[PATCH v2] docs: dt-bindings: add DTS Coding Style document

[Krzysztof Kozlowski]

On 20/11/2023 15:01, Michal Simek wrote:
> 
> 
> On 11/20/23 09:40, Krzysztof Kozlowski wrote:
>> Document preferred coding style for Devicetree sources (DTS and DTSI),
>> to bring consistency among all (sub)architectures and ease in reviews.
>>
>> Cc: Andrew Davis <afd at ti.com>
>> Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno at collabora.com>
>> Cc: Arnd Bergmann <arnd at arndb.de>
>> Cc: Bjorn Andersson <andersson at kernel.org>
>> Cc: Geert Uytterhoeven <geert+renesas at glider.be>
>> Cc: Heiko Stuebner <heiko at sntech.de>
>> Cc: Konrad Dybcio <konrad.dybcio at linaro.org>
>> Cc: Matthias Brugger <matthias.bgg at gmail.com>
>> Cc: Michal Simek <michal.simek at amd.com>
>> Cc: Neil Armstrong <neil.armstrong at linaro.org>
>> Cc: Nishanth Menon <nm at ti.com>
>> Cc: Olof Johansson <olof at lixom.net>
>> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski at linaro.org>
>>
>> ---
>>
>> Merging idea: Rob/DT bindings
>>
>> Changes in v2
>> =============
>> 1. Hopefully incorporate entire feedback from comments:
>> a. Fix \ { => / { (Rob)
>> b. Name: dts-coding-style (Rob)
>> c. Exceptions for ordering nodes by name for Renesas and pinctrl (Geert,
>>     Konrad)
>> d. Ordering properties by common/vendor (Rob)
>> e. Array entries in <> (Rob)
>>
>> 2. New chapter: Organizing DTSI and DTS
>>
>> 3. Several grammar fixes (missing articles)
>>
>> Cc: linux-rockchip at lists.infradead.org
>> Cc: linux-mediatek at lists.infradead.org
>> Cc: linux-samsung-soc at vger.kernel.org
>> Cc: linux-amlogic at lists.infradead.org
>> Cc: linux-arm-kernel at lists.infradead.org
>> Cc: linux-arm-msm at vger.kernel.org
>> ---
>>   .../devicetree/bindings/dts-coding-style.rst  | 163 ++++++++++++++++++
>>   Documentation/devicetree/bindings/index.rst   |   1 +
>>   2 files changed, 164 insertions(+)
>>   create mode 100644 Documentation/devicetree/bindings/dts-coding-style.rst
>>
>> diff --git a/Documentation/devicetree/bindings/dts-coding-style.rst b/Documentation/devicetree/bindings/dts-coding-style.rst
>> new file mode 100644
>> index 000000000000..cc7e3b4d1b92
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/dts-coding-style.rst
>> @@ -0,0 +1,163 @@
>> +.. SPDX-License-Identifier: GPL-2.0
>> +.. _dtscodingstyle:
>> +
>> +=====================================
>> +Devicetree Sources (DTS) Coding Style
>> +=====================================
>> +
>> +When writing Devicetree Sources (DTS) please observe below guidelines.  They
>> +should be considered complementary to any rules expressed already in Devicetree
>> +Specification and dtc compiler (including W=1 and W=2 builds).
>> +
>> +Individual architectures and sub-architectures can add additional rules, making
>> +the style stricter.
>> +
>> +Naming and Valid Characters
>> +---------------------------
>> +
>> +1. Node and property names are allowed to use only:
>> +
>> +   * lowercase characters: [a-z]
>> +   * digits: [0-9]
>> +   * dash: -
> 
> device-tree specification v0.4. Chapter 2.2.1/Table 2.1 is describing much more
> valid characters for node names.
> It means above description is not accurate or DT spec should be updated.

Spec allows way to much. dtc doesn't. One thing is the spec, second
thing is coding style.

> 
> 
>> +
>> +2. Labels are allowed to use only:
>> +
>> +   * lowercase characters: [a-z]
>> +   * digits: [0-9]
>> +   * underscore: _
> 
> based on dt spec uppercase is also valid char in label.

Which we do not want in the DTS.

> 
> 
>> +
>> +3. Unit addresses should use lowercase hex, without leading zeros (padding).
>> +
>> +4. Hex values in properties, e.g. "reg", should use lowercase hex.  The address
>> +   part can be padded with leading zeros.
>> +
>> +Example::
>> +
>> +	gpi_dma2: dma-controller at 800000 {
>> +		compatible = "qcom,sm8550-gpi-dma", "qcom,sm6350-gpi-dma";
>> +		reg = <0x0 0x00800000 0x0 0x60000>;
> 
> Is 0x0 recommended or 0 is enough?

I don't want to impose any rule on that, because someone would like to
argue that hex should be also in SPI chip-select reg.

