On 9/21/17 11:20 AM, Roopa Prabhu wrote: > this patch seems fine...but ideally I would have assumed > netdev_upper_dev_unlink which > is eventually called in both paths would generate the RTN_NEWLINK > IFF_MASTER in response > to the NETDEV_CHANGEUPPER notifier. If we add it there now, it might > need some more fixes > to not generate redundant notifications for the netlink case.
Agreed and it is another pandora's box
