netdev - Re: [PATCH 0/6] realtek: fix management of mdb entries

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <b6223705-e34d-4f0e-8d42-021fa467d773@lunn.ch>
Date:   Tue, 21 Mar 2023 20:02:06 +0100
From:   Andrew Lunn <andrew@...n.ch>
To:     Vladimir Oltean <olteanv@...il.com>
Cc:     Jan Hoffmann <jan@....eu>,
        Arınç ÜNAL <arinc.unal@...nc9.com>,
        netdev@...r.kernel.org, Florian Fainelli <f.fainelli@...il.com>,
        openwrt-devel@...ts.openwrt.org,
        Sander Vanheule <sander@...nheule.net>,
        erkin.bozoglu@...ont.com
Subject: Re: [PATCH 0/6] realtek: fix management of mdb entries

> > If the DSA subsystem could handle the "merging" instead and also call
> > port_mdb_add/port_mdb_del as appropriate for multicast router ports, the
> > individual drivers wouldn't have to deal with this particular issue at all.
> > 
> > > As a way to fix a bug quickly and get correct behavior, I guess there's
> > > also the option of stopping to process multicast packets in hardware,
> > > and configure the switch to always send any multicast to the CPU port
> > > only. As long as the tagger knows to leave skb->offload_fwd_mark unset,
> > > the bridge driver should know how to deal with those packets in
> > > software, and forward them only to whom is interested. But the drawback
> > > is that there is no forwarding acceleration involved. Maybe DSA should
> > > have done that from the get go for drivers which didn't care about
> > > multicast in particular, instead of ending up with this current situation
> > > which appears to be slightly chaotic.
> > 
> > Thanks,
> > Jan
> 
> Andrew, Florian, do you have any additional comments here?

I've no real experience with this either. But it sounds like this
merging should be pulled up into the bridge, if all switchdev drivers
need it.

     Andrew