On Wed, May 13, 2015 at 05:26:06PM +0800, Yingjoe Chen wrote:
> On Tue, 2015-05-12 at 17:05 +0100, Mark Brown wrote:
> > > > There were three compatible strings listed in the DT binding but only
> > > > two here.
> > > And I test it's ok on MT8135 platform. So I add struct of_device_id
> > > mtk_spi_of_match like this in spi driver code.
> > You should list all the compatibles documented in the binding here, if
> > some of them are the same just have them map to a single constant.
> Just for clarification. If we want to add spi support for a new soc, say
> mt8127, which we think is compatible to mt6589. Since it may turn out we
> need special handling for this soc latter, it is suggested to write
> compatible like this in mt8127.dtsi:
> compatible = "mediatek,mt8127-spi",
> "mediatek,mt6589-spi";
> Device tree binding should list all possible compatible string in .dts,
> so we'll have to add that to binding as well.
> - mediatek,mt6589-spi: for mt6589 platforms
> + - mediatek,mt8127-spi: for mt8127 platforms
> Then we'll also need to add this to source code to reflect this.
> { .compatible = "mediatek,mt6589-spi", .data = (void *)COMPAT_MT6589},
> + { .compatible = "mediatek,mt8127-spi", .data = (void *)COMPAT_MT6589},
> This seems to introduce lots of trivial patches just to add a compatible
> device to me...
Yes, but that's how DT works and as you say these patches are all
trivial so it's not like they take any appreciable effort.
Attachment:
signature.asc
Description: Digital signature
