixlen = 16,
	105	.label = 12,
2a8cc6c8 YH	106	},{ /* 2001::/32 */
	107	.prefix = &(struct in6_addr){{{ 0x20, 0x01 }}},
	108	.prefixlen = 32,
	109	.label = 6,
5fe47b8a JMT	110	},{ /* 2001:10::/28 */
	111	.prefix = &(struct in6_addr){{{ 0x20, 0x01, 0x00, 0x10 }}},
	112	.prefixlen = 28,
	113	.label = 7,
2a8cc6c8 YH	114	},{ /* ::ffff:0:0 */
	115	.prefix = &(struct in6_addr){{{ [10] = 0xff, [11] = 0xff }}},
	116	.prefixlen = 96,
	117	.label = 4,
	118	},{ /* ::/96 */
	119	.prefix = &in6addr_any,
	120	.prefixlen = 96,
	121	.label = 3,
	122	},{ /* ::1/128 */
	123	.prefix = &in6addr_loopback,
	124	.prefixlen = 128,
	125	.label = 0,
	126	}
	127	};
	128
	129	/* Object management */
	130	static inline void ip6addrlbl_free(struct ip6addrlbl_entry *p)
	131	{
3de23255 BT	132	#ifdef CONFIG_NET_NS
	133	release_net(p->lbl_net);
	134	#endif
2a8cc6c8 YH	135	kfree(p);
	136	}
	137
85040bcb YH	138	static void ip6addrlbl_free_rcu(struct rcu_head *h)
	139	{
	140	ip6addrlbl_free(container_of(h, struct ip6addrlbl_entry, rcu));
	141	}
	142
a50feda5	143	static bool ip6addrlbl_hold(struct ip6addrlbl_entry *p)
2a8cc6c8 YH	144	{
	145	return atomic_inc_not_zero(&p->refcnt);
	146	}
	147
	148	static inline void ip6addrlbl_put(struct ip6addrlbl_entry *p)
	149	{
	150	if (atomic_dec_and_test(&p->refcnt))
85040bcb	151	call_rcu(&p->rcu, ip6addrlbl_free_rcu);
2a8cc6c8 YH	152	}
	153
	154	/* Find label */
a50feda5 ED	155	static bool __ip6addrlbl_match(struct net *net,
	156	const struct ip6addrlbl_entry *p,
	157	const struct in6_addr *addr,
	158	int addrtype, int ifindex)
2a8cc6c8	159	{
3de23255	160	if (!net_eq(ip6addrlbl_net(p), net))
a50feda5	161	return false;
2a8cc6c8	162	if (p->ifindex && p->ifindex != ifindex)
a50feda5	163	return false;
2a8cc6c8	164	if (p->addrtype && p->addrtype != addrtype)
a50feda5	165	return false;
2a8cc6c8	166	if (!ipv6_prefix_equal(addr, &p->prefix, p->prefixlen))
a50feda5 ED	167	return false;
a50feda5 ED	168	return true;
2a8cc6c8 YH	169	}
2a8cc6c8 YH	170
3de23255 BT	171	static struct ip6addrlbl_entry __ipv6_addr_label(struct net net,
3de23255 BT	172	const struct in6_addr *addr,
2a8cc6c8 YH	173	int type, int ifindex)
2a8cc6c8 YH	174	{
2a8cc6c8	175	struct ip6addrlbl_entry *p;
b67bfe0d	176	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
3de23255	177	if (__ip6addrlbl_match(net, p, addr, type, ifindex))
2a8cc6c8 YH	178	return p;
	179	}
	180	return NULL;
	181	}
	182
3de23255 BT	183	u32 ipv6_addr_label(struct net *net,
3de23255 BT	184	const struct in6_addr *addr, int type, int ifindex)
2a8cc6c8 YH	185	{
	186	u32 label;
	187	struct ip6addrlbl_entry *p;
	188
	189	type &= IPV6_ADDR_MAPPED \| IPV6_ADDR_COMPATv4 \| IPV6_ADDR_LOOPBACK;
	190
	191	rcu_read_lock();
3de23255	192	p = __ipv6_addr_label(net, addr, type, ifindex);
2a8cc6c8 YH	193	label = p ? p->label : IPV6_ADDR_LABEL_DEFAULT;
	194	rcu_read_unlock();
	195
5b095d98	196	ADDRLABEL(KERN_DEBUG "%s(addr=%pI6, type=%d, ifindex=%d) => %08x\n",
0c6ce78a	197	__func__, addr, type, ifindex, label);
2a8cc6c8 YH	198
	199	return label;
	200	}
	201
	202	/* allocate one entry */
