On Tue, Jul 20, 2021 at 07:47:18PM +0000, Vladimir Oltean wrote:

On Tue, Jul 20, 2021 at 05:25:08PM +0300, Ido Schimmel wrote:

quoted

If you don't want to change the order, then at least make the
replay/cleanup optional and set it to 'false' for mlxsw. This should
mean that the only change in mlxsw should be adding calls to
switchdev_bridge_port_offload() / switchdev_bridge_port_unoffload() in
mlxsw_sp_bridge_port_create() / mlxsw_sp_bridge_port_destroy(),
respectively.

Is there any specific reason why you suggested me to move the
switchdev_bridge_port_offload() call from the top-level
mlxsw_sp_port_bridge_join() into mlxsw_sp_bridge_port_create()
(and similarly, from _pre_bridge_leave to _destroy)?

Even if you don't support replays right now, you'd need to move a bunch
of code around before you would get them to work. As far as I can see,
mlxsw_sp_bridge_port_create() is a bit too early and
mlxsw_sp_bridge_port_destroy() is a bit too late. The port needs to be
fairly up and running to be able to process the switchdev notifiers at
that stage.

Do you mind if I keep the hooks where they are, which is what I do for
all drivers? I don't think I am missing to handle any case.

I want to avoid introducing unnecessary changes. I will do them when
needed.

Thanks