A race in register_netdevice()

[Kalle Valo <kvalo@adurom.com>]

Hi,

there seems to be a race in register_netdevice(), which is reported here:

https://bugzilla.kernel.org/show_bug.cgi?id=15606

This is visible at least with flimflam and ath6kl. Basically what
happens is this:

Apr 29 00:21:35 roska flimflamd[2598]: src/udev.c:add_net_device() 
Apr 29 00:21:35 roska flimflamd[2598]: connman_inet_ifname: SIOCGIFNAME(index
4): No such device
Apr 29 00:21:45 roska flimflamd[2598]: src/rtnl.c:rtnl_message() buf
0xbfefda3c len 1004
Apr 29 00:21:45 roska flimflamd[2598]: src/rtnl.c:rtnl_message()
NEWLINK len 1004 type 16 flags 0x0000 seq 0

(ignore the 10 s delay, I added that to reproduce the issue easily)

There are two ways to fix this, first is to move kobject registration
after the call to list_netdevice():

--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -5425,11 +5425,6 @@ int register_netdevice(struct net_device *dev)
        if (ret)
                goto err_uninit;
 
-       ret = netdev_register_kobject(dev);
-       if (ret)
-               goto err_uninit;
-       dev->reg_state = NETREG_REGISTERED;
-
        netdev_update_features(dev);
 
        /*
@@ -5443,6 +5438,11 @@ int register_netdevice(struct net_device *dev)
        dev_hold(dev);
        list_netdevice(dev);
 
+       ret = netdev_register_kobject(dev);
+       if (ret)
+               goto err_uninit;
+       dev->reg_state = NETREG_REGISTERED;
+
        /* Notify protocols, that a new device appeared. */
        ret = call_netdevice_notifiers(NETDEV_REGISTER, dev);
        ret = notifier_to_errno(ret);

Other option, noticed by Jouni Malinen, is to take rtnl for
SIOCGIFNAME. For some reason it's currently unprotected:

--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4917,8 +4917,12 @@ int dev_ioctl(struct net *net, unsigned int
cmd, void __user *arg)
          rtnl_unlock();
                return ret;
                }
-               if (cmd == SIOCGIFNAME)
-                  return dev_ifname(net, (struct ifreq __user *)arg);
+                  if (cmd == SIOCGIFNAME) {
+                     rtnl_lock();
+                       ret = dev_ifname(net, (struct ifreq __user
-                  *)arg);
+                       rtnl_unlock();
+                               return ret;
+                               }
 
        if (copy_from_user(&ifr, arg, sizeof(struct ifreq)))
           return -EFAULT;

I have confirmed that both of these patches fix the issue. Now I'm
wondering which one is the best way forward. Or is there a better way
to fix this?

-- 
Kalle Valo