3de23255 BT	203	static struct ip6addrlbl_entry ip6addrlbl_alloc(struct net net,
3de23255 BT	204	const struct in6_addr *prefix,
40fee36e YH	205	int prefixlen, int ifindex,
40fee36e YH	206	u32 label)
2a8cc6c8 YH	207	{
	208	struct ip6addrlbl_entry *newp;
	209	int addrtype;
	210
5b095d98	211	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u)\n",
0c6ce78a	212	__func__, prefix, prefixlen, ifindex, (unsigned int)label);
2a8cc6c8 YH	213
	214	addrtype = ipv6_addr_type(prefix) & (IPV6_ADDR_MAPPED \| IPV6_ADDR_COMPATv4 \| IPV6_ADDR_LOOPBACK);
	215
	216	switch (addrtype) {
	217	case IPV6_ADDR_MAPPED:
	218	if (prefixlen > 96)
	219	return ERR_PTR(-EINVAL);
	220	if (prefixlen < 96)
	221	addrtype = 0;
	222	break;
	223	case IPV6_ADDR_COMPATv4:
	224	if (prefixlen != 96)
	225	addrtype = 0;
	226	break;
	227	case IPV6_ADDR_LOOPBACK:
	228	if (prefixlen != 128)
	229	addrtype = 0;
	230	break;
	231	}
	232
	233	newp = kmalloc(sizeof(*newp), GFP_KERNEL);
	234	if (!newp)
	235	return ERR_PTR(-ENOMEM);
	236
	237	ipv6_addr_prefix(&newp->prefix, prefix, prefixlen);
	238	newp->prefixlen = prefixlen;
	239	newp->ifindex = ifindex;
	240	newp->addrtype = addrtype;
	241	newp->label = label;
	242	INIT_HLIST_NODE(&newp->list);
3de23255 BT	243	#ifdef CONFIG_NET_NS
	244	newp->lbl_net = hold_net(net);
	245	#endif
2a8cc6c8 YH	246	atomic_set(&newp->refcnt, 1);
	247	return newp;
	248	}
	249
	250	/* add a label */
40fee36e	251	static int __ip6addrlbl_add(struct ip6addrlbl_entry *newp, int replace)
2a8cc6c8	252	{
639739b5 HFS	253	struct hlist_node *n;
639739b5 HFS	254	struct ip6addrlbl_entry last = NULL, p = NULL;
2a8cc6c8 YH	255	int ret = 0;
2a8cc6c8 YH	256
639739b5 HFS	257	ADDRLABEL(KERN_DEBUG "%s(newp=%p, replace=%d)\n", __func__, newp,
639739b5 HFS	258	replace);
2a8cc6c8	259
639739b5 HFS	260	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
	261	if (p->prefixlen == newp->prefixlen &&
	262	net_eq(ip6addrlbl_net(p), ip6addrlbl_net(newp)) &&
	263	p->ifindex == newp->ifindex &&
	264	ipv6_addr_equal(&p->prefix, &newp->prefix)) {
	265	if (!replace) {
	266	ret = -EEXIST;
2a8cc6c8 YH	267	goto out;
2a8cc6c8 YH	268	}
639739b5 HFS	269	hlist_replace_rcu(&p->list, &newp->list);
	270	ip6addrlbl_put(p);
	271	goto out;
	272	} else if ((p->prefixlen == newp->prefixlen && !p->ifindex) \|\|
	273	(p->prefixlen < newp->prefixlen)) {
	274	hlist_add_before_rcu(&newp->list, &p->list);
	275	goto out;
2a8cc6c8	276	}
639739b5	277	last = p;
2a8cc6c8	278	}
639739b5 HFS	279	if (last)
	280	hlist_add_after_rcu(&last->list, &newp->list);
	281	else
	282	hlist_add_head_rcu(&newp->list, &ip6addrlbl_table.head);
2a8cc6c8 YH	283	out:
	284	if (!ret)
	285	ip6addrlbl_table.seq++;
	286	return ret;
	287	}
	288
	289	/* add a label */
3de23255 BT	290	static int ip6addrlbl_add(struct net *net,
3de23255 BT	291	const struct in6_addr *prefix, int prefixlen,
40fee36e	292	int ifindex, u32 label, int replace)
2a8cc6c8 YH	293	{
	294	struct ip6addrlbl_entry *newp;
	295	int ret = 0;
	296
5b095d98	297	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u, replace=%d)\n",
