Re: Question about an assignment in handle_ing()

[jamal <hadi@cyberus.ca>]

On Tue, 2010-05-25 at 20:26 +1000, Herbert Xu wrote:

> In that case you should be checking whether the skb is cloned.

That is the general rule used (and what i specify to do in the docs)..
but i recall there were issues if the packet path emanated from ingress
and included multiple netdevices (earlier ex with mirror applies). There
may have been bugs then, eg I could not assume that it i had any ptype
at all that the ptype will clone the packet. Does tcpdump guarantee
skb->clone being set? I will try to test some scenarios when i am back
+settled. 

> After all, tcpdump might have simply filtered the packet out.

True - but i think thats an acceptable compromise.

> BTW, this is the case whenever you run a DHCP client/server.  So
> on most boxes your optimisation will never kick in as is.  

"Most" for people running serious firewalls or routers is not to run
DHCP servers;-> They may client, but thats a short-lived session.

> Also
> the skb may still be cloned even if there is no AF_PACKET listener.
> In that case your optimisation may be incorrect.

Did you mean that as long as there are other ptypes - which may or 
not be doing af packet? 

cheers,
jamal