[PATCH net-2.6.26 0/8][GRE]: Make IPGRE tunnel work in net namespaces.

[PATCH net-2.6.26 0/8][GRE]: Make IPGRE tunnel work in net namespaces.

[Pavel Emelyanov]

This one is mostly the same as the ipip one.

Hm... these sets are more likely ~8 patches each, not
~7 as I promised earlier...

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

[PATCH net-2.6.26 1/8][GRE]: Introduce empty ipgre_net structure and net init/exit ops.

[Pavel Emelyanov]

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |   49 +++++++++++++++++++++++++++++++++++++++++++++++++
 1 files changed, 49 insertions(+), 0 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 50972b3..d729ca8 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -39,6 +39,8 @@
 #include <net/dsfield.h>
 #include <net/inet_ecn.h>
 #include <net/xfrm.h>
+#include <net/net_namespace.h>
+#include <net/netns/generic.h>
 
 #ifdef CONFIG_IPV6
 #include <net/ipv6.h>
@@ -122,6 +124,10 @@ static void ipgre_tunnel_setup(struct net_device *dev);
 
 static int ipgre_fb_tunnel_init(struct net_device *dev);
 
+static int ipgre_net_id;
+struct ipgre_net {
+};
+
 static struct net_device *ipgre_fb_tunnel_dev;
 
 /* Tunnel hash table */
@@ -1275,6 +1281,40 @@ static struct net_protocol ipgre_protocol = {
 	.err_handler	=	ipgre_err,
 };
 
+static int ipgre_init_net(struct net *net)
+{
+	int err;
+	struct ipgre_net *ign;
+
+	err = -ENOMEM;
+	ign = kmalloc(sizeof(struct ipgre_net), GFP_KERNEL);
+	if (ign == NULL)
+		goto err_alloc;
+
+	err = net_assign_generic(net, ipgre_net_id, ign);
+	if (err < 0)
+		goto err_assign;
+
+	return 0;
+
+err_assign:
+	kfree(ign);
+err_alloc:
+	return err;
+}
+
+static void ipgre_exit_net(struct net *net)
+{
+	struct ipgre_net *ign;
+
+	ign = net_generic(net, ipgre_net_id);
+	kfree(ign);
+}
+
+static struct pernet_operations ipgre_net_ops = {
+	.init = ipgre_init_net,
+	.exit = ipgre_exit_net,
+};
 
 /*
  *	And now the modules code and kernel interface.
@@ -1302,6 +1342,10 @@ static int __init ipgre_init(void)
 
 	if ((err = register_netdev(ipgre_fb_tunnel_dev)))
 		goto err2;
+
+	err = register_pernet_gen_device(&ipgre_net_id, &ipgre_net_ops);
+	if (err < 0)
+		goto err3;
 out:
 	return err;
 err2:
@@ -1309,6 +1353,9 @@ err2:
 err1:
 	inet_del_protocol(&ipgre_protocol, IPPROTO_GRE);
 	goto out;
+err3:
+	unregister_netdevice(ipgre_fb_tunnel_dev);
+	goto err1;
 }
 
 static void __exit ipgre_destroy_tunnels(void)
@@ -1333,6 +1380,8 @@ static void __exit ipgre_fini(void)
 	rtnl_lock();
 	ipgre_destroy_tunnels();
 	rtnl_unlock();
+
+	unregister_pernet_gen_device(ipgre_net_id, &ipgre_net_ops);
 }
 
 module_init(ipgre_init);
-- 
1.5.3.4

[PATCH net-2.6.26 2/8][GRE]: Add net/gre_net argument to some functions.

[Pavel Emelyanov]

The fallback device and hashes are to become per-net, but many
code doesn't have anything to get the struct net pointer from.

