[PATCH v2 0/2] gpio: create the /sys/class/gpio mount point with GPIO_SYSFS disabled

[Bartosz Golaszewski]

Greg, Rafael,

The first patch in this series is small but may be seen as controversial
so here's a little backstory.

We have two sets of GPIO APIs currently in the kernel: the legacy one
based on numbers and the modern one using descriptors. Our goal is to
remove the old one from the kernel to which there are two obstacles: the
first one is easier and consists of converting all remaining in-kernel
users to the preferred API. This is tedious but it's all within our
control, just demands a lot of effort. The second obstacle is much harder
as it involves removing an existing kernel uABI that is the GPIO sysfs
interface at /sys/class/gpio.

Despite providing a number of user-space tools making using the GPIO
character device easier, it's become clear that some users just prefer
how the sysfs interface works and want to keep using it. Unless we can
provide a drop-in replacement, they will protest any attempts at
removing it from the kernel. As the GPIO sysfs module is the main user
of the global GPIO numberspace, we will not be able to remove it from
the kernel either.

I am working on a FUSE-based libgpiod-to-sysfs compatibility layer that
could replace the in-kernel sysfs and keep all the user-space programs
running but in order to keep it fully compatible, we need to be able to
mount it at /sys/class/gpio. We can't create directories in sysfs from
user-space and with GPIO_SYSFS disabled, the directory is simply not
there.

I would like to do what we already do for /sys/kernel/debug,
/sys/kernel/config, etc. and create an always-empty mount point at
/sys/class/gpio. To that end, I need the address of the /sys/class
kobject and the first patch in this series exports it.

The second adds a Kconfig switch in GPIOLIB and code that creates the
mount point.

I proposed creating an empty class stub in v1 but figured that a proper
mount point will work better.

If this is ok with you, please consider leaving your Ack or taking
patch 1/2 through your tree and providing me with an immutable tag.

To: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
To: Rafael J. Wysocki <rafael@xxxxxxxxxx>
To: Linus Walleij <linus.walleij@xxxxxxxxxx>
To: Bartosz Golaszewski <brgl@xxxxxxxx>
To: Kent Gibson <warthog618@xxxxxxxxx>
Cc: linux-kernel@xxxxxxxxxxxxxxx
Cc: linux-gpio@xxxxxxxxxxxxxxx
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@xxxxxxxxxx>

Changes in v2:
- Don't create an empty class but use sysfs_create_mount_point() instead
- Link to v1: https://lore.kernel.org/all/20241014121047.103179-1-brgl@xxxxxxxx/

---
Bartosz Golaszewski (2):
      driver core: class: expose the class kobject
      gpio: create the /sys/class/gpio mount point with GPIO_SYSFS disabled

 drivers/base/class.c         |  6 ++++++
 drivers/gpio/Kconfig         | 18 ++++++++++++++++++
 drivers/gpio/gpiolib.c       |  6 ++++++
 include/linux/device/class.h |  6 ++++++
 4 files changed, 36 insertions(+)
---
base-commit: b852e1e7a0389ed6168ef1d38eb0bad71a6b11e8
change-id: 20241014-gpio-class-mountpoint-2da8c54fafbe

Best regards,
-- 
Bartosz Golaszewski <bartosz.golaszewski@xxxxxxxxxx>