Hi,

This is an old issue that I've been discussing with Patrick. Actually, 
he gave me the clue and I think that he's right. Anyway, consider this 
as something experimental. I would like to discuss this issue to know 
more opinions.

This patch reduces locking in ip_queue based on the fact that the user 
process which uses libipq only runs on a CPU at a time, so there's no 
need to lock ip_queue configuration vars (peer_pid, copy_range,...) 
because the process itself serializes all the operations.

On the other hand, we keep locking to enqueue and dequeue because more 
ip_queue.c could run on more than one processor.

regards,
Pablo