[PATCH v6 0/3] hwmon: it87: add IT8613E support

[Jerome Tollet <jerome.tollet@gmail.com>]

This is a respin of Benoit Masson's v4 series, with his permission.
It replaces my fan-input-only RFC with the broader hardware monitoring
support requested in review.

The v4 series was dropped after a NULL pointer dereference was reported.
The IT8613E device entry did not initialize reg_pwm, while the probe path
unconditionally dereferences it. It also advertised six temperature
offsets but the register table contained only three entries.

Changes in v6:
- Drop the patch exposing additional temperature attributes. The existing
  type register handling only describes temperature inputs 1 to 3, so
  exposing temp4_type through temp6_type would not be safe.
- Remove the unused NUM_TEMP_OFFSET macro.
- Retain the legacy two groups of three temperature sources where
  appropriate, and honor the four global sources on IT8603E and IT8622E.
- Fix the legacy automatic/manual PWM control flow so the manual duty
  cache is not overwritten in automatic mode.
- Do not refresh PWM control state while changing modes. On newer
  controllers, write temperature mappings in both manual and automatic
  modes because the duty cycle uses a separate register. This keeps a
  mapping across cache refreshes and mode changes.
- Clarify that only three IT8613E temperature inputs are currently known.
  The six limit, offset, and PWM map resources do not imply six inputs,
  so FEAT_SIX_TEMP is intentionally not set.

Tested on an IT8613E at 0xa30, revision 12. The complete series builds as
an external module with W=1 and passes checkpatch --strict with no
warnings or errors.

The module exposes fan2_input and PWM2, PWM4, and PWM5 on this system.
The deferred temperature mapping path was tested by switching PWM2 to
manual mode, changing pwm2_auto_channels_temp from 4 to 1, waiting for a
cache refresh, and switching back to automatic mode. The value remained
1 after both the refresh and mode change. The original mapping, duty, and
automatic mode were then restored.

The PWM2 write path was tested by switching from automatic control to
manual full duty (255). Fan2 increased from 2789 RPM to 4821 RPM. The
original duty (65), automatic mode, and temperature mapping were then
restored.

An S3 suspend/resume cycle with an RTC wakeup completed successfully.
The PWM configuration and sensor readings were preserved, with no it87 or
hwmon warnings in the kernel log.

Link: https://lore.kernel.org/linux-hwmon/20260114221210.98071-1-yahoo@perenite.com/
Link: https://lore.kernel.org/linux-hwmon/20260829170545.33401-1-jerome.tollet@gmail.com/

benoit.masson (3):
  hwmon: it87: describe per-chip temperature resources
  hwmon: it87: prepare for extended PWM temp maps
  hwmon: it87: add IT8613E support

 Documentation/hwmon/it87.rst |   8 +
 drivers/hwmon/it87.c         | 418 ++++++++++++++++++++++++++++-------
 2 files changed, 342 insertions(+), 84 deletions(-)


base-commit: 75f2c0b3690702c90863c2e138cb5520670845ea
-- 
2.55.0