0c6ce78a HH	298	__func__, prefix, prefixlen, ifindex, (unsigned int)label,
0c6ce78a HH	299	replace);
2a8cc6c8	300
3de23255	301	newp = ip6addrlbl_alloc(net, prefix, prefixlen, ifindex, label);
2a8cc6c8 YH	302	if (IS_ERR(newp))
	303	return PTR_ERR(newp);
	304	spin_lock(&ip6addrlbl_table.lock);
	305	ret = __ip6addrlbl_add(newp, replace);
	306	spin_unlock(&ip6addrlbl_table.lock);
	307	if (ret)
	308	ip6addrlbl_free(newp);
	309	return ret;
	310	}
	311
	312	/* remove a label */
3de23255 BT	313	static int __ip6addrlbl_del(struct net *net,
3de23255 BT	314	const struct in6_addr *prefix, int prefixlen,
40fee36e	315	int ifindex)
2a8cc6c8 YH	316	{
2a8cc6c8 YH	317	struct ip6addrlbl_entry *p = NULL;
b67bfe0d	318	struct hlist_node *n;
2a8cc6c8 YH	319	int ret = -ESRCH;
2a8cc6c8 YH	320
5b095d98	321	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
0c6ce78a	322	__func__, prefix, prefixlen, ifindex);
2a8cc6c8	323
b67bfe0d	324	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
2a8cc6c8	325	if (p->prefixlen == prefixlen &&
3de23255	326	net_eq(ip6addrlbl_net(p), net) &&
2a8cc6c8 YH	327	p->ifindex == ifindex &&
	328	ipv6_addr_equal(&p->prefix, prefix)) {
	329	hlist_del_rcu(&p->list);
	330	ip6addrlbl_put(p);
2a8cc6c8 YH	331	ret = 0;
	332	break;
	333	}
	334	}
	335	return ret;
	336	}
	337
3de23255 BT	338	static int ip6addrlbl_del(struct net *net,
3de23255 BT	339	const struct in6_addr *prefix, int prefixlen,
40fee36e	340	int ifindex)
2a8cc6c8 YH	341	{
	342	struct in6_addr prefix_buf;
	343	int ret;
	344
5b095d98	345	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
0c6ce78a	346	__func__, prefix, prefixlen, ifindex);
2a8cc6c8 YH	347
	348	ipv6_addr_prefix(&prefix_buf, prefix, prefixlen);
	349	spin_lock(&ip6addrlbl_table.lock);
3de23255	350	ret = __ip6addrlbl_del(net, &prefix_buf, prefixlen, ifindex);
2a8cc6c8 YH	351	spin_unlock(&ip6addrlbl_table.lock);
	352	return ret;
	353	}
	354
	355	/* add default label */
3de23255	356	static int __net_init ip6addrlbl_net_init(struct net *net)
2a8cc6c8 YH	357	{
	358	int err = 0;
	359	int i;
	360
f3213831	361	ADDRLABEL(KERN_DEBUG "%s\n", __func__);
2a8cc6c8 YH	362
2a8cc6c8 YH	363	for (i = 0; i < ARRAY_SIZE(ip6addrlbl_init_table); i++) {
3de23255 BT	364	int ret = ip6addrlbl_add(net,
3de23255 BT	365	ip6addrlbl_init_table[i].prefix,
2a8cc6c8 YH	366	ip6addrlbl_init_table[i].prefixlen,
	367	0,
	368	ip6addrlbl_init_table[i].label, 0);
	369	/* XXX: should we free all rules when we catch an error? */
	370	if (ret && (!err \|\| err != -ENOMEM))
	371	err = ret;
	372	}
	373	return err;
	374	}
	375
3de23255 BT	376	static void __net_exit ip6addrlbl_net_exit(struct net *net)
	377	{
	378	struct ip6addrlbl_entry *p = NULL;
b67bfe0d	379	struct hlist_node *n;
3de23255 BT	380
	381	/* Remove all labels belonging to the exiting net */
	382	spin_lock(&ip6addrlbl_table.lock);
b67bfe0d	383	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
3de23255 BT	384	if (net_eq(ip6addrlbl_net(p), net)) {
	385	hlist_del_rcu(&p->list);
	386	ip6addrlbl_put(p);
	387	}
	388	}
	389	spin_unlock(&ip6addrlbl_table.lock);
