On 21/04/2023 11:58, Varadarajan Narayanan wrote: > On Mon, Apr 17, 2023 at 10:05:11AM +0200, Johan Hovold wrote: >> On Thu, Apr 06, 2023 at 09:41:49AM +0200, Krzysztof Kozlowski wrote: >>> On 05/04/2023 13:41, Varadarajan Narayanan wrote: >>>> Add dt-bindings for USB3 PHY found on Qualcomm IPQ9574 >>>> >>>> Signed-off-by: Varadarajan Narayanan <quic_varada@xxxxxxxxxxx> >>>> --- >>>> Changes in v8: >>>> - Update clock names for ipq9574 >>>> >>>> Changes in v6: >>>> - Made power-domains optional >>>> >>>> Note: In the earlier patch sets, had used the (legacy) >>>> specification available in qcom,msm8996-qmp-usb3-phy.yaml. Moved >>>> to newer specification in qcom,sc8280xp-qmp-usb3-uni-phy.yaml >>>> --- >>>> .../phy/qcom,sc8280xp-qmp-usb3-uni-phy.yaml | 43 +++++++++++++++++++--- >>>> 1 file changed, 37 insertions(+), 6 deletions(-) >> >>>> + clock-names: >>>> + items: >>>> + - const: aux >>>> + - const: ref >>>> + - const: com_aux >>> >>> Can anyone explain me why do we name these (here and other Qualcomm >>> bindings) based on clock name, not input? Just because different clock >>> is fed to the block, does not necessarily mean the input should be named >>> differently. >> >> I guess part of the answer is that this has just been copied from the >> vendor dts and (almost) no one but Qualcomm has access to the >> documentation. What would the input names be here? >> >> Also note that there are SoCs that enable both 'cfg_ahb' and 'com_aux' >> (e.g. sc7180). > > The clock name definitions are auto-generated based on the clock > tree definitions provided by the h/w team. We followed the naming > pattern done in the previous SoCs. Are you sure? We talk about clock inputs here. Best regards, Krzysztof
