Thread (8 messages) flat view 8 messages, 2 authors, 4d ago
COOLING4d

[PATCH v1 net-next 6/6] ip6_gre: Support per-netns device unregistration.

From: Kuniyuki Iwashima <kuniyu@google.com>
Date: 2026-09-16 23:04:03
Subsystem: networking [general], networking [ipv4/ipv6], the rest · Maintainers: "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, David Ahern, Ido Schimmel, Linus Torvalds 

ip6gre_exit_rtnl_net() iterates ip6gre devices and variants whose
link_net is dying and queues them for destruction.

The devices may reside in different netns.

Let's use unregister_netdevice_queue_net() to support per-netns
device unregistration.

Even after ip6gre_exit_rtnl_net() queues a cross-netns ip6gre
device, ip6gre_changelink_common(), ip6gre_dellink(), and
ip6gre_tunnel_siocdevprivate() could be called concurrently for
it (once RTNL is removed).  In such a case, __rtnl_net_unlock()
will perform the unregistration.

Also, ip6gre_tunnel_siocdevprivate() needs to check check_net(t->net),
otherwise it could create a new dev in dying netns after
ip6gre_exit_rtnl_net().

In the example below, we can see the fallback tunnel device
(ip6gre0) and the cross-netns device (ip6gre1) are unregistered
by different processes:

  # bpftrace -e '#include <linux/netdevice.h>
  kprobe:ip6gre_tunnel_uninit {
      $dev = (struct net_device *)arg0;
      printf("PID: %d | DEV: %s%s\n", pid, $dev->name, kstack());
  }
  kprobe:ip6gre_exit_rtnl_net {
      printf("PID: %d%s\n", pid, kstack());
  }' &
  # ip netns add ns1
  # ip netns add ns2
  # ip -n ns1 link add name ip6gre1 link-netns ns2 \
    type ip6gre local 2001:9292::1 remote 2001:9292::2
  # ip netns del ns2

  PID: 12
          ip6gre_exit_rtnl_net+5
          ops_undo_list+702
          cleanup_net+1122
  ...
  PID: 12 | DEV: ip6gre0  <------ fallback device (itn->fb_tunnel_dev).
          ip6gre_tunnel_uninit+5
          unregister_netdevice_many_notify+7137
          unregister_netdevice_many_net+1050
          __rtnl_net_unlock+37
          ops_undo_list+754
          cleanup_net+1122
  ...
  PID: 10 | DEV: ip6gre1
          ip6gre_tunnel_uninit+5
          unregister_netdevice_many_notify+7137
          unregister_netdevice_many_net+1050
          rtnl_net_work_func+136

Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
 net/ipv6/ip6_gre.c | 51 ++++++++++++++++++++++++++++++++--------------
 1 file changed, 36 insertions(+), 15 deletions(-)

diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
index ac1087330465..c851af22b9fe 100644
--- a/net/ipv6/ip6_gre.c
+++ b/net/ipv6/ip6_gre.c
@@ -80,7 +80,8 @@ static void ip6gre_tunnel_setup(struct net_device *dev);
 static void ip6gre_tunnel_link(struct ip6gre_net *ign, struct ip6_tnl *t);
 static void ip6gre_tnl_link_config(struct ip6_tnl *t, int set_mtu);
 static void ip6erspan_tnl_link_config(struct ip6_tnl *t, int set_mtu);
-static void __ip6gre_dellink(struct net_device *dev, struct list_head *head);
+static void __ip6gre_dellink(struct net *net, struct net_device *dev,
+			     struct list_head *head);
 
 /* Tunnel hash table */
 
@@ -285,6 +286,11 @@ static void ip6gre_tunnel_unlink(struct ip6gre_net *ign, struct ip6_tnl *t)
 	hlist_del_init_rcu(&t->hash_node);
 }
 
+static bool ip6gre_tunnel_unregistering(struct ip6_tnl *t)
+{
+	return hlist_unhashed(&t->hash_node);
+}
+
 static struct ip6_tnl *ip6gre_tunnel_find(struct net *net,
 					   const struct __ip6_tnl_parm *parms,
 					   int type)
@@ -1248,6 +1254,7 @@ static int ip6gre_tunnel_siocdevprivate(struct net_device *dev,
 					int cmd)
 {
 	struct ip6_tnl *t = netdev_priv(dev);
+	struct net *orig_net = dev_net(dev);
 	struct __ip6_tnl_parm p1 = {};
 	LIST_HEAD(dev_kill_list);
 	struct net *net = t->net;
@@ -1261,6 +1268,11 @@ static int ip6gre_tunnel_siocdevprivate(struct net_device *dev,
 
 	mutex_lock(&ign->tunnels_lock);
 
