Re: [PATCH v6 4/6] clk: samsung: add Exynos ACPM clock driver

[Tudor Ambarus <tudor.ambarus@linaro.org>]

On 11/11/25 8:24 AM, Tudor Ambarus wrote:
> 
> 
> On 11/11/25 3:39 AM, Stephen Boyd wrote:
> 
> Hi, Stephen!
> 
>> Quoting Tudor Ambarus (2025-10-20 00:45:58)
>>>
>>>
>>> On 10/20/25 7:54 AM, Krzysztof Kozlowski wrote:
>>>>> diff --git a/drivers/clk/samsung/Kconfig b/drivers/clk/samsung/Kconfig
>>>>> index 76a494e95027af26272e30876a87ac293bd56dfa..70a8b82a0136b4d0213d8ff95e029c52436e5c7f 100644
>>>>> --- a/drivers/clk/samsung/Kconfig
>>>>> +++ b/drivers/clk/samsung/Kconfig
>>>>> @@ -95,6 +95,16 @@ config EXYNOS_CLKOUT
>>>>>        status of the certains clocks from SoC, but it could also be tied to
>>>>>        other devices as an input clock.
>>>>>  
>>>>> +config EXYNOS_ACPM_CLK
>>>>> +    tristate "Clock driver controlled via ACPM interface"
>>>>> +    depends on EXYNOS_ACPM_PROTOCOL || (COMPILE_TEST && !EXYNOS_ACPM_PROTOCOL)
>>>>
>>>> I merged the patches but I don't get why we are not enabling it by
>>>> default, just like every other clock driver. What is so special here?
>>>
>>> Thanks! Are you referring to the depends on line? I needed it otherwise
>>> on randconfigs where COMPILE_TEST=y and EXYNOS_ACPM_PROTOCOL=n I get:
>>>
>>> ERROR: modpost: "devm_acpm_get_by_node" [drivers/clk/samsung/clk-acpm.ko] undefined!
>>>
>>
>> I don't understand that part. The depends on statement "COMPILE_TEST &&
>> !EXYNOS_ACPM_PROTOCOL" is equivalent to COMPILE_TEST=y and
>> EXYNOS_ACPM_PROTOCOL=n, so are you trying to avoid
>> EXYNOS_ACPM_PROTOCOL=y when COMPILE_TEST=y?
> 
> My previous comment was misleading.
> The depends on line allows CONFIG_EXYNOS_ACPM_CLK to be selected in two
> main scenarios:
> 1/ if EXYNOS_ACPM_PROTOCOL is enabled the clock driver that uses it can
>    be enabled (the normal case).
> 2/ COMPILE_TEST is enabled AND EXYNOS_ACPM_PROTOCOL is NOT enabled. This
>    is the special scenario for build testing. I want to build test the
>    clock driver even if EXYNOS_ACPM_PROTOCOL is NOT enabled. For that I
>    also needed the following patch:
> 
> https://lore.kernel.org/linux-samsung-soc/20251021-fix-acpm-clk-build-test-v1-1-236a3d6db7f5@linaro.org/
> 

What I described in 2/ EXYNOS_ACPM_PROTOCOL [=n] && EXYNOS_ACPM_CLK [=y] 
can be achieved with a more relaxed:
depends on EXYNOS_ACPM_PROTOCOL || COMPILE_TEST
because of the stub (dummy method) that I referenced in the link above.

It's really what Krzysztof explained in his reply, I wanted to avoid
the link failure for COMPILE_TEST [=y] when
EXYNOS_ACPM_PROTOCOL [=m] && EXYNOS_ACPM_CLK [=y].

We have the following possibilities with:
depends on EXYNOS_ACPM_PROTOCOL || (COMPILE_TEST && !EXYNOS_ACPM_PROTOCOL)
1/ CONMPILE_TEST=n
EXYNOS_ACPM_PROTOCOL=n EXYNOS_ACPM_CLK=n
EXYNOS_ACPM_PROTOCOL=m EXYNOS_ACPM_CLK=n,m
EXYNOS_ACPM_PROTOCOL=y EXYNOS_ACPM_CLK=n,m,y

2/COMPILE_TEST=y
EXYNOS_ACPM_PROTOCOL=n EXYNOS_ACPM_CLK=n,m,y
EXYNOS_ACPM_PROTOCOL=m EXYNOS_ACPM_CLK=n,m
EXYNOS_ACPM_PROTOCOL=y EXYNOS_ACPM_CLK=n,m,y

We have the following possibilities with:
depends on EXYNOS_ACPM_PROTOCOL || COMPILE_TEST
1/ CONMPILE_TEST=n
EXYNOS_ACPM_PROTOCOL=n EXYNOS_ACPM_CLK=n
EXYNOS_ACPM_PROTOCOL=m EXYNOS_ACPM_CLK=n,m
EXYNOS_ACPM_PROTOCOL=y EXYNOS_ACPM_CLK=n,m,y

2/COMPILE_TEST=y
EXYNOS_ACPM_PROTOCOL=n EXYNOS_ACPM_CLK=n,m,y
EXYNOS_ACPM_PROTOCOL=m EXYNOS_ACPM_CLK=n,m,y <- link failure when y
EXYNOS_ACPM_PROTOCOL=y EXYNOS_ACPM_CLK=n,m,y

Thanks,
ta