[PATCH v3 2/2] dt-bindings: simplefb-sunxi: Add sunxi simplefb extensions

Hans de Goede hdegoede at redhat.com
Tue Nov 25 05:45:51 PST 2014


Hi,

On 11/25/2014 02:38 PM, Tomi Valkeinen wrote:
> On 25/11/14 15:21, Hans de Goede wrote:
>
>>> Shouldn't the compatible then be "allwinner,sunxi-simple-framebuffer",
>>> to differentiate from some other SoC Allwinner has or might create in
>>> the future? That is, presuming you're confident enough that a single
>>> compatible string covers all the current and forthcoming sunxi SoCs.
>>
>> This was discussed in an earlier thread, we (Ian Campbell, Grant and me)
>
> Okay. Sorry for not having time at the moment to follow the discussions
> properly. =)
>
>> decided to settle on allwinner,simple-framebuffer to make it clear that
>> these are allwinner extensions to the standard simple-framebuffer bindings,
>> and that the node otherwise is simple-framebuffer compatible.
>>
>> We were afraid that e.g. sun4i-simple-framebuffer would signal that it
>> is not a normal simple-framebuffer node, so we decided to go with just
>> the allwinner, prefix to indicate that it uses allwinner specific
>> extensions.
>
> Wouldn't
>
> compatible = "allwinner,sun4i-simple-framebuffer", "simple-framebuffer";
>
> tell that it's a simple-framebuffer, with allwinner's sun4i extensions?
>
> I guess you can have just "allwinner,simple-framebuffer", and then if a
> new Allwinner SoC has a totally different display controller, the
> documentation would specify that this property is for that SoC, and this
> another property is for that another SoC. But isn't the compatible
> string what's supposed to use in cases like this?

The only soc specific thing in the binding is the pipeline property string
values, and we can always add new values to that, the rest is all generic,
as simplefb is generic.

As said Ian Campbell, Grant and me have decided on using this, and currently
patches are already queued up for both the dts files and u-boot to use this,
so unless there are really strong reasons to change it at this point I would
prefer to keep this as is.

Regards,

Hans



More information about the linux-arm-kernel mailing list