Hi Jiri,
Apologies for possibly newbie questions here, but:
> +bool swdev_dev_check(const struct net_device *dev)
> +{
> + return dev->priv_flags & IFF_SWITCH;
> +}
> +EXPORT_SYMBOL(swdev_dev_check);
[...]
> + BUG_ON(!swdev_dev_check(dev));
[...]
> +int __swdev_register(struct net_device *dev)
> +{
> + if (dev->priv_flags & IFF_SWITCH) {
> + netdev_err(dev, "Device is already registered as a switch
> device\n");
> + return -EBUSY;
> + }
[...]
1. What's the benefit of requiring and policing the IFF_SWITCH flag, and
requiring a switch netdev to call swdev_register, as opposed to just checking
whether the netdev provides the relevant ops?
> +int swdev_register(struct net_device *dev)
> +{
> + int err;
> +
> + rtnl_lock();
> + err = __swdev_register(dev);
> + rtnl_unlock();
> + return err;
> +}
> +EXPORT_SYMBOL(swdev_register);
2. Why grab and release the rtnl lock, given that there appears to be nothing
rtnl-related in between?
Regards,
Neil
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
