Re: [PATCH net-next v2 2/5] net: gro: only merge packets with incrementing or fixed outer ids

[Jakub Kicinski <kuba@kernel.org>]

On Wed, 20 Aug 2025 14:27:12 +0200 Richard Gobert wrote:
> Jakub Kicinski wrote:
> > On Tue, 19 Aug 2025 10:46:01 -0400 Willem de Bruijn wrote:  
> >> It's a bit unclear what the meaning of inner and outer are in the
> >> unencapsulated (i.e., normal) case. In my intuition outer only exists
> >> if encapsulated, but it seems you reason the other way around: inner
> >> is absent unless encapsulated.   
> > 
> > +1, whether the header in unencapsulted packet is inner or outer
> > is always a source of unnecessary confusion. I would have also
> > preferred your suggestion on v1 to use _ENCAP in the name.  
> 
> Yeah, I guess that was the source of confusion. IMO, it makes more sense that
> INNER is absent unless encapsulated since that seems to be the convention in
> the rest of the network stack. (e.g. inner_network_header for both skb and
> napi_gro_cb is only relevant for encapsulation)
> 
> I could rename the OUTER variant to simply SKB_GSO_TCP_FIXEDID so that it's
> clearer that it's the default (resembling network_header). WDYT?

Yup! That'd match the skb fields so SGTM!