From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andrew Lunn <andrew@lunn.ch>
Subject: Re: [PATCH] net: bridge: add max_fdb_count
Date: Thu, 16 Nov 2017 03:25:40 +0100
Message-ID: <20171116022540.GF2130@lunn.ch>
References: <1510774027-2468-1-git-send-email-srn@prgmr.com>
 <20171115120452.3b426442@xeon-e3>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Sarah Newman <srn@prgmr.com>, netdev@vger.kernel.org
To: Stephen Hemminger <stephen@networkplumber.org>
Return-path: <netdev-owner@vger.kernel.org>
Received: from vps0.lunn.ch ([185.16.172.187]:56174 "EHLO vps0.lunn.ch"
        rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
        id S1750750AbdKPCZm (ORCPT <rfc822;netdev@vger.kernel.org>);
        Wed, 15 Nov 2017 21:25:42 -0500
Content-Disposition: inline
In-Reply-To: <20171115120452.3b426442@xeon-e3>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

> Also what do the vendors using bridge for L2 offload to switch think?

The Marvell L2 switches which DSA supports have 8K FDB/MDB entries. So
maybe 1024 is a bit low?

How big is an FDB entry? Even an OpenWRT/LEDE class devices with 512MB
RAM can probably support 1MB of RAM for FDB entries.

> Many switches need to clone table, and similar limits must be in
> other versions.

DSA does not maintain synchronisation between the hardware tables and
the software tables. So no cloning. We expect both bridges to perform
learning based on the frames. This means we should be able to handle
one bridge flooding because its table is full, while the other has an
entry and forwards out the correct port.

	 Andrew