Re: [PATCH] ASoC: Intel: boards: fix unmet dependency on PINCTRL

[Arnd Bergmann]

On Sun, Mar 22, 2026, at 20:48, Julian Braha wrote:
> Currently, SND_SOC_INTEL_SOUNDWIRE_SOF_MACH selects PINCTRL_CS42L43
> without also selecting or depending on PINCTRL, despite PINCTRL_CS42L43
> depending on PINCTRL.
>
> See the following Kbuild warning:
>
> WARNING: unmet direct dependencies detected for PINCTRL_CS42L43
>   Depends on [n]: PINCTRL [=n] && MFD_CS42L43 [=m]
>   Selected by [m]:
>   - SND_SOC_INTEL_SOUNDWIRE_SOF_MACH [=m] && SOUND [=y] && SND [=m] && 
> SND_SOC [=m] && SND_SOC_INTEL_MACH [=y] && (SND_SOC_SOF_INTEL_COMMON 
> [=m] || !SND_SOC_SOF_INTEL_COMMON [=m]) && SND_SOC_SOF_INTEL_SOUNDWIRE 
> [=m] && I2C [=y] && SPI_MASTER [=y] && ACPI [=y] && (MFD_INTEL_LPSS 
> [=n] || COMPILE_TEST [=y]) && (SND_SOC_INTEL_USER_FRIENDLY_LONG_NAMES 
> [=n] || COMPILE_TEST [=y]) && SOUNDWIRE [=m]
>
> This unmet dependency was detected by kconfirm, a static analysis
> tool for Kconfig.
>
> Fixes: c073f0757663 ("ASoC: Intel: sof_sdw: select PINCTRL_CS42L43 and 
> SPI_CS42L43")
> Signed-off-by: Julian Braha <julianbraha@xxxxxxxxx>

I don't see any compile-time dependency between sof_sdw and
the PINCTRL_CS42L43 driver and I wonder if it would be better
to instead revert that patch.

It is generally a bad idea for an individual device driver
to force-enable another subsystem or a driver from another
subsystem, as that increases the risk for dependency loops
and makes it harder to turn subsystems off entirely.

       Arnd