So pass the proper net there with an extra argument.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |   51 ++++++++++++++++++++++++++++++++-------------------
 1 files changed, 32 insertions(+), 19 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index d729ca8..6209ab3 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -162,7 +162,8 @@ static DEFINE_RWLOCK(ipgre_lock);
 
 /* Given src, dst and key, find appropriate for input tunnel. */
 
-static struct ip_tunnel * ipgre_tunnel_lookup(__be32 remote, __be32 local, __be32 key)
+static struct ip_tunnel * ipgre_tunnel_lookup(struct net *net,
+		__be32 remote, __be32 local, __be32 key)
 {
 	unsigned h0 = HASH(remote);
 	unsigned h1 = HASH(key);
@@ -198,7 +199,8 @@ static struct ip_tunnel * ipgre_tunnel_lookup(__be32 remote, __be32 local, __be3
 	return NULL;
 }
 
-static struct ip_tunnel **__ipgre_bucket(struct ip_tunnel_parm *parms)
+static struct ip_tunnel **__ipgre_bucket(struct ipgre_net *ign,
+		struct ip_tunnel_parm *parms)
 {
 	__be32 remote = parms->iph.daddr;
 	__be32 local = parms->iph.saddr;
@@ -216,14 +218,15 @@ static struct ip_tunnel **__ipgre_bucket(struct ip_tunnel_parm *parms)
 	return &tunnels[prio][h];
 }
 
-static inline struct ip_tunnel **ipgre_bucket(struct ip_tunnel *t)
+static inline struct ip_tunnel **ipgre_bucket(struct ipgre_net *ign,
+		struct ip_tunnel *t)
 {
-	return __ipgre_bucket(&t->parms);
+	return __ipgre_bucket(ign, &t->parms);
 }
 
-static void ipgre_tunnel_link(struct ip_tunnel *t)
+static void ipgre_tunnel_link(struct ipgre_net *ign, struct ip_tunnel *t)
 {
-	struct ip_tunnel **tp = ipgre_bucket(t);
+	struct ip_tunnel **tp = ipgre_bucket(ign, t);
 
 	t->next = *tp;
 	write_lock_bh(&ipgre_lock);
@@ -231,11 +234,11 @@ static void ipgre_tunnel_link(struct ip_tunnel *t)
 	write_unlock_bh(&ipgre_lock);
 }
 
-static void ipgre_tunnel_unlink(struct ip_tunnel *t)
+static void ipgre_tunnel_unlink(struct ipgre_net *ign, struct ip_tunnel *t)
 {
 	struct ip_tunnel **tp;
 
-	for (tp = ipgre_bucket(t); *tp; tp = &(*tp)->next) {
+	for (tp = ipgre_bucket(ign, t); *tp; tp = &(*tp)->next) {
 		if (t == *tp) {
 			write_lock_bh(&ipgre_lock);
 			*tp = t->next;
@@ -245,7 +248,8 @@ static void ipgre_tunnel_unlink(struct ip_tunnel *t)
 	}
 }
 
-static struct ip_tunnel * ipgre_tunnel_locate(struct ip_tunnel_parm *parms, int create)
+static struct ip_tunnel * ipgre_tunnel_locate(struct net *net,
+		struct ip_tunnel_parm *parms, int create)
 {
 	__be32 remote = parms->iph.daddr;
 	__be32 local = parms->iph.saddr;
@@ -253,8 +257,9 @@ static struct ip_tunnel * ipgre_tunnel_locate(struct ip_tunnel_parm *parms, int
 	struct ip_tunnel *t, **tp, *nt;
 	struct net_device *dev;
 	char name[IFNAMSIZ];
+	struct ipgre_net *ign = net_generic(net, ipgre_net_id);
 
