> On Mon, Apr 07, 2025 at 04:18:30PM +0200, Lorenzo Bianconi wrote:
> > Introduce l2_flows rhashtable in airoha_ppe struct in order to
> > store L2 flows committed by upper layers of the kernel. This is a
> > preliminary patch in order to offload L2 traffic rules.
> > 
> > Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
> 
> The patch logic and coding style looks OK to me.
> Just one question inline.
> 
> Thanks,
> Michal
> 
> > ---
> >  drivers/net/ethernet/airoha/airoha_eth.h | 15 ++++++-
> >  drivers/net/ethernet/airoha/airoha_ppe.c | 67 +++++++++++++++++++++++++++-----
> >  2 files changed, 72 insertions(+), 10 deletions(-)
> > 
> > diff --git a/drivers/net/ethernet/airoha/airoha_eth.h b/drivers/net/ethernet/airoha/airoha_eth.h
> > index ec8908f904c61988c3dc973e187596c49af139fb..57925648155b104021c10821096ba267c9c7cef6 100644
> > --- a/drivers/net/ethernet/airoha/airoha_eth.h
> > +++ b/drivers/net/ethernet/airoha/airoha_eth.h
> > @@ -422,12 +422,23 @@ struct airoha_flow_data {
> >  	} pppoe;
> >  };
> >  
> > +enum airoha_flow_entry_type {
> > +	FLOW_TYPE_L4,
> 
> I didn't find any usage of L4 flow type in the series.
> Is that reserved for future series? Shouldn't it be added together with
> its usage then?

Hi Michal,

FLOW_TYPE_L4 is equal to 0 so it is the default value for
airoha_flow_table_entry type when not set explicitly.
It is done this way to reduce code changes.

Regards,
Lorenzo

> 
> > +	FLOW_TYPE_L2,
> > +	FLOW_TYPE_L2_SUBFLOW,
> > +};
> > +