On Sun, Jan 07, 2018 at 09:26:44PM +0530, Keerthy wrote:

On 1/5/2018 4:17 AM, Ladislav Michl wrote:

quoted

On Tue, Jan 02, 2018 at 03:39:59PM +0530, Keerthy wrote:

quoted

From: Ladislav Michl <redacted>

Invalid prescaler value is silently ignored. Fix that
by returning -EINVAL in such case. As invalid value
disabled use of the prescaler, use -1 explicitely for
that purpose.

Signed-off-by: Ladislav Michl <redacted>
---
  drivers/clocksource/timer-dm.c | 3 +++
  1 file changed, 3 insertions(+)

diff --git a/drivers/clocksource/timer-dm.c b/drivers/clocksource/timer-dm.c
index 60db173..01a9cb0 100644
--- a/drivers/clocksource/timer-dm.c
+++ b/drivers/clocksource/timer-dm.c

@@ -672,6 +672,9 @@ int omap_dm_timer_set_prescaler(struct omap_dm_timer *timer, int prescaler)
  	if (prescaler >= 0x00 && prescaler <= 0x07) {
  		l |= OMAP_TIMER_CTRL_PRE;
  		l |= prescaler << 2;
+	} else {
+		if (prescaler != -1)
+			return -EINVAL;

Argh... This is actually wrong, as it leaves timer enabled.
I suggest simply dropping this patch and I'll rethink whole
approach a bit later (and better).

Okay. I hope the rest 9 patches work well for you.

Yes. I rebased event capture patches on top of this serie (based on
linux-next) and will post them during next week for review. Fixed
patch will be included (note, it is not really needed for your serie
as noone is calling this function).

quoted

quoted

  	}
  	omap_dm_timer_write_reg(timer, OMAP_TIMER_CTRL_REG, l);

Sorry for the noise,
	ladis