	390	}
	391
	392	static struct pernet_operations ipv6_addr_label_ops = {
	393	.init = ip6addrlbl_net_init,
	394	.exit = ip6addrlbl_net_exit,
	395	};
	396
2a8cc6c8 YH	397	int __init ipv6_addr_label_init(void)
	398	{
	399	spin_lock_init(&ip6addrlbl_table.lock);
	400
3de23255	401	return register_pernet_subsys(&ipv6_addr_label_ops);
2a8cc6c8 YH	402	}
2a8cc6c8 YH	403
2cc6d2bf NH	404	void ipv6_addr_label_cleanup(void)
	405	{
	406	unregister_pernet_subsys(&ipv6_addr_label_ops);
	407	}
	408
2a8cc6c8 YH	409	static const struct nla_policy ifal_policy[IFAL_MAX+1] = {
	410	[IFAL_ADDRESS] = { .len = sizeof(struct in6_addr), },
	411	[IFAL_LABEL] = { .len = sizeof(u32), },
	412	};
	413
661d2967	414	static int ip6addrlbl_newdel(struct sk_buff skb, struct nlmsghdr nlh)
2a8cc6c8	415	{
3b1e0a65	416	struct net *net = sock_net(skb->sk);
2a8cc6c8 YH	417	struct ifaddrlblmsg *ifal;
	418	struct nlattr *tb[IFAL_MAX+1];
	419	struct in6_addr *pfx;
	420	u32 label;
	421	int err = 0;
	422
	423	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
	424	if (err < 0)
	425	return err;
	426
	427	ifal = nlmsg_data(nlh);
	428
	429	if (ifal->ifal_family != AF_INET6 \|\|
	430	ifal->ifal_prefixlen > 128)
	431	return -EINVAL;
	432
2a8cc6c8 YH	433	if (!tb[IFAL_ADDRESS])
2a8cc6c8 YH	434	return -EINVAL;
2a8cc6c8	435	pfx = nla_data(tb[IFAL_ADDRESS]);
2a8cc6c8 YH	436
	437	if (!tb[IFAL_LABEL])
	438	return -EINVAL;
	439	label = nla_get_u32(tb[IFAL_LABEL]);
	440	if (label == IPV6_ADDR_LABEL_DEFAULT)
	441	return -EINVAL;
	442
	443	switch(nlh->nlmsg_type) {
	444	case RTM_NEWADDRLABEL:
0771275b FW	445	if (ifal->ifal_index &&
	446	!__dev_get_by_index(net, ifal->ifal_index))
	447	return -EINVAL;
	448
3de23255	449	err = ip6addrlbl_add(net, pfx, ifal->ifal_prefixlen,
2a8cc6c8 YH	450	ifal->ifal_index, label,
	451	nlh->nlmsg_flags & NLM_F_REPLACE);
	452	break;
	453	case RTM_DELADDRLABEL:
3de23255	454	err = ip6addrlbl_del(net, pfx, ifal->ifal_prefixlen,
2a8cc6c8 YH	455	ifal->ifal_index);
	456	break;
	457	default:
	458	err = -EOPNOTSUPP;
	459	}
	460	return err;
	461	}
	462
a50feda5 ED	463	static void ip6addrlbl_putmsg(struct nlmsghdr *nlh,
a50feda5 ED	464	int prefixlen, int ifindex, u32 lseq)
2a8cc6c8 YH	465	{
	466	struct ifaddrlblmsg *ifal = nlmsg_data(nlh);
	467	ifal->ifal_family = AF_INET6;
	468	ifal->ifal_prefixlen = prefixlen;
	469	ifal->ifal_flags = 0;
	470	ifal->ifal_index = ifindex;
	471	ifal->ifal_seq = lseq;
	472	};
	473
	474	static int ip6addrlbl_fill(struct sk_buff *skb,
	475	struct ip6addrlbl_entry *p,
	476	u32 lseq,
15e47304	477	u32 portid, u32 seq, int event,
2a8cc6c8 YH	478	unsigned int flags)
2a8cc6c8 YH	479	{
15e47304	480	struct nlmsghdr *nlh = nlmsg_put(skb, portid, seq, event,
2a8cc6c8 YH	481	sizeof(struct ifaddrlblmsg), flags);
	482	if (!nlh)
	483	return -EMSGSIZE;
	484
	485	ip6addrlbl_putmsg(nlh, p->prefixlen, p->ifindex, lseq);
	486
	487	if (nla_put(skb, IFAL_ADDRESS, 16, &p->prefix) < 0 \|\|
	488	nla_put_u32(skb, IFAL_LABEL, p->label) < 0) {
	489	nlmsg_cancel(skb, nlh);
	490	return -EMSGSIZE;
	491	}
	492
	493	return nlmsg_end(skb, nlh);
