Re: [PATCH 6/9] x86/intel_pstate: the main boby of the intel_pstate driver

[Julien Grall <julien.grall@citrix.com>]

On 26/04/2015 05:32, Wang, Wei W wrote:
> Hi Julien,

Hi Wei,

> On 24/04/2015 20:57, Julien Grall wrote
>> On 23/04/2016 18:58, Wei Wang wrote:
>>> diff --git a/xen/include/acpi/cpufreq/processor_perf.h
>>> b/xen/include/acpi/cpufreq/processor_perf.h
>>> index d8a1ba6..ebff11d 100644
>>> --- a/xen/include/acpi/cpufreq/processor_perf.h
>>> +++ b/xen/include/acpi/cpufreq/processor_perf.h
>>> @@ -7,6 +7,7 @@
>>>
>>>    #define XEN_PX_INIT 0x80000000
>>>
>>> +int intel_pstate_init(void);
>>
>> The intel pstate driver is x86 specific. Although xen/include/acpi contains
>> common headers for common code.
>
> Thanks for your comments. But I saw "int powernow_cpufreq_init(void);" is put there.

FWIW, this prototype doesn't have any implementation even on x86.

While currently some drivers (such as the x86 powernow) may define 
prototype in the common header. This is wrong, the common code should 
not be able to call those functions.

There is an ongoing support on ACPI for ARM (an RFC has been sent a 
couple of months ago). Adding new x86 prototype in this directory 
complicate the splitting. Please help us to at least avoid adding new 
x86 specific prototype/code in the common code when it's possible.

We will take care of moving the current x86 prototype/code in the 
arch-specific directories.

Although, I'm not a maintainer. They may have a different opinion on 
this point.

Regards,

-- 
Julien Grall