Re: [RESEND PATCH 2/4] cpufreq: amd-pstate: Add test module for amd-pstate driver

[Shuah Khan <skhan@linuxfoundation.org>]

On 8/28/22 18:09, Huang Rui wrote:
> On Sat, Aug 27, 2022 at 06:36:30AM +0800, Shuah Khan wrote:
>> On 8/26/22 3:47 PM, Shuah Khan wrote:
>>> On 8/16/22 9:46 PM, Meng Li wrote:
>>>> Add amd-pstate-ut test module, this module is used by kselftest
>>>> to unit test amd-pstate functionality. This module will be
>>>> expected by some of selftests to be present and loaded.
>>>>
>>>> Signed-off-by: Meng Li <li.meng@amd.com>
>>>> Acked-by: Huang Rui <ray.huang@amd.com>
>>>> Reviewed-by: Shuah Khan <skhan@linuxfoundation.org>
>>>> ---
>>>>    drivers/cpufreq/Kconfig.x86     |   7 +
>>>>    drivers/cpufreq/Makefile        |   1 +
>>>>    drivers/cpufreq/amd-pstate-ut.c | 293 ++++++++++++++++++++++++++++++++
>>>>    3 files changed, 301 insertions(+)
>>>>    create mode 100644 drivers/cpufreq/amd-pstate-ut.c
>>>>
>>>> diff --git a/drivers/cpufreq/Kconfig.x86 b/drivers/cpufreq/Kconfig.x86
>>>> index 55516043b656..fdd819069d72 100644
>>>> --- a/drivers/cpufreq/Kconfig.x86
>>>> +++ b/drivers/cpufreq/Kconfig.x86
>>>> @@ -51,6 +51,13 @@ config X86_AMD_PSTATE
>>>>          If in doubt, say N.
>>>> +config X86_AMD_PSTATE_UT
>>>> +    tristate "selftest for AMD Processor P-State driver"
>>>> +    depends on X86 && ACPI_PROCESSOR
>>
>> This has to specify dependency on X86_AMD_PSTATE
>>
>>>> +    default n
>>>> +    help
>>>> +      This kernel module is used for testing. It's safe to say M here.
>>>> +
>>>
>>> Shouldn't this X86_AMD_PSTATE_UT depend on X86_AMD_PSTATE?
>>> I am running a few tests and when I have X86_AMD_PSTATE_UT
>>> enabled as built-in and X86_AMD_PSTATE is disabled, test
>>> fails saying incorrect cpufreq driver?
>>>
>>> Skipped: Test can only run on amd-pstate driver.
>>>
>>> So it sounds like X86_AMD_PSTATE_UT depends on X86_AMD_PSTATE.
>>>
>>
>> Once I enabled X86_AMD_PSTATE and X86_AMD_PSTATE_UT=m, the test
>> ran correctly.
>>
>> Please fix the dependencies and send me a new version. Send all
>> the patches so it is easier to apply them all at once.
>>
> 
> Hi Shuah,
> 
> Thanks to accept the patch series. Actually, we want amd-pstate-ut can be
> modprobed even without amd-pstate module, because it can tell the users
> they loaded a wrong module like "acpi-cpufreq" and needs to replace it with
> amd-pstate. :-)
> 

I see. The message could be expanded to let the user know which
config option needs to be enabled to run the test.

You can send this change as a separate patch. I will apply this series
as is.

thanks,
-- Shuah