Re: [PATCH] apple-gmux: Correctly depend on apple_bl for building

[Arun Raghavan <arun.raghavan@collabora.co.uk>]

On Thu, 2012-06-28 at 08:48 -0500, Seth Forshee wrote:
> On Thu, Jun 28, 2012 at 06:53:35PM +0530, Arun Raghavan wrote:
> > On Thu, 2012-06-28 at 08:20 -0500, Seth Forshee wrote:
> > > On Thu, Jun 28, 2012 at 04:06:36PM +0530, Arun Raghavan wrote:
> > > > apple_bl_register/unregister() are unconditionally used now, so we need
> > > > to make sure that apple_bl is selected appropriately (built-in/module)
> > > > if apple-gmux is being built.
> > > 
> > > apple_bl.h provides stubs for these functions when apple_bl is not
> > > built, so this shouldn't be necessary. Are you encoutering a situation
> > > that causes build problems?
> > 
> > Yes, I am. With CONFIG_BACKLIGHT_APPLE=m and CONFIG_APPLE_GMUX=y, I see
> > the following error:
> > 
> >   LD      init/built-in.o
> > drivers/built-in.o: In function `gmux_probe':
> > apple-gmux.c:(.devinit.text+0xa769): undefined reference to `apple_bl_unregister'
> > drivers/built-in.o: In function `gmux_remove':
> > apple-gmux.c:(.devexit.text+0x46d): undefined reference to `apple_bl_register'
> 
> Yuck. In that case I don't think your patch probably goes far enough to
> fix your situation.
> 
> As far as I know select isn't recursive, i.e. it won't select the
> dependencies of the thing you select. 

I missed the stubs, so the patch should probably just be adding this
line:

depends on APPLE_BACKLIGHT = n || APPLE_BACKLIGHT

but that conflicts with the 'select BACKLIGHT_CLASS_DEVICE' (a recursive
dependency is detected).

-- Arun