Re: [PATCH 4/4] arm64: dts: mediatek: Add Cudy WR3000 V1

[AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>]

Il 21/02/24 11:18, Rafał Miłecki ha scritto:
> On 21.02.2024 11:11, AngeloGioacchino Del Regno wrote:
>> Il 21/02/24 08:35, Rafał Miłecki ha scritto:
>>> From: Rafał Miłecki <rafal@milecki.pl>
>>>
>>> Cudy WR3000 V1 is an MT7981B (AKA Filogic 820) based wireless router. It
>>> has 256 MiB of RAM, some LEDs & buttons and (not described yet) 4
>>> Ethernet ports.
>>>
>>> Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
>>> ---
>>>   arch/arm64/boot/dts/mediatek/Makefile         |  1 +
>>>   .../dts/mediatek/mt7981b-cudy-wr3000-v1.dts   | 74 +++++++++++++++++++
>>>   2 files changed, 75 insertions(+)
>>>   create mode 100644 arch/arm64/boot/dts/mediatek/mt7981b-cudy-wr3000-v1.dts
>>>
>>> diff --git a/arch/arm64/boot/dts/mediatek/Makefile 
>>> b/arch/arm64/boot/dts/mediatek/Makefile
>>> index 37b4ca3a87c9..96da4ad640aa 100644
>>> --- a/arch/arm64/boot/dts/mediatek/Makefile
>>> +++ b/arch/arm64/boot/dts/mediatek/Makefile
>>> @@ -8,6 +8,7 @@ dtb-$(CONFIG_ARCH_MEDIATEK) += mt6797-evb.dtb
>>>   dtb-$(CONFIG_ARCH_MEDIATEK) += mt6797-x20-dev.dtb
>>>   dtb-$(CONFIG_ARCH_MEDIATEK) += mt7622-rfb1.dtb
>>>   dtb-$(CONFIG_ARCH_MEDIATEK) += mt7622-bananapi-bpi-r64.dtb
>>> +dtb-$(CONFIG_ARCH_MEDIATEK) += mt7981b-cudy-wr3000-v1.dtb
>>>   dtb-$(CONFIG_ARCH_MEDIATEK) += mt7981b-xiaomi-ax3000t.dtb
>>>   dtb-$(CONFIG_ARCH_MEDIATEK) += mt7986a-acelink-ew-7886cax.dtb
>>>   dtb-$(CONFIG_ARCH_MEDIATEK) += mt7986a-bananapi-bpi-r3.dtb
>>> diff --git a/arch/arm64/boot/dts/mediatek/mt7981b-cudy-wr3000-v1.dts 
>>> b/arch/arm64/boot/dts/mediatek/mt7981b-cudy-wr3000-v1.dts
>>> new file mode 100644
>>> index 000000000000..cb36a089518a
>>> --- /dev/null
>>> +++ b/arch/arm64/boot/dts/mediatek/mt7981b-cudy-wr3000-v1.dts
>>> @@ -0,0 +1,74 @@
>>> +// SPDX-License-Identifier: GPL-2.0-only OR MIT
>>> +
>>> +/dts-v1/;
>>> +#include <dt-bindings/input/input.h>
>>> +#include <dt-bindings/gpio/gpio.h>
>>> +#include <dt-bindings/leds/common.h>
>>> +
>>> +#include "mt7981b.dtsi"
>>> +
>>> +/ {
>>> +    compatible = "cudy,wr3000-v1", "mediatek,mt7981b";
>>> +    model = "Cudy WR3000 V1";
>>> +
>>> +    memory@40000000 {
>>> +        reg = <0 0x40000000 0 0x10000000>;
>>> +        device_type = "memory";
>>> +    };
>>> +
>>> +    keys {
>>> +        compatible = "gpio-keys";
>>> +
>>> +        key-wps {
>>> +            label = "WPS";
>>> +            gpios = <&pio 0 GPIO_ACTIVE_LOW>;
>>> +            linux,code = <KEY_WPS_BUTTON>;
>>> +        };
>>> +
>>> +        key-reset {
>>> +            label = "RESET";
>>> +            gpios = <&pio 1 GPIO_ACTIVE_LOW>;
>>> +            linux,code = <KEY_RESTART>;
>>> +        };
>>> +    };
>>> +
>>> +    leds {
>>> +        compatible = "gpio-leds";
>>> +
>>> +        led-0 {
>>> +            function = LED_FUNCTION_WAN;
>>> +            color = <LED_COLOR_ID_BLUE>;
>>> +            gpios = <&pio 5 GPIO_ACTIVE_LOW>;
>>
>> Can we please order those properties alphabetically, as it doesn't impact
>> on human readability in any way?
>> Just a nitpick, anyway.
>>
>> color = <LED_COLOR_ID_BLUE>;
>> function = LED_FUNCTION_WAN;
>> gpios = <&pio 5 GPIO_ACTIVE_LOW>;
> 
> If I understand [1] correctly alphabetical / alpha-numerical order
> applies to non-addresses nodes only.
> 
> Here I put "function" first as that seemed more important than a color.
> 
> Perhaps you could suggest an addition to Devicetree Sources (DTS) Coding
> Style or point me to the missed part, please?
> 
> [1] 
> https://www.kernel.org/doc/html/next/devicetree/bindings/dts-coding-style.html#order-of-properties-in-device-node
> 
> 

