[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20231218083345.GA6288@kernel.org>
Date: Mon, 18 Dec 2023 08:33:45 +0000
From: Simon Horman <horms@...nel.org>
To: Ioana Ciornei <ioana.ciornei@....com>
Cc: davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org,
pabeni@...hat.com, netdev@...r.kernel.org
Subject: Re: [PATCH net-next v2 6/8] dpaa2-switch: reorganize the
[pre]changeupper events
On Fri, Dec 15, 2023 at 02:08:51PM +0200, Ioana Ciornei wrote:
> On Fri, Dec 15, 2023 at 11:49:39AM +0000, Simon Horman wrote:
> > On Wed, Dec 13, 2023 at 02:14:09PM +0200, Ioana Ciornei wrote:
...
> > > if (!dpaa2_switch_port_dev_check(netdev))
> > > - return NOTIFY_DONE;
> > > + return 0;
> > >
> > > extack = netdev_notifier_info_to_extack(&info->info);
> > > -
> > > - switch (event) {
> > > - case NETDEV_PRECHANGEUPPER:
> > > - upper_dev = info->upper_dev;
> > > - if (!netif_is_bridge_master(upper_dev))
> > > - break;
> > > -
> > > + upper_dev = info->upper_dev;
> > > + if (netif_is_bridge_master(upper_dev)) {
> > > err = dpaa2_switch_prechangeupper_sanity_checks(netdev,
> > > upper_dev,
> > > extack);
> > > if (err)
> > > - goto out;
> > > + return err;
> > >
> > > if (!info->linking)
> > > dpaa2_switch_port_pre_bridge_leave(netdev);
> > > + }
> >
> > FWIIW, I think that a more idomatic flow would be to return if
> > netif_is_bridge_master() is false. Something like this (completely untested!):
> >
> > if (!netif_is_bridge_master(upper_dev))
> > return 0;
> >
> > err = dpaa2_switch_prechangeupper_sanity_checks(netdev, upper_dev,
> > extack);
> > if (err)
> > return err;
> >
> > if (!info->linking)
> > dpaa2_switch_port_pre_bridge_leave(netdev);
> >
>
> It looks better but I don't think this it's easily extensible.
>
> I am planning to add support for LAG offloading which would mean that I
> would have to revert to the initial flow and extend it to something
> like:
>
> if (netif_is_bridge_master(upper_dev)) {
> ...
> } else if (netif_is_lag_master(upper_dev)) {
> ...
> }
>
> The same thing applies to the dpaa2_switch_port_changeupper() function
> below.
Understood. If this is going somewhere then don't let me derail it.
,,,
Powered by blists - more mailing lists