[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 {
	possible_net_t lbl_net;
	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)
{
	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);
	write_pnet(&newp->lbl_net, net);
	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_behind_rcu(&newp->list, &last->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_in6_addr(skb, IFAL_ADDRESS, &p->prefix) < 0 ||
	    nla_put_u32(skb, IFAL_LABEL, p->label) < 0) {
		nlmsg_cancel(skb, nlh);
		return -EMSGSIZE;
	}

	nlmsg_end(skb, nlh);
	return 0;
}

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)) {
			err = ip6addrlbl_fill(skb, p,
					      ip6addrlbl_table.seq,
					      NETLINK_CB(cb->skb).portid,
					      cb->nlh->nlmsg_seq,
					      RTM_NEWADDRLABEL,
					      NLM_F_MULTI);
			if (err < 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;
	}

	skb = nlmsg_new(ip6addrlbl_msgsize(), GFP_KERNEL);
	if (!skb) {
		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:
1c8669e0	9	* YOSHIFUJI Hideaki @ USAGI/WIDE Project <yoshfuji@linux-ipv6.org>
2a8cc6c8 YH	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
1c8669e0	25	#define ADDRLABEL(x...) do { ; } while (0)
2a8cc6c8 YH	26	#endif
	27
	28	/*
	29	* Policy Table
	30	*/
22b285d6	31	struct ip6addrlbl_entry {
0c5c9fb5	32	possible_net_t lbl_net;
2a8cc6c8 YH	33	struct in6_addr prefix;
	34	int prefixlen;
	35	int ifindex;
	36	int addrtype;
	37	u32 label;
	38	struct hlist_node list;
	39	atomic_t refcnt;
	40	struct rcu_head rcu;
	41	};
	42
	43	static struct ip6addrlbl_table
	44	{
	45	struct hlist_head head;
	46	spinlock_t lock;
	47	u32 seq;
	48	} ip6addrlbl_table;
	49
3de23255 BT	50	static inline
	51	struct net ip6addrlbl_net(const struct ip6addrlbl_entry lbl)
	52	{
c2d9ba9b	53	return read_pnet(&lbl->lbl_net);
3de23255 BT	54	}
3de23255 BT	55
2a8cc6c8	56	/*
417962a0	57	* Default policy table (RFC6724 + extensions)
2a8cc6c8 YH	58	*
	59	* prefix addr_type label
	60	* -------------------------------------------------------------------------
	61	* ::1/128 LOOPBACK 0
	62	* ::/0 N/A 1
	63	* 2002::/16 N/A 2
	64	* ::/96 COMPATv4 3
	65	* ::ffff:0:0/96 V4MAPPED 4
	66	* fc00::/7 N/A 5 ULA (RFC 4193)
	67	* 2001::/32 N/A 6 Teredo (RFC 4380)
5fe47b8a	68	* 2001:10::/28 N/A 7 ORCHID (RFC 4843)
417962a0 YH	69	* fec0::/10 N/A 11 Site-local
	70	* (deprecated by RFC3879)
	71	* 3ffe::/16 N/A 12 6bone
2a8cc6c8 YH	72	*
2a8cc6c8 YH	73	* Note: 0xffffffff is used if we do not have any policies.
417962a0	74	* Note: Labels for ULA and 6to4 are different from labels listed in RFC6724.
