Enable the CMT0 device and configure channel 0 as a clock event
provider.
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@xxxxxxxxxxxxxxxx>
diff --git a/arch/arm/mach-shmobile/include/mach/r8a7790.h
b/arch/arm/mach-shmobile/include/mach/r8a7790.h index 0b95bab..62b31f3
100644 --- a/arch/arm/mach-shmobile/include/mach/r8a7790.h
+++ b/arch/arm/mach-shmobile/include/mach/r8a7790.h
@@ -29,7 +29,6 @@ enum {
};
void r8a7790_add_standard_devices(void);
-void r8a7790_add_dt_devices(void);
void r8a7790_clock_init(void);
void r8a7790_pinmux_init(void);
void r8a7790_pm_init(void);
diff --git a/arch/arm/mach-shmobile/setup-r8a7790.c
b/arch/arm/mach-shmobile/setup-r8a7790.c index 3e5813f..462c81f 100644
--- a/arch/arm/mach-shmobile/setup-r8a7790.c
+++ b/arch/arm/mach-shmobile/setup-r8a7790.c
@@ -294,11 +294,6 @@ static struct resource cmt0_resources[] = {
&cmt##idx##_platform_data, \
sizeof(struct sh_timer_config))
-void __init r8a7790_add_dt_devices(void)
-{
- r8a7790_register_cmt(0);
-}
-
void __init r8a7790_add_standard_devices(void)
{
r8a7790_register_scif(0);
@@ -311,7 +306,7 @@ void __init r8a7790_add_standard_devices(void)
r8a7790_register_scif(7);
r8a7790_register_scif(8);
r8a7790_register_scif(9);
- r8a7790_add_dt_devices();
+ r8a7790_register_cmt(0);
r8a7790_register_irqc(0);
r8a7790_register_thermal();
}
IMHO, these 2 files should be split into a separate patch.
That could easily be done, but why ?
It does not seem necessary to combine these changes in one patch.
Remember, Simon has separate branches for boards and SoCs. So finally
it's up to him to decide on this...
Right, but in this case I need to remove the CMT platform device
registration from r8a7790_add_dt_devices()
You're not removing anything in these 2 files, you're just replacing
"indirect" call to r8a7790_register_cmt(0) with direct.
OK, I see what you mean now. Given that the patch removes the
r8a7790_add_dt_devices() call from arch/arm/mach-shmobile/board-lager-
reference.c, I could indeed split those two changes into a separate patch as
there's no-one calling the function anymore. That would introduce a dependency
between the branches, which might not be better. I'll let Simon comment on
what he would prefer and will act accordingly.