[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <5568EA4F.2040705@roeck-us.net>
Date: Fri, 29 May 2015 15:38:07 -0700
From: Guenter Roeck <linux@...ck-us.net>
To: Vivien Didelot <vivien.didelot@...oirfairelinux.com>,
netdev@...r.kernel.org
CC: Florian Fainelli <f.fainelli@...il.com>,
Andrew Lunn <andrew@...n.ch>,
Jerome Oufella <jerome.oufella@...oirfairelinux.com>,
Chris Healy <cphealy@...il.com>, Jiri Pirko <jiri@...nulli.us>,
Scott Feldman <sfeldma@...il.com>
Subject: Re: [RFC 2/3] net: dsa: mv88e6xxx: add support for VTU operations
Hi Vivien,
On 05/28/2015 02:37 PM, Vivien Didelot wrote:
> This commit implements the port_vlan_add, port_vlan_kill, and
> port_bridge_setlink dsa_switch_driver functions to access the VTU, and
> thus add support for adding, removing VLANs, and joining ports to them.
>
> Signed-off-by: Vivien Didelot <vivien.didelot@...oirfairelinux.com>
Pretty much similar to what I have, except for the stu part
which I have completely missing, and I use the default port FIDs.
I wonder if we can really use 'fid = DSA_MAX_PORTS + vid'.
Problem I see is that there may be multiple bridge groups on a switch,
and some ports may not be part of a bridge group. If the same fid is used
for the same vid on multiple ports which belong to different bridge groups,
don't we get into trouble with the address database ?
My assumption was that we have to use a separate fid for each vid per
bridge group (or port if the port is not in a bridge group) if 802.1s
is used, and that we should be able to use the port fid otherwise.
Is that wrong ?
When reporting addresses, how do we associate fdb entries to a vid ?
Or, in other words, how do we handle the fdb entries associated
with a fid in __mv88e6xxx_port_getnext ?
Thanks,
Guenter
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists