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: <d51b7f8d-3681-e19d-3ebb-7e021fc50403@linaro.org>
Date:   Mon, 24 Oct 2022 20:15:25 -0400
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.0
Subject: Re: [PATCH v1 2/5] dt-bindings: soc: hpe: Add hpe,gxp-plreg
Content-Language: en-US
To:     "Hawkins, Nick" <nick.hawkins@hpe.com>
Cc:     "Verdun, Jean-Marie" <verdun@hpe.com>,
        "krzysztof.kozlowski+dt@linaro.org" 
        <krzysztof.kozlowski+dt@linaro.org>,
        "linux@armlinux.org.uk" <linux@armlinux.org.uk>,
        "devicetree@vger.kernel.org" <devicetree@vger.kernel.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "linux-arm-kernel@lists.infradead.org" 
        <linux-arm-kernel@lists.infradead.org>,
        Rob Herring <robh+dt@kernel.org>
References: <20221011185525.94210-1-nick.hawkins@hpe.com>
 <20221011185525.94210-3-nick.hawkins@hpe.com>
 <CAL_Jsq+xb2Ltfne4mQMXQAde-eHS7TsO73YZ-vhE7nK1Z_M0gw@mail.gmail.com>
 <DM4PR84MB192795B45639710259E9B19D88229@DM4PR84MB1927.NAMPRD84.PROD.OUTLOOK.COM>
 <820095a2-3722-5c3a-77fb-5a6b6b44e1c3@linaro.org>
 <DM4PR84MB19275A47D7006BD5664E1AF488319@DM4PR84MB1927.NAMPRD84.PROD.OUTLOOK.COM>
From:   Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
In-Reply-To: <DM4PR84MB19275A47D7006BD5664E1AF488319@DM4PR84MB1927.NAMPRD84.PROD.OUTLOOK.COM>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 24/10/2022 20:03, Hawkins, Nick wrote:
>> I don't think DT place is to describe register or memory layout, with some exceptions like MTD partitions or nvmem cells. Basically you are representing here a device register map inside DT, just because it is a CPLD.
> 
>> Every regular multi-functional device handles its register map in the driver itself and uses Linux framework to expose the internals. CPLD should not be different, except that is programmable.
> Hi Krzysztof,
> 
> Thank you for your time and feedback. We are looking for a place to describe differences within our CPLD implementation due to our memory mapping not being consistent. The idea we are pursuing is to use the device tree to serve as an input to Linux to prevent driver code fragmentation from multiple platforms needing their own specific offsets. 

I understand what you want to achieve, but Devicetree does not seem a
tool for that. DT describes the hardware characteristics, but not is
exact memory map.

Although your goal differs than for example goal of some developer of
I2C or MMIO device drivers, but essentially devices are similar. We do
not describe memory map of MMIO or register map of I2C devices in DT.


> If this is not acceptable to do through the device tree, should we rely on having an include file for each platform instead?

I would say use rather standard Linux subsystems and problem is gone.
You have fan? Sure, we have subsystem for fans. You have power supply or
battery - we have stuff for this as well.

Best regards,
Krzysztof