I didn't say that this was *mandatory*, but since alphanumeric ordering in device
trees is actually a thing for nodes, this could be as well applied to properties
inside when this doesn't impact on human readability.

As far as I'm aware, I'm not the only maintainer asking for that (am I?!)

If you've got any *strong opinion* about having `function` as the first property,
I'm full ears, and I will be okay with that. No problem on my side, I'm open for
alternatives when those make sense.

> 
>>> +        };
>>> +
>>> +        led-1 {
>>> +            function = LED_FUNCTION_WLAN_2GHZ;
>>> +            color = <LED_COLOR_ID_BLUE>;
>>> +            gpios = <&pio 6 GPIO_ACTIVE_LOW>;
>>> +        };
>>> +
>>> +        led-2 {
>>> +            function = LED_FUNCTION_WLAN_5GHZ;
>>> +            color = <LED_COLOR_ID_BLUE>;
>>> +            gpios = <&pio 7 GPIO_ACTIVE_LOW>;
>>> +        };
>>> +
>>> +        led-3 {
>>> +            function = LED_FUNCTION_LAN;
>>> +            color = <LED_COLOR_ID_BLUE>;
>>> +            gpios = <&pio 9 GPIO_ACTIVE_LOW>;
>>> +        };
>>> +
>>> +        led-4 {
>>> +            function = LED_FUNCTION_STATUS;
>>> +            color = <LED_COLOR_ID_BLUE>;
>>> +            gpios = <&pio 10 GPIO_ACTIVE_LOW>;
>>> +        };
>>> +
>>> +        led-5 {
>>> +            function = "online";
>>
>> Uhm, what does "online" mean?
>>
>> Are you sure that you can't use any of the LED_FUNCTION_XXX standard definitions?
> 
> This device has:
> 1. "WAN Port" LED that indicates "WAN Port has connection" state

Just a curiosity: is the WAN port ethernet, fiber or what?
(to entirely clarify in my brain if that LED is a carrier detect indicator,
but doesn't matter for this patch, it's really just a personal curiosity and
nothing else)

> 2. "Internet" LED that indicates "Connected to the Internet" state
> 
> I couldn't find any define that would fit Internet access LED case.

I was imagining something like that, but wasn't sure, so asking you was the best
choice.

I wonder if it makes sense to add a "wan-online" or "wan-connected" or .. I don't
know, some definition that makes common sense, to dt-bindings/leds/common.h to
somewhat standardize the function for that usecase?

One thing I'm (mostly) sure of, it makes sense to have the "wan-" prefix before
any word (because you can have sorts of authenticated LANs as well).

I don't think that there's anything hugely restrictive that would prevent you
from doing that: there are even "player-[1-5]" definitions for game controllers,
and yours is not the first router that I physically see with my own eyes that
has such a LED for that state.

After all, what's going to turn on that LED is userspace, and if we can get one
that works for all (without custom functions for each router), why not :-)