On Wed, 21 Sep 2016 10:57:34 +0300
Tariq Toukan <ttoukan.li...@gmail.com> wrote:
> On 20/09/2016 6:40 PM, Alexei Starovoitov wrote:
> > On Tue, Sep 20, 2016 at 03:53:10PM +0300, Tariq Toukan wrote:
> >>>>> + case XDP_ABORTED:
> >>>> It is not clearly defined, but I believe XDP_ABORTED should also result
> >>>> in a warning (calling bpf_warn_invalid_xdp_action(act)).
> >> I'll add this.
> > Certainly NOT.
> > XDP_ABORTED is an exception case when program does divide by zero.
> > It should NOT do bpf_warn. It must drop the packet.
> > We discussed it several months ago.
> > See mlx4/en_rx.c for canonical implementation.
> This is also the example given here:
Nice, the documentation is already useful :-)))
I actually adjusted the code example after this feedback. Which I also
wrote in this email thread about updating the documentation. I already
pointed to this commit, but here is a more specific link:
> I prefer to align with the documentation (and with current mlx4 driver
> code), which means keeping the XDP_ABORTED w/o a warning.
> Anyway, I don't think this should block the coming V2. If you decide to
> change documentation/specification, we will simply adjust our drivers
This is not blocking your V2, please resend so we can all start working
on a common code base for mlx5 (I'm currently running mlx5 with my own
one-page-per-packet patch... and my page_pool on-top)
As discussed in this thread, the outcome will likely be a new interface
for Troubleshooting and Monitoring, as documented and summarized here
so we don't forget:
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
Author of http://www.iptv-analyzer.org