-	for (tp = __ipgre_bucket(parms); (t = *tp) != NULL; tp = &t->next) {
+	for (tp = __ipgre_bucket(ign, parms); (t = *tp) != NULL; tp = &t->next) {
 		if (local == t->parms.iph.saddr && remote == t->parms.iph.daddr) {
 			if (key == t->parms.i_key)
 				return t;
@@ -285,7 +290,7 @@ static struct ip_tunnel * ipgre_tunnel_locate(struct ip_tunnel_parm *parms, int
 		goto failed_free;
 
 	dev_hold(dev);
-	ipgre_tunnel_link(nt);
+	ipgre_tunnel_link(ign, nt);
 	return nt;
 
 failed_free:
@@ -295,7 +300,10 @@ failed_free:
 
 static void ipgre_tunnel_uninit(struct net_device *dev)
 {
-	ipgre_tunnel_unlink(netdev_priv(dev));
+	struct net *net = dev_net(dev);
+	struct ipgre_net *ign = net_generic(net, ipgre_net_id);
+
+	ipgre_tunnel_unlink(ign, netdev_priv(dev));
 	dev_put(dev);
 }
 
@@ -369,7 +377,9 @@ static void ipgre_err(struct sk_buff *skb, u32 info)
 	}
 
 	read_lock(&ipgre_lock);
-	t = ipgre_tunnel_lookup(iph->daddr, iph->saddr, (flags&GRE_KEY) ? *(((__be32*)p) + (grehlen>>2) - 1) : 0);
+	t = ipgre_tunnel_lookup(&init_net, iph->daddr, iph->saddr,
+			(flags&GRE_KEY) ?
+			*(((__be32*)p) + (grehlen>>2) - 1) : 0);
 	if (t == NULL || t->parms.iph.daddr == 0 ||
 	    ipv4_is_multicast(t->parms.iph.daddr))
 		goto out;
@@ -602,7 +612,8 @@ static int ipgre_rcv(struct sk_buff *skb)
 	}
 
 	read_lock(&ipgre_lock);
-	if ((tunnel = ipgre_tunnel_lookup(iph->saddr, iph->daddr, key)) != NULL) {
+	if ((tunnel = ipgre_tunnel_lookup(&init_net,
+					iph->saddr, iph->daddr, key)) != NULL) {
 		secpath_reset(skb);
 
 		skb->protocol = *(__be16*)(h + 2);
@@ -960,6 +971,8 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 	int err = 0;
 	struct ip_tunnel_parm p;
 	struct ip_tunnel *t;
+	struct net *net = dev_net(dev);
+	struct ipgre_net *ign = net_generic(net, ipgre_net_id);
 
 	switch (cmd) {
 	case SIOCGETTUNNEL:
@@ -969,7 +982,7 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 				err = -EFAULT;
 				break;
 			}
-			t = ipgre_tunnel_locate(&p, 0);
+			t = ipgre_tunnel_locate(net, &p, 0);
 		}
 		if (t == NULL)
 			t = netdev_priv(dev);
@@ -1001,7 +1014,7 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 		if (!(p.o_flags&GRE_KEY))
 			p.o_key = 0;
 
-		t = ipgre_tunnel_locate(&p, cmd == SIOCADDTUNNEL);
+		t = ipgre_tunnel_locate(net, &p, cmd == SIOCADDTUNNEL);
 
 		if (dev != ipgre_fb_tunnel_dev && cmd == SIOCCHGTUNNEL) {
 			if (t != NULL) {
@@ -1023,14 +1036,14 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 					err = -EINVAL;
 					break;
 				}
-				ipgre_tunnel_unlink(t);
+				ipgre_tunnel_unlink(ign, t);
 				t->parms.iph.saddr = p.iph.saddr;
 				t->parms.iph.daddr = p.iph.daddr;
 				t->parms.i_key = p.i_key;
 				t->parms.o_key = p.o_key;
 				memcpy(dev->dev_addr, &p.iph.saddr, 4);
 				memcpy(dev->broadcast, &p.iph.daddr, 4);
-				ipgre_tunnel_link(t);
+				ipgre_tunnel_link(ign, t);
 				netdev_state_change(dev);
 			}
 		}
@@ -1063,7 +1076,7 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 			if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
 				goto done;
 			err = -ENOENT;
-			if ((t = ipgre_tunnel_locate(&p, 0)) == NULL)
+			if ((t = ipgre_tunnel_locate(net, &p, 0)) == NULL)
 				goto done;
 			err = -EPERM;
 			if (t == netdev_priv(ipgre_fb_tunnel_dev))
-- 
1.5.3.4

[PATCH net-2.6.26 3/8][GRE]: Use proper net in hash-lookup functions.

