On Wed, Mar 19, 2014 at 03:18:20PM +0100, Hans-Christian Egtvedt wrote:
> Around Wed 19 Mar 2014 14:03:22 +0100 or thereabout, Alexandre Belloni wrote:
> > Switch to the pwm/pwm-atmel driver instead of misc/atmel_pwm
> >
> > Signed-off-by: Alexandre Belloni <alexandre.belloni@xxxxxxxxxxxxxxxxxx>
>
> Acked-by: Hans-Christian Egtvedt <egtvedt@xxxxxxxxxxxx>
>
> > ---
> > arch/avr32/mach-at32ap/at32ap700x.c | 7 ++-----
> > 1 file changed, 2 insertions(+), 5 deletions(-)
> >
> > diff --git a/arch/avr32/mach-at32ap/at32ap700x.c b/arch/avr32/mach-at32ap/at32ap700x.c
> > index a1f4d1e91b52..db85b5ec3351 100644
> > --- a/arch/avr32/mach-at32ap/at32ap700x.c
> > +++ b/arch/avr32/mach-at32ap/at32ap700x.c
> > @@ -1553,7 +1553,7 @@ static struct resource atmel_pwm0_resource[] __initdata = {
> > IRQ(24),
> > };
> > static struct clk atmel_pwm0_mck = {
> > - .name = "pwm_clk",
> > + .name = "at91sam9rl-pwm",
>
> I found this a tiny bit weird, but found the matching instance in
> drivers/pwm/pwm-atmel.c
>
> IMHO it would initially have been better to call it atpwm_v1 or something
> similar, more generic. But I guess that is too late to change at this point.
Generally when you call clk_get(dev, NULL), you'll pass in the consumer
device as "dev". Now the avr32 clock implementation compares that to the
device associated with the clock, which for the case of the PWM pwm0_mck
is the PWM controller (at32_add_device_pwm() for reference). So it seems
to be set up properly (the PWM controller is the consumer).
Also since the consumer ID isn't specified to the clk_get() call, the
name of the clock (what's being changed in the above hunk) shouldn't
matter because it won't match anyway.
Now that I look at it more closely, I wonder how clk_get(dev, NULL) can
even work on avr32, since there's no check for id == NULL and therefore
NULL will be blindly passed into strcmp() and then crash.
Thierry
Attachment:
pgp_CkuxGsyvM.pgp
Description: PGP signature
