Re: [PATCH 2/2] arm64: dts: ti: k3-am654-base-board: Add MMC/SD support

From: Sekhar Nori
Date: Mon Dec 10 2018 - 07:41:13 EST


Hi Nishanth,

On 10/12/18 5:36 PM, Nishanth Menon wrote:
> On 13:33-20181210, Sekhar Nori wrote:
>> On 08/12/18 9:24 PM, Nishanth Menon wrote:
>>> On 14:12-20181207, Faiz Abbas wrote:
>>>
>>>> +
>>>> +&sdhci0 {
>>>> + status = "okay";
>>>> + pinctrl-names = "default";
>>>> + pinctrl-0 = <&main_mmc0_pins_default>;
>>>> + bus-width = <8>;
>>>> + non-removable;
>>>> + ti,driver-strength-ohm = <50>;
>>>
>>> ^^
>>>
>>>> +};
>>>> +
>>>> +&sdhci1 {
>>>> + status = "okay";
>>>> + pinctrl-names = "default";
>>>> + pinctrl-0 = <&main_mmc1_pins_default>;
>>>> + ti,driver-strength-ohm = <50>;
>>>
>>> NAK.
>>>
>>> $ git checkout next-20181207
>>> $ git grep ti,driver-strength-ohm Documentation
>>> $
>>>
>>> Nada.. And.. I think "new phy binding" probably introduces this.
>>> [1] https://patchwork.kernel.org/project/linux-mmc/list/?series=53185
>>>
>>> If your patches are'nt really ready, please send them as RFC, I am not
>>> really in a mood to track the status of every single driver subsystem.
>>>
>>> If your binding is not in linux next at the baremin, as far as I am
>>> concerned, this is not ready, and should be RFC.
>>
>> No, RFC does not say "do not merge" or "this has dependencies". RFC is
>> used to invite a stronger review when introducing a new concept. Its
>> fair game to apply patches marked RFC if maintainer is okay with the
>> content.
>
> True, fair enough.. RFC is request for comments. Anyways, that is
> besides the point.
>>
>> Dependencies are either noted in cover-letter or below the patch
>> tear-line. With what you are asking, looks like patches need to be
>> resubmitted once dependencies are cleared, even if there is no change in
>> the content itself. This will be additional work.
>
> Yes please. There would be other dts changes that are probably ready and
> I really wont be tracking everything happening on other drivers. If the
> binding is present at least in next, it is a good indication of things
> clean and ready to go.

Agree that bindings should be in linux-next before device-tree files are
merged.

>
>>
>> That said, if it makes life convenient for you, you can impose such a
>> rule for patches you need to handle. But I think it will take some
>> getting used for developers who send patches to you as I don't think
>> this is a norm elsewhere.
>>
>> Adding Tony and Arnd as well, in case I have missed some recently
>> accepted convention.
>
>
> I have'nt looked at any conventions, The style I prefer to follow when I do
> submissions: It is my job to get the bindings in, until then my actual
> dts is just "request for comments". Only after the bindings are merged
> do I formally submit dts - simply because I dont expect dts maintainer
> to track what happened to my driver's binding and discussions there of.

Ok.

>
> Seriously, is'nt it really reasonable for dts maintainer to check every
> single driver's development status in 15 different mailing lists?
> Because, it sounds like what you are asking. At least I wont have time
> for it..
>
>
> I really am curious how Arnd / Tony actually pull this one off.. If they
> have continous cron job for checking if your patch is ready... I doubt
> it..

I think you can rely on the author to tell you when something is
actually ready to be merged (and you can tell him/her to remind you).

For the review itself, doing it by having a look at the dependencies
mentioned in the cover letter (like available for this series) should be
good enough (I feel).

I am not sure if there is a need to post an "RFC version", and then
follow it up with an actual "PATCH version" once dependencies are
cleared though.

Thanks,
Sekhar