2a8cc6c8 YH	75	*/
	76
	77	#define IPV6_ADDR_LABEL_DEFAULT 0xffffffffUL
	78
04a6f82c	79	static const __net_initconst struct ip6addrlbl_init_table
2a8cc6c8 YH	80	{
	81	const struct in6_addr *prefix;
	82	int prefixlen;
	83	u32 label;
	84	} ip6addrlbl_init_table[] = {
	85	{ /* ::/0 */
	86	.prefix = &in6addr_any,
	87	.label = 1,
1c8669e0 WY	88	}, { /* fc00::/7 */
1c8669e0 WY	89	.prefix = &(struct in6_addr){ { { 0xfc } } } ,
2a8cc6c8 YH	90	.prefixlen = 7,
2a8cc6c8 YH	91	.label = 5,
1c8669e0 WY	92	}, { /* fec0::/10 */
1c8669e0 WY	93	.prefix = &(struct in6_addr){ { { 0xfe, 0xc0 } } },
417962a0 YH	94	.prefixlen = 10,
417962a0 YH	95	.label = 11,
1c8669e0 WY	96	}, { /* 2002::/16 */
1c8669e0 WY	97	.prefix = &(struct in6_addr){ { { 0x20, 0x02 } } },
2a8cc6c8 YH	98	.prefixlen = 16,
2a8cc6c8 YH	99	.label = 2,
1c8669e0 WY	100	}, { /* 3ffe::/16 */
1c8669e0 WY	101	.prefix = &(struct in6_addr){ { { 0x3f, 0xfe } } },
417962a0 YH	102	.prefixlen = 16,
417962a0 YH	103	.label = 12,
1c8669e0 WY	104	}, { /* 2001::/32 */
1c8669e0 WY	105	.prefix = &(struct in6_addr){ { { 0x20, 0x01 } } },
2a8cc6c8 YH	106	.prefixlen = 32,
2a8cc6c8 YH	107	.label = 6,
1c8669e0 WY	108	}, { /* 2001:10::/28 */
1c8669e0 WY	109	.prefix = &(struct in6_addr){ { { 0x20, 0x01, 0x00, 0x10 } } },
5fe47b8a JMT	110	.prefixlen = 28,
5fe47b8a JMT	111	.label = 7,
1c8669e0 WY	112	}, { /* ::ffff:0:0 */
1c8669e0 WY	113	.prefix = &(struct in6_addr){ { { [10] = 0xff, [11] = 0xff } } },
2a8cc6c8 YH	114	.prefixlen = 96,
2a8cc6c8 YH	115	.label = 4,
1c8669e0	116	}, { /* ::/96 */
2a8cc6c8 YH	117	.prefix = &in6addr_any,
	118	.prefixlen = 96,
	119	.label = 3,
1c8669e0	120	}, { /* ::1/128 */
2a8cc6c8 YH	121	.prefix = &in6addr_loopback,
	122	.prefixlen = 128,
	123	.label = 0,
	124	}
	125	};
	126
	127	/* Object management */
	128	static inline void ip6addrlbl_free(struct ip6addrlbl_entry *p)
	129	{
	130	kfree(p);
	131	}
	132
85040bcb YH	133	static void ip6addrlbl_free_rcu(struct rcu_head *h)
	134	{
	135	ip6addrlbl_free(container_of(h, struct ip6addrlbl_entry, rcu));
	136	}
	137
a50feda5	138	static bool ip6addrlbl_hold(struct ip6addrlbl_entry *p)
2a8cc6c8 YH	139	{
	140	return atomic_inc_not_zero(&p->refcnt);
	141	}
	142
	143	static inline void ip6addrlbl_put(struct ip6addrlbl_entry *p)
	144	{
	145	if (atomic_dec_and_test(&p->refcnt))
85040bcb	146	call_rcu(&p->rcu, ip6addrlbl_free_rcu);
2a8cc6c8 YH	147	}
	148
	149	/* Find label */
a50feda5 ED	150	static bool __ip6addrlbl_match(struct net *net,
	151	const struct ip6addrlbl_entry *p,
	152	const struct in6_addr *addr,
	153	int addrtype, int ifindex)
2a8cc6c8	154	{
3de23255	155	if (!net_eq(ip6addrlbl_net(p), net))
a50feda5	156	return false;
2a8cc6c8	157	if (p->ifindex && p->ifindex != ifindex)
a50feda5	158	return false;
