On Fri, Sep 20, 2013 at 01:39:30PM +0100, Lee Jones wrote:
Personally I think the device should have sub-nodes. It's how we
dealt with our DB8500 and AB8500 and it works very well. At the end of
the day these all-in-one chips do house very different devices. I see
no harm in describing them that way. I'm sure other OSes can make use
of the sub-devices in similar ways.
For some bits of the device it makes sense - the regulators are a good
example with this device - but for others it's more questionable if the
Linux division of the device into subsystems is general enough (or even
static for Linux - like this one where there's the 32kHz clock being
handled by the RTC driver binding in the last patch posted).