> 
>> +	}
>> +
>> +Order of Nodes
>> +--------------
>> +
>> +1. Nodes within any bus, thus using unit addresses for children, shall be
>> +   ordered incrementally by unit address.
>> +   Alternatively for some sub-architectures, nodes of the same type can be
>> +   grouped together (e.g. all I2C controllers one after another even if this
>> +   breaks unit address ordering).
>> +
>> +2. Nodes without unit addresses should be ordered alpha-numerically by the node
>> +   name.  For a few types of nodes, they can be ordered by the main property
>> +   (e.g. pin configuration states ordered by value of "pins" property).
>> +
>> +3. When extending nodes in the board DTS via &label, the entries should be
>> +   ordered alpha-numerically.
>> +
>> +Example::
>> +
>> +	// SoC DTSI
> 
> Same comment about /* */ as was mentioned in another thread.
> 
>> +
>> +	/ {
>> +		cpus {
>> +			// ...
>> +		};
>> +
>> +		psci {
>> +			// ...
>> +		};
>> +
>> +		soc@ {
>> +			dma: dma-controller at 10000 {
>> +				// ...
>> +			};
>> +
>> +			clk: clock-controller at 80000 {
>> +				// ...
>> +			};
>> +		};
>> +	};
>> +
>> +	// Board DTS
>> +
>> +	&clk {
>> +		// ...
>> +	};
>> +
>> +	&dma {
>> +		// ...
>> +	};
>> +
>> +
>> +Order of Properties in Device Node
>> +----------------------------------
>> +
>> +Following order of properties in device nodes is preferred:
>> +
>> +1. compatible
>> +2. reg
>> +3. ranges
>> +4. Standard/common properties (defined by common bindings, e.g. without
>> +   vendor-prefixes)
>> +5. Vendor-specific properties
>> +6. status (if applicable)
>> +7. Child nodes, where each node is preceded with a blank line
> 
> Isn't the order already defined in DT spec in 2.3 in chapters?

Where is it defined as this is preferred order?

> compatible
> model
> status
> #address/size cells
> reg
> virtual-reg
> ranges
> dma-ranges
> dma-coherent
> dma-non-coherent
> 
> Again I am fine with whatever order but I think we should reflect it in the spec 

Spec is not a coding style.

> too. Especially status property is for my taste too low simply because you start 
> to read and then you will find that IP is disabled.

Which is exactly what you want. status is irrelevant for hardware
description, so should be the last item.

> 
> And are you describing all properties starting with # as standard properties?

Yes.

> 
> 
>> +
>> +The "status" property is by default "okay", thus it can be omitted.
>> +
>> +Example::
>> +
>> +	// SoC DTSI
> 
> 
> /* */
> 
>> +
>> +	usb_1_hsphy: phy at 88e3000 {
>> +		compatible = "qcom,sm8550-snps-eusb2-phy";
>> +		reg = <0x0 0x088e3000 0x0 0x154>;
>> +		#phy-cells = <0>;
>> +		resets = <&gcc GCC_QUSB2PHY_PRIM_BCR>;
>> +		status = "disabled";
>> +	};
>> +
>> +	// Board DTS
>> +
>> +	&usb_1_hsphy {
>> +		clocks = <&tcsr TCSR_USB2_CLKREF_EN>;
>> +		clock-names = "ref";
>> +		status = "okay";
>> +	};
>> +
>> +
>> +Indentation
>> +-----------
>> +
>> +1. Use indentation according to :ref:`codingstyle`.
>> +2. For arrays spanning across lines, it is preferred to align the continued
>> +   entries with opening < from the first line.
>> +3. Each entry in arrays with multiple cells (e.g. "reg" with two IO addresses)
>> +   should be enclosed in <>.
>> +
>> +Example::
>> +
>> +	thermal-sensor at c271000 {
>> +		compatible = "qcom,sm8550-tsens", "qcom,tsens-v2";
>> +		reg = <0x0 0x0c271000 0x0 0x1000>,
>> +		      <0x0 0x0c222000 0x0 0x1000>;
>> +	};
>> +
>> +Organizing DTSI and DTS
>> +-----------------------
>> +
>> +The DTSI and DTS files should be organized in a way representing the common
>> +(and re-usable) parts of the hardware.  Typically this means organizing DTSI
>> +and DTS files into several files:
>> +
>> +1. DTSI with contents of the entire SoC (without nodes for hardware not present
>> +   on the SoC).
>> +2. If applicable: DTSI with common or re-usable parts of the hardware (e.g.
>> +   entire System-on-Module).
> 
> DTS/DTSI - SOMs can actually run as they are that's why it is fair to say that
> there doesn't need to be DTS representing the board.

I have never seen a SoM which can run without elaborate hardware-hacking
(e.g. connecting multiple wires to the SoM pins). The definition of the
SoM is that it is a module. Module can be re-used, just like SoC.

Best regards,
Krzysztof