2a8cc6c8	159	if (p->addrtype && p->addrtype != addrtype)
a50feda5	160	return false;
2a8cc6c8	161	if (!ipv6_prefix_equal(addr, &p->prefix, p->prefixlen))
a50feda5 ED	162	return false;
a50feda5 ED	163	return true;
2a8cc6c8 YH	164	}
2a8cc6c8 YH	165
3de23255 BT	166	static struct ip6addrlbl_entry __ipv6_addr_label(struct net net,
3de23255 BT	167	const struct in6_addr *addr,
2a8cc6c8 YH	168	int type, int ifindex)
2a8cc6c8 YH	169	{
2a8cc6c8	170	struct ip6addrlbl_entry *p;
b67bfe0d	171	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
3de23255	172	if (__ip6addrlbl_match(net, p, addr, type, ifindex))
2a8cc6c8 YH	173	return p;
	174	}
	175	return NULL;
	176	}
	177
3de23255 BT	178	u32 ipv6_addr_label(struct net *net,
3de23255 BT	179	const struct in6_addr *addr, int type, int ifindex)
2a8cc6c8 YH	180	{
	181	u32 label;
	182	struct ip6addrlbl_entry *p;
	183
	184	type &= IPV6_ADDR_MAPPED \| IPV6_ADDR_COMPATv4 \| IPV6_ADDR_LOOPBACK;
	185
	186	rcu_read_lock();
3de23255	187	p = __ipv6_addr_label(net, addr, type, ifindex);
2a8cc6c8 YH	188	label = p ? p->label : IPV6_ADDR_LABEL_DEFAULT;
	189	rcu_read_unlock();
	190
5b095d98	191	ADDRLABEL(KERN_DEBUG "%s(addr=%pI6, type=%d, ifindex=%d) => %08x\n",
0c6ce78a	192	__func__, addr, type, ifindex, label);
2a8cc6c8 YH	193
	194	return label;
	195	}
	196
	197	/* allocate one entry */
3de23255 BT	198	static struct ip6addrlbl_entry ip6addrlbl_alloc(struct net net,
3de23255 BT	199	const struct in6_addr *prefix,
40fee36e YH	200	int prefixlen, int ifindex,
40fee36e YH	201	u32 label)
2a8cc6c8 YH	202	{
	203	struct ip6addrlbl_entry *newp;
	204	int addrtype;
	205
5b095d98	206	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u)\n",
0c6ce78a	207	__func__, prefix, prefixlen, ifindex, (unsigned int)label);
2a8cc6c8 YH	208
	209	addrtype = ipv6_addr_type(prefix) & (IPV6_ADDR_MAPPED \| IPV6_ADDR_COMPATv4 \| IPV6_ADDR_LOOPBACK);
	210
	211	switch (addrtype) {
	212	case IPV6_ADDR_MAPPED:
	213	if (prefixlen > 96)
	214	return ERR_PTR(-EINVAL);
	215	if (prefixlen < 96)
	216	addrtype = 0;
	217	break;
	218	case IPV6_ADDR_COMPATv4:
	219	if (prefixlen != 96)
	220	addrtype = 0;
	221	break;
	222	case IPV6_ADDR_LOOPBACK:
	223	if (prefixlen != 128)
	224	addrtype = 0;
	225	break;
	226	}
	227
	228	newp = kmalloc(sizeof(*newp), GFP_KERNEL);
	229	if (!newp)
	230	return ERR_PTR(-ENOMEM);
	231
	232	ipv6_addr_prefix(&newp->prefix, prefix, prefixlen);
	233	newp->prefixlen = prefixlen;
	234	newp->ifindex = ifindex;
	235	newp->addrtype = addrtype;
	236	newp->label = label;
	237	INIT_HLIST_NODE(&newp->list);
0c5c9fb5	238	write_pnet(&newp->lbl_net, net);
2a8cc6c8 YH	239	atomic_set(&newp->refcnt, 1);
	240	return newp;
	241	}
	242
	243	/* add a label */
40fee36e	244	static int __ip6addrlbl_add(struct ip6addrlbl_entry *newp, int replace)
