On Thu, Feb 13, 2025 at 2:20 PM Kuniyuki Iwashima [off-list ref] wrote:

From: Xiao Liang <redacted>
Date: Mon, 10 Feb 2025 21:29:56 +0800

quoted

When link_net is set, use it as link netns instead of dev_net(). This
prepares for rtnetlink core to create device in target netns directly,
in which case the two namespaces may be different.

Convert common ip_tunnel_newlink() to accept an extra link netns
argument. Don't overwrite ip_tunnel.net in ip_tunnel_init().

Why... ?  see a comment below.


[...]

quoted

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 1fe9b13d351c..26d15f907551 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c

@@ -1413,7 +1413,8 @@ static int ipgre_newlink(struct net_device *dev,
      err = ipgre_netlink_parms(dev, data, tb, &p, &fwmark);
      if (err < 0)
              return err;
-     return ip_tunnel_newlink(dev, tb, &p, fwmark);
+     return ip_tunnel_newlink(params->link_net ? : dev_net(dev), dev, tb, &p,

This is duplicate at all call sites, let's move it into
ip_tunnel_newlink() by passing params.

Existing tunnels use `params->link_net ? : dev_net(dev)` for
backward compatibility. But I think we can leave the choice of netns
to future tunnel drivers because rtnl_newlink_link_net() is preferred
in general.

quoted

+                              fwmark);
 }

 static int erspan_newlink(struct net_device *dev,

diff --git a/net/ipv4/ip_tunnel.c b/net/ipv4/ip_tunnel.c
index 09b73acf037a..618a50d5c0c2 100644
--- a/net/ipv4/ip_tunnel.c
+++ b/net/ipv4/ip_tunnel.c

@@ -1213,11 +1213,11 @@ void ip_tunnel_delete_nets(struct list_head *net_list, unsigned int id,
 }
 EXPORT_SYMBOL_GPL(ip_tunnel_delete_nets);

-int ip_tunnel_newlink(struct net_device *dev, struct nlattr *tb[],
-                   struct ip_tunnel_parm_kern *p, __u32 fwmark)
+int ip_tunnel_newlink(struct net *net, struct net_device *dev,
+                   struct nlattr *tb[], struct ip_tunnel_parm_kern *p,
+                   __u32 fwmark)
 {
      struct ip_tunnel *nt;
-     struct net *net = dev_net(dev);
      struct ip_tunnel_net *itn;
      int mtu;
      int err;

@@ -1326,7 +1326,9 @@ int ip_tunnel_init(struct net_device *dev)
      }

      tunnel->dev = dev;
-     tunnel->net = dev_net(dev);
+     if (!tunnel->net)
+             tunnel->net = dev_net(dev);

Isn't tunnel->net always non-NULL ?

ip_tunnel_newlink
-> netdev_priv(dev)->net = net
-> register_netdevice(dev)
  -> dev->netdev_ops->ndo_init(dev)
    -> ip_tunnel_init(dev)
      -> netdev_priv(dev)->net = dev_net(dev)

Didn't find a path that can leave tunnel->net to NULL either.
I think we can remove this.

Thanks.