[PATCH] reset: Correct dependencies for RESET_RZG2L_USBPHY_CTRL

[Marek Vasut]

On 3/21/26 6:10 PM, Tom Rini wrote:
> On Sat, Mar 21, 2026 at 03:23:35AM +0100, Marek Vasut wrote:
>> On 3/20/26 2:56 PM, Tom Rini wrote:
>>> On Fri, Mar 20, 2026 at 04:58:13AM +0100, Marek Vasut wrote:
>>>> On 3/19/26 2:41 PM, Tom Rini wrote:
>>>>
>>>> [...]
>>>>
>>>>>>> So I think the answer is if you fire up menuconfig, and turn off
>>>>>>> RESET_RZG2L_USBPHY_CTRL and REGULATOR_RZG2L_USBPHY and DM_REGULATOR and
>>>>>>> then enable RESET_RZG2L_USBPHY_CTRL things will become mad because
>>>>>>> RESET_RZG2L_USBPHY_CTRL has select'd REGULATOR_RZG2L_USBPHY but we don't
>>>>>>> have DM_REGULATOR.
>>>>>> How is it possible to enable RESET_RZG2L_USBPHY_CTRL without enabling its
>>>>>> dependencies ? This is what I do not understand.
>>>>>
>>>>> The way I said in the part you quoted? If you don't like the way I'm
>>>>> solving the problem with the dependency graph here please propose an
>>>>> alternate. But it is possible to encounter this problem, or randconfig
>>>>> wouldn't expose it.
>>>> What I do not understand is, how is it possible to enable Kconfig symbol
>>>> without enabling its dependencies. Is that a bug in Kconfig ?
>>>
>>> There is nothing ensuring that when you enable RESET_RZG2L_USBPHY_CTRL
>>> today that DM_REGULATOR is enabled. But since you're telling it to
>>> select REGULATOR_RZG2L_USBPHY it then will give an error.
>> Hmmm, I see it now. If I have both CONFIG_REGULATOR_RZG2L_USBPHY=y and
>> CONFIG_RESET_RZG2L_USBPHY_CTRL=y in .config, and I delete
>> CONFIG_DM_REGULATOR=y from .config , then I can select CONFIG_DM_REGULATOR=n
>> .
> 
> Right.
> 
>> But isn't the right fix the elimination of 'select' ? Linux Kconfig
>> documentation now hints that 'select' should only be used for hidden
> 
> As a much longer term goal, yes, this is largely where we should go. But
> we're still at the point where I'm catching these kind of dependency
> issues in the Kconfig side of things, let alone build time problems.
> 
>> symbols. So how about this fix instead , that should retain the transitive
>> dependency on REGULATOR_RZG2L_USBPHY , which depends on DM_REGULATOR :
>>
>> "
>> diff --git a/drivers/reset/Kconfig b/drivers/reset/Kconfig
>> index 2fd91d6299c..a7acd5cfa69 100644
>> --- a/drivers/reset/Kconfig
>> +++ b/drivers/reset/Kconfig
>> @@ -239,8 +239,7 @@ config RESET_AT91
>>
>>   config RESET_RZG2L_USBPHY_CTRL
>>          bool "Enable support for Renesas RZ/G2L USB 2.0 PHY control"
>> -       depends on DM_RESET
>> -       select REGULATOR_RZG2L_USBPHY
>> +       depends on DM_RESET && REGULATOR_RZG2L_USBPHY
>>          help
>>            Enable support for controlling USB 2.0 PHY resets on the Renesas
>>            RZ/G2L SoC. This is required for USB 2.0 functionality to work on
> 
> This is fine too, yes
Excellent, can you please do a V2 ? I'll RB it then.

Thank you !