Re: [Xen-devel] [RFC PATCH v2 24/25] NUMA: Move CONFIG_NUMA to common Kconfig

[Vijay Kilari <vijay.kilari@xxxxxxxxx>]

On Wed, May 31, 2017 at 4:07 PM, Jan Beulich <JBeulich@xxxxxxxx> wrote:
>>>> On 31.05.17 at 12:18, <julien.grall@xxxxxxx> wrote:
>> On 31/05/17 11:04, Jan Beulich wrote:
>>>>>> On 28.03.17 at 17:53, <vijay.kilari@xxxxxxxxx> wrote:
>>>> --- a/xen/common/Kconfig
>>>> +++ b/xen/common/Kconfig
>>>> @@ -41,6 +41,10 @@ config HAS_GDBSX
>>>>  config HAS_IOPORTS
>>>>     bool
>>>>
>>>> +config NUMA
>>>> +   def_bool y
>>>> +   depends on HAS_PDX
>>>
>>> What makes necessary this dependency?
>>
>> IIRC, this is because the numa code is using PDX helpers.
>
> Well, these helpers should have 1:1 translation equivalents for
> the non-PDX case; I don't see the need for the dependency.

PDX is necessary. Without that xen fails to compile for ARM.
IMO, there is no equivalent non-PDX support available.

As it is mandatory config, I propose to remove this dependency with
NUMA config. ok?

>
>>>> --- a/xen/drivers/acpi/Kconfig
>>>> +++ b/xen/drivers/acpi/Kconfig
>>>> @@ -4,6 +4,3 @@ config ACPI
>>>>
>>>>  config ACPI_LEGACY_TABLES_LOOKUP
>>>>     bool
>>>> -
>>>> -config NUMA
>>>> -   bool
>>>
>>> This makes clear that so far this is an option which architectures
>>> are expected to select. I think we want it to remain that way, but
>>> if we didn't you should remove the existing select(s).
>>>
>>> Also, does it really matter much whether this is under drivers/acpi/
>>> or common/? After all ACPI appears to be a prereq on ARM too.
>>
>> ACPI is not a prereq for NUMA. You can use it with Device Tree too.
>
> Oh, okay. That should be said in the commit message then.
>
> Jan
>

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel