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;
Date:   Tue, 26 May 2020 01:46:23 +0200
From:   Andrew Lunn <andrew@lunn.ch>
To:     Jeremy Linton <jeremy.linton@arm.com>
Cc:     Russell King - ARM Linux admin <linux@armlinux.org.uk>,
        netdev@vger.kernel.org, davem@davemloft.net, f.fainelli@gmail.com,
        hkallweit1@gmail.com, madalin.bucur@oss.nxp.com,
        calvin.johnson@oss.nxp.com, linux-kernel@vger.kernel.org
Subject: Re: [RFC 04/11] net: phy: Handle c22 regs presence better
Message-ID: <20200525234623.GG768009@lunn.ch>
References: <20200523183731.GZ1551@shell.armlinux.org.uk>
 <f85e4d86-ff58-0ed2-785b-c51626916140@arm.com>
 <20200525100612.GM1551@shell.armlinux.org.uk>
 <63ca13e3-11ea-3ddf-e1c7-90597d4a5f8c@arm.com>
 <20200525220127.GO1551@shell.armlinux.org.uk>
 <a9490c28-ebe1-ed6d-e65e-2e1d0a06386b@arm.com>
 <20200525230946.GR1551@shell.armlinux.org.uk>
 <ab756571-b269-ba7f-8e23-053098d9f470@arm.com>
 <20200525233335.GT1551@shell.armlinux.org.uk>
 <269de1a4-57cf-b175-3184-2f4604255bf7@arm.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <269de1a4-57cf-b175-3184-2f4604255bf7@arm.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

> Right, which is what I sort of expected. Because its falling back to a
> device list of 0xFFFFFFFF, which means probe every single MMD.
> 
> Combined with the lack of filtering means that your getting a bunch of MMD
> IDs that potentially are invalid, along with any that happen to be valid.
> Its that behavior (and some others) which were what blew this set up from a
> couple lines of tweaks into this mess.
> 
> I don't really see a way to guess at all the "wrong" ids that are being
> pushed into the system. Which is why I started to think about a "strict"
> mode later in the set. Maybe at this point the only way around some of these
> bugs/side effects/etc is just a second c45 probe routine if we don't think
> its possible to implement a variable strictness scanner in this code path
> without losing phys that previously were detected.

Hi Jeremy

I really think we need to find one of those boards which have a
cortina and see what it actually does. Can you get in contact with NXP
and see if they can find one?

    Thanks
        Andrew