On 10/26/16 2:31 PM, Mahesh Bandewar (महेश बंडेवार) wrote: > The hook insertion in sk_alloc() may not solve all control-path checks as not > much can be done (probably apart for changing sk_bound_dev_if) during > allocation but hooks in bind(), listen(), setsockopt() etc. (not a complete > list) will be needed. Also pushing this change (changing sk_bound_dev_if) > later (than sk_alloc()) you might avoid the case dumazet@ has mentioned.
For v2 I moved the running of the filter to the end of inet{6}_create.