On Tue, 1 Sep 2009 18:55:34 +0300
Octavian Purdila [off-list ref] wrote:

On Tuesday 01 September 2009 09:50:17 Eric Dumazet wrote:

quoted

Stephen Hemminger a écrit :

quoted

Looking at the neighbour table, it should be possible to get
rid of the two reader/writer locks.  The hash table lock is pretty
amenable to RCU, but the dynamic resizing makes it non-trivial.
Thinking of using a combination of RCU and sequence counts so that the
reader would just rescan if resize was in progress.

I am not sure neigh_tbl_lock rwlock should be changed, I did not
see any contention on it.

Speaking about neighbour optimizations, here is a RFC patch which makes the 
tables double linked, for constant time deletion. It has given us a significant 
performance improvement - in less then usual setups though, with lots of 
neighbours.

Would something like this be acceptable for upstream? (pardon the p4 diff dump 
:) - but I think it will give a rough idea, if acceptable will clean it up and 
properly submit it)

BTW, would switching to list_head be better?

Use hlist for the neighbour table. It has the right properties
and makes future RCU easier.