pm_runtime_get_sync will increment pm usage counter even it
failed. Forgetting to putting operation will result in a
reference leak here.
A new function pm_runtime_resume_and_get is introduced in
[0] to keep usage counter balanced. So We fix the reference
leak by replacing it with new funtion.
[0] 
dd8088d5a896 ("PM: runtime: Add  pm_runtime_resume_and_get to deal with usage counter")
Fixes: c2df3de0d07e ("gpio: zynq: properly support runtime PM for GPIO used as interrupts")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Qinglang Miao <miaoqinglang@huawei.com>
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
        struct gpio_chip *chip = irq_data_get_irq_chip_data(d);
        int ret;
 
-       ret = pm_runtime_get_sync(chip->parent);
+       ret = pm_runtime_resume_and_get(chip->parent);
        if (ret < 0)
                return ret;
 
 
        pm_runtime_set_active(&pdev->dev);
        pm_runtime_enable(&pdev->dev);
-       ret = pm_runtime_get_sync(&pdev->dev);
+       ret = pm_runtime_resume_and_get(&pdev->dev);
        if (ret < 0)
                goto err_pm_dis;