On Fri, 2016-08-19 at 09:36 -0500, Rob Herring wrote:

On Fri, Aug 19, 2016 at 10:14:10PM +0930, Andrew Jeffery wrote:

quoted

Signed-off-by: Andrew Jeffery <redacted>
---

Since v1:

Rob: I haven't added your Acked-by here as I've made the following changes and
wanted to get your input:

* Remove interrupt-controller as an optional property
* Defer to interrupt-controller bindings document for sub-node properties

I had a discussion with Joel about whether the interrupt-controller capability
should be optional and the conclusion was that it should always be configured
by the driver. This makes an optional interrupt-controller property feel
redundant (and possibly inaccurate if left out) so I've removed it.

I don't follow. What do you mean byt "configured by the driver". If the 
block supports interrupts, then it should be marked as an 
interrupt-controller. It never should have been optional. The OS can 
ignore the interrupt properties if it chooses. 

Right, clearly there was some confusion on my part. I will fix that up.
Thanks for clarifying.

Andrew