2a8cc6c8	245	{
639739b5 HFS	246	struct hlist_node *n;
639739b5 HFS	247	struct ip6addrlbl_entry last = NULL, p = NULL;
2a8cc6c8 YH	248	int ret = 0;
2a8cc6c8 YH	249
639739b5 HFS	250	ADDRLABEL(KERN_DEBUG "%s(newp=%p, replace=%d)\n", __func__, newp,
639739b5 HFS	251	replace);
2a8cc6c8	252
639739b5 HFS	253	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
	254	if (p->prefixlen == newp->prefixlen &&
	255	net_eq(ip6addrlbl_net(p), ip6addrlbl_net(newp)) &&
	256	p->ifindex == newp->ifindex &&
	257	ipv6_addr_equal(&p->prefix, &newp->prefix)) {
	258	if (!replace) {
	259	ret = -EEXIST;
2a8cc6c8 YH	260	goto out;
2a8cc6c8 YH	261	}
639739b5 HFS	262	hlist_replace_rcu(&p->list, &newp->list);
	263	ip6addrlbl_put(p);
	264	goto out;
	265	} else if ((p->prefixlen == newp->prefixlen && !p->ifindex) \|\|
	266	(p->prefixlen < newp->prefixlen)) {
	267	hlist_add_before_rcu(&newp->list, &p->list);
	268	goto out;
2a8cc6c8	269	}
639739b5	270	last = p;
2a8cc6c8	271	}
639739b5	272	if (last)
1d023284	273	hlist_add_behind_rcu(&newp->list, &last->list);
639739b5 HFS	274	else
639739b5 HFS	275	hlist_add_head_rcu(&newp->list, &ip6addrlbl_table.head);
2a8cc6c8 YH	276	out:
	277	if (!ret)
	278	ip6addrlbl_table.seq++;
	279	return ret;
	280	}
	281
	282	/* add a label */
3de23255 BT	283	static int ip6addrlbl_add(struct net *net,
3de23255 BT	284	const struct in6_addr *prefix, int prefixlen,
40fee36e	285	int ifindex, u32 label, int replace)
2a8cc6c8 YH	286	{
	287	struct ip6addrlbl_entry *newp;
	288	int ret = 0;
	289
5b095d98	290	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d, label=%u, replace=%d)\n",
0c6ce78a HH	291	__func__, prefix, prefixlen, ifindex, (unsigned int)label,
0c6ce78a HH	292	replace);
2a8cc6c8	293
3de23255	294	newp = ip6addrlbl_alloc(net, prefix, prefixlen, ifindex, label);
2a8cc6c8 YH	295	if (IS_ERR(newp))
	296	return PTR_ERR(newp);
	297	spin_lock(&ip6addrlbl_table.lock);
	298	ret = __ip6addrlbl_add(newp, replace);
	299	spin_unlock(&ip6addrlbl_table.lock);
	300	if (ret)
	301	ip6addrlbl_free(newp);
	302	return ret;
	303	}
	304
	305	/* remove a label */
3de23255 BT	306	static int __ip6addrlbl_del(struct net *net,
3de23255 BT	307	const struct in6_addr *prefix, int prefixlen,
40fee36e	308	int ifindex)
2a8cc6c8 YH	309	{
2a8cc6c8 YH	310	struct ip6addrlbl_entry *p = NULL;
b67bfe0d	311	struct hlist_node *n;
2a8cc6c8 YH	312	int ret = -ESRCH;
2a8cc6c8 YH	313
5b095d98	314	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
0c6ce78a	315	__func__, prefix, prefixlen, ifindex);
2a8cc6c8	316
b67bfe0d	317	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
2a8cc6c8	318	if (p->prefixlen == prefixlen &&
3de23255	319	net_eq(ip6addrlbl_net(p), net) &&
2a8cc6c8 YH	320	p->ifindex == ifindex &&
