[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <ZJqEsoFLPBqkgs6c@localhost.localdomain>
Date: Tue, 27 Jun 2023 08:41:54 +0200
From: Michal Swiatkowski <michal.swiatkowski@...ux.intel.com>
To: Jakub Kicinski <kuba@...nel.org>
Cc: Tony Nguyen <anthony.l.nguyen@...el.com>, davem@...emloft.net,
pabeni@...hat.com, edumazet@...gle.com, netdev@...r.kernel.org,
Wojciech Drewek <wojciech.drewek@...el.com>, jiri@...nulli.us,
ivecera@...hat.com, simon.horman@...igine.com,
Sujai Buvaneswaran <sujai.buvaneswaran@...el.com>,
Vlad Buslov <vladbu@...dia.com>
Subject: Re: [PATCH net-next 06/12] ice: Implement basic eswitch bridge setup
On Mon, Jun 26, 2023 at 10:35:42AM -0700, Jakub Kicinski wrote:
> On Mon, 26 Jun 2023 16:26:15 +0200 Michal Swiatkowski wrote:
> > We found out that adding VF and corresponding port representor to the
> > bridge cause loop in the bridge. Packets are looping through the bridge.
> > I know that it isn't valid configuration, howevere, it can happen and
> > after that the server is quite unstable.
> >
> > Does mellanox validate the port for this scenario? Or we should assume
> > that user will add port wisely? I was looking at your code, but didn't
> > find that. You are using NETDEV_PRECHANGEUPPER, do you think we should
> > validate if user is trying to add VF when his PR is currently added?
>
> Can you try to plug two ends of a veth into a bridge and see if the
> same thing happens? My instinct is that this is a classic bridge
> problem and the answer is STP.
Yeah, the same happens with veth connected to the bridge with both ends.
Turning STP on on the bridge fix the problem. Thanks for help :)
Michal
Powered by blists - more mailing lists