Re: [PATCH net 3/5] fm10k: Implement ndo_gso_check()

["Vick, Matthew" <matthew.vick@intel.com>]

On 11/7/14, 2:35 PM, "Joe Stringer" <joestringer@nicira.com> wrote:

>Sure, I think fm10k_tx_encap_offload() is a good place for the header
>length 
>check. Separately, my question above was regarding the idea of a helper
>for 
>SKB_GSO_{GRE,UDP_TUNNEL}. The only reason it might be useful for the
>fm10k 
>driver is because all encap is checked in the fm10k_tx_encap_offload()
>function. 
>Other drivers don't seem to handle different tunnels together like this
>though, 
>so I'm inclined to stick with the below for now.
>                  
>      
>static bool fm10k_gso_check(struct sk_buff *skb, struct net_device *dev)
>      
>{                 
>      
>        return (!(skb_shinfo(skb)->gso_type &
>      
>                  (SKB_GSO_UDP_TUNNEL | SKB_GSO_GRE)) ||
>      
>                fm10k_tx_encap_offload(skb));
>      
>}
>
>Cheers,
>Joe

I agree. I think that makes the most sense.

Cheers,
Matthew