On Tue, 3 Apr 2018 11:00:20 -0600 David Ahern <dsah...@gmail.com> wrote:
> On 4/3/18 10:41 AM, John Fastabend wrote: > > On 04/03/2018 08:07 AM, David Ahern wrote: > >> On 4/2/18 12:16 PM, Alexei Starovoitov wrote: > >>> On Mon, Apr 02, 2018 at 12:09:44PM -0600, David Ahern wrote: > >>>> On 4/2/18 12:03 PM, John Fastabend wrote: > >>>>> [...] > >>>> That's what I have here: > >>>> > >>>> https://github.com/dsahern/linux/commit/bab42f158c0925339f7519df7fb2cde8eac33aa8 > >>>> > >>> > >>> was wondering what's up with the delay and when are you going to > >>> submit them officially... I'm also eager to see these go upstream! :-) [...] > >> > >> 2. Jesper had concerns about xdp redirect to any netdev. e.g., How does > >> the lookup know the egress netdev supports xdp? Right now you can try > >> and the packet is dropped if it is not supported. > >> > > > > There should probably be a tracepoint there if not already. Otherwise > > I think the orchestration/loader layer should be ensuring that xdp > > support is sufficient. I don't think we need anything specific in the > > XDP/BPF code to handle unsupported devices. > > ok. I am fine with starting with that. David this concern was not related to your code. Your code need to stay generic, to be usable from both XDP and clsact. As John writes the error cases can be troubleshooted through XDP tracepoints. My request/concern was actually performance related. I wanted to add a flag to the helper, to allow for a later optimization. Today, we have pushed the responsibility, of knowing if an ifindex is XDP-xmit capable, to the BPF program(mer). This implies that the BPF-prog need extra code/lookup into a map to validate the ifindex returned from your helper. Later, we might get some XDP feature bits, and a flag could specify to only return the looked up device if XDP-xmit is enabled. Thus, simplifying the work need in the bpf-prog. -- Best regards, Jesper Dangaard Brouer MSc.CS, Principal Kernel Engineer at Red Hat LinkedIn: http://www.linkedin.com/in/brouer
