Re: [PATCH 1/4] ARM: dts: add support for hdmi to exynos4

[Sylwester Nawrocki <sylvester.nawrocki@gmail.com>]

Hi Rahul,

On 01/28/2013 07:19 AM, Rahul Sharma wrote:
>>> --- a/arch/arm/boot/dts/exynos4.dtsi
>>> +++ b/arch/arm/boot/dts/exynos4.dtsi
>>> @@ -289,4 +289,11 @@
>>>                          interrupts =<0 34 0>;
>>>                  };
>>>          };
>>> +
>>> +       hdmi {
>>> +               compatible = "samsung,exynos4-hdmi14";
>>
>>
>> What "-hdmi14" signifies here ? And to which Exynos4 SoC revisions this
>> compatible property is supposed to apply ?
>>
>
> -hdmi14 signifies 1.4 compliant Hdmi IP like in Exynos4412, while
> Exynos4210 has 1.3
> compliant HDMI block. I have done a mistake here. I should have kept
> it in 4412.dtsi. OR
> I can move it to 4x12.dtsi but not sure whether all 4x12 socs have 1.4 hdmi.

Exynos4210 also have the HDMI IP. I'm not sure how much common those 
IP's are
among Exynos4 series SoCs. I suppose it makes sense to create a hdmi node in
exynos4.dtsi and override what's required either in exynos4x12.dtsi or
exynos4412.dtsi file.

What is worring me is that you're trying to just append the supported
HDMI standard version to the string, and there might be more differences
in the IPs across all Exynos4 SoC series. What if there are two different
SoCs that support e.g. HDMI v1.4 but there are some differences in the
HDMI IPs ? You would likely need to invent new properties and pass them
to the driver. Appending specific SoC name to the compatible string looks
like a better alternative to me.

>>> +               reg =<0x12D00000 0x70000>;
>>> +               interrupts =<0 95 0>;
>>> +               hpd-gpio =<&gpx3 7 0xf 1 3>;
>>> +       };
>>
>>
>> New bindings need to be documented in Documentation/devicetree/bindings/.
>>
> It is present at Documentation/devicetree/bindings/drm/exynos/hdmi.txt.

Thanks. Sorry, I've missed that. But it seems to be only for exynos5 SoCs ?
Also the location is a bit unexpected to me, if we want to have Exynos4 and
Exynos5 specific IP documented in this file. As I mentioned in previous
e-mail this IP is not only handled by the DRM API.

--

Regards,
Sylwester