> Subject: Re: [PATCH v4 1/4] dt-bindings: mfd: Document RZ/G2L MTU3a > bindings > > On 11/10/2022 15:23, Biju Das wrote: > >> Subject: Re: [PATCH v4 1/4] dt-bindings: mfd: Document RZ/G2L MTU3a > >> bindings > >> > >> On 11/10/2022 10:55, Biju Das wrote: > >>> > >>>>> .../bindings/mfd/renesas,rz-mtu3.yaml | 305 > >>>> ++++++++++++++++++ > >>>>> 1 file changed, 305 insertions(+) create mode 100644 > >>>>> Documentation/devicetree/bindings/mfd/renesas,rz-mtu3.yaml > >>>> > >>>> This should not be in MFD. Just because some device has few > >> features, > >>>> does not mean it should go to MFD... Choose either timer or pwm. > >>> > >>> MFD is for multifunction device. This IP supports multiple > functions > >>> like timer, pwm, clock source/events. That is the reason I have > >> added > >>> here. MFD is core which provides register access for client > devices. > >>> > >>> For me moving it to pwm or counter is not a big problem. > >>> Why do you think it cannot be MFD? > >> > >> > >> Because it makes MFD a dump for everything where author did not > want > >> to think about real device aspects, but instead represented driver > >> design (MFD driver). > > > > Core driver is MFD, just provides resources to child devices and is > > not aware of any real device aspects. > > > >> > >> MFDs are pretty often combining unrelated features, e.g. PMICs > which > >> have wakeup and system power control, regulator, 32 kHz clocks, RTC > >> and some USB connector. > > > > Here also same right? pwm, counter and clock are 3 unrelated > features. > > That is the reason we have separate subsystems for these features. > > These are quite similar features of a similar piece of hardware. > Sometimes called timer. > > > > >> > >> Just because you will have clocksource driver, PWM driver and timer > >> driver does not make it a MFD. > > > > MFD is multi function device. > > No. MFD is a Linux subsystem name. Not a device type. The bindings are > located in respective type. > > > So are are you agreeing Clock source, PWM and timer are different > > functionalities or not? If not, why do we have 3 subsystems, if it > is > > same? > > Linux subsystems? We can have millions of them and it is not related > to bindings. OK. > > > > Where do keep these bindings as there is only single "rz_mtu" > bindings for these 3 different functionalities? > > Again, focus on hardware not on Linux drivers. Hardware is called MTU > - Multi-Function TIMER Unit. Timer. OK > > > pwm or counter or mfd? > > Not MFD. I already proposed where to put it. Other Timer/PWM/Counter > units are also in timer. > I guess for counter/pwm maintainers, it is ok to model MTU3 as a single binding "rz-mtu3" in timer that binds against counter and pwm functionalities as well?? Cheers, Biju
