Re: [PATCH v2] sunxi-irq: Fix Kconfig dependency on GENERIC_IRQ_CHIP

[Andre Przywara]

Hi,

On 14/04/16 09:56, Suzuki K Poulose wrote:
> On 14/04/16 09:47, Suzuki K Poulose wrote:
>> On 13/04/16 21:07, Maxime Ripard wrote:
>>> Hi,
>>>
>>> On Wed, Apr 13, 2016 at 04:59:27PM +0100, Suzuki K Poulose wrote:
>>>> The sunxi-nmi and sun4i irq chips are selected by ARCH_SUNXI on both
>>>> arm
>>>> and arm64 (recently added). However on arm64, we don't explicitly
>>>> select
>>>> GENERIC_IRQ_CHIP which is required by the sunxi-nmi irq chip driver and
>>>> causes build breaks on 4.6-rc1 :
>>
>>>> diff --git a/drivers/irqchip/Makefile b/drivers/irqchip/Makefile
>>>> index b03cfcb..c9626f4 100644
>>>> --- a/drivers/irqchip/Makefile
>>>> +++ b/drivers/irqchip/Makefile
>>>> @@ -19,8 +19,8 @@ obj-$(CONFIG_CLPS711X_IRQCHIP)        +=
>>>> irq-clps711x.o
>>>>   obj-$(CONFIG_OR1K_PIC)            += irq-or1k-pic.o
>>>>   obj-$(CONFIG_ORION_IRQCHIP)        += irq-orion.o
>>>>   obj-$(CONFIG_OMAP_IRQCHIP)        += irq-omap-intc.o
>>>> -obj-$(CONFIG_ARCH_SUNXI)        += irq-sun4i.o
>>>> -obj-$(CONFIG_ARCH_SUNXI)        += irq-sunxi-nmi.o
>>>> +obj-$(CONFIG_SUNXI_IRQCHIP)        += irq-sun4i.o
>>>> +obj-$(CONFIG_SUNXI_IRQCHIP)        += irq-sunxi-nmi.o
>>>
>>> Those are two different drivers, that should be selected by different
>>> SoCs (MACH_SUN4I and MACH_SUN5I for the former, SUN6I, SUN7I and SUN8I
>>> for the latter)

But actually they aren't at the moment - and this is a post-rc1 fix, so
I'd prefer to not introduce this right now.

I can bake a proper patch for 4.7 which separates the two, if people
want this change.

>> OK. I had thought about that, but wasn't sure who needs what. So went
>> with
>> what existed already. I can respin it.

I would rather not do this right now.
Maxime, can't we just go with this patch? That seems like the proper fix
for this part of the release cycle to me.

> As a follow up question, what does the arm64 port need ? just sunxi-nmi
> ? or both ?

Currently none of them, frankly.
irq-sun4i will definitely not be needed, and I don't see the NMI
controller documented in the manual (which does not mean too much, though).

So my hunch is that we may need the nmi driver in the future, but then
we can still enable it once we merge the respective DT node or add
support for a peripheral that needs it.

Cheers,
Andre.

> I would be happy if someone who has better understanding of the
> dependencies can fix
> it up, rather than me trying to figure out all the dependencies.