On Thu, Jun 28, 2012 at 07:00:54AM +0200, Eric Dumazet wrote:
> [1] : LLTX drivers case 
>    since ndo_start_xmit() can be run concurrently by many cpus, safely
> updating an "unsigned long" requires additional hassle :
> 
>    1) Use of a spinlock to protect the update.
>    2) Use atomic_long_t instead of "unsigned long"
>    3) Use percpu data
> 
> 3) is overkill for devices with light traffic, because it consumes lot
> of RAM on machines with 2048 possible cpus, _and_ the reader must fold
> the data of all possible values.

Thanks Eric.

So am I right in thinking that a v3 patch which uses atomic_long_t for
the statistics would be the correct way forwards?

-- 
Tom Parkin
Katalix Systems Ltd
http://www.katalix.com
Catalysts for your Embedded Linux software development