[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <87lgjhhg0j.fsf@weeman.i-did-not-set--mail-host-address--so-tickle-me>
Date: Tue, 07 Nov 2017 17:37:32 -0500
From: Vivien Didelot <vivien.didelot@...oirfairelinux.com>
To: Andrew Lunn <andrew@...n.ch>,
Florian Fainelli <f.fainelli@...il.com>
Cc: David Miller <davem@...emloft.net>, netdev <netdev@...r.kernel.org>
Subject: Re: [PATCH v3 net-next 0/5] IGMP snooping for local traffic
Hi Andrew,
Andrew Lunn <andrew@...n.ch> writes:
>> In a switch case, they all translate to programming a MDB entry for
>> a given switch port, right?
>
> No, in fact it is the exact opposite.
Yes, they do. The proof is you call dsa_port_mdb_add.
> A normal switchdev MDB says send traffic for a group OUT this port.
>
> A host switchdev MDB says send traffic coming IN from a port to the
> CPU.
Still, what I see here _from a switch driver point of view_ is either
program an MDB entry on a user port, or on its CPU port.
There's no direction. There's only an association between a given switch
port and a MAC address (in a given VLAN ID for sure.)
> This is why i think there should be two different switchdev calls, but
> Vivien wants one call, and to look carefully at the parameters to
> determine which of the two above to do.
I do not "want one call". I'm not fighting you. I'm just trying to
understand and avoid adding new object when it isn't necessary. Maybe
your proposal is the correct way to go, maybe translating orig_dev to
the CPU port is just enough. The thing is it is still unclear to me.
Maybe I'm stupid.
Vivien
Powered by blists - more mailing lists