Re: [PATCH v2] dt-bindings: mfd: rk808: Convert bindings to yaml

["Heiko Stübner" <heiko@sntech.de>]

Hi Chris,

Am Freitag, 17. September 2021, 03:31:28 CEST schrieb Chris Morgan:
> On Thu, Sep 16, 2021 at 07:49:06PM -0500, Rob Herring wrote:
> > On Thu, 16 Sep 2021 15:19:47 -0500, Chris Morgan wrote:
> > > From: Chris Morgan <macromorgan@hotmail.com>
> > > 
> > > Convert the rk808 bindings into yaml format. Please note that currently
> > > there are a few errors that appear when performing a make dtbs_check.
> > > However, after looking at the errors it appears in most cases it occurs
> > > on device trees which are not following the current rk808.txt document
> > > today. For example for the rk808 there are multiple errors regarding
> > > vcc13-supply, vcc14-supply, and vddio-supply; however these supplies
> > > are not listed in the current driver or cared for in any way.
> > > 
> > > For the moment the rk817 is the only MFD that will support a battery.
> > > I believe the rk818 also supports a batter but I do not have one to
> > > test or write the code for. When it is supported we can split off
> > > the battery to its own document. Note that the battery is being added
> > > in a separate commit series.
> > > 
> > > Changes from V1:
> > > 
> > >  - Removed generic descriptions.
> > >  - Added maxItems to clock-output-names. Max items is 2 per the driver.
> > >  - Added unevaluatedProperties as false to regulators.
> > >  - Correct i2c node.
> > >  - Added note about the battery.
> > > 
> > > Signed-off-by: Chris Morgan <macromorgan@hotmail.com>
> > > ---
> > >  .../devicetree/bindings/mfd/rk808.txt         | 465 ------------------
> > >  .../bindings/mfd/rockchip,rk805.yaml          |  84 ++++
> > >  .../bindings/mfd/rockchip,rk808.yaml          | 253 ++++++++++
> > >  .../bindings/mfd/rockchip,rk809.yaml          |  98 ++++
> > >  .../bindings/mfd/rockchip,rk817.yaml          | 362 ++++++++++++++
> > >  .../bindings/mfd/rockchip,rk818.yaml          | 106 ++++
> > >  6 files changed, 903 insertions(+), 465 deletions(-)
> > >  delete mode 100644 Documentation/devicetree/bindings/mfd/rk808.txt
> > >  create mode 100644 Documentation/devicetree/bindings/mfd/rockchip,rk805.yaml
> > >  create mode 100644 Documentation/devicetree/bindings/mfd/rockchip,rk808.yaml
> > >  create mode 100644 Documentation/devicetree/bindings/mfd/rockchip,rk809.yaml
> > >  create mode 100644 Documentation/devicetree/bindings/mfd/rockchip,rk817.yaml
> > >  create mode 100644 Documentation/devicetree/bindings/mfd/rockchip,rk818.yaml
> > > 
> > 
> > My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check'
> > on your patch (DT_CHECKER_FLAGS is new in v5.13):
> > 
> > yamllint warnings/errors:
> > 
> > dtschema/dtc warnings/errors:
> > /builds/robherring/linux-dt-review/Documentation/devicetree/bindings/mfd/rockchip,rk808.example.dt.yaml: pmic@1b: 'vddio-supply' does not match any of the regexes: 'pinctrl-[0-9]+'
> > 	From schema: /builds/robherring/linux-dt-review/Documentation/devicetree/bindings/mfd/rockchip,rk808.yaml
> > 
> > doc reference errors (make refcheckdocs):
> > Documentation/devicetree/bindings/pinctrl/pinctrl-rk805.txt: Documentation/devicetree/bindings/mfd/rk808.txt
> > 
> > See https://patchwork.ozlabs.org/patch/1529027
> > 
> > This check can fail if there are any dependencies. The base for a patch
> > series is generally the most recent rc1.
> > 
> > If you already ran 'make dt_binding_check' and didn't see the above
> > error(s), then make sure 'yamllint' is installed and dt-schema is up to
> > date:
> > 
> > pip3 install dtschema --upgrade
> > 
> > Please check and re-submit.
> 
> What would be the best way to handle this? I can confirm that there
> is no vddio cared for in the driver, however the datasheet appears
> to show that one exists (vddio appears to be the supply for the
> BUCK1 and BUCK2 DVS voltage). Should I update the yaml to reflect what
> already exists today in the various device trees, should I only allow
> what the driver cares for today, or should I "meet halfway" and allow
> that which the datasheet permits (such as this) even if it's not
> implemented in the driver? Note that there will also be some other
> errors that are expected, because the existing device trees didn't
> always follow the previous rk808.txt file (there is at least one board
> with the wrong clock-cells value).

For devicetree in general, the hardware is always the defining element.

So a Linux "implementation detail" does not matter for a dt-binding, but
instead it should always be what the hardware datasheet says.

So if there is a vddio supply specified, it should be in the binding.
I looked it up and vddio is the "digital i/o power supply" of the chip
itself, so should definitly be specified I think :-)

Heiko