On Mon, Apr 03, 2023 at 10:33:11AM +0200, Pablo Neira Ayuso wrote:
> On Mon, Apr 03, 2023 at 10:29:47AM +0200, Florian Westphal wrote:
> > Sven Auhagen <Sven.Auhagen@xxxxxxxxxxxx> wrote:
> > > Change from v4:
> > > * use per cpu counters instead of an atomic variable
> >
> > > diff --git a/include/net/netns/flow_table.h b/include/net/netns/flow_table.h
> > > index 1c5fc657e267..1496a6af6ac4 100644
> > > --- a/include/net/netns/flow_table.h
> > > +++ b/include/net/netns/flow_table.h
> > > @@ -6,6 +6,8 @@ struct nf_flow_table_stat {
> > > unsigned int count_wq_add;
> > > unsigned int count_wq_del;
> > > unsigned int count_wq_stats;
> > > + unsigned int count_flowoffload_add;
> > > + unsigned int count_flowoffload_del;
> >
> > Do we really need new global stats for this?
> >
> > Would it be possible to instead expose the existing ht->nelems during
> > flowtable netlink dumps?
> >
> > This way we do not need any new counters.
>
> I would prefer a netlink interface for this too.
I can post a sketch code to make it easier for v6.
