2012/10/23 Kukjin Kim [off-list ref]:

Tushar Behera wrote:

quoted

On 09/28/2012 03:23 PM, Sangsu Park wrote:

quoted

On Thu, Sep 27, 2012 at 07:30 PM, Padmavathi Venna wrote:

[...]

quoted

quoted

quoted

@@ -63,7 +63,7 @@ config SND_SOC_SAMSUNG_SMDK_WM8580

 config SND_SOC_SAMSUNG_SMDK_WM8994
   tristate "SoC I2S Audio support for WM8994 on SMDK"
-  depends on SND_SOC_SAMSUNG && (MACH_SMDKV310 || MACH_SMDKC210 ||

MACH_SMDK4212)
+  depends on SND_SOC_SAMSUNG && (MACH_SMDKV310 || MACH_SMDKC210 ||
MACH_SMDK4212 || MACH_SMDK4412 || MACH_EXYNOS5_DT)
   depends on I2C=y && GENERIC_HARDIRQS
   select MFD_WM8994
   select SND_SOC_WM8994

I'm not sure about this since MACH_EXYNOS5_DT can have others instead of

SMDK.

quoted

But there is no other way we can differentiate between two EXYNOS5_DT
based boards in the kernel. So, support for all possible codecs and
sound-cards should be enabled in the kernel. The instantiation of the
drivers should depend on the device-tree entries.

Hmm...any idea on this?

Tushar is right. In particular, there is absolutely no reason to
enforce driver selection based on platform enablement, as long as it
still builds no matter what (and if it doesn't, that should probably
be fixed).

It should be possible to enable a superset of all drivers, and only
have the ones that are actually instantiated by platform devices or
device tree entries probe and configure. That's how the kernel works
on nearly all platforms.


So, removing as much "depends on" as possible (probably only keep
SND_SOC_SAMSUNG here) seems like a good choice, and fixing any
compilation errors caused by it.


-Olof