Re: [PATCH v3 4/4] arm64: dts: ti: Add support for J784S4 EVM board

[Andrew Davis <afd@ti.com>]

On 11/18/22 11:47 AM, Nishanth Menon wrote:
> On 11:40-20221118, Andrew Davis wrote:
>> On 11/16/22 7:04 AM, Apurva Nandan wrote:
> 
> [...]
> 
>>> +#include <dt-bindings/net/ti-dp83867.h>
>>> +#include <dt-bindings/gpio/gpio.h>
>>> +#include "k3-j784s4.dtsi"
>>> +
>>> +/ {
>>> +	compatible = "ti,j784s4-evm", "ti,j784s4";
>>> +	model = "Texas Instruments J784S4 EVM";
>>> +
>>> +	chosen {
>>> +		stdout-path = "serial2:115200n8";
>>> +	};
>>> +
>>> +	aliases {
>>> +		serial2 = &main_uart8;
>>
>> This feels hacky. Your chosen node picks serial2 as that is usually
>> the one that is wired up on K3 boards. But on this board it is main_uart8.
>> So why not have this be serial10, then choose
>>
>> stdout-path = "serial10:115200n8";
>>
>> Also, I've made comments on previous version of this series, it is
>> nice to include folks who have commented before in the CC for future
>> versions, that way our filters don't hide these away and we can more
>> easily check that our comments have been addressed.
> 
> Please stick with the standard of serial2 as the linux console standard.
> We ended up with that to ease up capabilities of various distros to
> uniformly work across SoC and board variants.
> 

The chosen "stdout-path" is for setting the kernel's default output terminal.
Distros and other userspaces need to use their own policy mechanisms for
picking what serial port to run getty on or whatever the issue may be.

Some look at the kernel command line, and our bootloader provides
that too, so still no reason to fake alias names here.

Andrew

> I do agree that phandle is the wrong approach here (baud etc information
> missing). "serial2:115200n8" is probably the way to do this right.
>