[Bug 222138] hp-wmi: HPWMI_GM queries fail with AE_AML_OPERAND_TYPE on OmniBook 5 16-ag1xxx (8DF4); firmware WHCM never calls GMCF

[bugzilla-daemon@kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=222138

--- Comment #4 from lorenzocollevecchio17@gmail.com ---
(In reply to Krishna Chomal from comment #2)
> Created attachment 311148 [details]
> Proposed patch for disabling fan support in hwmon
> 
> Hi Lorenzo,
> 
> Could you test the proposed patch and check if the WHCM errors appear only
> once during driver init and not regularly.
> 
> Also, all hwmon attributes have been disabled, this includes:
> - pwm1_enable
> - pwm1
> - pwm2
> - fan1_input
> - fan2_input
> 
> Thanks

Hi Krishna,

Tested your patch on the HP OmniBook 5 16-ag1xxx (board 8DF4, BIOS F.12).

Base: stable v7.2.10 (drivers/platform/x86/hp/hp-wmi.c), built out of tree
against 7.2.10-1-cachyos. Your diff does not apply to v7.2.10: hunks 2 and 4
fail, and the helpers hp_wmi_fan_control_supported(),
hp_wmi_fan_table_supported() and hp_wmi_get_active_fan_speed() do not exist in
this tree. I ported the same logic by hand:
  - added int fan_count to struct hp_wmi_hwmon_priv
  - in hp_wmi_hwmon_is_visible(): return 0 for hwmon_pwm and hwmon_fan when
priv->fan_count <= 0
  - in hp_wmi_setup_fan_settings(): priv->fan_count =
hp_wmi_get_fan_count_userdefine_trigger() right after priv->mode =
PWM_MODE_AUTO
  - in hp_wmi_hwmon_init(): the dev_info() and early return after the
IS_ERR(hwmon) check

Result after insmod of the patched module (stock hp_wmi unloaded first):

[ 1125.137950] hp_wmi: loading out-of-tree module taints kernel.
[ 1125.759817] input: HP WMI hotkeys as /devices/virtual/input/input27
[ 1126.171893] ACPI Error: Aborting method \_SB.WMID.WHCM due to previous error
(AE_AML_OPERAND_TYPE) (20260408/psparse-545)
[ 1126.172052] hp-wmi hp-wmi: No usable fans reported by WMI, disabling fan
support
[ 1126.582653] hp_wmi: Registered as platform profile handler

  - WHCM is logged once during init (the stock driver logged three, followed by
"Failed to apply initial fan settings: -22"). The -22 message is gone.
  - /sys/devices/platform/hp-wmi/hwmon/*/ contains only: device name power
subsystem uevent. No pwm1_enable, pwm1, pwm2, fan1_input or fan2_input.
  - Not regular: dmesg | grep -c WHCM was 4 (3 from boot + 1 from loading the
module) before and after calling sensors twice, and it stayed at 4.
  - The platform profile still registers (quiet balanced performance).

acpi_fan is unaffected by the patch: fan1_input is still present and readable
(it showed 0 RPM during a stress test at 80 °C).  Tested-by: Lorenzo
Collevecchio lorenzocollevecchio17@gmail.com

Thanks,
Lorenzo

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.