| Message ID | 20240708194808.1819185-1-gnstark@salutedevices.com |
|---|---|
| Headers | show |
| Series | amlogic SoC's power-domains fixes | expand |
On 08/07/2024 21:48, George Stark wrote: > On newer SoCs, the thermal hardware can require a power domain to > operate so add corresponding property as optional by default and as > required for Meson A1 due to it's temperature sensor has dedicated > power domain. > That's again ABI break with a reason "there is power domain", so not really sufficient. Best regards, Krzysztof
On Mon, 08 Jul 2024 22:48:05 +0300, George Stark wrote: > SPI Flash Controller has dedicated power domain so make the > corresponding property required. > > Signed-off-by: George Stark <gnstark@salutedevices.com> > --- > Documentation/devicetree/bindings/spi/amlogic,a1-spifc.yaml | 4 ++++ > 1 file changed, 4 insertions(+) > My bot found errors running 'make dt_binding_check' on your patch: yamllint warnings/errors: dtschema/dtc warnings/errors: /builds/robherring/dt-review-ci/linux/Documentation/devicetree/bindings/spi/amlogic,a1-spifc.example.dtb: spi@fd000400: Unevaluated properties are not allowed ('power-domains' was unexpected) from schema $id: http://devicetree.org/schemas/spi/amlogic,a1-spifc.yaml# doc reference errors (make refcheckdocs): See https://patchwork.ozlabs.org/project/devicetree-bindings/patch/20240708194808.1819185-2-gnstark@salutedevices.com The base for the series is generally the latest rc1. A different dependency should be noted in *this* patch. 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 after running the above command yourself. Note that DT_SCHEMA_FILES can be set to your schema file to speed up checking your schema. However, it must be unset to test all examples with your schema.
Hi, On 08/07/2024 21:48, George Stark wrote: > Here's some fixes to the bindings and device tree related to Amlogic A1 SoC. > The SoC provides dedicated power domain for for almost all periphery. > > George Stark (4): > dt-bindings: spi: amlogic,a1-spifc: make power-domains required > dt-bindings: i2c: amlogic,meson6-i2c: add optional power-domains > dt-bindings: thermal: amlogic,thermal: add power-domains > arm64: dts: meson: a1: bind power domain to temperature sensor > > .../bindings/i2c/amlogic,meson6-i2c.yaml | 3 +++ > .../devicetree/bindings/spi/amlogic,a1-spifc.yaml | 4 ++++ > .../bindings/thermal/amlogic,thermal.yaml | 14 ++++++++++++++ > arch/arm64/boot/dts/amlogic/meson-a1.dtsi | 1 + > 4 files changed, 22 insertions(+) > > -- > 2.25.1 > You can drop patch 1 & drop required on patch 3, and it will be good to go. Thanks, Neil
Hello Krzysztof, Neil, Rob Thanks for your reviews. I'd just like to clarify things a little on that matter. On A1 most of periphery has either dedicated power domain (ir, spifc, spi, cpu etc) or several peripheries share single PD e.g. pwm, temp sensor, i2c, adc share PD 12. Appropriate PD must be turned on before the periphery is touched and be on while periphery is used. Experiments confirm it. So power-domains property in dts nodes is really required. It makes kernel to control PDs when bound periphery is used. If I understand bindings thing correctly we should make power-domains property required in bindings too in that case, at least for those peripheries which has *a1* compatibles (some peripheries use compatible from older SoCs without PDs). What you think? On 7/8/24 22:51, Krzysztof Kozlowski wrote: > On 08/07/2024 21:48, George Stark wrote: >> SPI Flash Controller has dedicated power domain so make the >> corresponding property required. > > That's an ABI break. Being part of power domain does not necessarily > mean it must be required, so your commit msg lacks rationale. Especially > on the ABI part. > > > > Best regards, > Krzysztof >
On 10/07/2024 18:23, George Stark wrote: > Hello Krzysztof, Neil, Rob > > Thanks for your reviews. > > I'd just like to clarify things a little on that matter. On A1 most of > periphery has either dedicated power domain (ir, spifc, spi, cpu etc) or > several peripheries share single PD e.g. pwm, temp sensor, i2c, adc > share PD 12. Appropriate PD must be turned on before the periphery is > touched and be on while periphery is used. Experiments confirm it. So > power-domains property in dts nodes is really required. It makes kernel > to control PDs when bound periphery is used. > > If I understand bindings thing correctly we should make power-domains > property required in bindings too in that case, at least for those > peripheries which has *a1* compatibles (some peripheries use compatible > from older SoCs without PDs). > > What you think? Yes, but if the compatible has already been defined without a required power-domain, then it's too late to mark it as required since it's an ABI break. So next time a new compatible is added, it's the appropriate time to mark it as required. But in any case it's not a big deal because if the power-domain is not specified it won't work, like if you use wrong register address or the wrong interrupt line. The fact power-domain can be specified is enough. Neil > > On 7/8/24 22:51, Krzysztof Kozlowski wrote: >> On 08/07/2024 21:48, George Stark wrote: >>> SPI Flash Controller has dedicated power domain so make the >>> corresponding property required. >> >> That's an ABI break. Being part of power domain does not necessarily >> mean it must be required, so your commit msg lacks rationale. Especially >> on the ABI part. >> >> >> >> Best regards, >> Krzysztof >> >