Re: [PATCH] leds-gpio: Fix error handling and memory leak

[Corey Minyard <minyard@acm.org>]

On 03/26/2015 08:20 PM, Bryan Wu wrote:
> On Mon, Mar 9, 2015 at 5:43 PM,  <minyard@acm.org> wrote:
>> From: Corey Minyard <cminyard@mvista.com>
>>
>> The leds-gpio driver would not clean up properly if it failed in some
>> places, and it wasn't freeing its private data.
>>
>> Signed-off-by: Corey Minyard <cminyard@mvista.com>
>> ---
>>  drivers/leds/leds-gpio.c | 13 +++++++++----
>>  1 file changed, 9 insertions(+), 4 deletions(-)
>>
>> diff --git a/drivers/leds/leds-gpio.c b/drivers/leds/leds-gpio.c
>> index d26af0a..32f7642 100644
>> --- a/drivers/leds/leds-gpio.c
>> +++ b/drivers/leds/leds-gpio.c
>> @@ -198,8 +198,10 @@ static struct gpio_leds_priv *gpio_leds_create(struct platform_device *pdev)
>>                 } else {
>>                         if (IS_ENABLED(CONFIG_OF) && !led.name && np)
>>                                 led.name = np->name;
>> -                       if (!led.name)
>> -                               return ERR_PTR(-EINVAL);
>> +                       if (!led.name) {
>> +                               ret = -EINVAL;
>> +                               goto err;
>> +                       }
>>                 }
>>                 fwnode_property_read_string(child, "linux,default-trigger",
>>                                             &led.default_trigger);
>> @@ -217,19 +219,21 @@ static struct gpio_leds_priv *gpio_leds_create(struct platform_device *pdev)
>>                 if (fwnode_property_present(child, "retain-state-suspended"))
>>                         led.retain_state_suspended = 1;
>>
>> -               ret = create_gpio_led(&led, &priv->leds[priv->num_leds++],
>> +               ret = create_gpio_led(&led, &priv->leds[priv->num_leds],
> Why need this change? it's correct. And your add one more line
> "priv->num_leds++"

That's actually the major source of the problem.  The value of
priv->num_leds was not correct if it failed before this point, and there
was already one "goto err" above this code and I added another to
properly handle not allocating the led name.  If it failed there it
would leave an LED lying around but free the memory underneath it.  So
instead, modify the failure recovery code to be priv->num_leds-1 instead
of priv->num_leds-2 and don't increment priv->num_leds until you have
success.

>>                                       dev, NULL);
>>                 if (ret < 0) {
>>                         fwnode_handle_put(child);
>>                         goto err;
>>                 }
>> +               priv->num_leds++;
> Why need this?

See above.

>>         }
>>
>>         return priv;
>>
>>  err:
>> -       for (count = priv->num_leds - 2; count >= 0; count--)
>> +       for (count = priv->num_leds - 1; count >= 0; count--)
>>                 delete_gpio_led(&priv->leds[count]);
>> +       devm_kfree(dev, priv);
> priv is created by devm_kzalloc(), so if driver probing return error,
> it will be freed automatically, you don't need call devm_free();

Ah, ok.  Then this is unnecessary.  Do want a new patch?

Thanks,

-corey

>>         return ERR_PTR(ret);
>>  }
>>
>> @@ -283,6 +287,7 @@ static int gpio_led_remove(struct platform_device *pdev)
>>
>>         for (i = 0; i < priv->num_leds; i++)
>>                 delete_gpio_led(&priv->leds[i]);
>> +       devm_kfree(&pdev->dev, priv);
> No need this during remove.
>
>>         return 0;
>>  }
>> --
>> 1.8.3.1
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-leds" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html