	321	ipv6_addr_equal(&p->prefix, prefix)) {
	322	hlist_del_rcu(&p->list);
	323	ip6addrlbl_put(p);
2a8cc6c8 YH	324	ret = 0;
	325	break;
	326	}
	327	}
	328	return ret;
	329	}
	330
3de23255 BT	331	static int ip6addrlbl_del(struct net *net,
3de23255 BT	332	const struct in6_addr *prefix, int prefixlen,
40fee36e	333	int ifindex)
2a8cc6c8 YH	334	{
	335	struct in6_addr prefix_buf;
	336	int ret;
	337
5b095d98	338	ADDRLABEL(KERN_DEBUG "%s(prefix=%pI6, prefixlen=%d, ifindex=%d)\n",
0c6ce78a	339	__func__, prefix, prefixlen, ifindex);
2a8cc6c8 YH	340
	341	ipv6_addr_prefix(&prefix_buf, prefix, prefixlen);
	342	spin_lock(&ip6addrlbl_table.lock);
3de23255	343	ret = __ip6addrlbl_del(net, &prefix_buf, prefixlen, ifindex);
2a8cc6c8 YH	344	spin_unlock(&ip6addrlbl_table.lock);
	345	return ret;
	346	}
	347
	348	/* add default label */
3de23255	349	static int __net_init ip6addrlbl_net_init(struct net *net)
2a8cc6c8 YH	350	{
	351	int err = 0;
	352	int i;
	353
f3213831	354	ADDRLABEL(KERN_DEBUG "%s\n", __func__);
2a8cc6c8 YH	355
2a8cc6c8 YH	356	for (i = 0; i < ARRAY_SIZE(ip6addrlbl_init_table); i++) {
3de23255 BT	357	int ret = ip6addrlbl_add(net,
3de23255 BT	358	ip6addrlbl_init_table[i].prefix,
2a8cc6c8 YH	359	ip6addrlbl_init_table[i].prefixlen,
	360	0,
	361	ip6addrlbl_init_table[i].label, 0);
	362	/* XXX: should we free all rules when we catch an error? */
	363	if (ret && (!err \|\| err != -ENOMEM))
	364	err = ret;
	365	}
	366	return err;
	367	}
	368
3de23255 BT	369	static void __net_exit ip6addrlbl_net_exit(struct net *net)
	370	{
	371	struct ip6addrlbl_entry *p = NULL;
b67bfe0d	372	struct hlist_node *n;
3de23255 BT	373
	374	/* Remove all labels belonging to the exiting net */
	375	spin_lock(&ip6addrlbl_table.lock);
b67bfe0d	376	hlist_for_each_entry_safe(p, n, &ip6addrlbl_table.head, list) {
3de23255 BT	377	if (net_eq(ip6addrlbl_net(p), net)) {
	378	hlist_del_rcu(&p->list);
	379	ip6addrlbl_put(p);
	380	}
	381	}
	382	spin_unlock(&ip6addrlbl_table.lock);
	383	}
	384
	385	static struct pernet_operations ipv6_addr_label_ops = {
	386	.init = ip6addrlbl_net_init,
	387	.exit = ip6addrlbl_net_exit,
	388	};
	389
2a8cc6c8 YH	390	int __init ipv6_addr_label_init(void)
	391	{
	392	spin_lock_init(&ip6addrlbl_table.lock);
	393
3de23255	394	return register_pernet_subsys(&ipv6_addr_label_ops);
2a8cc6c8 YH	395	}
2a8cc6c8 YH	396
2cc6d2bf NH	397	void ipv6_addr_label_cleanup(void)
	398	{
	399	unregister_pernet_subsys(&ipv6_addr_label_ops);
	400	}
	401
2a8cc6c8 YH	402	static const struct nla_policy ifal_policy[IFAL_MAX+1] = {
	403	[IFAL_ADDRESS] = { .len = sizeof(struct in6_addr), },
	404	[IFAL_LABEL] = { .len = sizeof(u32), },
	405	};
	406
661d2967	407	static int ip6addrlbl_newdel(struct sk_buff skb, struct nlmsghdr nlh)
