[meta-freescale] Fwd: new mx28evk board support

[Fabio Estevam]

Adding the list on Cc.


---------- Forwarded message ----------
From: Fabio Estevam <festevam at gmail.com>
Date: Mon, Nov 9, 2015 at 1:07 PM
Subject: Re: [meta-freescale] new mx28evk board support
To: Ivan Sergio Borgonovo <ivan at webthatworks.it>


On Mon, Nov 9, 2015 at 1:02 PM, Ivan Sergio Borgonovo
<ivan at webthatworks.it> wrote:
> On 11/09/2015 03:38 PM, Fabio Estevam wrote:
>
>> Why don't you generate a patch against mainline kernel and submit it
>> upstream?
>>
>> Then your hardware can work out of the box in future kernel releases.
>>
>> I see you are patching imx28-evk.dts, but you should better create a
>> new dts for this new board.
>
>
> That exactly the kind of feedback I was looking for. The main reasons I
> didn't create a new dts are:
> - I didn't think it was worth to have "one more board" for just a patch

>From a kernel perspective a new board is just a dts file, so yes the
correct approach is to add a new board dts file even though it is
similar to mx28evk.

> - the patch is related to peripherals that don't affect the boot

dts is about the hardware description. You can start with a minimal
dts if you want.

>
> Furthermore the manufacturer offers the board with some "configurable" pins
> that makes supporting new boards just a matter of small tweaks in the dts
> for things that still don't affect the boot.
>
> As a user I would certainly find nice if all I needed to do to boot a board
> was to set the MACHINE but well there are so many ways to remap IO that
> don't affect the boot I was really wondering if such a proliferation of
> boards is what I would really like.
>
> I was thinking about dt overlays etc... but at the moment I don't find
> enough strong motivation to start the journey, and that's another reason I
> was asking for some feedback.
>
> I've recently read Octavio talking about SPL to support multiple boards.

We do this for wandboard, cuboxi, where we can support multiple boards
with a single U-boot binary.

> I actually don't have a clear idea about the whole subject, that makes
> harder to get an opinion about what I'd enjoy as a user, let alone being
> able to implement it.
> But as I wrote from the beginning I see few value in a quick hack of a dts,
> not to mention escalate to mainline kernel.
>
> In fact I'm here to ask ;)

My suggestions:

- Upstream your board dts
- Upstream U-boot for your board
- Then upstream it to meta-fsl-arm-extra