Re: [PATCH v20 2/2] power: reset: reboot-mode: Expose sysfs for registered reboot_modes

[Bartosz Golaszewski]

On Sun, Nov 30, 2025 at 7:21 PM Shivendra Pratap
<shivendra.pratap@xxxxxxxxxxxxxxxx> wrote:
>
> Currently, there is no standardized mechanism for userspace to discover
> which reboot-modes are supported on a given platform. This limitation
> forces tools and scripts to rely on hardcoded assumptions about the
> supported reboot-modes.
>
> Create a class 'reboot-mode' and a device under it to expose a sysfs
> interface to show the available reboot mode arguments to userspace. Use
> the driver_name field of the struct reboot_mode_driver to create the
> device.  For device-based drivers, configure the device driver name as
> driver_name.
>
> This results in the creation of:
>   /sys/class/reboot-mode/<driver>/reboot_modes
>

This should also be documented under Documentation/ABI/.

> This read-only sysfs file will exposes the list of supported reboot
> modes arguments provided by the driver, enabling userspace to query the
> list of arguments.
>
> Signed-off-by: Shivendra Pratap <shivendra.pratap@xxxxxxxxxxxxxxxx>
> ---
>  drivers/power/reset/reboot-mode.c | 76 +++++++++++++++++++++++++++++++++++++--
>  include/linux/reboot-mode.h       |  6 ++++
>  2 files changed, 80 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/power/reset/reboot-mode.c b/drivers/power/reset/reboot-mode.c
> index fba53f638da04655e756b5f8b7d2d666d1379535..ae03f2d96a84477f1e9f281bf3110911d7044a70 100644
> --- a/drivers/power/reset/reboot-mode.c
> +++ b/drivers/power/reset/reboot-mode.c
> @@ -3,6 +3,8 @@
>   * Copyright (c) 2016, Fuzhou Rockchip Electronics Co., Ltd
>   */
>
> +#define pr_fmt(fmt)    "reboot-mode: " fmt
> +
>  #include <linux/device.h>
>  #include <linux/init.h>
>  #include <linux/kernel.h>
> @@ -19,6 +21,56 @@ struct mode_info {
>         struct list_head list;
>  };
>
> +static ssize_t reboot_modes_show(struct device *dev, struct device_attribute *attr, char *buf)
> +{
> +       struct reboot_mode_driver *reboot;

This is not related to this patch but please consider proposing
renaming of this structure to struct reboot_mode_devicd because
calling it a driver is quite confusing where in reality it's a device.

> +       struct mode_info *info;
> +       ssize_t size = 0;
> +
> +       reboot = container_of(dev, struct reboot_mode_driver, reboot_mode_device);
> +       if (!reboot)
> +               return -ENODATA;

container_of(p, t, m) returns the address of the structure of type t
containing the member pointed to by p known under the name m in the
type of t. It just calculates the offset between the address of m in t
and p. It's not possible for it to return NULL.

> +       scoped_guard(mutex, &reboot->reboot_mode_mutex) {
> +               list_for_each_entry(info, &reboot->head, list)
> +                       size += sysfs_emit_at(buf, size, "%s ", info->mode);
> +       }
> +
> +       if (!size)
> +               return -ENODATA;
> +
> +       return size + sysfs_emit_at(buf, size - 1, "\n");
> +}
> +static DEVICE_ATTR_RO(reboot_modes);
> +
> +static struct attribute *reboot_mode_attrs[] = {
> +       &dev_attr_reboot_modes.attr,
> +       NULL,
> +};
> +ATTRIBUTE_GROUPS(reboot_mode);
> +
> +static const struct class reboot_mode_class = {
> +       .name = "reboot-mode",
> +       .dev_groups = reboot_mode_groups,
> +};
> +
> +static void reboot_mode_device_release(struct device *dev)
> +{
> +    /* place holder to avoid warning on device_unregister. nothing to free */
> +}
> +
> +static void reboot_mode_register_device(struct reboot_mode_driver *reboot)
> +{
> +       int ret;
> +
> +       reboot->reboot_mode_device.class = &reboot_mode_class;
> +       reboot->reboot_mode_device.release = reboot_mode_device_release;
> +       dev_set_name(&reboot->reboot_mode_device, reboot->driver_name);
> +       /* Check return value to avoid compiler warning */
> +       ret = device_register(&reboot->reboot_mode_device);
> +       if (ret)
> +               pr_debug("device_register failed for %s : %d\n", reboot->driver_name, ret);
> +}

I'm not sure if this has been addressed before but why is this even
optional? Why don't we just return -1 here if we fail to register the
device?

And just for the record: I'm still convinced that using
device_create() here will result in nicer code and allow us to contain
the associated reboot_mode_device inside this compilation unit.

If Bjorn really insists on keeping it this way though, then you need
at least a call to device_initialize() here.

> +
>  static unsigned int get_reboot_mode_magic(struct reboot_mode_driver *reboot,
>                                           const char *cmd)
>  {
> @@ -76,6 +128,7 @@ int reboot_mode_register(struct reboot_mode_driver *reboot)
>         size_t len = strlen(PREFIX);
>         int ret;
>
> +       mutex_init(&reboot->reboot_mode_mutex);
>         INIT_LIST_HEAD(&reboot->head);
>
>         for_each_property_of_node(np, prop) {
> @@ -112,6 +165,7 @@ int reboot_mode_register(struct reboot_mode_driver *reboot)
>
>         reboot->reboot_notifier.notifier_call = reboot_mode_notify;
>         register_reboot_notifier(&reboot->reboot_notifier);
> +       reboot_mode_register_device(reboot);
>
>         return 0;
>
> @@ -132,9 +186,13 @@ int reboot_mode_unregister(struct reboot_mode_driver *reboot)
>         struct mode_info *info;
>
>         unregister_reboot_notifier(&reboot->reboot_notifier);
> +       if (device_is_registered(&reboot->reboot_mode_device))
> +               device_unregister(&reboot->reboot_mode_device);

If you bail out of reboot_mode_register_device(), you don't need the
above check anymore because the device could

Bart