Re: [GIT PULL 4/4] arm64: dts: exynos: DT64 for v4.11, third round

[Sylwester Nawrocki]

On 02/09/2017 07:56 PM, Krzysztof Kozlowski wrote:
> On Thu, Feb 09, 2017 at 04:19:57PM +0100, Arnd Bergmann wrote:
>> On Thursday, February 2, 2017 8:30:08 PM CET Krzysztof Kozlowski wrote:
>>>  arch/arm64/boot/dts/exynos/exynos5433-tm2.dts      |  34 +++
>>>  arch/arm64/boot/dts/exynos/exynos5433-tm2e.dts     |  34 +++
>>>  arch/arm64/boot/dts/exynos/exynos7-espresso.dts    |  49 +++-
>>>  arch/arm64/boot/dts/exynos/exynos7-pinctrl.dtsi    | 302 +++++++++++----------
>>>  arch/arm64/boot/dts/exynos/exynos7.dtsi            |  34 +++
>>>  drivers/clk/samsung/clk-exynos5433.c               |   8 +-
>>>  include/dt-bindings/clock/exynos5433.h             |   5 +-
>>>  include/dt-bindings/pinctrl/samsung.h              |   8 +
>>>
>> It seems a drivers/clk change slipped in there in addition to the header changes.
>> We are trying to keep the next/dt branches free of driver changes, are you
>> able to rework the branch without this?
>
> +Cc Sylwester,
> 
> This came from clk tree from Sylwester. It contains two changes in
> clocks:
> 1. Mew clock ID - we need only the header but Sylwester provided also
>    the source file change,

That's a one patch made so there is no runtime regression.  If we stick
to the next/dt branches being clean of driver changes I'm afraid we need
to be prepared for the bisection breaks.

> 2. Update of frequency (used in DTS as assigned clock rate) - this
>    is strictly source file change and we need it.
> 
> Overall, rework would not be trivial and it would involve Sylwester.
> I think it is better to stay with this as is. I will remember the policy
> for future dependencies.

The clk changes are already pulled into clk tree, we would get conflicts
with commits that are already in the clock tree if the branch is reworked
now.  This might not be a big issue but bisection would be broken for sure.

I'll also try to keep in future include/dt-bindings changes separate.

-- 
Regards,
Sylwester