Re: [PATCH] backlight: pwm_bl: Fix uninitialized variable

[Marcel Ziswiler]

On Wed, 2018-07-18 at 09:09 +0100, Lee Jones wrote:
> On Mon, 16 Jul 2018, Daniel Thompson wrote:
> 
> > Currently, if the DT does not define num-interpolated-steps then
> > num_steps is undefined and the interpolation code will deploy
> > randomly.
> > Fix this.
> > 
> > Fixes: 573fe6d1c25c ("backlight: pwm_bl: Linear interpolation
> > between
> > brightness-levels")
> > Reported-by: Marcel Ziswiler <marcel.ziswiler@xxxxxxxxxxx>
> > Signed-off-by: Daniel Thompson <daniel.thompson@xxxxxxxxxx>
> > Signed-off-by: Marcel Ziswiler <marcel.ziswiler@xxxxxxxxxxx>
> 
> This line is confusing.  Did you guys author this patch together?

Yes, I reported it and we came to a conclusion together.

> My guess is that this line should be dropped and the RB and TB tags
> should remain?  If it was reviewed too, perhaps an AB too?

I'm OK either way and do not need any explicit authorship to be
expressed for me.

> > Tested-by: Marcel Ziswiler <marcel.ziswiler@xxxxxxxxxxx>
> > ---
> >  drivers/video/backlight/pwm_bl.c | 17 ++++++++---------
> >  1 file changed, 8 insertions(+), 9 deletions(-)
> > 
> > diff --git a/drivers/video/backlight/pwm_bl.c
> > b/drivers/video/backlight/pwm_bl.c
> > index 9ee4c1b735b2..e3c22b79fbcd 100644
> > --- a/drivers/video/backlight/pwm_bl.c
> > +++ b/drivers/video/backlight/pwm_bl.c
> > @@ -299,15 +299,14 @@ static int pwm_backlight_parse_dt(struct
> > device *dev,
> >  		 * interpolation between each of the values of
> > brightness levels
> >  		 * and creates a new pre-computed table.
> >  		 */
> > -		of_property_read_u32(node, "num-interpolated-
> > steps",
> > -				     &num_steps);
> > -
> > -		/*
> > -		 * Make sure that there is at least two entries in
> > the
> > -		 * brightness-levels table, otherwise we can't
> > interpolate
> > -		 * between two points.
> > -		 */
> > -		if (num_steps) {
> > +		if ((of_property_read_u32(node, "num-interpolated-
> > steps",
> > +					  &num_steps) == 0) &&
> > num_steps) {
> 
> This is pretty ugly, and isn't it suffering from over-bracketing?  My
> suggestion would be to break out the invocation of
> of_property_read_u32() from the if and test only the result.
> 
> 		of_property_read_u32(node, "num-interpolated-steps", 
> &num_steps);

you mean:

		ret = of_property_read_u32(node, "num-interpolated-
steps", &num_steps);

> 		if (!ret && num_steps) {
> 
> I haven't checked the underling code, but is it even feasible for
> of_property_read_u32() to not succeed AND for num_steps to be set?
> 
> If not, the check for !ret if superfluous and you can drop it.

No, then we are back to the initial issue of num_steps potentially not
being initialised. We really want both of_property_read_u32() to
succeed AND num_steps to actually be set.

> > +			/*
> > +			 * Make sure that there is at least two
> > entries in the
> 
> s/is/are/
> 
> > +			 * brightness-levels table, otherwise we
> > can't
> > +			 * interpolate
> 
> Why break the line here?

That's probably a remnant of going back and forth plus quoting on the
mailing list.

> > +			 * between two points.
> > +			 */
> >  			if (data->max_brightness < 2) {
> >  				dev_err(dev, "can't
> > interpolate\n");
> >  				return -EINVAL;