ipv6 netns: Address labels per namespace

This pacth makes IPv6 address labels per network namespace.
It keeps the global label tables, ip6addrlbl_table, but
adds a 'net' member to each ip6addrlbl_entry.
This new member is taken into account when matching labels.

Changelog
=========
* v1: Initial version
* v2:
* Minize the penalty when network namespaces are not configured:
* the 'net' member is added only if CONFIG_NET_NS is
defined. This saves space when network namespaces are not
configured.
* 'net' value is retrieved with the inlined function
ip6addrlbl_net() that always return &init_net when
CONFIG_NET_NS is not defined.
* 'net' member in ip6addrlbl_entry renamed to the less generic
'lbl_net' name (helps code search).

Signed-off-by: Benjamin Thery <[e-mail]>
Signed-off-by: YOSHIFUJI Hideaki <[e-mail]>

include/net/addrconf.h

···
  */
 extern int			ipv6_addr_label_init(void);
 extern void			ipv6_addr_label_rtnl_register(void);
-extern u32			ipv6_addr_label(const struct in6_addr *addr,
+extern u32			ipv6_addr_label(struct net *net,
+						const struct in6_addr *addr,
 						int type, int ifindex);
 
 /*

net/ipv6/addrconf.c

···
 	return 0;
 }
 
-static int ipv6_get_saddr_eval(struct ipv6_saddr_score *score,
+static int ipv6_get_saddr_eval(struct net *net,
+			       struct ipv6_saddr_score *score,
 			       struct ipv6_saddr_dst *dst,
 			       int i)
 {
···
 		break;
 	case IPV6_SADDR_RULE_LABEL:
 		/* Rule 6: Prefer matching label */
-		ret = ipv6_addr_label(&score->ifa->addr, score->addr_type,
+		ret = ipv6_addr_label(net,
+				      &score->ifa->addr, score->addr_type,
 				      score->ifa->idev->dev->ifindex) == dst->label;
 		break;
 #ifdef CONFIG_IPV6_PRIVACY
···
 	dst.addr = daddr;
 	dst.ifindex = dst_dev ? dst_dev->ifindex : 0;
 	dst.scope = __ipv6_addr_src_scope(dst_type);
-	dst.label = ipv6_addr_label(daddr, dst_type, dst.ifindex);
+	dst.label = ipv6_addr_label(net, daddr, dst_type, dst.ifindex);
 	dst.prefs = prefs;
 
 	hiscore->rule = -1;
···
 			for (i = 0; i < IPV6_SADDR_RULE_MAX; i++) {
 				int minihiscore, miniscore;
 
-				minihiscore = ipv6_get_saddr_eval(hiscore, &dst, i);
-				miniscore = ipv6_get_saddr_eval(score, &dst, i);
+				minihiscore = ipv6_get_saddr_eval(net, hiscore, &dst, i);
+				miniscore = ipv6_get_saddr_eval(net, score, &dst, i);
 
 				if (minihiscore > miniscore) {
 					if (i == IPV6_SADDR_RULE_SCOPE &&

net/ipv6/addrlabel.c

···
  */
 struct ip6addrlbl_entry
 {
+#ifdef CONFIG_NET_NS
+	struct net *lbl_net;
+#endif
 	struct in6_addr prefix;
 	int prefixlen;
 	int ifindex;
···
 	spinlock_t lock;
 	u32 seq;
 } ip6addrlbl_table;
+
+static inline
+struct net *ip6addrlbl_net(const struct ip6addrlbl_entry *lbl)
+{
+#ifdef CONFIG_NET_NS
+	return lbl->lbl_net;
+#else
+	return &init_net;
+#endif
+}
 
