Re: linux-next: Tree for Aug 2 (drivers/platform/x86/apple-gmux.c)

[Seth Forshee <seth.forshee@canonical.com>]

On Thu, Aug 02, 2012 at 11:36:16AM -0500, Seth Forshee wrote:
> On Thu, Aug 02, 2012 at 11:31:54AM -0500, Seth Forshee wrote:
> > On Thu, Aug 02, 2012 at 09:19:30AM -0700, Randy Dunlap wrote:
> > > On 08/02/2012 09:15 AM, Seth Forshee wrote:
> > > 
> > > > On Thu, Aug 02, 2012 at 09:07:04AM -0700, Randy Dunlap wrote:
> > > >> On 08/01/2012 09:15 PM, Stephen Rothwell wrote:
> > > >>
> > > >>> Hi all,
> > > >>>
> > > >>> Please do not add anything to linux-next included branches/series that is
> > > >>> destined for v3.7 until after v3.6-rc1 is released.
> > > >>>
> > > >>> Reminder: do not rebase your branches before asking Linus to pull them ...
> > > >>>
> > > >>> Changes since 20120731:
> > > >>>
> > > >>
> > > >>
> > > >> when CONFIG_ACPI is not enabled: (on i386)
> > > >>
> > > >> drivers/platform/x86/apple-gmux.c: In function 'gmux_probe':
> > > >> drivers/platform/x86/apple-gmux.c:196:2: error: implicit declaration of function 'acpi_video_dmi_promote_vendor'
> > > >> drivers/platform/x86/apple-gmux.c: In function 'gmux_remove':
> > > >> drivers/platform/x86/apple-gmux.c:219:2: error: implicit declaration of function 'acpi_video_dmi_demote_vendor'
> > > >>
> > > >>
> > > >> Full randconfig file is attached.
> > > > 
> > > > I sent Matthew a patch that should fix this [1], but he doesn't seem to
> > > > have picked it up yet.
> > > > 
> > > > Seth
> > > > 
> > > > [1] https://lkml.org/lkml/2012/7/2/257
> > > > 
> > > 
> > > 
> > > That patch does not fix this build error.
> > > There is still an implicit assumption in Kconfig and
> > > the driver source that ACPI is enabled.
> > 
> > The functions in question are built or not based off of ACPI_VIDEO,
> > which in turn depends on ACPI. So if ACPI=n, then ACPI_VIDEO=n and you
> > can't configure APPLE_GMUX=(m|y). Is that right, or am I missing
> > something?
> 
> Nevermind, I see what I missed.
> 
> This is kind of messy -- maybe those declartions should be extracted
> from the #ifdef CONFIG_ACPI to avoid this kind of Kconfig madness?

Eh, the driver doesn't make any sense without ACPI anyway, so I'll
refresh the patch with that dependency.

Seth