+	if (!check_net(net)) {
+		err = -EBUSY;
+		goto done;
+	}
+
 	switch (cmd) {
 	case SIOCGETTUNNEL:
 		if (dev == ign->fb_tunnel_dev) {
@@ -1302,7 +1314,7 @@ static int ip6gre_tunnel_siocdevprivate(struct net_device *dev,
 		t = ip6gre_tunnel_locate(net, &p1, cmd == SIOCADDTUNNEL);
 
 		if (dev != ign->fb_tunnel_dev && cmd == SIOCCHGTUNNEL) {
-			if (t) {
+			if (t && !ip6gre_tunnel_unregistering(t)) {
 				if (t->dev != dev) {
 					err = -EEXIST;
 					break;
@@ -1310,23 +1322,26 @@ static int ip6gre_tunnel_siocdevprivate(struct net_device *dev,
 			} else {
 				t = netdev_priv(dev);
 
-				ip6gre_tunnel_unlink(ign, t);
-				synchronize_net();
-				ip6gre_tnl_change(t, &p1, 1);
-				ip6gre_tunnel_link(ign, t);
-				netdev_state_change(dev);
+				if (!ip6gre_tunnel_unregistering(t)) {
+					ip6gre_tunnel_unlink(ign, t);
+					synchronize_net();
+					ip6gre_tnl_change(t, &p1, 1);
+					ip6gre_tunnel_link(ign, t);
+					netdev_state_change(dev);
+				}
 			}
 		}
 
-		if (t) {
+		if (t && !ip6gre_tunnel_unregistering(t)) {
 			err = 0;
 
 			memset(&p, 0, sizeof(p));
 			ip6gre_tnl_parm_to_user(&p, &t->parms);
 			if (copy_to_user(data, &p, sizeof(p)))
 				err = -EFAULT;
-		} else
+		} else {
 			err = (cmd == SIOCADDTUNNEL ? -ENOBUFS : -ENOENT);
+		}
 		break;
 
 	case SIOCDELTUNNEL:
@@ -1349,7 +1364,8 @@ static int ip6gre_tunnel_siocdevprivate(struct net_device *dev,
 			dev = t->dev;
 		}
 
-		__ip6gre_dellink(dev, &dev_kill_list);
+		if (!ip6gre_tunnel_unregistering(t))
+			__ip6gre_dellink(orig_net, dev, &dev_kill_list);
 		err = 0;
 		break;
 
@@ -1570,7 +1586,7 @@ static void __net_exit ip6gre_exit_rtnl_net(struct net *net,
 			struct ip6_tnl *t;
 
 			hlist_for_each_entry_safe(t, tmp, head, hash_node)
-				__ip6gre_dellink(t->dev, dev_kill_list);
+				__ip6gre_dellink(net, t->dev, dev_kill_list);
 		}
 	}
 
@@ -2049,6 +2065,9 @@ ip6gre_changelink_common(struct net_device *dev, struct nlattr *tb[],
 		t = nt;
 	}
 
+	if (ip6gre_tunnel_unregistering(t))
+		return ERR_PTR(-ENODEV);
+
 	return t;
 }
 
@@ -2085,7 +2104,8 @@ static int ip6gre_changelink(struct net_device *dev, struct nlattr *tb[],
 	return err;
 }
 
-static void __ip6gre_dellink(struct net_device *dev, struct list_head *head)
+static void __ip6gre_dellink(struct net *net, struct net_device *dev,
+			     struct list_head *head)
 {
 	struct ip6_tnl *t = netdev_priv(dev);
 	struct ip6gre_net *ign;
@@ -2098,7 +2118,7 @@ static void __ip6gre_dellink(struct net_device *dev, struct list_head *head)
 		ip6gre_tunnel_unlink_md(ign, t);
 
 	ip6gre_tunnel_unlink(ign, t);
-	unregister_netdevice_queue(dev, head);
+	unregister_netdevice_queue_net(net, dev, head);
 }
 
 static void ip6gre_dellink(struct net_device *dev, struct list_head *head)
@@ -2110,8 +2130,9 @@ static void ip6gre_dellink(struct net_device *dev, struct list_head *head)
 
 	mutex_lock(&ign->tunnels_lock);
 
-	if (dev != ign->fb_tunnel_dev)
-		__ip6gre_dellink(dev, head);
+	if (dev != ign->fb_tunnel_dev &&
+	    !ip6gre_tunnel_unregistering(t))
+		__ip6gre_dellink(dev_net(dev), dev, head);
 
 	mutex_unlock(&ign->tunnels_lock);
 }
-- 
2.55.0.1082.g2b9226bbc0-goog
Other recent patches touching these files
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help