David S. Miller wrote:

>Why don't you bypass all the cleanup diffs and just do the functionality
>change instead?  When you mix whitespace and coding style cleanups
>with real changes, it puts your real changes at risk if we think your
>cleanups are ugly or bogus since you've created a patch dependency.
>  
>

Ok, sorry for the nuisance. I'll try to avoid such things in future. 
Thanks davem.

Back to what I wanted to introduce...

The following patches introduce a new function that check that the 
netlink messages received are clean. Actually some people performs such 
checkings, some don't and others simply do some half of them. I think 
that we must unify the behaviour of a netlink socket when it has to 
reply to malformed messages.

01process_skb.patch:
      introduces the new function called netlink_process_skb that does 
the sanity checkings for received messages.
02xfrm.patch:
      the modification to make xfrm_user use such new function.
03rtnetlink.patch:
      same thing for rtnetlink.

The 02 and 03 patches are straight forward conversions. If you are ok 
with this, I could post more patches to make other netlink sockets use 
this new function.

--
Pablo