	494	}
	495
	496	static int ip6addrlbl_dump(struct sk_buff skb, struct netlink_callback cb)
	497	{
3b1e0a65	498	struct net *net = sock_net(skb->sk);
2a8cc6c8	499	struct ip6addrlbl_entry *p;
2a8cc6c8 YH	500	int idx = 0, s_idx = cb->args[0];
	501	int err;
	502
	503	rcu_read_lock();
b67bfe0d	504	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
3de23255 BT	505	if (idx >= s_idx &&
3de23255 BT	506	net_eq(ip6addrlbl_net(p), net)) {
2a8cc6c8 YH	507	if ((err = ip6addrlbl_fill(skb, p,
2a8cc6c8 YH	508	ip6addrlbl_table.seq,
15e47304	509	NETLINK_CB(cb->skb).portid,
2a8cc6c8 YH	510	cb->nlh->nlmsg_seq,
	511	RTM_NEWADDRLABEL,
	512	NLM_F_MULTI)) <= 0)
	513	break;
	514	}
	515	idx++;
	516	}
	517	rcu_read_unlock();
	518	cb->args[0] = idx;
	519	return skb->len;
	520	}
	521
	522	static inline int ip6addrlbl_msgsize(void)
	523	{
a02cec21	524	return NLMSG_ALIGN(sizeof(struct ifaddrlblmsg))
2a8cc6c8	525	+ nla_total_size(16) /* IFAL_ADDRESS */
a02cec21	526	+ nla_total_size(4); /* IFAL_LABEL */
2a8cc6c8 YH	527	}
2a8cc6c8 YH	528
22b285d6	529	static int ip6addrlbl_get(struct sk_buff in_skb, struct nlmsghdr nlh)
2a8cc6c8	530	{
3b1e0a65	531	struct net *net = sock_net(in_skb->sk);
2a8cc6c8 YH	532	struct ifaddrlblmsg *ifal;
	533	struct nlattr *tb[IFAL_MAX+1];
	534	struct in6_addr *addr;
	535	u32 lseq;
	536	int err = 0;
	537	struct ip6addrlbl_entry *p;
	538	struct sk_buff *skb;
	539
	540	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
	541	if (err < 0)
	542	return err;
	543
	544	ifal = nlmsg_data(nlh);
	545
	546	if (ifal->ifal_family != AF_INET6 \|\|
	547	ifal->ifal_prefixlen != 128)
	548	return -EINVAL;
	549
	550	if (ifal->ifal_index &&
3de23255	551	!__dev_get_by_index(net, ifal->ifal_index))
2a8cc6c8 YH	552	return -EINVAL;
	553
	554	if (!tb[IFAL_ADDRESS])
	555	return -EINVAL;
2a8cc6c8	556	addr = nla_data(tb[IFAL_ADDRESS]);
2a8cc6c8 YH	557
2a8cc6c8 YH	558	rcu_read_lock();
3de23255	559	p = __ipv6_addr_label(net, addr, ipv6_addr_type(addr), ifal->ifal_index);
2a8cc6c8 YH	560	if (p && ip6addrlbl_hold(p))
	561	p = NULL;
	562	lseq = ip6addrlbl_table.seq;
	563	rcu_read_unlock();
	564
	565	if (!p) {
	566	err = -ESRCH;
	567	goto out;
	568	}
	569
	570	if (!(skb = nlmsg_new(ip6addrlbl_msgsize(), GFP_KERNEL))) {
	571	ip6addrlbl_put(p);
	572	return -ENOBUFS;
	573	}
	574
	575	err = ip6addrlbl_fill(skb, p, lseq,
15e47304	576	NETLINK_CB(in_skb).portid, nlh->nlmsg_seq,
2a8cc6c8 YH	577	RTM_NEWADDRLABEL, 0);
	578
	579	ip6addrlbl_put(p);
	580
	581	if (err < 0) {
	582	WARN_ON(err == -EMSGSIZE);
	583	kfree_skb(skb);
	584	goto out;
	585	}
	586
15e47304	587	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
2a8cc6c8 YH	588	out:
	589	return err;
	590	}
	591
	592	void __init ipv6_addr_label_rtnl_register(void)
	593	{
c7ac8679 GR	594	__rtnl_register(PF_INET6, RTM_NEWADDRLABEL, ip6addrlbl_newdel,
	595	NULL, NULL);
	596	__rtnl_register(PF_INET6, RTM_DELADDRLABEL, ip6addrlbl_newdel,
	597	NULL, NULL);
	598	__rtnl_register(PF_INET6, RTM_GETADDRLABEL, ip6addrlbl_get,
	599	ip6addrlbl_dump, NULL);
2a8cc6c8 YH	600	}
2a8cc6c8 YH	601