[PATCH RFC 2/5] dt-bindings: soc: Describe the BCM6846 Processor Monitor Bus

[Krzysztof Kozlowski]

On 19/10/2026 16:03, Linus Walleij wrote:
> On Fri, Oct 16, 2026 at 4:20 PM Krzysztof Kozlowski <krzk at kernel.org> wrote:
> 
>>> +    /* Called PMB Processor Monitor Bus in vendor tree */
>>> +    bus at ffb20100 {
>>> +        compatible = "brcm,bcm6846-pmb";
>>> +        reg = <0xffb20100 0x200>;
>>> +
>>> +        reset-controller {
>>> +            compatible = "brcm,bcm6846-pmb-reset";
>>> +            #reset-cells = <1>;
>>
>> Anyway this tells - no resources here. This must be part of the parent
>> node, not a separate device.
> 
> It so happens that this bus (PMB) has several children just not defined
> in this series. (thermal, clocks, power domains).
> 
> One of the children is the thermal monitor, which in turn is using a
> reset from the reset-controller node, to reset just that hardware
> which is under the PMB.
> 
> At this point if we put both #reset-cells and #thernal-sensor-cells
> in the same PMB root node, the reset resource would come from
> one part of the PMB to another part.

This would not be a problem but hint that these are the same device and
we try not to encode device internals in DT, at least for simple devices.

> 
> They really are different devices under PMB bus, but I'll just go
> and implement thermal sensing in the same patch series so
> that it becomes obvious what is going on, why not.

If they are different devices then it is fine. Usually people cannot
actually distinguish different devices and they represent driver model
in DT. Bindings are also supposed to be complete, which then avoids such
questions/comments as mine. Commit msg or binding description should
clarify such points.

Best regards,
Krzysztof