Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   DENG Qingfang <dqfext@gmail.com>
To:     Vladimir Oltean <olteanv@gmail.com>
Cc:     Andrew Lunn <andrew@lunn.ch>,
        Vivien Didelot <vivien.didelot@gmail.com>,
        Florian Fainelli <f.fainelli@gmail.com>,
        "David S. Miller" <davem@davemloft.net>,
        Jakub Kicinski <kuba@kernel.org>,
        Russell King <linux@armlinux.org.uk>,
        "open list:NETWORKING DRIVERS" <netdev@vger.kernel.org>,
        open list <linux-kernel@vger.kernel.org>,
        Ansuel Smith <ansuelsmth@gmail.com>,
        Jonathan McDowell <noodles@earth.li>,
        Michal =?utf-8?B?Vm9rw6HEjQ==?= <vokac.m@gmail.com>,
        Christian Lamparter <chunkeey@gmail.com>,
        Nishka Dasgupta <nishkadg.linux@gmail.com>,
        John Crispin <john@phrozen.org>,
        Stefan Lippers-Hollmann <s.l-h@gmx.de>,
        Hannu Nyman <hannu.nyman@iki.fi>,
        Imran Khan <gururug@gmail.com>,
        Frank Wunderlich <frank-w@public-files.de>,
        Nick Lowe <nick.lowe@gmail.com>,
        =?iso-8859-1?Q?Andr=E9?= Valentin <avalentin@marcant.net>
Subject: Re: [RFC net-next 2/3] net: dsa: qca8k: enable assisted learning on CPU port
Date:   Mon,  9 Aug 2021 00:05:03 +0800
Message-Id: <20210808160503.227880-1-dqfext@gmail.com>
In-Reply-To: <20210807222555.y6r7qxhdyy6d3esx@skbuf>
References: <20210807120726.1063225-1-dqfext@gmail.com> <20210807120726.1063225-3-dqfext@gmail.com> <20210807222555.y6r7qxhdyy6d3esx@skbuf>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: 8bit
Precedence: bulk

On Sun, Aug 08, 2021 at 01:25:55AM +0300, Vladimir Oltean wrote:
> On Sat, Aug 07, 2021 at 08:07:25PM +0800, DENG Qingfang wrote:
> > Enable assisted learning on CPU port to fix roaming issues.
> 
> 'roaming issues' implies to me it suffered from blindness to MAC
> addresses learned on foreign interfaces, which appears to not be true
> since your previous patch removes hardware learning on the CPU port
> (=> hardware learning on the CPU port was supported, so there were no
> roaming issues)

The datasheet says learning is enabled by default, but if that's true,
the driver won't have to enable it manually.

Others have reported roaming issues as well:
https://github.com/Ansuel/openwrt/pull/3

As I don't have the hardware to test, I don't know what the default
value really is, so I just disable learning to make sure.

> 
> > 
> > Although hardware learning is available, it won't work well with
> > software bridging fallback or multiple CPU ports.
> 
> This part is potentially true however, but it would need proof. I am not
> familiar enough with the qca8k driver to say for sure that it suffers
> from the typical problem with bridging with software LAG uppers (no FDB
> isolation for standalone ports => attempt to shortcircuit the forwarding
> through the CPU port and go directly towards the bridged port, which
> would result in drops), and I also can't say anything about its support
> for multiple CPU ports.

QCA8337 supports disabling learning and FDB lookup on a per-VLAN basis,
so we could assign all standalone ports to a reserved VLAN (ID 0 or 4095)
with learning and FDB lookup disabled.

Ansuel has a patch set for multiple CPU ports.