[PATCH 10/11] irqchip / GICv2 / ACPI: Consolidate GICv2 ACPI related init code

[hanjun.guo@linaro.org (Hanjun Guo)]

Hi Marc,

On 06/11/2015 12:29 AM, Marc Zyngier wrote:
> On 21/05/15 15:27, Hanjun Guo wrote:
>> On 2015?05?21? 04:44, Tomasz Nowicki wrote:
>>> Hi Hanjun,
>>>
>>> On 05/18/2015 02:59 PM, Hanjun Guo wrote:
>>>> Move GICv2 ACPI related init code in irq-gic.c to irq-gic-acpi.c,
>>>> this can make the ACPI related GIC init code slef-contained.
>>>>
>>>> Introduce set_acpi_core_irqdomain() to set acpi_irqdomain then
>>>> it will be no need to make gic_data[] as a global value, and
>>>> it will save the confilcts with GICv3's gic_data in the later
>>>> patch.
>>>>
>>>> acpi_gic_parse_distributor() have the same function as
>>>> gic_acpi_parse_madt_distributor() to get the GIC distributor
>>>> physical base address, so just remove the duplicate one, and
>>>> only get the GIC version when it is unknown.
>>>>
>>>> Signed-off-by: Hanjun Guo <hanjun.guo@linaro.org>
>>>> ---
>>>>    drivers/irqchip/irq-gic-acpi.c       |  95
>>>> +++++++++++++++++++++++++++++++-
>>>>    drivers/irqchip/irq-gic.c            | 103
>>>> +----------------------------------
>>>>    include/linux/irqchip/arm-gic-acpi.h |   5 ++
>>>>    3 files changed, 101 insertions(+), 102 deletions(-)
>>>>
>>>> diff --git a/drivers/irqchip/irq-gic-acpi.c
>>>> b/drivers/irqchip/irq-gic-acpi.c
>>>> index 1388d9e..8463e48 100644
>>>> --- a/drivers/irqchip/irq-gic-acpi.c
>>>> +++ b/drivers/irqchip/irq-gic-acpi.c
>>>> @@ -13,12 +13,16 @@
>>>>
>>>>    #include <linux/acpi.h>
>>>>    #include <linux/init.h>
>>>> +#include <linux/irqchip/arm-gic.h>
>>>>    #include <linux/irqchip/arm-gic-acpi.h>
>>>>    #include <linux/irqchip/arm-gic-v3.h>
>>>
>>> arm-gic.h and arm-gic-v3.h describe register map for respective drivers
>>> and should be used separately within parent driver only.
>>
>> Seems that there is no duplicate macros in that two head
>> file, but yes, it will confuse people.
>>
>> Consolidating all ACPI GIC code is an improvement to make
>> ACPI related code self-contained, but also have some drawbacks,
>>
>> Marc, what do you think?
>
> What I think is "Over my dead body".
>
> These include files are private to the respective interrupt controller
> code, and the only reason they are in linux/irqchip is because the
> corresponding KVM support code uses them too.

So for this patch, we are totally in the wrong direction, I will
remove the last two patches in next version.

Thanks for your constructive comments, much appreciated :)

Thanks
Hanjun