Re: [PATCH v2 1/9] gpio: sysfs: add a parallel class device for each GPIO chip using device IDs

[Andy Shevchenko <andriy.shevchenko@intel.com>]

On Mon, Jun 23, 2025 at 10:59:49AM +0200, Bartosz Golaszewski wrote:
> 
> In order to enable moving away from the global GPIO numberspace-based
> exporting of lines over sysfs: add a parallel, per-chip entry under
> /sys/class/gpio/ for every registered GPIO chip, denoted by device ID
> in the file name and not its base GPIO number.
> 
> Compared to the existing chip group: it does not contain the "base"
> attribute as the goal of this change is to not refer to GPIOs by their
> global number from user-space anymore. It also contains its own,
> per-chip export/unexport attribute pair which allow to export lines by
> their hardware offset within the chip.
> 
> Caveat #1: the new device cannot be a link to (or be linked to by) the
> existing "gpiochip<BASE>" entry as we cannot create links in
> /sys/class/xyz/.
> 
> Caveat #2: the new entry cannot be named "gpiochipX" as it could
> conflict with devices whose base is statically defined to a low number.
> Let's go with "chipX" instead.
> 
> While at it: the chip label is unique so update the untrue statement
> when extending the docs.

...

>  struct gpiodev_data {
>  	struct gpio_device *gdev;
>  	struct device *cdev_base; /* Class device by GPIO base */
> +	struct device *cdev_id; /* Class device by GPIO device ID */

I would add it in the middle in a way of the possible drop or conditional
compiling of the legacy access in the future.

>  };

...

> +static int export_gpio_desc(struct gpio_desc *desc)
> +{
> +	int offset, ret;

Why offset is signed?

> +	CLASS(gpio_chip_guard, guard)(desc);
> +	if (!guard.gc)
> +		return -ENODEV;
> +
> +	offset = gpio_chip_hwgpio(desc);
> +	if (!gpiochip_line_is_valid(guard.gc, offset)) {
> +		pr_debug_ratelimited("%s: GPIO %d masked\n", __func__,
> +				     gpio_chip_hwgpio(desc));

Can we use gdev here? (IIRC we can't due to some legacy corner cases)

> +		return -EINVAL;
> +	}
> +
> +	/*
> +	 * No extra locking here; FLAG_SYSFS just signifies that the
> +	 * request and export were done by on behalf of userspace, so
> +	 * they may be undone on its behalf too.
> +	 */
> +
> +	ret = gpiod_request_user(desc, "sysfs");
> +	if (ret)
> +		return ret;
> +
> +	ret = gpiod_set_transitory(desc, false);
> +	if (ret) {
> +		gpiod_free(desc);
> +		return ret;
> +	}
> +
> +	ret = gpiod_export(desc, true);
> +	if (ret < 0) {
> +		gpiod_free(desc);
> +	} else {
> +		set_bit(FLAG_SYSFS, &desc->flags);
> +		gpiod_line_state_notify(desc, GPIO_V2_LINE_CHANGED_REQUESTED);
> +	}
> +
> +	return ret;
> +}

...

> +static struct device_attribute dev_attr_export = __ATTR(export, 0200, NULL,
> +							chip_export_store);

__ATTR_WO()

...

> +static struct device_attribute dev_attr_unexport = __ATTR(unexport, 0200,
> +							  NULL,
> +							  chip_unexport_store);

Ditto.

...

> +static struct attribute *gpiochip_ext_attrs[] = {
> +	&dev_attr_label.attr,
> +	&dev_attr_ngpio.attr,
> +	&dev_attr_export.attr,
> +	&dev_attr_unexport.attr,
> +	NULL,

No comma for the terminator, please.

> +};

...

> +	data->cdev_id = device_create_with_groups(&gpio_class, parent,
> +						  MKDEV(0, 0), data,
> +						  gpiochip_ext_groups,
> +						  "chip%d", gdev->id);
> +	if (IS_ERR(data->cdev_id)) {
> +		device_unregister(data->cdev_base);
> +		kfree(data);

UAF

> +		return PTR_ERR(data->cdev_id);
> +	}

-- 
With Best Regards,
Andy Shevchenko