RE: [PATCH 1/4] dt-bindings: mfd: nxp,bbnsm: Add binding for nxp bbnsm

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



> Subject: Re: [PATCH 1/4] dt-bindings: mfd: nxp,bbnsm: Add binding for nxp
> bbnsm
> 
> On 21/11/2022 11:26, Jacky Bai wrote:
> >> Subject: Re: [PATCH 1/4] dt-bindings: mfd: nxp,bbnsm: Add binding for
> >> nxp bbnsm
> >>
> >> On 21/11/2022 07:51, Jacky Bai wrote:
> >>> Add binding for NXP BBNSM(Battery-Backed Non-Secure Module).
> >>>
> >>> Signed-off-by: Jacky Bai <ping.bai@xxxxxxx>
> >>> ---
> >
> > ...
> >
> >>> +
> >>> +    properties:
> >>> +      compatible:
> >>> +        const: nxp,bbnsm-rtc
> >>
> >>
> >> Missing ref to rtc.yaml.
> >
> > Ok will include in v2.
> >>
> >>> +
> >>> +      regmap:
> >>
> >> Use vendor prefix, descriptive name and description. Where is the
> >> type of 'regmap' defined?
> >
> > Type is missed. Will add a description and type define if necessary.
> >
> >>
> >>> +        maxItems: 1
> >>
> >> I don't think this is correct. Rob explained the simple-mfd means
> >> children
> > do
> >> not depend on anything from the parent, but taking a regmap from its
> > parent
> >> contradicts it.
> >
> > For this BBNSM module, basically, it provides two sperate & different
> > function: RTC and ON/OFF button control. But no separate register
> > offset range for each of these functions. For example, the interrupt
> > enable control, Interrupt status and basic function control are mixed
> > in the same registers'
> > different bit.
> >
> > Any good suggestion on how to handle such case? ^_^
> 
> The solution for more complex common parts, dedicated device driver (MFD
> driver) with its own binding. However I understand why it would be overshoot
> here.
>

Is it ok to keep current implementation rather than reimplement a new dedicate MFD wrapper driver?

BR
> >
> >>
> >>> +
> >>> +      interrupts:
> >>> +        maxItems: 1
> >>
> >> You have same interrupt and same address space used by two devices.
> >>
> >> Both arguments (depending on parent regmap, sharing interrupt)
> >> suggests that this is one device block, so describing it with
> >> simple-mfd is quite unflexible.
> >>
> >
> > It is depends on how SoC integrates this BBNSM module. From the BBNSM
> > side, it has separate IRQ lines for RTC function and ON/OFF function.
> > Different IRQ lines can be used for RTC and ON/OFF button. But in
> > current i.MX93 SoC, those interrupts are ORed together at SoC level.
> > That's why same interrupt in the example.
> 
> It's fine then.
> 
> Best regards,
> Krzysztof





[Index of Archives]     [Linux Media Devel]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [Linux Wireless Networking]     [Linux Omap]

  Powered by Linux