On Thu, Oct 17, 2013 at 7:05 AM, Mark Rutland [off-list ref] wrote:

On Wed, Oct 16, 2013 at 10:47:08PM +0100, Tim Kryger wrote:

quoted

When an clock handle is specified in the device tree, enable it and use
it to determine the external clock frequency.

I'd drop handle here and just say "When a clock is specified".

This will need a binding document update.

Can you be more specific about what you are looking for here?

Shall I just say that the frequency property is now deprecated since
it actually belongs to the external clock whose association with a
kona timer instance might be declared in the DT using the common clock
bindings but could also be declared through other means?

quoted

Signed-off-by: Tim Kryger <redacted>
Reviewed-by: Markus Mayer <redacted>
Reviewed-by: Matt Porter <redacted>
---
 drivers/clocksource/bcm_kona_timer.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/drivers/clocksource/bcm_kona_timer.c b/drivers/clocksource/bcm_kona_timer.c
index 0d7d8c3..fd11f96 100644
--- a/drivers/clocksource/bcm_kona_timer.c
+++ b/drivers/clocksource/bcm_kona_timer.c

@@ -17,6 +17,7 @@
 #include <linux/jiffies.h>
 #include <linux/clockchips.h>
 #include <linux/types.h>
+#include <linux/clk.h>

 #include <linux/io.h>
 #include <asm/mach/time.h>

@@ -107,11 +108,18 @@ static const struct of_device_id bcm_timer_ids[] __initconst = {
 static void __init kona_timers_init(struct device_node *node)
 {
      u32 freq;
+     struct clk *external_clk;

-     if (!of_property_read_u32(node, "clock-frequency", &freq))
+     external_clk = of_clk_get_by_name(node, NULL);

Is there only a single external clock input to the kona timer, or is
only one relevant here?

Are there any other inputs currently not modelled (e.g. regulators)?

The timer block relies upon a single clock and does not use any regulators.

quoted

+
+     if (!IS_ERR(external_clk)) {
+             arch_timer_rate = clk_get_rate(external_clk);
+             clk_prepare_enable(external_clk);
+     } else if (!of_property_read_u32(node, "clock-frequency", &freq)) {
              arch_timer_rate = freq;
-     else
-             panic("clock-frequency not set in the .dts file");
+     } else {
+             panic("neither clock-frequency or clocks handle in .dts file");

Nit: it's not a handle, it's a phandle+args pair.

Why not just "Unable to determine clock frequency"?

Sure that sounds fine.

Do we need to panic here? Might a system have other clocks it could use
to continue?

I completely agree that this driver could be improved.  However, this
is unrelated to the topic of this series so I will address it later in
a separate patch.