 /*
  * Default policy table (RFC3484 + extensions)
···
 
 #define IPV6_ADDR_LABEL_DEFAULT	0xffffffffUL
 
-static const __initdata struct ip6addrlbl_init_table
+static const __net_initdata struct ip6addrlbl_init_table
 {
 	const struct in6_addr *prefix;
 	int prefixlen;
···
 /* Object management */
 static inline void ip6addrlbl_free(struct ip6addrlbl_entry *p)
 {
+#ifdef CONFIG_NET_NS
+	release_net(p->lbl_net);
+#endif
 	kfree(p);
 }
 
···
 }
 
 /* Find label */
-static int __ip6addrlbl_match(struct ip6addrlbl_entry *p,
+static int __ip6addrlbl_match(struct net *net,
+			      struct ip6addrlbl_entry *p,
 			      const struct in6_addr *addr,
 			      int addrtype, int ifindex)
 {
+	if (!net_eq(ip6addrlbl_net(p), net))
+		return 0;
 	if (p->ifindex && p->ifindex != ifindex)
 		return 0;
 	if (p->addrtype && p->addrtype != addrtype)
···
 	return 1;
 }
 
-static struct ip6addrlbl_entry *__ipv6_addr_label(const struct in6_addr *addr,
+static struct ip6addrlbl_entry *__ipv6_addr_label(struct net *net,
+						  const struct in6_addr *addr,
 						  int type, int ifindex)
 {
 	struct hlist_node *pos;
 	struct ip6addrlbl_entry *p;
 	hlist_for_each_entry_rcu(p, pos, &ip6addrlbl_table.head, list) {
-		if (__ip6addrlbl_match(p, addr, type, ifindex))
+		if (__ip6addrlbl_match(net, p, addr, type, ifindex))
 			return p;
 	}
 	return NULL;
 }
 
-u32 ipv6_addr_label(const struct in6_addr *addr, int type, int ifindex)
+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(addr, type, ifindex);
+	p = __ipv6_addr_label(net, addr, type, ifindex);
 	label = p ? p->label : IPV6_ADDR_LABEL_DEFAULT;
 	rcu_read_unlock();
 
···
 }
 
 /* allocate one entry */
-static struct ip6addrlbl_entry *ip6addrlbl_alloc(const struct in6_addr *prefix,
+static struct ip6addrlbl_entry *ip6addrlbl_alloc(struct net *net,
+						 const struct in6_addr *prefix,
 						 int prefixlen, int ifindex,
 						 u32 label)
 {
···
 	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;
 }
···
 		hlist_for_each_entry_safe(p, pos, 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) {
···
 }
 
 /* add a label */
-static int ip6addrlbl_add(const struct in6_addr *prefix, int prefixlen,
+static int ip6addrlbl_add(struct net *net,
+			  const struct in6_addr *prefix, int prefixlen,
 			  int ifindex, u32 label, int replace)
 {
 	struct ip6addrlbl_entry *newp;
···
 			(unsigned int)label,
 			replace);
 
-	newp = ip6addrlbl_alloc(prefix, prefixlen, ifindex, label);
+	newp = ip6addrlbl_alloc(net, prefix, prefixlen, ifindex, label);
 	if (IS_ERR(newp))
 		return PTR_ERR(newp);
 	spin_lock(&ip6addrlbl_table.lock);
···
 }
 
 /* remove a label */
-static int __ip6addrlbl_del(const struct in6_addr *prefix, int prefixlen,
+static int __ip6addrlbl_del(struct net *net,
+			    const struct in6_addr *prefix, int prefixlen,
 			    int ifindex)
 {
 	struct ip6addrlbl_entry *p = NULL;
···
 
 	hlist_for_each_entry_safe(p, pos, 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);
···
 	return ret;
 }
 
