Re: [PATCH v3 net-next 08/15] ipv4: Replace &arp_tbl with arp_table(net).

On Tue, Aug 11, 2026 at 8:42 AM David Ahern <[email protected]> wrote:
>
> On 8/10/26 8:23 PM, Kuniyuki Iwashima wrote:
> > We will allocate per-netns neigh_table in net->neigh_tables[].
> >
> > Let's replace &arp_tbl with arp_table(net) and remove extern
> > definition for arp_tbl.
> >
> > Three notes:
> >
> >   1. mlx5e_rep_netevent_event() and nfp_tun_neigh_event_handler()
> >      have code assuming neigh_table other than &arp_tbl and
> >      &nb_tbl, and the part is removed as it will be false once
> >      per-netns table is allocated.
> >
> >   2. prestera and rocker uses init_net because they set
> >      dev->netns_immutable to true.
> >
> >   3. mlx5e_tc_update_neigh_used_value() dereferences nhe->neigh_dev
> >      in trace_mlx5e_tc_update_neigh_used_value().
> >
> > Signed-off-by: Kuniyuki Iwashima <[email protected]>
> > ---
> > v2: Add note about mlx5e_tc_update_neigh_used_value()
>
>
> Seems like this is not a bisectable patch set. References for arp_tbl
> and nd_tbl need to be fixed up before switching to per network namespace
> tables.

It's still bisectable, and this comment is just to make human review
(against AI)
easier because Sashiko keeps asking "Is it safe to dereference dev here ?
previously only the pointer value was used for comparison", which is false.
Sashiko just misses that trace_mlx5e_tc_update_neigh_used_value() does.