Re: [PATCH] gpio: pxa: propagate pxa_gpio_probe_dt() error code

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Low] gpio: pxa: probe returns -EPERM (-1) when CONFIG_OF is disabled
--

commit e354788ab87bb0e86721190d9d799a35572742ca
Author: Haotian Zhang <vulab@iscas.ac.cn>

gpio: pxa: propagate pxa_gpio_probe_dt() error code

pxa_gpio_probe_dt() returns the negative errno produced by
devm_irq_alloc_descs(), which may be -ENOMEM, -EINVAL or -EEXIST. The
caller in pxa_gpio_probe() overwrites that value with a fixed -EINVAL, so
the actual reason for the failure is lost. Return irq_base so that the
original error code is propagated.

> diff --git a/drivers/gpio/gpio-pxa.c b/drivers/gpio/gpio-pxa.c
> index 3ff21e8992788..7d513932d09f0 100644
> --- a/drivers/gpio/gpio-pxa.c
> +++ b/drivers/gpio/gpio-pxa.c
> @@ -632,7 +632,7 @@ static int pxa_gpio_probe(struct platform_device *pdev)
>  	} else {
>  		irq_base = pxa_gpio_probe_dt(pdev, pchip);
>  		if (irq_base < 0)
> -			return -EINVAL;
> +			return irq_base;

[Severity: Low]
Does this unintentionally return -EPERM when CONFIG_OF is disabled?

When CONFIG_OF is not enabled, pxa_gpio_probe_dt() is defined as a stub that
returns -1:

#else
#define pxa_gpio_probe_dt(pdev, pchip)		(-1)
#endif

With this patch, pxa_gpio_probe() will now directly propagate that -1, which
translates to -EPERM. This seems like a misleading error code for missing
platform data or device tree configuration compared to the previous -EINVAL.

>  	}
>  
>  	if (!pxa_last_gpio)

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261009050604.3128124-1-vulab@iscas.ac.cn?part=1