Re: [PATCH 01/12] mmc: dt-bindings: update Mediatek MMC bindings

[Ulf Hansson]

[...]

>> > >> >  Required properties:
>> > >> > -- compatible: Should be "mediatek,mt8173-mmc","mediatek,mt8135-mmc"
>> > >> > +- compatible: value should be either of the following.
>> > >> > +   "mediatek,mt8135-mmc": for mmc host ip compatible with mt8135
>> > >> > +   "mediatek,mt8173-mmc": for mmc host ip compatible with mt8173
>> > >> > +   "mediatek,mt2701-mmc": for mmc host ip compatible with mt2701
>> > >> > +   "mediatek,mt2712-mmc": for mmc host ip compatible with mt2712
>> > >> > +- reg: physical base address of the controller and length
>> > >> >  - interrupts: Should contain MSDC interrupt number
>> > >> > -- clocks: MSDC source clock, HCLK
>> > >> > -- clock-names: "source", "hclk"
>> > >> > +- clocks: MSDC source clock, HCLK, source_cg
>> > >> > +- clock-names: "source", "hclk", "source_cg"
>> > >>
>> > >> All chips support source_cg? That's not backwards compatible for
>> > >> existing compatible strings if the driver requires it.
>> > > Not all chips support source_cg, for chips which do not support
>> > > source_cg, no need source_cg here, and the driver will parse it
>> > > to know if current chip support it.
>> >
>> > In such case you must not add add a required binding for it. I think
>> > that is what Rob is trying to point out for you.
>> >
>> > [...]
>> >
>> > Kind regards
>> > Uffe
>> The source_cg is required(MUST) at MT2712 and future SoCs, but not
>> required(do not have it) at previous SoCs, so that put it at required
>> properties, let the driver to handle it.

Then you must explain that in the binding...

On 29 September 2017 at 03:56, Chaotian Jing <chaotian.jing@xxxxxxxxxxxx> wrote:
> On Wed, 2017-09-27 at 09:18 +0800, Chaotian Jing wrote:
>> On Wed, 2017-09-27 at 00:33 +0200, Ulf Hansson wrote:
>> > On 14 September 2017 at 04:10, Chaotian Jing <chaotian.jing@xxxxxxxxxxxx> wrote:
>> > > On Wed, 2017-09-13 at 09:10 -0500, Rob Herring wrote:
>> > >> On Tue, Sep 12, 2017 at 05:07:41PM +0800, Chaotian Jing wrote:
>> > >> > Change the comptiable for support of multi-platform
>> > >> > Add description for reg
>> > >> > Add description for source_cg
>> > >> > Add description for mediatek,latch-ck
>> > >>
>> > >> This is at least the 3rd patch with exactly the same vague subject.
>> > >> Please make the subject somewhat unique.
>> > >>
>> > > Thx, will change the subject at next version
>> > >> >
>> > >> > Signed-off-by: Chaotian Jing <chaotian.jing@xxxxxxxxxxxx>
>> > >> > ---
>> > >> >  Documentation/devicetree/bindings/mmc/mtk-sd.txt | 13 ++++++++++---
>> > >> >  1 file changed, 10 insertions(+), 3 deletions(-)
>> > >> >
>> > >> > diff --git a/Documentation/devicetree/bindings/mmc/mtk-sd.txt b/Documentation/devicetree/bindings/mmc/mtk-sd.txt
>> > >> > index 4182ea3..405cd06 100644
>> > >> > --- a/Documentation/devicetree/bindings/mmc/mtk-sd.txt
>> > >> > +++ b/Documentation/devicetree/bindings/mmc/mtk-sd.txt
>> > >> > @@ -7,10 +7,15 @@ This file documents differences between the core properties in mmc.txt
>> > >> >  and the properties used by the msdc driver.
>> > >> >
>
> Any other comments about it ? still must not add a required binding for
> it ? if add a optional binding for it, how to add it ? as cannot
> duplicate "clocks" & "clock-names" in one node.
>
>

I suggest you keep the description of the new clock name part of the
"Required properties:" header. However, after each clock name,
explicit state when the clock is required/optional. Something along
the lines as sdhci-msm does it.
Documentation/devicetree/bindings/mmc/sdhci-msm.txt

Kind regards
Uffe