mbox series

[v2,00/10] regulator/mfd/clock: dt-bindings: Samsung S2M and S5M to dtschema

Message ID 20211001094106.52412-1-krzysztof.kozlowski@canonical.com (mailing list archive)
Headers show
Series regulator/mfd/clock: dt-bindings: Samsung S2M and S5M to dtschema | expand

Message

Krzysztof Kozlowski Oct. 1, 2021, 9:40 a.m. UTC
Hi All,

Changes since v1
================
1. Drop DTS patches - applied.
2. Fully remove bindings/regulator/samsung,s5m8767.txt .
3. Minor subject reformatting and few typos in text.


Intro
=====
This patchset converts all devicetree bindings of Samsung S2M and S5M
PMIC devices from txt to dtschema.

It includes also two fixes because later conversion depends on it
(contextually).


Merging/dependencies
====================
1. Regulator related binding changes depend on first two commits (the
   fixes), because of context.
2. The mfd bindings depend on clock and regulator bindings.

The fixes and bindings changes (patches 1-10) should go via the same
tree.  For example regulator or mfd tree.  I propose the regulator tree,
since it will have also one driver change (the fix, first commit).


Overview of devices
===================
Essentially all Samsung S2M and S5M PMICs are very similar devices. They
provide the same functionality: regulators, RTC, 2 or 3 clocks and main
power management (e.g. power cut to SoC).

The differences are mostly in registers layout and number of regulators.

The drivers are built around one common part, mfd/sec-core.c, and share
some drivers between devices:
1. MFD sec-core for all devices,
1. one clock driver for most of devices,
2. one RTC driver for all devices,
3. three regulator drivers.

The regulator drivers were implementing slightly different features,
therefore one regulator binding for all devices does not make much
sense.  However the clock device binding can be shared.

The final dtschema bindings try to implement this - share only the clock
bindings.

Best regards,
Krzysztof

Krzysztof Kozlowski (10):
  regulator: s5m8767: do not use reset value as DVS voltage if GPIO DVS
    is disabled
  regulator: dt-bindings: samsung,s5m8767: correct
    s5m8767,pmic-buck-default-dvs-idx property
  dt-bindings: clock: samsung,s2mps11: convert to dtschema
  regulator: dt-bindings: samsung,s2m: convert to dtschema
  regulator: dt-bindings: samsung,s2mpa01: convert to dtschema
  regulator: dt-bindings: samsung,s5m8767: convert to dtschema
  dt-bindings: mfd: samsung,s2mps11: convert to dtschema
  dt-bindings: mfd: samsung,s2mpa01: convert to dtschema
  dt-bindings: mfd: samsung,s5m8767: convert to dtschema
  dt-bindings: mfd: samsung,s5m8767: document buck and LDO supplies

 .../bindings/clock/samsung,s2mps11.txt        |  49 ---
 .../bindings/clock/samsung,s2mps11.yaml       |  45 +++
 .../bindings/mfd/samsung,s2mpa01.yaml         |  91 ++++++
 .../bindings/mfd/samsung,s2mps11.yaml         | 267 +++++++++++++++
 .../bindings/mfd/samsung,s5m8767.yaml         | 307 ++++++++++++++++++
 .../bindings/mfd/samsung,sec-core.txt         |  86 -----
 .../bindings/regulator/samsung,s2mpa01.txt    |  79 -----
 .../bindings/regulator/samsung,s2mpa01.yaml   |  69 ++++
 .../bindings/regulator/samsung,s2mps11.txt    | 102 ------
 .../bindings/regulator/samsung,s2mps11.yaml   |  52 +++
 .../bindings/regulator/samsung,s2mps13.yaml   |  52 +++
 .../bindings/regulator/samsung,s2mps14.yaml   |  52 +++
 .../bindings/regulator/samsung,s2mps15.yaml   |  52 +++
 .../bindings/regulator/samsung,s2mpu02.yaml   |  52 +++
 .../bindings/regulator/samsung,s5m8767.txt    | 145 ---------
 .../bindings/regulator/samsung,s5m8767.yaml   |  83 +++++
 MAINTAINERS                                   |   9 +-
 drivers/regulator/s5m8767.c                   |  21 +-
 18 files changed, 1136 insertions(+), 477 deletions(-)
 delete mode 100644 Documentation/devicetree/bindings/clock/samsung,s2mps11.txt
 create mode 100644 Documentation/devicetree/bindings/clock/samsung,s2mps11.yaml
 create mode 100644 Documentation/devicetree/bindings/mfd/samsung,s2mpa01.yaml
 create mode 100644 Documentation/devicetree/bindings/mfd/samsung,s2mps11.yaml
 create mode 100644 Documentation/devicetree/bindings/mfd/samsung,s5m8767.yaml
 delete mode 100644 Documentation/devicetree/bindings/mfd/samsung,sec-core.txt
 delete mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mpa01.txt
 create mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mpa01.yaml
 delete mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mps11.txt
 create mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mps11.yaml
 create mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mps13.yaml
 create mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mps14.yaml
 create mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mps15.yaml
 create mode 100644 Documentation/devicetree/bindings/regulator/samsung,s2mpu02.yaml
 delete mode 100644 Documentation/devicetree/bindings/regulator/samsung,s5m8767.txt
 create mode 100644 Documentation/devicetree/bindings/regulator/samsung,s5m8767.yaml

