On Wed, Sep 09, 2026 at 07:25:38AM +0200, Peter Rosin wrote:
> Den Tue, Sep 08, 2026 at 06:47:01PM +0100, skrev Conor Dooley:
> > On Tue, Sep 08, 2026 at 06:01:45PM +0530, Shiv Prakash Gupta wrote:
> > > Add device tree binding schema for the NXP PCA9641 2-to-1 I2C bus
> > > master arbiter.
> > > 
> > > The PCA9641 arbitrates between two upstream I2C masters competing for a
> > > single downstream slave bus using a lock/grant ownership model. The
> > > binding supports an optional 'interrupts' property for interrupt-assisted
> > > arbitration.
> > > 
> > > Signed-off-by: Shiv Prakash Gupta <shivprakash.gupta@nxp.com>
> > > ---
> 
> ...
> 
> > > +  i2c-arb:
> > > +    type: object
> > > +    $ref: /schemas/i2c/i2c-controller.yaml
> > > +    unevaluatedProperties: false
> > > +    description:
> > > +      I2C bus node representing the downstream slave bus controlled by the
> > > +      PCA9641. Downstream slave devices are declared as child nodes here.
> > 
> > Why does this node need to exist? Can't the i2c-arbiter node itself
> > contain the child devices?
> 
> For symmetry with the pre-existing I2C arbs, which all have the extra
> node for historical implementation reasons? I don't know if that's
> considered a valid reason? I prefer to have all arbs follow the same
> structure...

If this is just how these devices are described in general, I think
that's probably fine.

Acked-by: Conor Dooley <conor.dooley@microchip.com>
pw-bot: not-applicable

(Also, good to see you back active)

Cheers.,
Conor.

> 
> The "documentation" describing common I2C arbitration details went missing
> here:
> 
> https://lore.kernel.org/all/20230731163833.319258-1-krzysztof.kozlowski@linaro.org/
> 
> Cheers,
> Peter