On Mon, Jan 20, 2020 at 09:09:17PM +0100, Uwe Kleine-König wrote:
> Hello Thierry,
> 
> On Mon, Jan 20, 2020 at 03:32:06PM +0100, Thierry Reding wrote:
> > GCC can't always determine that the duty, period and prescaler values
> > are initialized when returning from sun4i_pwm_calculate(), so help out a
> > little by initializing them to 0.
> 
> Is it worth mentioning the gcc version you're using?

I could, but what good is that going to be? I don't think this is
something that's limited to one specific version but I don't know
exactly which ones are impacted. Stating just one specific version
isn't all that useful in that case.

> > Signed-off-by: Thierry Reding <thierry.reding@gmail.com>
> > ---
> >  drivers/pwm/pwm-sun4i.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/pwm/pwm-sun4i.c b/drivers/pwm/pwm-sun4i.c
> > index 0decc7cde133..3e3efa6c768f 100644
> 
> I don't find this object (0decc7cde133) in my tree or next. Which
> version is this?

I made this on top of my local pwm/for-next when I was build-testing,
which I usually do before pushing, so it's not surprising that you
don't have this in your tree.

> 
> > --- a/drivers/pwm/pwm-sun4i.c
> > +++ b/drivers/pwm/pwm-sun4i.c
> > @@ -234,9 +234,9 @@ static int sun4i_pwm_apply(struct pwm_chip *chip, struct pwm_device *pwm,
> >  {
> >  	struct sun4i_pwm_chip *sun4i_pwm = to_sun4i_pwm_chip(chip);
> >  	struct pwm_state cstate;
> > -	u32 ctrl, duty, period, val;
> > +	u32 ctrl, duty = 0, period = 0, val;
> 
> +     u32 ctrl, uninitialized_var(duty), uninitialized_var(period), val;
> 
> should fix the warnings, too, and additionally explicitly documents that
> it's just the compiler that doesn't see there is no problem.

I haven't convinced myself fully yet that there really isn't a problem.
I'm fairly sure it's safe, but always initializing to 0 doesn't hurt.

Thierry