On Tue, Oct 18, 2022 at 06:56:09PM +0200, Hans J. Schultz wrote:
> Add a 'blackhole' fdb flag, ensuring that no forwarding from any port
> to a destination MAC that has a FDB entry with this flag on will occur.
> The packets will thus be dropped.
>
> When the blackhole fdb flag is set, the 'local' flag will also be enabled
> as blackhole entries are not associated with any port.
It reads as if the kernel will enable the 'local' flag automatically,
which is not true anymore. The bridge driver enforces that
'NUD_PERMANENT' is set if 'NTF_EXT_BLACKHOLE' is specified.
>
> Thus the command will be alike to:
> bridge fdb add MAC dev br0 local blackhole
>
> Signed-off-by: Hans J. Schultz <netdev@xxxxxxxxxxxxxxxxxxxx>
Looks OK to me. See one comment below.
Reviewed-by: Ido Schimmel <idosch@xxxxxxxxxx>
[...]
> @@ -1140,7 +1148,7 @@ static int __br_fdb_add(struct ndmsg *ndm, struct net_bridge *br,
> err = br_fdb_external_learn_add(br, p, addr, vid, true);
> } else {
> spin_lock_bh(&br->hash_lock);
> - err = fdb_add_entry(br, p, addr, ndm, nlh_flags, vid, nfea_tb);
> + err = fdb_add_entry(br, p, addr, ndm, nlh_flags, vid, ext_flags, nfea_tb);
I believe the preference is to wrap to 80 columns when possible.
> spin_unlock_bh(&br->hash_lock);
> }
