Re: [PATCH v4 5/9] iio: light: gp2ap020a00f: Return directly from the switch cases

[Andy Shevchenko]

On Thu, Feb 19, 2026 at 01:32:20PM -0600, Ethan Tidmore wrote:
> On Tue Feb 17, 2026 at 10:37 PM CST, Ethan Tidmore wrote:

...

> > By doing this, it appears to have fixed a preexisting bug, the variable
> > err in  gp2ap020a00f_buffer_predisable() is only checked once for errors
> > after the for loop is finished. This could allow errors to be swallowed.

...

> >  static int gp2ap020a00f_buffer_postenable(struct iio_dev *indio_dev)
> >  {
> >  	struct gp2ap020a00f_data *data = iio_priv(indio_dev);
> > -	int i, err = 0;
> > +	int i, err;
> >  
> >  	guard(mutex)(&data->lock);
> >  
> > @@ -1400,15 +1378,15 @@ static int gp2ap020a00f_buffer_postenable(struct iio_dev *indio_dev)
> >  
> >  	data->buffer = kmalloc(indio_dev->scan_bytes, GFP_KERNEL);
> >  	if (!data->buffer)
> > -		err = -ENOMEM;
> > +		return -ENOMEM;
> >  
> > -	return err;
> > +	return 0;
> >  }
> 
> Looking over the code again it looks like 
> gp2ap020a00f_buffer_postenable() contains the same bug you mentioned
> where the error check should have been inside of the loop?
> 
> 
> 	iio_for_each_active_channel(indio_dev, i) {
> 		switch (i) {
> 		case GP2AP020A00F_SCAN_MODE_LIGHT_CLEAR:
> 			err = gp2ap020a00f_exec_cmd(data,
> 					GP2AP020A00F_CMD_TRIGGER_CLEAR_EN);
> 			break;
> 		case GP2AP020A00F_SCAN_MODE_LIGHT_IR:
> 			err = gp2ap020a00f_exec_cmd(data,
> 					GP2AP020A00F_CMD_TRIGGER_IR_EN);
> 			break;
> 		case GP2AP020A00F_SCAN_MODE_PROXIMITY:
> 			err = gp2ap020a00f_exec_cmd(data,
> 					GP2AP020A00F_CMD_TRIGGER_PROX_EN);
> 			break;
> 		}
> 	}
> 
> 	if (err < 0)
> 		goto error_unlock;
> 
> Just wanted to confirm before putting it in the v5.

Good catch! So, these two should be split into a separate change explaining
the change in behaviour.

-- 
With Best Regards,
Andy Shevchenko