[Pavel Emelyanov]

This is the part#2 of the patch #2 - get the proper net for
these functions. This change in a separate patch in order not
to get lost in a large previous patch.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |    4 ++--
 1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 6209ab3..1a17c5b 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -377,7 +377,7 @@ static void ipgre_err(struct sk_buff *skb, u32 info)
 	}
 
 	read_lock(&ipgre_lock);
-	t = ipgre_tunnel_lookup(&init_net, iph->daddr, iph->saddr,
+	t = ipgre_tunnel_lookup(dev_net(skb->dev), iph->daddr, iph->saddr,
 			(flags&GRE_KEY) ?
 			*(((__be32*)p) + (grehlen>>2) - 1) : 0);
 	if (t == NULL || t->parms.iph.daddr == 0 ||
@@ -612,7 +612,7 @@ static int ipgre_rcv(struct sk_buff *skb)
 	}
 
 	read_lock(&ipgre_lock);
-	if ((tunnel = ipgre_tunnel_lookup(&init_net,
+	if ((tunnel = ipgre_tunnel_lookup(dev_net(skb->dev),
 					iph->saddr, iph->daddr, key)) != NULL) {
 		secpath_reset(skb);
 
-- 
1.5.3.4

[PATCH net-2.6.26 4/8][GRE]: Make the fallback tunnel device per-net.

[Pavel Emelyanov]

Everything is prepared for this change now. Create on in
init callback, use it over the code and destroy on net exit.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |   63 +++++++++++++++++++++++++++--------------------------
 1 files changed, 32 insertions(+), 31 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 1a17c5b..a8ec090 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -126,10 +126,9 @@ static int ipgre_fb_tunnel_init(struct net_device *dev);
 
 static int ipgre_net_id;
 struct ipgre_net {
+	struct net_device *fb_tunnel_dev;
 };
 
-static struct net_device *ipgre_fb_tunnel_dev;
-
 /* Tunnel hash table */
 
 /*
@@ -168,6 +167,7 @@ static struct ip_tunnel * ipgre_tunnel_lookup(struct net *net,
 	unsigned h0 = HASH(remote);
 	unsigned h1 = HASH(key);
 	struct ip_tunnel *t;
+	struct ipgre_net *ign = net_generic(net, ipgre_net_id);
 
 	for (t = tunnels_r_l[h0^h1]; t; t = t->next) {
 		if (local == t->parms.iph.saddr && remote == t->parms.iph.daddr) {
@@ -194,8 +194,8 @@ static struct ip_tunnel * ipgre_tunnel_lookup(struct net *net,
 			return t;
 	}
 
-	if (ipgre_fb_tunnel_dev->flags&IFF_UP)
-		return netdev_priv(ipgre_fb_tunnel_dev);
+	if (ign->fb_tunnel_dev->flags&IFF_UP)
+		return netdev_priv(ign->fb_tunnel_dev);
 	return NULL;
 }
 
@@ -977,7 +977,7 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 	switch (cmd) {
 	case SIOCGETTUNNEL:
 		t = NULL;
-		if (dev == ipgre_fb_tunnel_dev) {
+		if (dev == ign->fb_tunnel_dev) {
 			if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p))) {
 				err = -EFAULT;
 				break;
@@ -1016,7 +1016,7 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 
 		t = ipgre_tunnel_locate(net, &p, cmd == SIOCADDTUNNEL);
 
-		if (dev != ipgre_fb_tunnel_dev && cmd == SIOCCHGTUNNEL) {
+		if (dev != ign->fb_tunnel_dev && cmd == SIOCCHGTUNNEL) {
 			if (t != NULL) {
 				if (t->dev != dev) {
 					err = -EEXIST;
@@ -1071,7 +1071,7 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 		if (!capable(CAP_NET_ADMIN))
 			goto done;
 
