On Wed, Oct 23, 2019 at 12:27 PM Krzysztof Kozlowski <krzk@xxxxxxxxxx> wrote:
> On Thu, Oct 10, 2019 at 10:29:49PM +0200, Arnd Bergmann wrote:
> > diff --git a/arch/arm/mach-s5pv210/pm.c b/arch/arm/mach-s5pv210/pm.c
> > index b336df0c57f3..efdb5a27c060 100644
> > --- a/arch/arm/mach-s5pv210/pm.c
> > +++ b/arch/arm/mach-s5pv210/pm.c
> > @@ -99,8 +99,6 @@ static int s5pv210_suspend_enter(suspend_state_t state)
> > u32 eint_wakeup_mask = s5pv210_read_eint_wakeup_mask();
> > int ret;
> >
> > - s3c_pm_debug_init();
>
> Your patch is not equivalent here. If there is a reason behind removal
> of UART init (e.g. not needed), I prefer to make it in separate patch.
It is equivalent, but the reason is a bit subtle:
The definition looks like
#ifdef CONFIG_SAMSUNG_ATAGS
#include <plat/pm.h>
#include <mach/pm-core.h>
#else
static inline void s3c_pm_debug_init_uart(void) {}
#endif
void s3c_pm_debug_init(void)
{
/* restart uart clocks so we can use them to output */
s3c_pm_debug_init_uart();
}
On s5pv210, CONFIG_SAMSUNG_ATAGS is never set, so this
function does not do anything. Splitting the change out into a
separate patch to explain that is a good idea, I'll do this.
Arnd
