[PATCH 0/2] HID: fix battery queries with nonzero field offsets

[Mason Camara <ping@masoncamara.com>]

Magic Mouse 2 input report 0x90 places a status byte before the capacity
field. The synchronous battery query treats that byte as capacity and
reports 4% in sysfs when the device reports 95%.

Use the field offset and width from the report descriptor when processing
GET_REPORT responses, and add a UHID regression test.

Tool use:
  Codex (GPT-5) reviewed the completed series for submission-process,
  formatting, commit-message issues, and suggested edits to the commit
  messages. The code and test were written without AI assistance. I
  reviewed all resulting changes and take responsibility for the series.

Testing:
  - W=1 builds of drivers/hid/hid-input.o with
    CONFIG_HID_BATTERY_STRENGTH=y and =n
  - v7.2-rc2 plus patch 2 only: FAIL (capacity 4, expected 95)
  - v7.2-rc2 plus the complete series: PASS

Mason Camara (2):
  HID: input: honor report field offsets in battery queries
  selftests/hid: test battery queries with nonzero field offsets

 drivers/hid/hid-input.c                       | 39 ++++++++++----
 include/linux/hid.h                           |  6 +++
 tools/testing/selftests/hid/config            |  1 +
 .../testing/selftests/hid/tests/test_mouse.py | 54 +++++++++++++++++++
 4 files changed, 90 insertions(+), 10 deletions(-)


base-commit: 8cdeaa50eae8dad34885515f62559ee83e7e8dda
-- 
2.55.0