Hello,
On Wed, 4 Nov 2015, Johannes Berg wrote:
> On Wed, 2015-11-04 at 22:59 +0200, Julian Anastasov wrote:
> >
> > } else if (unlikely(skb->pkt_type != PACKET_HOST)) {
> >
> > May be such check can save some cycles because
> > it is more common to see PACKET_HOST packets...
>
> I had thought about that based on your earlier comments, but then I
> didn't quite see the point. However, perhaps we could invert the check
> below to check the pkt_type first, hoping it'll have some kind of cache
> effect?
Yes, looks like it would be better if the checks
are reordered, the IN_DEV_* methods look more complex recently.
Regards
--
Julian Anastasov <ja@xxxxxx>