2a8cc6c8	408	{
3b1e0a65	409	struct net *net = sock_net(skb->sk);
2a8cc6c8 YH	410	struct ifaddrlblmsg *ifal;
	411	struct nlattr *tb[IFAL_MAX+1];
	412	struct in6_addr *pfx;
	413	u32 label;
	414	int err = 0;
	415
	416	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
	417	if (err < 0)
	418	return err;
	419
	420	ifal = nlmsg_data(nlh);
	421
	422	if (ifal->ifal_family != AF_INET6 \|\|
	423	ifal->ifal_prefixlen > 128)
	424	return -EINVAL;
	425
2a8cc6c8 YH	426	if (!tb[IFAL_ADDRESS])
2a8cc6c8 YH	427	return -EINVAL;
2a8cc6c8	428	pfx = nla_data(tb[IFAL_ADDRESS]);
2a8cc6c8 YH	429
	430	if (!tb[IFAL_LABEL])
	431	return -EINVAL;
	432	label = nla_get_u32(tb[IFAL_LABEL]);
	433	if (label == IPV6_ADDR_LABEL_DEFAULT)
	434	return -EINVAL;
	435
1c8669e0	436	switch (nlh->nlmsg_type) {
2a8cc6c8	437	case RTM_NEWADDRLABEL:
0771275b FW	438	if (ifal->ifal_index &&
	439	!__dev_get_by_index(net, ifal->ifal_index))
	440	return -EINVAL;
	441
3de23255	442	err = ip6addrlbl_add(net, pfx, ifal->ifal_prefixlen,
2a8cc6c8 YH	443	ifal->ifal_index, label,
	444	nlh->nlmsg_flags & NLM_F_REPLACE);
	445	break;
	446	case RTM_DELADDRLABEL:
3de23255	447	err = ip6addrlbl_del(net, pfx, ifal->ifal_prefixlen,
2a8cc6c8 YH	448	ifal->ifal_index);
	449	break;
	450	default:
	451	err = -EOPNOTSUPP;
	452	}
	453	return err;
	454	}
	455
a50feda5 ED	456	static void ip6addrlbl_putmsg(struct nlmsghdr *nlh,
a50feda5 ED	457	int prefixlen, int ifindex, u32 lseq)
2a8cc6c8 YH	458	{
	459	struct ifaddrlblmsg *ifal = nlmsg_data(nlh);
	460	ifal->ifal_family = AF_INET6;
	461	ifal->ifal_prefixlen = prefixlen;
	462	ifal->ifal_flags = 0;
	463	ifal->ifal_index = ifindex;
	464	ifal->ifal_seq = lseq;
	465	};
	466
	467	static int ip6addrlbl_fill(struct sk_buff *skb,
	468	struct ip6addrlbl_entry *p,
	469	u32 lseq,
15e47304	470	u32 portid, u32 seq, int event,
2a8cc6c8 YH	471	unsigned int flags)
2a8cc6c8 YH	472	{
15e47304	473	struct nlmsghdr *nlh = nlmsg_put(skb, portid, seq, event,
2a8cc6c8 YH	474	sizeof(struct ifaddrlblmsg), flags);
	475	if (!nlh)
	476	return -EMSGSIZE;
	477
	478	ip6addrlbl_putmsg(nlh, p->prefixlen, p->ifindex, lseq);
	479
930345ea	480	if (nla_put_in6_addr(skb, IFAL_ADDRESS, &p->prefix) < 0 \|\|
2a8cc6c8 YH	481	nla_put_u32(skb, IFAL_LABEL, p->label) < 0) {
	482	nlmsg_cancel(skb, nlh);
	483	return -EMSGSIZE;
	484	}
	485
053c095a JB	486	nlmsg_end(skb, nlh);
053c095a JB	487	return 0;
2a8cc6c8 YH	488	}
	489
	490	static int ip6addrlbl_dump(struct sk_buff skb, struct netlink_callback cb)
	491	{
3b1e0a65	492	struct net *net = sock_net(skb->sk);
2a8cc6c8	493	struct ip6addrlbl_entry *p;
2a8cc6c8 YH	494	int idx = 0, s_idx = cb->args[0];
	495	int err;
	496
	497	rcu_read_lock();
b67bfe0d	498	hlist_for_each_entry_rcu(p, &ip6addrlbl_table.head, list) {
