On 04/18/2012 02:07 AM, Sascha Hauer wrote:
On Tue, Apr 17, 2012 at 03:13:49PM -0700, Turquette, Mike wrote:
quoted
On Mon, Apr 9, 2012 at 1:24 PM, Rob Herring [off-list ref] wrote:
quoted
From: Rob Herring <redacted>
Using the common clock infrastructure without the common clkdev code makes
little sense, so select CLKDEV_LOOKUP for COMMON_CLK.
Signed-off-by: Rob Herring <redacted>
---
drivers/clk/Kconfig | 1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/drivers/clk/Kconfig b/drivers/clk/Kconfig
index 165e1fe..f05a60d 100644
--- a/drivers/clk/Kconfig
+++ b/drivers/clk/Kconfig
@@ -12,6 +12,7 @@ config HAVE_MACH_CLKDEV
config COMMON_CLK
bool
select HAVE_CLK_PREPARE
+ select CLKDEV_LOOKUP
---help---
The common clock framework is a single definition of struct
clk, useful across many platforms, as well as an
Hi Rob,
I agree that the common clk framework is not particularly useful
without clkdev. However the core code has no dependency on clkdev.
Why not just select it from an arch Kconfig, or even make it a
dependency based on your own platform clock data/code?
Selecting it from COMMON_CLK means that the compiler/linker will point
everybody trying to implement it without CLKDEV into the right
direction.
Right. That was part of my rationale as well.
The DT clock support is dependent on both options, so without this patch
we'll have to add an otherwise unneeded OF_CLOCK option to select both.
Rob