net: do not create fallback tunnels for non-default namespaces

fallback tunnels (like tunl0, gre0, gretap0, erspan0, sit0,
ip6tnl0, ip6gre0) are automatically created when the corresponding
module is loaded.

These tunnels are also automatically created when a new network
namespace is created, at a great cost.

In many cases, netns are used for isolation purposes, and these
extra network devices are a waste of resources. We are using
thousands of netns per host, and hit the netns creation/delete
bottleneck a lot. (Many thanks to Kirill for recent work on this)

Add a new sysctl so that we can opt-out from this automatic creation.

Note that these tunnels are still created for the initial namespace,
to be the least intrusive for typical setups.

Tested:
lpk43:~# cat add_del_unshare.sh
for i in `seq 1 40`
do
 (for j in `seq 1 100` ; do  unshare -n /bin/true >/dev/null ; done) &
done
wait

lpk43:~# echo 0 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh

real	0m37.521s
user	0m0.886s
sys	7m7.084s
lpk43:~# echo 1 >/proc/sys/net/core/fb_tunnels_only_for_init_net
lpk43:~# time ./add_del_unshare.sh

real	0m4.761s
user	0m0.851s
sys	1m8.343s
lpk43:~#
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Documentation/sysctl/net.txt

@@ -270,6 +270,18 @@ optmem_max
 Maximum ancillary buffer size allowed per socket. Ancillary data is a sequence
 of struct cmsghdr structures with appended data.
 
+fb_tunnels_only_for_init_net
+----------------------------
+
+Controls if fallback tunnels (like tunl0, gre0, gretap0, erspan0,
+sit0, ip6tnl0, ip6gre0) are automatically created when a new
+network namespace is created, if corresponding tunnel is present
+in initial network namespace.
+If set to 1, these devices are not automatically created, and
+user space is responsible for creating them if needed.
+
+Default : 0  (for compatibility reasons)
+
 2. /proc/sys/net/unix - Parameters for Unix domain sockets
 -------------------------------------------------------
 

include/linux/netdevice.h

@@ -585,6 +585,13 @@ struct netdev_queue {
 #endif
 } ____cacheline_aligned_in_smp;
 
+extern int sysctl_fb_tunnels_only_for_init_net;
+
+static inline bool net_has_fallback_tunnels(const struct net *net)
+{
+	return net == &init_net || !sysctl_fb_tunnels_only_for_init_net;
+}
+
 static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
 {
 #if defined(CONFIG_XPS) && defined(CONFIG_NUMA)

include/net/ip_tunnels.h

@@ -180,8 +180,10 @@ struct tnl_ptk_info {
 
 struct ip_tunnel_net {
 	struct net_device *fb_tunnel_dev;
+	struct rtnl_link_ops *rtnl_link_ops;
 	struct hlist_head tunnels[IP_TNL_HASH_SIZE];
 	struct ip_tunnel __rcu *collect_md_tun;
+	int type;
 };
 
 static inline void ip_tunnel_key_init(struct ip_tunnel_key *key,

net/core/sysctl_net_core.c

@@ -32,6 +32,9 @@ static int max_skb_frags = MAX_SKB_FRAGS;
 
 static int net_msg_warn;	/* Unused, but still a sysctl */
 
+int sysctl_fb_tunnels_only_for_init_net __read_mostly = 0;
+EXPORT_SYMBOL(sysctl_fb_tunnels_only_for_init_net);
+
 #ifdef CONFIG_RPS
 static int rps_sock_flow_sysctl(struct ctl_table *table, int write,
 				void __user *buffer, size_t *lenp, loff_t *ppos)
@@ -513,6 +516,15 @@ static struct ctl_table net_core_table[] = {
 		.proc_handler	= proc_dointvec_minmax,
 		.extra1		= &zero,
 	},
+	{
+		.procname	= "fb_tunnels_only_for_init_net",
+		.data		= &sysctl_fb_tunnels_only_for_init_net,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &one,
+	},
 	{ }
 };
 

net/ipv4/ip_tunnel.c

@@ -347,8 +347,7 @@ static struct ip_tunnel *ip_tunnel_create(struct net *net,
 	struct net_device *dev;
 	int t_hlen;
 
-	BUG_ON(!itn->fb_tunnel_dev);
-	dev = __ip_tunnel_create(net, itn->fb_tunnel_dev->rtnl_link_ops, parms);
+	dev = __ip_tunnel_create(net, itn->rtnl_link_ops, parms);
 	if (IS_ERR(dev))
 		return ERR_CAST(dev);
 
@@ -822,7 +821,6 @@ int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd)
 	struct net *net = t->net;
 	struct ip_tunnel_net *itn = net_generic(net, t->ip_tnl_net_id);
 
-	BUG_ON(!itn->fb_tunnel_dev);
 	switch (cmd) {
 	case SIOCGETTUNNEL:
 		if (dev == itn->fb_tunnel_dev) {
@@ -847,7 +845,7 @@ int ip_tunnel_ioctl(struct net_device *dev, struct ip_tunnel_parm *p, int cmd)
 				p->o_key = 0;
 		}
 
