[RFC PATCH v5 0/3] iio: position: add Rust driver for ams AS5600

[Muchamad Coirul Anwar <muchamadcoirulanwar@gmail.com>]

This is v5 of the Rust driver for the ams AS5600 12-bit magnetic rotary
position sensor.

Link: https://lore.kernel.org/linux-iio/20260707151542.91997-1-muchamadcoirulanwar@gmail.com/

Base tree and dependencies:

  This series is based on driver-core-testing [1], not vanilla rust-next.
  It depends on Gary Guo's io_projection-v6 [2] for IoBackend, IoBase,
  Region, and KnownSize.

  FallibleIoCapable is included in patch 1/3 following Danilo's
  suggestion [3] to carry it as a prerequisite until it lands upstream.

  [1] https://git.kernel.org/pub/scm/linux/kernel/git/driver-core/driver-core.git/log/?h=driver-core-testing
  [2] https://lore.kernel.org/driver-core/20260706-io_projection-v6-0-72cd5d055d54@garyguo.net/
  [3] https://lore.kernel.org/all/DJVQ852J7SOH.26YBIJTQ9B66G@kernel.org/

Changes since RFC v4:

  I2C abstraction:
  - Switch to FallibleIoCapable — IoCapable is infallible by design,
    but I2C can fail at transport level (Igor)
  - Restrict I/O ops to I2cClient<Bound> (Danilo)
  - Add smbus_read_word() and smbus_read_word_swapped() for odd offsets
  - Add FallibleIoCapable trait to io.rs with blanket impl
  - Add bit_usize() to bits.rs

  IIO abstraction:
  - Replace raw isize mask with IioChanInfo enum (Nuno)
  - Expand PinnedDrop SAFETY comment re: kernfs_drain() (Danilo)
  - build_device() takes modes parameter instead of hardcoding
  - channels() now returns &'static slice

  Driver:
  - Use smbus_read_word_swapped() instead of manual swap_bytes()
  - Use ARef<I2cClient<Bound>> (Danilo)
  - Drop pr_info! debug logging
  - Kconfig cleanup (Jonathan)

  Known limitations (unchanged):
  - No power management
  - No write_raw, buffer, or trigger support

Design notes:

  The IIO abstraction uses iio_device_alloc (not devm_*) so the Rust
  Drop controls cleanup ordering: unregister, drop driver data, then
  free iio_dev.

  iio_device_unregister() drains in-flight sysfs reads via kernfs_drain().
  This is sufficient for INDIO_DIRECT_MODE without buffer/trigger.
  Character device paths need separate analysis.

  Module ownership via __iio_device_register(), not iio_info.owner.

Muchamad Coirul Anwar (3):
  i2c: rust: implement SMBus access via IoBackend and FallibleIoCapable
  rust: add minimal IIO subsystem abstractions
  iio: position: add Rust driver for ams AS5600

 drivers/iio/position/Kconfig    |  11 +
 drivers/iio/position/Makefile   |   1 +
 drivers/iio/position/as5600.rs  | 189 +++++++++++++++++
 rust/bindings/bindings_helper.h |   2 +
 rust/kernel/bits.rs             |  29 +++
 rust/kernel/error.rs            |   1 +
 rust/kernel/i2c.rs              | 302 +++++++++++++++++++++++++++
 rust/kernel/iio.rs              | 384 +++++++++++++++++++++++++++++++++++
 rust/kernel/io.rs               |  66 ++++--
 rust/kernel/lib.rs              |   2 +
 10 files changed, 967 insertions(+), 20 deletions(-)
 create mode 100644 drivers/iio/position/as5600.rs
 create mode 100644 rust/kernel/iio.rs

---
Tested on BeagleBone Black (AM335x), kernel 7.2.0-rc1+,
AS5600 on i2c-2 (0x36) at 3.3V, 6mm diametric neodymium magnet.
Full test session: 2026-08-17.

Build: make ARCH=arm CROSS_COMPILE=arm-linux-gnueabihf- modules
       (zero warnings, zero errors)

Functional tests:

  1. Probe and registration:
     $ sudo insmod as5600.ko
     $ echo "as5600 0x36" > /sys/bus/i2c/devices/i2c-2/new_device
     $ cat /sys/bus/iio/devices/iio:device0/name
     as5600
     $ ls /sys/bus/i2c/devices/2-0036/driver
     2-0036  bind  module  uevent  unbind

  2. Raw angle and scale (magnet present):
     $ cat /sys/bus/iio/devices/iio:device0/in_angl_raw
     3366
     $ cat /sys/bus/iio/devices/iio:device0/in_angl_scale
     0.001533981
     20 consecutive reads, all within 0-4095.
     Computed: 3366 * 0.001533981 = 5.163 rad (~295.9 degrees).

  3. Unbind/rebind lifecycle (PinnedDrop with ARef cleanup):
     $ echo "2-0036" > /sys/bus/i2c/devices/2-0036/driver/unbind
     $ ls /sys/bus/iio/devices/iio:device0 2>&1
     ls: cannot access '...': No such file or directory
     $ echo "2-0036" > /sys/bus/i2c/drivers/as5600/bind
     $ cat /sys/bus/iio/devices/iio:device0/name
     as5600
     $ dmesg | grep -i "oops\|panic\|bug:"
     (empty)

  4. Concurrent stress (Mutex serialization under contention):
     10 parallel readers hammering in_angl_raw for 5 seconds.
     Repeated as 20-cycle unbind/rebind loop with readers active
     throughout (8 seconds total).
     $ dmesg | grep -i "oops\|panic\|bug:\|rcu"
     (empty)

  5. Module removal under active I/O:
     a) rmmod while flood readers are running — no crash.
        iio_device_unregister() drains in-flight read_raw callbacks
        before PinnedDrop proceeds; subsequent reads return ENOENT.
     b) rmmod while a sysfs fd is held open — no crash.
        iio_dev kref not released until fd is closed.
     $ dmesg | grep -i "oops\|panic\|bug:"
     (empty)

  6. Lifecycle stress:
     50x rapid unbind/rebind — no crash, device functional after all
     cycles.
     50x insmod/rmmod — no crash.
     $ dmesg | grep -i "oops\|panic\|bug:"
     (empty)

  7. I2C bus disconnect:
     SCL/SDA physically pulled while read loop is running. Driver
     returns "Remote I/O error" immediately on each failed transfer;
     no hang, no internal retry loop. Cable reconnected — reads resume
     from the next iteration without rmmod (about 2 seconds downtime).
     rmmod issued while bus still in error state — exits cleanly.
     ARef<I2cClient> drop is put_device() only, no bus transaction.
     $ dmesg | grep -i "oops\|panic\|bug:"
     (empty)

  8. Memory and locking:
     dmesg contains no strings matching "KASAN:" or "possible deadlock".
     The test kernel was not built with CONFIG_KASAN or CONFIG_PROVE_LOCKING;
     the above is a pattern match against kernel log output, not sanitizer
     or lockdep instrumentation.
     kmemleak not available (CONFIG_DEBUG_KMEMLEAK not set).

-- 
2.50.0