On Tue, Feb 9, 2016 at 8:24 PM, Jassi Brar <jassisinghbrar@xxxxxxxxx> wrote:
> On Tue, Feb 9, 2016 at 6:01 PM, Nishanth Menon <nm@xxxxxx> wrote:
>> On 02/08/2016 10:14 PM, Jassi Brar wrote:
>>
>>>> +
>>> I think we should get rid of consumer specifics from the provider node...
>>
>>
>> If I get rid of the consumer nodes, how do you propose I describe the rx
>> queue interrupt(s) in the msmgr dt node (Every Rx queue will have it's
>> own interrupt - and it cannot be reverse computed from queue ID, proxy ID)?
>>
> One option is to have controller driver construct interrupt name from
> queue and proxy ids like
>
> msgmgr: msgmgr@02a00000 {
> ....
> interrupt-names = "irq_5_2", "irq_0_0"; /* irq_<queue-id>_<proxy-id> */
> interrupts = <GIC_SPI 324 IRQ_TYPE_EDGE_RISING>,
> <GIC_SPI 325 IRQ_TYPE_EDGE_RISING>;
> }
>
> and have the consumer specify queue and proxy ids in mboxes property like
> pmmc {
> ....
> mbox-names = "tx", "rx";
> mboxes = <&msgmgr 0 0>
> <&msgmgr 5 2>;
> };
>
However if the queue+proxy+interrupt tuple is a hard property of a
channel (which it seems to me now), then probably your original scheme
of chile node phandle is just as fine.
Thanks
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