-		t = ip_tunnel_find(itn, p, itn->fb_tunnel_dev->type);
+		t = ip_tunnel_find(itn, p, itn->type);
 
 		if (cmd == SIOCADDTUNNEL) {
 			if (!t) {
@@ -991,10 +989,15 @@ int ip_tunnel_init_net(struct net *net, unsigned int ip_tnl_net_id,
 	struct ip_tunnel_parm parms;
 	unsigned int i;
 
+	itn->rtnl_link_ops = ops;
 	for (i = 0; i < IP_TNL_HASH_SIZE; i++)
 		INIT_HLIST_HEAD(&itn->tunnels[i]);
 
-	if (!ops) {
+	if (!ops || !net_has_fallback_tunnels(net)) {
+		struct ip_tunnel_net *it_init_net;
+
+		it_init_net = net_generic(&init_net, ip_tnl_net_id);
+		itn->type = it_init_net->type;
 		itn->fb_tunnel_dev = NULL;
 		return 0;
 	}
@@ -1012,6 +1015,7 @@ int ip_tunnel_init_net(struct net *net, unsigned int ip_tnl_net_id,
 		itn->fb_tunnel_dev->features |= NETIF_F_NETNS_LOCAL;
 		itn->fb_tunnel_dev->mtu = ip_tunnel_bind_dev(itn->fb_tunnel_dev);
 		ip_tunnel_add(itn, netdev_priv(itn->fb_tunnel_dev));
+		itn->type = itn->fb_tunnel_dev->type;
 	}
 	rtnl_unlock();
 
@@ -1019,10 +1023,10 @@ int ip_tunnel_init_net(struct net *net, unsigned int ip_tnl_net_id,
 }
 EXPORT_SYMBOL_GPL(ip_tunnel_init_net);
 
-static void ip_tunnel_destroy(struct ip_tunnel_net *itn, struct list_head *head,
+static void ip_tunnel_destroy(struct net *net, struct ip_tunnel_net *itn,
+			      struct list_head *head,
 			      struct rtnl_link_ops *ops)
 {
-	struct net *net = dev_net(itn->fb_tunnel_dev);
 	struct net_device *dev, *aux;
 	int h;
 
@@ -1054,7 +1058,7 @@ void ip_tunnel_delete_nets(struct list_head *net_list, unsigned int id,
 	rtnl_lock();
 	list_for_each_entry(net, net_list, exit_list) {
 		itn = net_generic(net, id);
-		ip_tunnel_destroy(itn, &list, ops);
+		ip_tunnel_destroy(net, itn, &list, ops);
 	}
 	unregister_netdevice_many(&list);
 	rtnl_unlock();

net/ipv6/ip6_gre.c

@@ -236,7 +236,7 @@ static struct ip6_tnl *ip6gre_tunnel_lookup(struct net_device *dev,
 		return t;
 
 	dev = ign->fb_tunnel_dev;
-	if (dev->flags & IFF_UP)
+	if (dev && dev->flags & IFF_UP)
 		return netdev_priv(dev);
 
 	return NULL;
@@ -1472,6 +1472,8 @@ static int __net_init ip6gre_init_net(struct net *net)
 	struct ip6gre_net *ign = net_generic(net, ip6gre_net_id);
 	int err;
 
+	if (!net_has_fallback_tunnels(net))
+		return 0;
 	ign->fb_tunnel_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6gre0",
 					  NET_NAME_UNKNOWN,
 					  ip6gre_tunnel_setup);

net/ipv6/ip6_tunnel.c

@@ -2205,6 +2205,8 @@ static int __net_init ip6_tnl_init_net(struct net *net)
 	ip6n->tnls[0] = ip6n->tnls_wc;
 	ip6n->tnls[1] = ip6n->tnls_r_l;
 
+	if (!net_has_fallback_tunnels(net))
+		return 0;
 	err = -ENOMEM;
 	ip6n->fb_tnl_dev = alloc_netdev(sizeof(struct ip6_tnl), "ip6tnl0",
 					NET_NAME_UNKNOWN, ip6_tnl_dev_setup);

net/ipv6/sit.c

@@ -182,7 +182,7 @@ static void ipip6_tunnel_clone_6rd(struct net_device *dev, struct sit_net *sitn)
 #ifdef CONFIG_IPV6_SIT_6RD
 	struct ip_tunnel *t = netdev_priv(dev);
 
-	if (dev == sitn->fb_tunnel_dev) {
+	if (dev == sitn->fb_tunnel_dev || !sitn->fb_tunnel_dev) {
 		ipv6_addr_set(&t->ip6rd.prefix, htonl(0x20020000), 0, 0, 0);
 		t->ip6rd.relay_prefix = 0;
 		t->ip6rd.prefixlen = 16;
@@ -1835,6 +1835,9 @@ static int __net_init sit_init_net(struct net *net)
 	sitn->tunnels[2] = sitn->tunnels_r;
 	sitn->tunnels[3] = sitn->tunnels_r_l;
 
+	if (!net_has_fallback_tunnels(net))
+		return 0;
+
 	sitn->fb_tunnel_dev = alloc_netdev(sizeof(struct ip_tunnel), "sit0",
 					   NET_NAME_UNKNOWN,
 					   ipip6_tunnel_setup);