Re: [PATCH v3 1/4] gpio: kunit: fix the label of the dummy GPIO chip

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] gpio: resource leak of gdev on KUnit test assertion failure
--

commit b7e40967a53cb1589f5bac90dea8c80d8d27bce0
Author: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>

gpio: kunit: fix the label of the dummy GPIO chip

The label of the test GPIO provider is currently set to the consumer
name. Use the provider name instead.

> diff --git a/drivers/gpio/gpiolib-kunit.c b/drivers/gpio/gpiolib-kunit.c
> --- a/drivers/gpio/gpiolib-kunit.c
> +++ b/drivers/gpio/gpiolib-kunit.c

[ ... ]

> @@ -780,7 +779,7 @@ static void gpio_hog_assert(struct kunit *test, unsigned int offset,
>  
>  	/* The hog must have configured the line with the expected direction. */
>  	struct gpio_device *gdev __free(gpio_device_put) =
> -		gpio_device_find_by_label(GPIO_CONSUMER_NAME);
> +			gpio_device_find_by_label(GPIO_TEST_PROVIDER);
>  	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, gdev);
>  
>  	desc = gpio_device_get_desc(gdev, offset);
>  	KUNIT_ASSERT_NOT_ERR_OR_NULL(test, desc);
>  
>  	ret = gpiod_get_direction(desc);
>  	KUNIT_ASSERT_EQ(test, ret, expected_direction);

[Severity: Medium]
Does this code leak the gdev reference when KUnit assertions fail?

If KUNIT_ASSERT_NOT_ERR_OR_NULL(test, desc) or the subsequent KUNIT_ASSERT_EQ()
fails in gpio_hog_assert(), KUnit aborts the test thread immediately using
kthread_exit().

Because kthread_exit() bypasses normal C scope exit, compiler-generated
cleanup handlers like __free(gpio_device_put) are skipped, causing a reference
count leak of the gpio_device.

This leaked reference could prevent the dummy device from being properly freed,
which might wedge subsequent test runs or module unloads.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261013-gpio-kunit-plat-lookup-v3-0-3907f913e964@oss.qualcomm.com?part=1