-		if (dev == ipgre_fb_tunnel_dev) {
+		if (dev == ign->fb_tunnel_dev) {
 			err = -EFAULT;
 			if (copy_from_user(&p, ifr->ifr_ifru.ifru_data, sizeof(p)))
 				goto done;
@@ -1079,7 +1079,7 @@ ipgre_tunnel_ioctl (struct net_device *dev, struct ifreq *ifr, int cmd)
 			if ((t = ipgre_tunnel_locate(net, &p, 0)) == NULL)
 				goto done;
 			err = -EPERM;
-			if (t == netdev_priv(ipgre_fb_tunnel_dev))
+			if (t == netdev_priv(ign->fb_tunnel_dev))
 				goto done;
 			dev = t->dev;
 		}
@@ -1270,7 +1270,7 @@ static int ipgre_tunnel_init(struct net_device *dev)
 	return 0;
 }
 
-static int __init ipgre_fb_tunnel_init(struct net_device *dev)
+static int ipgre_fb_tunnel_init(struct net_device *dev)
 {
 	struct ip_tunnel *tunnel = netdev_priv(dev);
 	struct iphdr *iph = &tunnel->parms.iph;
@@ -1308,8 +1308,25 @@ static int ipgre_init_net(struct net *net)
 	if (err < 0)
 		goto err_assign;
 
+	ign->fb_tunnel_dev = alloc_netdev(sizeof(struct ip_tunnel), "gre0",
+					   ipgre_tunnel_setup);
+	if (!ign->fb_tunnel_dev) {
+		err = -ENOMEM;
+		goto err_alloc_dev;
+	}
+
+	ign->fb_tunnel_dev->init = ipgre_fb_tunnel_init;
+	dev_net_set(ign->fb_tunnel_dev, net);
+
+	if ((err = register_netdev(ign->fb_tunnel_dev)))
+		goto err_reg_dev;
+
 	return 0;
 
+err_reg_dev:
+	free_netdev(ign->fb_tunnel_dev);
+err_alloc_dev:
+	/* nothing */
 err_assign:
 	kfree(ign);
 err_alloc:
@@ -1321,6 +1338,10 @@ static void ipgre_exit_net(struct net *net)
 	struct ipgre_net *ign;
 
 	ign = net_generic(net, ipgre_net_id);
+	rtnl_lock();
+	if (net != &init_net)
+		unregister_netdevice(ign->fb_tunnel_dev);
+	rtnl_unlock();
 	kfree(ign);
 }
 
@@ -1344,31 +1365,11 @@ static int __init ipgre_init(void)
 		return -EAGAIN;
 	}
 
-	ipgre_fb_tunnel_dev = alloc_netdev(sizeof(struct ip_tunnel), "gre0",
-					   ipgre_tunnel_setup);
-	if (!ipgre_fb_tunnel_dev) {
-		err = -ENOMEM;
-		goto err1;
-	}
-
-	ipgre_fb_tunnel_dev->init = ipgre_fb_tunnel_init;
-
-	if ((err = register_netdev(ipgre_fb_tunnel_dev)))
-		goto err2;
-
 	err = register_pernet_gen_device(&ipgre_net_id, &ipgre_net_ops);
 	if (err < 0)
-		goto err3;
-out:
+		inet_del_protocol(&ipgre_protocol, IPPROTO_GRE);
+
 	return err;
-err2:
-	free_netdev(ipgre_fb_tunnel_dev);
-err1:
-	inet_del_protocol(&ipgre_protocol, IPPROTO_GRE);
-	goto out;
-err3:
-	unregister_netdevice(ipgre_fb_tunnel_dev);
-	goto err1;
 }
 
 static void __exit ipgre_destroy_tunnels(void)
-- 
1.5.3.4

[PATCH net-2.6.26 5/8][GRE]: Make tunnels hashes per-net.

