Hi Arnd,
On 14 May 2014 10:26, Arnd wrote:
> On Wednesday 14 May 2014 09:13:45 Phil Edworthy wrote:
> > > Because of those differences I've pushed the irq properties of the
> > > binding into the SoC specific part.
> > > If we can enumerate all irq signals from the DW core we could maybe
> push
> > > this stuff back into common doc/code, but I'm not sure about how to
> > > handle this without breaking the Exynos binding. i.MX won't care as the
> > > only irq it uses besides the legacy irqs, which are mapped through DT,
> > > is now a named irq.
> > I don’t have a lot of experience with DT bindings, though it appears as
> > though it would be good if we could specify interrupts by their name,
> > rather than the order in which they are listed. Of course, that would add
> > some overhead to DT parsing.
>
> In general, it's preferred to specify a particular order and use that
> as the primary identification, with the names being an additional way
> to identify them. This is done for historic reasons: traditional OF
> does not have an interrupt-names property, and requires the order to
> be fixed. The names are a Linux extension originally added to allow
> platform drivers to keep working when they already used named interrupts.
Thanks for the clarification, I wondered why we had the interrupt-names
property.
> > I can imagine this sort of problem happening with lots of other IP blocks.
>
> It's usually not as bad, but there are other IP blocks that are confused
> e.g. about the number of clocks that get routed into them, based on
> which documentation (or vendor sourcecode) you read.
>
> The other problem with IntA/IntB/IntC/IntD on PCI is you can wire
> parallel PCI devices to have their interrupts connected directly to
> the GIC. With PCIe devices this does not happen (the interrupts are
> messages sent to the host bridge), but nothing prevents a board
> design from adding a PCIe-to-PCI bridge chip with hardwired PCI
> devices or actual slots. This means we have to describe them in
> the interrupt-map property that can actually handle them.
Phil
��.n��������+%������w��{.n�����{���"�)�� jg��� �����ݢj����G���� ���j:+v���w�m������w�������h�����٥
