From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andrew Lunn <andrew@lunn.ch>
Subject: Re: [PATCH] net: phy: fix PHY_RUNNING in phy_state_machine
Date: Thu, 17 Mar 2016 16:12:40 +0100
Message-ID: <20160317151240.GD26019@lunn.ch>
References: <1439526220-31458-1-git-send-email-shh.xie@gmail.com>
 <CAGm1_kufFr0SsVwVcABXkL0ABJahiwz4EZ70=ywrtC9Bffbfng@mail.gmail.com>
 <20160316161848.GE2393@lunn.ch>
 <CAGm1_kuMQrECdb6_NpZ83Nm4RFSn6sQMhMK6392PLyaFFxCs1A@mail.gmail.com>
 <20160316230505.GA18161@lunn.ch>
 <CAGm1_kv7eia4LPWT5TxUJp1NV+eFZj6gBQc-pHwxbygoF03EYg@mail.gmail.com>
 <20160317134134.GA24913@lunn.ch>
 <CAGm1_kvJn0Pq38iqOmOcyNpe62EEUmq_s5G2_HXps4Gx+xB_=Q@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: shaohui ??? <shh.xie@gmail.com>, netdev <netdev@vger.kernel.org>,
	David Miller <davem@davemloft.net>,
	Shaohui Xie <Shaohui.Xie@freescale.com>,
	Florian Fainelli <f.fainelli@gmail.com>,
	"N, Mugunthan V" <mugunthanvnm@ti.com>, drivshin@allworx.com
To: Yegor Yefremov <yegorslists@googlemail.com>
Return-path: <netdev-owner@vger.kernel.org>
Received: from vps0.lunn.ch ([178.209.37.122]:49386 "EHLO vps0.lunn.ch"
	rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
	id S932614AbcCQPMp (ORCPT <rfc822;netdev@vger.kernel.org>);
	Thu, 17 Mar 2016 11:12:45 -0400
Content-Disposition: inline
In-Reply-To: <CAGm1_kvJn0Pq38iqOmOcyNpe62EEUmq_s5G2_HXps4Gx+xB_=Q@mail.gmail.com>
Sender: netdev-owner@vger.kernel.org
List-ID: <netdev.vger.kernel.org>

> After changing cpsw_emac0 entry to:
> 
> &cpsw_emac0 {
>         phy-mode = "rmii";
>         dual_emac_res_vlan = <1>;
>         fixed-link {
>                 speed = <100>;
>                 full-duplex;
>         };
> };
> 
> I've got packets running in both directions.

Great.
 
> Now I have another problem: I cannot disable ICPlus IP175D ports via
> SIOCSMIIREG as I could do previously. I get not ioctl errors [1], but
> the ports are always on.
> 
> [1] https://github.com/visionsystemsgmbh/libonrisc/blob/master/src/onrisc.c#L83

The MDIO bus is now logically not connected to eth0. Instead you have
the fixed-link mdio device connected to eth0. So SIOCSMIIREG calls on
eth0 now try to manipulate the fixed link phys.

However, i think you can still access the MDIO bus, just use eth1 in
your ioctl call.

You should however consider writing a DSA driver for the switch.

    Andrew