[Pavel Emelyanov]

Very similar to what was done for the IPIP code.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |   65 +++++++++++++++++++++++++---------------------------
 1 files changed, 31 insertions(+), 34 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index a8ec090..74d4c51 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -124,8 +124,12 @@ static void ipgre_tunnel_setup(struct net_device *dev);
 
 static int ipgre_fb_tunnel_init(struct net_device *dev);
 
+#define HASH_SIZE  16
+
 static int ipgre_net_id;
 struct ipgre_net {
+	struct ip_tunnel *tunnels[4][HASH_SIZE];
+
 	struct net_device *fb_tunnel_dev;
 };
 
@@ -147,15 +151,12 @@ struct ipgre_net {
    will match fallback tunnel.
  */
 
-#define HASH_SIZE  16
 #define HASH(addr) (((__force u32)addr^((__force u32)addr>>4))&0xF)
 
-static struct ip_tunnel *tunnels[4][HASH_SIZE];
-
-#define tunnels_r_l	(tunnels[3])
-#define tunnels_r	(tunnels[2])
-#define tunnels_l	(tunnels[1])
-#define tunnels_wc	(tunnels[0])
+#define tunnels_r_l	tunnels[3]
+#define tunnels_r	tunnels[2]
+#define tunnels_l	tunnels[1]
+#define tunnels_wc	tunnels[0]
 
 static DEFINE_RWLOCK(ipgre_lock);
 
@@ -169,19 +170,19 @@ static struct ip_tunnel * ipgre_tunnel_lookup(struct net *net,
 	struct ip_tunnel *t;
 	struct ipgre_net *ign = net_generic(net, ipgre_net_id);
 
-	for (t = tunnels_r_l[h0^h1]; t; t = t->next) {
+	for (t = ign->tunnels_r_l[h0^h1]; t; t = t->next) {
 		if (local == t->parms.iph.saddr && remote == t->parms.iph.daddr) {
 			if (t->parms.i_key == key && (t->dev->flags&IFF_UP))
 				return t;
 		}
 	}
-	for (t = tunnels_r[h0^h1]; t; t = t->next) {
+	for (t = ign->tunnels_r[h0^h1]; t; t = t->next) {
 		if (remote == t->parms.iph.daddr) {
 			if (t->parms.i_key == key && (t->dev->flags&IFF_UP))
 				return t;
 		}
 	}
-	for (t = tunnels_l[h1]; t; t = t->next) {
+	for (t = ign->tunnels_l[h1]; t; t = t->next) {
 		if (local == t->parms.iph.saddr ||
 		     (local == t->parms.iph.daddr &&
 		      ipv4_is_multicast(local))) {
@@ -189,7 +190,7 @@ static struct ip_tunnel * ipgre_tunnel_lookup(struct net *net,
 				return t;
 		}
 	}
-	for (t = tunnels_wc[h1]; t; t = t->next) {
+	for (t = ign->tunnels_wc[h1]; t; t = t->next) {
 		if (t->parms.i_key == key && (t->dev->flags&IFF_UP))
 			return t;
 	}
@@ -215,7 +216,7 @@ static struct ip_tunnel **__ipgre_bucket(struct ipgre_net *ign,
 		h ^= HASH(remote);
 	}
 
-	return &tunnels[prio][h];
+	return &ign->tunnels[prio][h];
 }
 
 static inline struct ip_tunnel **ipgre_bucket(struct ipgre_net *ign,
@@ -1274,6 +1275,7 @@ static int ipgre_fb_tunnel_init(struct net_device *dev)
 {
 	struct ip_tunnel *tunnel = netdev_priv(dev);
 	struct iphdr *iph = &tunnel->parms.iph;
+	struct ipgre_net *ign = net_generic(dev_net(dev), ipgre_net_id);
 
 	tunnel->dev = dev;
 	strcpy(tunnel->parms.name, dev->name);
@@ -1284,7 +1286,7 @@ static int ipgre_fb_tunnel_init(struct net_device *dev)
 	tunnel->hlen		= sizeof(struct iphdr) + 4;
 
 	dev_hold(dev);
-	tunnels_wc[0]		= tunnel;
+	ign->tunnels_wc[0]	= tunnel;
 	return 0;
 }
 
