Re: -tip: origin tree build failure (Was: [git pull] drm tree for merge window)

[Phil Turmel <philip@turmel.org>]

On 10/28/2010 09:12 AM, Ingo Molnar wrote:
> 
> * Dave Airlie <airlied@linux.ie> wrote:
> 
>>> Lee, Chun-Yi (1):
>>>       gpu: Add Intel GMA500(Poulsbo) Stub Driver
> 
> Today's -tip fails to build due to upstream commit e26fd11 ("gpu: Add Intel 
> GMA500(Poulsbo) Stub Driver"), committed two days ago and merged yesterday, on
> x86 allmodconfig with BACKLIGHT_CLASS_DEVICE disabled:
> 
>  drivers/built-in.o: In function `acpi_video_bus_put_one_device':
>  video.c:(.text+0x7d26f): undefined reference to `backlight_device_unregister'
>  drivers/built-in.o: In function `acpi_video_switch_brightness':
>  video.c:(.text+0x7d6f5): undefined reference to `backlight_force_update'
>  drivers/built-in.o: In function `acpi_video_device_find_cap':
>  video.c:(.text+0x7dfdb): undefined reference to `backlight_device_register'
> 
> drivers/gpu/stub/Kconfig selects ACPI_VIDEO, but ACPI_VIDEO is a complex interactive 
> Kconfig option with a lot of dependencies:
> 
>  config ACPI_VIDEO
> 	tristate "Video"
> 	depends on X86 && BACKLIGHT_CLASS_DEVICE && VIDEO_OUTPUT_CONTROL
> 	depends on INPUT
> 	select THERMAL
> 	help
> 	  This driver implements the ACPI Extensions For Display Adapters
> 
> and if any of its dependencies are not met, we get a build failure. This problem was 
> apparently realized in the driver at a certain stage:
> 
>  config STUB_POULSBO
>         tristate "Intel GMA500 Stub Driver"
>         depends on PCI
>         # Poulsbo stub depends on ACPI_VIDEO when ACPI is enabled
>         # but for select to work, need to select ACPI_VIDEO's dependencies, ick
>         select ACPI_VIDEO if ACPI
> 
> but not fully understood and not fully fixed.

I submitted a nouveau patch that does this as well.  Part of the same merge.  I was
copying from the Intel driver but didn't catch the comment on the dependencies.

(Added Francisco to the CC: list)

> As a quick fix select these secondary dependencies, like drivers/gpu/drm/Kconfig 
> does:
> 
>  config DRM_I915
> 	tristate "i915 driver"
> 	depends on AGP_INTEL
> 	select SHMEM
> 	select DRM_KMS_HELPER
> 	select FB_CFB_FILLRECT
> 	select FB_CFB_COPYAREA
> 	select FB_CFB_IMAGEBLIT
> 	# i915 depends on ACPI_VIDEO when ACPI is enabled
> 	# but for select to work, need to select ACPI_VIDEO's dependencies, ick
> 	select VIDEO_OUTPUT_CONTROL if ACPI
> 	select BACKLIGHT_CLASS_DEVICE if ACPI
> 	select INPUT if ACPI
> 	select ACPI_VIDEO if ACPI
> 	select ACPI_BUTTON if ACPI
> 	help
> 	  Choose this option if you have a system that has Intel 830M, 845G,
> 	  852GM, 855GM 865G or 915G integrated graphics.  If M is selected, the
> 
> But it's arguably not particularly nice looking, so maybe this area of code is ripe 
> for a Kconfig restructuring/cleanup.

I have no idea what such a cleanup would look like, but I'm guessing the following is
needed for nouveau in the meantime:

Signed-off-by: Philip J. Turmel <philip@turmel.org>
---

The build fix for nouveau's use of acpi_video_get_edid omitted the
dependencies of ACPI_VIDEO.

diff --git a/drivers/gpu/drm/nouveau/Kconfig b/drivers/gpu/drm/nouveau/Kconfig
index 72730e9..c33ecb6 100644
--- a/drivers/gpu/drm/nouveau/Kconfig
+++ b/drivers/gpu/drm/nouveau/Kconfig
@@ -11,6 +11,9 @@ config DRM_NOUVEAU
 	select FRAMEBUFFER_CONSOLE if !EMBEDDED
 	select FB_BACKLIGHT if DRM_NOUVEAU_BACKLIGHT
 	select ACPI_VIDEO if ACPI
+	select VIDEO_OUTPUT_CONTROL if ACPI
+	select BACKLIGHT_CLASS_DEVICE if ACPI
+	select INPUT if ACPI
 	help
 	  Choose this option for open-source nVidia support.