3de23255 BT	499	if (idx >= s_idx &&
3de23255 BT	500	net_eq(ip6addrlbl_net(p), net)) {
cb6e926e WY	501	err = ip6addrlbl_fill(skb, p,
	502	ip6addrlbl_table.seq,
	503	NETLINK_CB(cb->skb).portid,
	504	cb->nlh->nlmsg_seq,
	505	RTM_NEWADDRLABEL,
	506	NLM_F_MULTI);
053c095a	507	if (err < 0)
2a8cc6c8 YH	508	break;
	509	}
	510	idx++;
	511	}
	512	rcu_read_unlock();
	513	cb->args[0] = idx;
	514	return skb->len;
	515	}
	516
	517	static inline int ip6addrlbl_msgsize(void)
	518	{
a02cec21	519	return NLMSG_ALIGN(sizeof(struct ifaddrlblmsg))
2a8cc6c8	520	+ nla_total_size(16) /* IFAL_ADDRESS */
a02cec21	521	+ nla_total_size(4); /* IFAL_LABEL */
2a8cc6c8 YH	522	}
2a8cc6c8 YH	523
22b285d6	524	static int ip6addrlbl_get(struct sk_buff in_skb, struct nlmsghdr nlh)
2a8cc6c8	525	{
3b1e0a65	526	struct net *net = sock_net(in_skb->sk);
2a8cc6c8 YH	527	struct ifaddrlblmsg *ifal;
	528	struct nlattr *tb[IFAL_MAX+1];
	529	struct in6_addr *addr;
	530	u32 lseq;
	531	int err = 0;
	532	struct ip6addrlbl_entry *p;
	533	struct sk_buff *skb;
	534
	535	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
	536	if (err < 0)
	537	return err;
	538
	539	ifal = nlmsg_data(nlh);
	540
	541	if (ifal->ifal_family != AF_INET6 \|\|
	542	ifal->ifal_prefixlen != 128)
	543	return -EINVAL;
	544
	545	if (ifal->ifal_index &&
3de23255	546	!__dev_get_by_index(net, ifal->ifal_index))
2a8cc6c8 YH	547	return -EINVAL;
	548
	549	if (!tb[IFAL_ADDRESS])
	550	return -EINVAL;
2a8cc6c8	551	addr = nla_data(tb[IFAL_ADDRESS]);
2a8cc6c8 YH	552
2a8cc6c8 YH	553	rcu_read_lock();
3de23255	554	p = __ipv6_addr_label(net, addr, ipv6_addr_type(addr), ifal->ifal_index);
e459dfee	555	if (p && !ip6addrlbl_hold(p))
2a8cc6c8 YH	556	p = NULL;
	557	lseq = ip6addrlbl_table.seq;
	558	rcu_read_unlock();
	559
	560	if (!p) {
	561	err = -ESRCH;
	562	goto out;
	563	}
	564
cb6e926e WY	565	skb = nlmsg_new(ip6addrlbl_msgsize(), GFP_KERNEL);
cb6e926e WY	566	if (!skb) {
2a8cc6c8 YH	567	ip6addrlbl_put(p);
	568	return -ENOBUFS;
	569	}
	570
	571	err = ip6addrlbl_fill(skb, p, lseq,
15e47304	572	NETLINK_CB(in_skb).portid, nlh->nlmsg_seq,
2a8cc6c8 YH	573	RTM_NEWADDRLABEL, 0);
	574
	575	ip6addrlbl_put(p);
	576
	577	if (err < 0) {
	578	WARN_ON(err == -EMSGSIZE);
	579	kfree_skb(skb);
	580	goto out;
	581	}
	582
15e47304	583	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).portid);
2a8cc6c8 YH	584	out:
	585	return err;
	586	}
	587
	588	void __init ipv6_addr_label_rtnl_register(void)
	589	{
c7ac8679 GR	590	__rtnl_register(PF_INET6, RTM_NEWADDRLABEL, ip6addrlbl_newdel,
	591	NULL, NULL);
	592	__rtnl_register(PF_INET6, RTM_DELADDRLABEL, ip6addrlbl_newdel,
	593	NULL, NULL);
	594	__rtnl_register(PF_INET6, RTM_GETADDRLABEL, ip6addrlbl_get,
	595	ip6addrlbl_dump, NULL);
2a8cc6c8 YH	596	}
2a8cc6c8 YH	597