Re: [PATCH] fbdev: make FB_BACKLIGHT a tristate

[Arnd Bergmann <arnd@arndb.de>]

On Thu, Oct 11, 2018 at 3:16 AM Rob Clark <robdclark@gmail.com> wrote:
>
> On Wed, Oct 10, 2018 at 11:35 AM Arnd Bergmann <arnd@arndb.de> wrote:
> >
> > On 10/10/18, Rob Clark <robdclark@gmail.com> wrote:
> > > BACKLIGHT_CLASS_DEVICE is already tristate, but a dependency
> > > FB_BACKLIGHT prevents it from being built as a module.  There
> > > doesn't seem to be any particularly good reason for this, so
> > > switch FB_BACKLIGHT over to tristate.
> > >
> > > Signed-off-by: Rob Clark <robdclark@gmail.com>
> >
> > I don't see anything immediately wrong, but anything related to
> > BACKLIGHT_CLASS_DEVICE, BACKLIGHT_LCD_SUPPORT
> > and FB_BACKLIGHT is really fragile in Kconfig, because of the
> > way those interact with other options.
> >
> > I've applied your patch to my randconfig build tree for testing,
> > let's see what happens there before you apply it.
> >
>
> thanks.. tbh the fragility of backlight vs kconfig is why I've
> procrastinated on fixing this for a while.. in the end the solution
> seems not as bad as I feared (and after a iteration or two makes rhel
> kernel builds for various archs happy, at least).. but defn some
> randconfig build testing is in order.

Ok, my overnight testing shows no regression, so please add

Tested-by: Arnd Bergmann <arnd@arndb.de>

> PS. discovering that the thing you need to fix (a) never really worked
> as intended, and (b) involves backlight + fbdev.. is never a good way
> to start your day ;-)

I still have a bunch of related fixes in my tree that address
randconfig builds that never worked. I think at one point
I got a few 'Reviewed-by' replies from DRM folks, but then
nobody picked it up and subsequently it stopped applying.
I need to go back and dig out all the dependent patches from
my randconfig tree and resend that.

      Arnd