Re: [PATCH] gpio: stmpe: Use irqchip template

[Serge Semin <fancer.lancer@xxxxxxxxx>]

On Thu, Jul 16, 2020 at 12:06:38PM +0200, Linus Walleij wrote:
> This makes the driver use the irqchip template to assign
> properties to the gpio_irq_chip instead of using the
> explicit calls to gpiochip_irqchip_add_nested() and
> gpiochip_set_nested_irqchip(). The irqchip is instead
> added while adding the gpiochip.
> 
> Cc: Patrice Chotard <patrice.chotard@xxxxxx>
> Cc: Alexandre TORGUE <alexandre.torgue@xxxxxx>
> Signed-off-by: Linus Walleij <linus.walleij@xxxxxxxxxx>
> ---
>  drivers/gpio/gpio-stmpe.c | 24 +++++++++++-------------
>  1 file changed, 11 insertions(+), 13 deletions(-)
> 
> diff --git a/drivers/gpio/gpio-stmpe.c b/drivers/gpio/gpio-stmpe.c
> index 542706a852e6..395ee51445ea 100644
> --- a/drivers/gpio/gpio-stmpe.c
> +++ b/drivers/gpio/gpio-stmpe.c
> @@ -507,6 +507,8 @@ static int stmpe_gpio_probe(struct platform_device *pdev)
>  	}
>  
>  	if (irq > 0) {
> +		struct gpio_irq_chip *girq;
> +
>  		ret = devm_request_threaded_irq(&pdev->dev, irq, NULL,
>  				stmpe_gpio_irq, IRQF_ONESHOT,
>  				"stmpe-gpio", stmpe_gpio);
> @@ -514,20 +516,16 @@ static int stmpe_gpio_probe(struct platform_device *pdev)
>  			dev_err(&pdev->dev, "unable to get irq: %d\n", ret);
>  			goto out_disable;
>  		}
> -		ret =  gpiochip_irqchip_add_nested(&stmpe_gpio->chip,
> -						   &stmpe_gpio_irq_chip,
> -						   0,
> -						   handle_simple_irq,
> -						   IRQ_TYPE_NONE);
> -		if (ret) {
> -			dev_err(&pdev->dev,
> -				"could not connect irqchip to gpiochip\n");
> -			goto out_disable;
> -		}
>  
> -		gpiochip_set_nested_irqchip(&stmpe_gpio->chip,
> -					    &stmpe_gpio_irq_chip,
> -					    irq);

> +		girq = &stmpe_gpio->chip.irq;
> +		girq->chip = &stmpe_gpio_irq_chip;
> +		/* This will let us handle the parent IRQ in the driver */
> +		girq->parent_handler = NULL;
> +		girq->num_parents = 0;
> +		girq->parents = NULL;
> +		girq->default_type = IRQ_TYPE_NONE;
> +		girq->handler = handle_simple_irq;
> +		girq->threaded = true;

Hmm, the GPIO-irq-chip setting are initialized after the GPIO-chip has been
registered by calling the gpiochip_add_data() method earlier in this method.
That means the IRQ-chip won't be created by the GPIOlib, which will introduce a
bug. In order to fix the problem it's better to move the whole "if (irq > 0)
{...}" clause to be executed before the gpiochip_add_data() function invocation.

Though I could misunderstood something...

-Sergey

>  	}
>  
>  	platform_set_drvdata(pdev, stmpe_gpio);
> -- 
> 2.26.2
>