On 03/19/2013 12:05 AM, Stephen Warren wrote:
On 03/18/2013 04:27 PM, Rob Herring wrote:
On 03/18/2013 01:11 PM, Stephen Warren wrote:
On 03/18/2013 09:50 AM, Rob Herring wrote:
On 03/13/2013 05:42 PM, Sylwester Nawrocki wrote:
Rob,
On 03/13/2013 03:39 PM, Rob Herring wrote:
I fail to see what the hack is. The order of interrupt properties must
be defined by the binding. interrupt-names is auxiliary data and must
not be required by an OS.
Is that true for all foo-names properties, or only for interrupt-names?
I was under the impression that foo-names was specifically invented so
that the order of the entries didn't matter, and instead they could be
requested by name.
I think it depends on the specific name the property is tied too. For
interrupt and reg properties which have a long history and convention,
the order should be defined. IIRC, this was Grant's position too. For
new bindings, perhaps we can be more lenient.
OK, that makes sense for interrupts/reg. Can we decide that clock-namess
are new-style and that order is not significant? I guess gpio-names too?
I guess this should be documented in whatever binding describes the core
interrupts/reg-names/gpio-names/clock-names/dma-names properties.
It certainly would be useful to have it documented somewhere. Not sure if
resource-names.txt would be a good place to have more information about the
order for each property.
_______________________________________________
devicetree-discuss mailing list
[email protected]
https://lists.ozlabs.org/listinfo/devicetree-discuss
