On Sat, Jun 02 2018, Eric Dumazet wrote:

> On 06/02/2018 01:03 AM, Herbert Xu wrote:
>  
>> Yes the concept looks good to me.  But I would like to hear from
>> Eric/Dave as to whether this would be acceptable for existing
>> network hash tables such as the ones in inet.
>
>
> What about lockdep support ?

bitlocks don't have native lockdep support.
I would be fairly easy to add lockdep support to
rht_{lock,unlock,unlocked} if you think it is worthwhile.
It could only really help if a hash-function or cmp-function took a
lock, but it is not a great cost so we may as well just do it.
I'll try to have a patch in the next day or so.

Thanks,
NeilBrown