On Tue, Nov 09, 2010 at 04:22:02PM +0800, Axel Lin wrote:
> 2010/11/9 Jean Delvare <khali@linux-fr.org>:
> > On Tue, 09 Nov 2010 13:18:39 +0800, Axel Lin wrote:
> >> Remove sysfs entries before return -ENODEV.
> >>
> >> Signed-off-by: Axel Lin <axel.lin@gmail.com>
> >> ---
> >>  drivers/hwmon/gpio-fan.c |    4 +++-
> >>  1 files changed, 3 insertions(+), 1 deletions(-)
> >>
> >> diff --git a/drivers/hwmon/gpio-fan.c b/drivers/hwmon/gpio-fan.c
> >> index aa701a1..d2e66b4 100644
> >> --- a/drivers/hwmon/gpio-fan.c
> >> +++ b/drivers/hwmon/gpio-fan.c
> >> @@ -388,11 +388,13 @@ static int fan_ctrl_init(struct gpio_fan_data *fan_data,
> >>       fan_data->speed_index = get_fan_speed_index(fan_data);
> >>       if (fan_data->speed_index < 0) {
> >>               err = -ENODEV;
> >> -             goto err_free_gpio;
> >> +             goto err_remove_sysfs;
> >>       }
> >>
> >>       return 0;
> >>
> >> +err_remove_sysfs:
> >> +     sysfs_remove_group(&pdev->dev.kobj, &gpio_fan_ctrl_group);
> >>  err_free_gpio:
> >>       for (i = i - 1; i >= 0; i--)
> >>               gpio_free(ctrl[i]);
> >
> > Oh, and while you're here, fan_alarm_init() needs some love as well.
> > There's a "return 0" in the middle which clearly can't be correct.
> I think this part is correct. see the comment:
>         /*
>          * If the alarm GPIO don't support interrupts, just leave
>          * without initializing the fail notification support.
>          */
> 

Yes, GPIO interrupts could not be supported. That's not an error case.

Simon