Comments

Mark Brown Oct. 5, 2021, 12:14 p.m. UTC | #1
On Fri, Oct 01, 2021 at 11:40:56AM +0200, Krzysztof Kozlowski wrote:

> Merging/dependencies
> ====================
> 1. Regulator related binding changes depend on first two commits (the
>    fixes), because of context.
> 2. The mfd bindings depend on clock and regulator bindings.
> 
> The fixes and bindings changes (patches 1-10) should go via the same
> tree.  For example regulator or mfd tree.  I propose the regulator tree,
> since it will have also one driver change (the fix, first commit).

Lee, Stephen, Michael does Krzysztof's plan make sense to you?
Lee Jones Oct. 5, 2021, 1:14 p.m. UTC | #2
On Tue, 05 Oct 2021, Mark Brown wrote:

> On Fri, Oct 01, 2021 at 11:40:56AM +0200, Krzysztof Kozlowski wrote:
> 
> > Merging/dependencies
> > ====================
> > 1. Regulator related binding changes depend on first two commits (the
> >    fixes), because of context.
> > 2. The mfd bindings depend on clock and regulator bindings.
> > 
> > The fixes and bindings changes (patches 1-10) should go via the same
> > tree.  For example regulator or mfd tree.  I propose the regulator tree,
> > since it will have also one driver change (the fix, first commit).
> 
> Lee, Stephen, Michael does Krzysztof's plan make sense to you?

I tend to take cross subsystem patches.  MFD is usually in the centre
of these scenarios and I have tooling to easily set-up immutable
branches/pull-requests.

Always happy to discuss if others have different/better ideas though.
Krzysztof Kozlowski Oct. 6, 2021, 12:55 p.m. UTC | #3
On 05/10/2021 15:14, Lee Jones wrote:
> On Tue, 05 Oct 2021, Mark Brown wrote:
> 
>> On Fri, Oct 01, 2021 at 11:40:56AM +0200, Krzysztof Kozlowski wrote:
>>
>>> Merging/dependencies
>>> ====================
>>> 1. Regulator related binding changes depend on first two commits (the
>>>    fixes), because of context.
>>> 2. The mfd bindings depend on clock and regulator bindings.
>>>
>>> The fixes and bindings changes (patches 1-10) should go via the same
>>> tree.  For example regulator or mfd tree.  I propose the regulator tree,
>>> since it will have also one driver change (the fix, first commit).
>>
>> Lee, Stephen, Michael does Krzysztof's plan make sense to you?
> 
> I tend to take cross subsystem patches.  MFD is usually in the centre
> of these scenarios and I have tooling to easily set-up immutable
> branches/pull-requests.
> 
> Always happy to discuss if others have different/better ideas though.
> 

Another alternative is that regulator patches (1-2, 4-6) go via Mark who
later gives you a stable branch/tag to pull. Then the clock and MFD
bindings would go on top via MFD tree.

There is a comment from Rob which I will fix in v3.

Best regards,
Krzysztof
Lee Jones Oct. 6, 2021, 1:20 p.m. UTC | #4
On Wed, 06 Oct 2021, Krzysztof Kozlowski wrote:

> On 05/10/2021 15:14, Lee Jones wrote:
> > On Tue, 05 Oct 2021, Mark Brown wrote:
> > 
> >> On Fri, Oct 01, 2021 at 11:40:56AM +0200, Krzysztof Kozlowski wrote:
> >>
> >>> Merging/dependencies
> >>> ====================
> >>> 1. Regulator related binding changes depend on first two commits (the
> >>>    fixes), because of context.
> >>> 2. The mfd bindings depend on clock and regulator bindings.
> >>>
> >>> The fixes and bindings changes (patches 1-10) should go via the same
> >>> tree.  For example regulator or mfd tree.  I propose the regulator tree,
> >>> since it will have also one driver change (the fix, first commit).
> >>
> >> Lee, Stephen, Michael does Krzysztof's plan make sense to you?
> > 
> > I tend to take cross subsystem patches.  MFD is usually in the centre
> > of these scenarios and I have tooling to easily set-up immutable
> > branches/pull-requests.
> > 
> > Always happy to discuss if others have different/better ideas though.
> > 
> 
> Another alternative is that regulator patches (1-2, 4-6) go via Mark who
> later gives you a stable branch/tag to pull. Then the clock and MFD
> bindings would go on top via MFD tree.

It shouldn't matter where they are first applied.  Creating 2
immutable branches when just 1 will do would be a pain.

> There is a comment from Rob which I will fix in v3.

Sure.