@@ -1294,13 +1296,27 @@ static struct net_protocol ipgre_protocol = {
 	.err_handler	=	ipgre_err,
 };
 
+static void ipgre_destroy_tunnels(struct ipgre_net *ign)
+{
+	int prio;
+
+	for (prio = 0; prio < 4; prio++) {
+		int h;
+		for (h = 0; h < HASH_SIZE; h++) {
+			struct ip_tunnel *t;
+			while ((t = ign->tunnels[prio][h]) != NULL)
+				unregister_netdevice(t->dev);
+		}
+	}
+}
+
 static int ipgre_init_net(struct net *net)
 {
 	int err;
 	struct ipgre_net *ign;
 
 	err = -ENOMEM;
-	ign = kmalloc(sizeof(struct ipgre_net), GFP_KERNEL);
+	ign = kzalloc(sizeof(struct ipgre_net), GFP_KERNEL);
 	if (ign == NULL)
 		goto err_alloc;
 
@@ -1339,8 +1355,7 @@ static void ipgre_exit_net(struct net *net)
 
 	ign = net_generic(net, ipgre_net_id);
 	rtnl_lock();
-	if (net != &init_net)
-		unregister_netdevice(ign->fb_tunnel_dev);
+	ipgre_destroy_tunnels(ign);
 	rtnl_unlock();
 	kfree(ign);
 }
@@ -1372,29 +1387,11 @@ static int __init ipgre_init(void)
 	return err;
 }
 
-static void __exit ipgre_destroy_tunnels(void)
-{
-	int prio;
-
-	for (prio = 0; prio < 4; prio++) {
-		int h;
-		for (h = 0; h < HASH_SIZE; h++) {
-			struct ip_tunnel *t;
-			while ((t = tunnels[prio][h]) != NULL)
-				unregister_netdevice(t->dev);
-		}
-	}
-}
-
 static void __exit ipgre_fini(void)
 {
 	if (inet_del_protocol(&ipgre_protocol, IPPROTO_GRE) < 0)
 		printk(KERN_INFO "ipgre close: can't remove protocol\n");
 
-	rtnl_lock();
-	ipgre_destroy_tunnels();
-	rtnl_unlock();
-
 	unregister_pernet_gen_device(ipgre_net_id, &ipgre_net_ops);
 }
 
-- 
1.5.3.4

[PATCH net-2.6.26 6/8][GRE]: Use proper net in routing calls.

[Pavel Emelyanov]

As for the IPIP tunnel, there are some ip_route_output_key()
calls in there that require a proper net so give one to them.

And a proper net for the __get_dev_by_index hanging around.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |   12 ++++++------
 1 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 74d4c51..aa97381 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -493,7 +493,7 @@ out:
 	fl.fl4_dst = eiph->saddr;
 	fl.fl4_tos = RT_TOS(eiph->tos);
 	fl.proto = IPPROTO_GRE;
-	if (ip_route_output_key(&init_net, &rt, &fl)) {
+	if (ip_route_output_key(dev_net(skb->dev), &rt, &fl)) {
 		kfree_skb(skb2);
 		return;
 	}
@@ -506,7 +506,7 @@ out:
 		fl.fl4_dst = eiph->daddr;
 		fl.fl4_src = eiph->saddr;
 		fl.fl4_tos = eiph->tos;
-		if (ip_route_output_key(&init_net, &rt, &fl) ||
+		if (ip_route_output_key(dev_net(skb->dev), &rt, &fl) ||
 		    rt->u.dst.dev->type != ARPHRD_IPGRE) {
 			ip_rt_put(rt);
 			kfree_skb(skb2);
@@ -762,7 +762,7 @@ static int ipgre_tunnel_xmit(struct sk_buff *skb, struct net_device *dev)
 						.saddr = tiph->saddr,
 						.tos = RT_TOS(tos) } },
 				    .proto = IPPROTO_GRE };
-		if (ip_route_output_key(&init_net, &rt, &fl)) {
+		if (ip_route_output_key(dev_net(dev), &rt, &fl)) {
 			tunnel->stat.tx_carrier_errors++;
 			goto tx_error;
 		}
@@ -935,7 +935,7 @@ static void ipgre_tunnel_bind_dev(struct net_device *dev)
 						.tos = RT_TOS(iph->tos) } },
 				    .proto = IPPROTO_GRE };
 		struct rtable *rt;
-		if (!ip_route_output_key(&init_net, &rt, &fl)) {
+		if (!ip_route_output_key(dev_net(dev), &rt, &fl)) {
 			tdev = rt->u.dst.dev;
 			ip_rt_put(rt);
 		}
@@ -943,7 +943,7 @@ static void ipgre_tunnel_bind_dev(struct net_device *dev)
 	}
 
