On 11/28/2012 01:20 AM, Santosh Shilimkar wrote:

On Wednesday 28 November 2012 12:16 PM, Igor Grinberg wrote:

quoted

On 11/28/12 08:28, Santosh Shilimkar wrote:

quoted

On Wednesday 28 November 2012 07:45 AM, Jon Hunter wrote:

quoted

When compiling the kernel with configuration options ...

   # CONFIG_ARCH_OMAP2 is not set
   # CONFIG_ARCH_OMAP3 is not set
   # CONFIG_ARCH_OMAP4 is not set
   # CONFIG_SOC_OMAP5 is not set
   CONFIG_SOC_AM33XX=y

   ... the following build warning is seen.

    CC      arch/arm/mach-omap2/timer.o
    arch/arm/mach-omap2/timer.c:395:19: warning:
?omap2_sync32k_clocksource_init?
        defined but not used [-Wunused-function]

This issue was introduced by commit 6f80b3b (ARM: OMAP2+: timer: remove
CONFIG_OMAP_32K_TIMER) where the omap2_sync32k_clocksource_init() is no
longer referenced by the timer initialisation function for the AM335x
device as it has no 32k-sync timer.

Fix this by only including the omap2_sync32k_clocksource_init()
function
if either OMAP2, OMAP3, OMAP4 or OMAP5 devices are enabled.

Cc: Igor Grinberg <redacted>

Signed-off-by: Jon Hunter <redacted>
---
   arch/arm/mach-omap2/timer.c |    3 +++
   1 file changed, 3 insertions(+)

diff --git a/arch/arm/mach-omap2/timer.c b/arch/arm/mach-omap2/timer.c
index eb96712..085c7e7 100644
--- a/arch/arm/mach-omap2/timer.c
+++ b/arch/arm/mach-omap2/timer.c

@@ -386,6 +386,8 @@ static u32 notrace dmtimer_read_sched_clock(void)
       return 0;
   }

+#if defined(CONFIG_ARCH_OMAP2) || defined(CONFIG_ARCH_OMAP3) || \
+    defined(CONFIG_ARCH_OMAP4) || defined(CONFIG_SOC_OMAP5)

#ifndef CONFIG_SOC_AM33XX ?

#ifdef things are really ugly and needs constant patching and
hence something like CONFIG_HAS_32K kind of feature flags are
better. But that will undo certain part of f80b3b
(ARM: OMAP2+: timer: remove  CONFIG_OMAP_32K_TIMER).

Agreed on ugliness of ifdefs.
What about adding __maybe_unused to the function signature?
That will cover any future SoC also w/o the need to extend the ifdefs.

Sounds good to me.

Yes agree on the ugliness of this. However, my thought was these would
only remain until we migrate over to device-tree and then the detection
of the a 32k source can be determine via DT.

However, I can update to use __maybe_unused for now. We just need to
remember to remove this in the future if it becomes unnecessary :-)

Cheers
Jon