Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <c732e1ce-8c9c-b947-8d4b-78903920a5b2@gmail.com>
Date:   Wed, 2 Feb 2022 19:12:34 -0800
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:91.0) Gecko/20100101
 Thunderbird/91.5.0
Subject: Re: [PATCH net v3] net: phy: intel-xway: enable integrated led
 functions
Content-Language: en-US
To:     Andrew Lunn <andrew@lunn.ch>, Tim Harvey <tharvey@gateworks.com>
Cc:     Martin Schiller <ms@dev.tdt.de>, Hauke Mehrtens <hauke@hauke-m.de>,
        martin.blumenstingl@googlemail.com, hkallweit1@gmail.com,
        Russell King - ARM Linux <linux@armlinux.org.uk>,
        David Miller <davem@davemloft.net>, kuba@kernel.org,
        netdev <netdev@vger.kernel.org>,
        open list <linux-kernel@vger.kernel.org>
References: <20210421055047.22858-1-ms@dev.tdt.de>
 <CAJ+vNU1=4sDmGXEzPwp0SCq4_p0J-odw-GLM=Qyi7zQnVHwQRA@mail.gmail.com>
 <YfspazpWoKuHEwPU@lunn.ch>
From:   Florian Fainelli <f.fainelli@gmail.com>
In-Reply-To: <YfspazpWoKuHEwPU@lunn.ch>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 2/2/2022 5:01 PM, Andrew Lunn wrote:
>> As a person responsible for boot firmware through kernel for a set of
>> boards I continue to do the following to keep Linux from mucking with
>> various PHY configurations:
>> - remove PHY reset pins from Linux DT's to keep Linux from hard resetting PHY's
>> - disabling PHY drivers
>>
>> What are your thoughts about this?
> 
> Hi Tim
> 
> I don't like the idea that the bootloader is controlling the hardware,
> not linux.

This is really trying to take advantage of the boot loader setting 
things up in a way that Linux can play dumb by using the Generic PHY 
driver and being done with it. This works... until it stops, which 
happens very very quickly in general. The perfect counter argument to 
using the Generic PHY driver is when your system implements a low power 
mode where the PHY loses its power/settings, comes up from suspend and 
the strap configuration is insufficient and the boot loader is not part 
of the resume path *prior* to Linux. In that case Linux needs to restore 
the settings, but it needs a PHY driver for that.

If your concern Tim is with minimizing the amount of time the link gets 
dropped and re-established, then there is not really much that can be 
done that is compatible with Linux setting things up, short of 
minimizing the amount of register writes that do need the "commit phase" 
via BMCR.RESET.

I do agree that blindly imposing LED settings that are different than 
those you want is not great, and should be remedied. Maybe you can 
comment this part out in your downstream tree for a while until the LED 
binding shows up (we have never been so close I am told).
-- 
Florian