 	if (!tdev && tunnel->parms.link)
-		tdev = __dev_get_by_index(&init_net, tunnel->parms.link);
+		tdev = __dev_get_by_index(dev_net(dev), tunnel->parms.link);
 
 	if (tdev) {
 		hlen = tdev->hard_header_len;
@@ -1193,7 +1193,7 @@ static int ipgre_open(struct net_device *dev)
 						.tos = RT_TOS(t->parms.iph.tos) } },
 				    .proto = IPPROTO_GRE };
 		struct rtable *rt;
-		if (ip_route_output_key(&init_net, &rt, &fl))
+		if (ip_route_output_key(dev_net(dev), &rt, &fl))
 			return -EADDRNOTAVAIL;
 		dev = rt->u.dst.dev;
 		ip_rt_put(rt);
-- 
1.5.3.4

[PATCH net-2.6.26 7/8][GRE]: Allow to create IPGRE tunnels in net namespaces.

[Pavel Emelyanov]

I.e. set the proper net and mark as NETNS_LOCAL.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |    3 +++
 1 files changed, 3 insertions(+), 0 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index aa97381..7ff5262 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -278,6 +278,8 @@ static struct ip_tunnel * ipgre_tunnel_locate(struct net *net,
 	if (!dev)
 	  return NULL;
 
+	dev_net_set(dev, net);
+
 	if (strchr(name, '%')) {
 		if (dev_alloc_name(dev, name) < 0)
 			goto failed_free;
@@ -1236,6 +1238,7 @@ static void ipgre_tunnel_setup(struct net_device *dev)
 	dev->flags		= IFF_NOARP;
 	dev->iflink		= 0;
 	dev->addr_len		= 4;
+	dev->features		|= NETIF_F_NETNS_LOCAL;
 }
 
 static int ipgre_tunnel_init(struct net_device *dev)
-- 
1.5.3.4

[PATCH net-2.6.26 8/8][GRE]: Allow for IPPROTO_GRE protocol in namespaces.

[Pavel Emelyanov]

This one was also disabled by default for sanity.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

---
 net/ipv4/ip_gre.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 7ff5262..2ada033 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -1297,6 +1297,7 @@ static int ipgre_fb_tunnel_init(struct net_device *dev)
 static struct net_protocol ipgre_protocol = {
 	.handler	=	ipgre_rcv,
 	.err_handler	=	ipgre_err,
+	.netns_ok	=	1,
 };
 
 static void ipgre_destroy_tunnels(struct ipgre_net *ign)
-- 
1.5.3.4

Re: [PATCH net-2.6.26 8/8][GRE]: Allow for IPPROTO_GRE protocol in namespaces.

[David Miller]

From: Pavel Emelyanov <xemul@openvz.org>
Date: Tue, 15 Apr 2008 17:03:04 +0400

> This one was also disabled by default for sanity.
> 
> Signed-off-by: Pavel Emelyanov <xemul@openvz.org>

Applied and pushed to net-2.6.26

Keep up the great work Pavel.