Re: [PATCH] ARM: tegra: add DT binding for Tegra186 BPMP I2C

[Stephen Warren <swarren-3lzwWm7+Weoh9ZMKESR00Q@public.gmane.org>]

On 07/16/2016 03:07 PM, Rob Herring wrote:
> On Thu, Jul 07, 2016 at 01:37:11PM -0600, Stephen Warren wrote:
>> From: Stephen Warren <swarren-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
>>
>> In Tegra186, the BPMP (Boot and Power Management Processor) owns certain
>> HW devices, such as the I2C controller for the power management I2C bus.
>> Software running on other CPUs must perform IPC to the BPMP in order to
>> execute transactions on that I2C bus. This binding describes an I2C bus
>> that is accessed in such a fashion.
>>
>> Signed-off-by: Stephen Warren <swarren-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>
>> ---
>>   .../bindings/i2c/nvidia,tegra186-bpmp-i2c.txt      | 35 ++++++++++++++++++++++
>>   1 file changed, 35 insertions(+)
>>   create mode 100644 Documentation/devicetree/bindings/i2c/nvidia,tegra186-bpmp-i2c.txt
>>
>> diff --git a/Documentation/devicetree/bindings/i2c/nvidia,tegra186-bpmp-i2c.txt b/Documentation/devicetree/bindings/i2c/nvidia,tegra186-bpmp-i2c.txt
>> new file mode 100644
>> index 000000000000..eb9f70723ab7
>> --- /dev/null
>> +++ b/Documentation/devicetree/bindings/i2c/nvidia,tegra186-bpmp-i2c.txt
>> @@ -0,0 +1,35 @@
>> +NVIDIA Tegra186 BPMP I2C controller
>> +
>> +In Tegra186, the BPMP (Boot and Power Management Processor) owns certain HW
>> +devices, such as the I2C controller for the power management I2C bus. Software
>> +running on other CPUs must perform IPC to the BPMP in order to execute
>> +transactions on that I2C bus. This binding describes an I2C bus that is
>> +accessed in such a fashion.
>> +
>> +Required properties:
>> +- compatible:
>> +    Array of strings.
>> +    One of:
>> +    - "nvidia,tegra186-bpmp-i2c".
>> +- address-cells: Address cells for I2C device address.
>> +    Single-cell integer.
>> +    Must be <1>.
>> +- size-cells:
>> +    Single-cell integer.
>> +    Must be <0>.
>> +- nvidia,bpmp:
>> +    The phandle to the BPMP device.
>
> Any reason to not make this a sub-node of the BPMP device?

That would be possible too.

My thought process was along the lines of: The system has an I2C bus, 
which deserves a DT node. That fact seemed more important than the 
access mechanism; the fact it's accessed via BPMP rather than direct 
register access felt a bit more like an implementation detail. Still, I 
suppose we could flip it around and store the node underneath the BPMP 
if you want; let me know.