-static int ip6addrlbl_del(const struct in6_addr *prefix, int prefixlen,
+static int ip6addrlbl_del(struct net *net,
+			  const struct in6_addr *prefix, int prefixlen,
 			  int ifindex)
 {
 	struct in6_addr prefix_buf;
···
 
 	ipv6_addr_prefix(&prefix_buf, prefix, prefixlen);
 	spin_lock(&ip6addrlbl_table.lock);
-	ret = __ip6addrlbl_del(&prefix_buf, prefixlen, ifindex);
+	ret = __ip6addrlbl_del(net, &prefix_buf, prefixlen, ifindex);
 	spin_unlock(&ip6addrlbl_table.lock);
 	return ret;
 }
 
 /* add default label */
-static __init int ip6addrlbl_init(void)
+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(ip6addrlbl_init_table[i].prefix,
+		int ret = ip6addrlbl_add(net,
+					 ip6addrlbl_init_table[i].prefix,
 					 ip6addrlbl_init_table[i].prefixlen,
 					 0,
 					 ip6addrlbl_init_table[i].label, 0);
···
 	return err;
 }
 
+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,
+};
+
 int __init ipv6_addr_label_init(void)
 {
 	spin_lock_init(&ip6addrlbl_table.lock);
 
-	return ip6addrlbl_init();
+	return register_pernet_subsys(&ipv6_addr_label_ops);
 }
 
 static const struct nla_policy ifal_policy[IFAL_MAX+1] = {
···
 	u32 label;
 	int err = 0;
 
-	if (net != &init_net)
-		return 0;
-
 	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
 	if (err < 0)
 		return err;
···
 		return -EINVAL;
 
 	if (ifal->ifal_index &&
-	    !__dev_get_by_index(&init_net, ifal->ifal_index))
+	    !__dev_get_by_index(net, ifal->ifal_index))
 		return -EINVAL;
 
 	if (!tb[IFAL_ADDRESS])
···
 
 	switch(nlh->nlmsg_type) {
 	case RTM_NEWADDRLABEL:
-		err = ip6addrlbl_add(pfx, ifal->ifal_prefixlen,
+		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(pfx, ifal->ifal_prefixlen,
+		err = ip6addrlbl_del(net, pfx, ifal->ifal_prefixlen,
 				     ifal->ifal_index);
 		break;
 	default:
···
 	int idx = 0, s_idx = cb->args[0];
 	int err;
 
-	if (net != &init_net)
-		return 0;
-
 	rcu_read_lock();
 	hlist_for_each_entry_rcu(p, pos, &ip6addrlbl_table.head, list) {
-		if (idx >= s_idx) {
+		if (idx >= s_idx &&
+		    net_eq(ip6addrlbl_net(p), net)) {
 			if ((err = ip6addrlbl_fill(skb, p,
 						   ip6addrlbl_table.seq,
 						   NETLINK_CB(cb->skb).pid,
···
 	struct ip6addrlbl_entry *p;
 	struct sk_buff *skb;
 
-	if (net != &init_net)
-		return 0;
-
 	err = nlmsg_parse(nlh, sizeof(*ifal), tb, IFAL_MAX, ifal_policy);
 	if (err < 0)
 		return err;
···
 		return -EINVAL;
 
 	if (ifal->ifal_index &&
-	    !__dev_get_by_index(&init_net, ifal->ifal_index))
+	    !__dev_get_by_index(net, ifal->ifal_index))
 		return -EINVAL;
 
 	if (!tb[IFAL_ADDRESS])
···
 		return -EINVAL;
 
 	rcu_read_lock();
-	p = __ipv6_addr_label(addr, ipv6_addr_type(addr), ifal->ifal_index);
+	p = __ipv6_addr_label(net, addr, ipv6_addr_type(addr), ifal->ifal_index);
 	if (p && ip6addrlbl_hold(p))
 		p = NULL;
 	lseq = ip6addrlbl_table.seq;
···
 		goto out;
 	}
 
-	err = rtnl_unicast(skb, &init_net, NETLINK_CB(in_skb).pid);
+	err = rtnl_unicast(skb, net, NETLINK_CB(in_skb).pid);
 out:
 	return err;
 }