[PATCH 1/3] picoxcell: support for Picochip picoxcell devices

Grant Likely grant.likely at secretlab.ca
Sun Aug 21 00:37:39 EST 2011


On Fri, Aug 19, 2011 at 11:52 PM, Jamie Iles <jamie at jamieiles.com> wrote:
> Hi Grant,
>
> On Fri, Aug 19, 2011 at 11:31:40PM -0700, Grant Likely wrote:
>> On 11-08-09 03:10 AM, Jamie Iles wrote:
> [...]
>> >+VIC required properties:
>> >+    - compatible = "arm,pl192-vic-pair".
>> >+    - interrupt-controller.
>> >+    - reg : The register bank for the devices.  Picoxcell has two VIC's and the
>> >+      IRQ outputs are wire-OR'd together so we effectively have a combined
>> >+      controller that handles 64 IRQ's.  The first reg tuple is the register
>> >+      bank of the VIC that generates IRQ's 0->31 and the second tuple handles
>> >+      IRQ's 32->63.
>>
>> If they are two separate devices, then it probably should still be
>> represented as two separate device tree nodes.  Linux can take care
>> of assigning a range of linux irq numbers to each VIC.
>
> The reason I did this is because of the entry macro get_irqnr_and_base
> macro decoding.  If I used irq_domain_generate_simple() rather than
> irq_domain_add_simple() then I guess the ordering of the VIC's in the DT
> wont matter.  I'll give this a test when I have access to hardware again
> and respin.

Don't get too hung up on irq_domain_{generate,add}_simple().  Those
functions are temporary helpers until the vic driver itself directly
uses irq_domain.

g.


More information about the devicetree-discuss mailing list