On Mon, Oct 22, 2012 at 02:27:11PM +0100, Mark Brown wrote:

On Mon, Oct 22, 2012 at 02:05:57PM +0100, Russell King - ARM Linux wrote:

quoted

On Mon, Oct 22, 2012 at 02:02:50PM +0100, Mark Brown wrote:

quoted

@@ -327,6 +328,7 @@ config ARCH_AT91
 	select ARCH_REQUIRE_GPIOLIB
 	select CLKDEV_LOOKUP
 	select HAVE_CLK
+	select HAVE_CUSTOM_CLK

quoted

This is silly.  If you select "HAVE_CUSTOM_CLK" then isn't it true that
"HAVE_CLK" should also be selected?  If so, why not have "HAVE_CUSTOM_CLK"
do that selection and remove it from all these entries?

If we're worrying about that there's the larger point that the effect of
this patch is to make HAVE_CLK meaningless as there will be no platform
for which it's not true.  I was just leaving HAVE_CLK alone for now
ready to circle around on it if we ever manage to get the enabling bit
sorted.

Are you sure that all architectures are fine with having that permanently
enabled?  What about nommu architectures?