[PATCH] arm: sched_clock: Enable HAVE_SCHED_CLOCK support without init_sched_clock

[PATCH] arm: sched_clock: Enable HAVE_SCHED_CLOCK support without init_sched_clock

[Kyungmin Park]

From: Kyungmin Park <kyungmin.park@samsung.com>

On Samsung Exynos4 series, EVT0 chip uses the s5p_timer with HAVE_SCHED_CLOCK and remains use the MCT timer without init_sched_clock.

The similar combination will be happend with other SOCs. some board use HAVE_SCHED_CLOCK or others are not.

To co-exist the these boards. check the sched_clock_update_fn when sched_clock_postinit is called and return if not defined.

Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
---
diff --git a/arch/arm/kernel/sched_clock.c b/arch/arm/kernel/sched_clock.c
index 9a46370..12c6b6b 100644
--- a/arch/arm/kernel/sched_clock.c
+++ b/arch/arm/kernel/sched_clock.c
@@ -70,5 +70,8 @@ void __init init_sched_clock(struct clock_data *cd, void (*update)(void),
 
 void __init sched_clock_postinit(void)
 {
+	/* If init_sched_clock is not called, just return */
+	if (!sched_clock_update_fn)
+		return;
 	sched_clock_poll(sched_clock_timer.data);
 }

[PATCH] arm: sched_clock: Enable HAVE_SCHED_CLOCK support without init_sched_clock

[Russell King - ARM Linux]

On Thu, Sep 15, 2011 at 04:35:12PM +0900, Kyungmin Park wrote:
> From: Kyungmin Park <kyungmin.park@samsung.com>
> 
> On Samsung Exynos4 series, EVT0 chip uses the s5p_timer with HAVE_SCHED_CLOCK and remains use the MCT timer without init_sched_clock.

So in other words, sched_clock on exynos4 is broken because the update
function won't be called, and it'll wrap well below the 64-bit ns range
required of it.

NAK.

[PATCH] arm: sched_clock: Enable HAVE_SCHED_CLOCK support without init_sched_clock

[Kyungmin Park]

On Thu, Sep 15, 2011 at 5:20 PM, Russell King - ARM Linux
<linux@arm.linux.org.uk> wrote:
> On Thu, Sep 15, 2011 at 04:35:12PM +0900, Kyungmin Park wrote:
>> From: Kyungmin Park <kyungmin.park@samsung.com>
>>
>> On Samsung Exynos4 series, EVT0 chip uses the s5p_timer with HAVE_SCHED_CLOCK and remains use the MCT timer without init_sched_clock.
>
> So in other words, sched_clock on exynos4 is broken because the update
> function won't be called, and it'll wrap well below the 64-bit ns range
> required of it.

Right, even though MCT doesn't use the HAVE_SCHED_CLOCK, sched_clock
is override with others.
It mean it should be implement the sched_clock for all boards if there
are some boards use the HAVE_SCHED_CLOCK.

As I remember, there's attempt to solve the sched_clock registration
issue. I'll check it.

Thank you,
Kyungmin Park
>
> NAK.
>