On Tue, Nov 04, 2025 at 03:48:10PM +0100, Danilo Krummrich wrote:
> On Tue Nov 4, 2025 at 3:35 PM CET, Greg Kroah-Hartman wrote:
> > On Tue, Nov 04, 2025 at 11:04:49PM +0900, Alexandre Courbot wrote:
> >> diff --git a/drivers/base/firmware_loader/Kconfig 
> >> b/drivers/base/firmware_loader/Kconfig
> >> index 752b9a9bea03..15eff8a4b505 100644
> >> --- a/drivers/base/firmware_loader/Kconfig
> >> +++ b/drivers/base/firmware_loader/Kconfig
> >> @@ -38,7 +38,7 @@ config FW_LOADER_DEBUG
> >>  config RUST_FW_LOADER_ABSTRACTIONS
> >>    bool "Rust Firmware Loader abstractions"
> >>    depends on RUST
> >> -  depends on FW_LOADER=y
> >> +  select FW_LOADER
> >
> > Please no, select should almost never be used, it causes hard-to-debug
> > issues.
> 
> I agree that select can be very annoying at times, but in this case it seems 
> to
> be the correct thing to do?
> 
> For instance for something like:
> 
>       config MY_DRIVER
>               depends on PCI
>               depends on DRM
>               select AUXILIARY_BUS
>               select FW_LOADER
> 
> In this case MY_DRIVER is only available if PCI and DRM is enabled, which 
> makes
> sense, there is no reason to show users PCI and DRM drivers if both are
> disabled.
> 
> However, for things like AUXILIARY_BUS and FW_LOADER, I'd argue that they are
> implementation details of the driver and should be selected if the driver is
> selected.
> 
> Otherwise, wouldn't we expect users to know implementation details of drivers
> before being able to select them?

Ah, good point, I guess this is something like a "feature" that a driver
needs to work properly.  Ok, no objection from me (other than agreeing
that it needs to be split up as you already said.)

thanks,

greg k-h

Reply via email to