Re: Question about an assignment in handle_ing()

[jamal <hadi@cyberus.ca>]

On Thu, 2010-06-03 at 18:01 +1000, Herbert Xu wrote:
> On Sun, May 30, 2010 at 09:29:10AM -0400, jamal wrote:

> > The packet path is:
> > -->eth0-->tcpdump eth0-->pedit-->mirror to dummy0-->tcpdump dummy0
> 
> Well this doesn't guarantee a cloned packet at all.  Once af_packet
> receives the packet it'll wake up any listeners like tcpdump, if
> tcpdump gets to it before pedit runs then the packet won't be
> cloned anymore.

I may be misreading, but:
This is the point i have been trying to make, Herbert;-> There is no
_guarantee_ that the first tcpdump will see the packet that came out of
eth0 instead of seeing the packet that came out the pedit part of the
pipeline. I need to see the correct packet. I know with my check
this is guaranteed.

> Anyway, I don't see why actions are special.  Everybody else lives
> by the rule that cloned skbs are not writeable.  

Yes, if skb_cloned() is true but it is not as i said in my earlier
email.

> So if this was
> indeed buggy as you say it would have shown up a long time ago.

Things may have been buggy - I dont know; you just validated to me that
it _may_ happen. I will be more than happy to remove it if i can get a
guarantee.
So how do we fix this?
Does af_packet need to always clone? That way i can depend on it. I have
a feeling someone will be unhappy with that. I am avoiding to clone
every packet on my part because afaik this problem doesnt exist if i
dont use tcpdump/af_packet...

> Case in point, we had a bug in certain NIC drivers where they
> modified cloned skbs for TSO.  This quickly showed up as bogus
> packets in tcpdump and we fixed it.

I think this is different.

cheers,
jamal