Slimbook Compatibilty addition #12
Conversation
|
My apologies, I did it accidentally. Maybe we can come up with something in the future. |
|
This adds compatibility for:
Also adds silent-mode feature for this models. |
pdev.c
Outdated
| @@ -224,6 +292,8 @@ static DEVICE_ATTR_RW(fan_always_on); | |||
| static DEVICE_ATTR_RW(fan_reduced_duty_cycle); | |||
| static DEVICE_ATTR_RW(manual_control); | |||
| static DEVICE_ATTR_RW(super_key_lock); | |||
| static DEVICE_ATTR_RW(fan_boost); | |||
There was a problem hiding this comment.
I am not sure a new attribute is needed, see: https://github.com/torvalds/linux/blob/f2906aa863381afb0015a9eb7fefad885d4e5a56/Documentation/ABI/testing/sysfs-class-hwmon#L289 ? Enabling fan boost is already implemented via the pwm1_enable attribute of the hwmon device created in hwmon_pwm.c, which uses fan.c:qc71_fan_set_mode() to turn it on, although that writes the FAN_CTRL register directly. I am now wondering if using the TRIGGER_1 register might be better?
There was a problem hiding this comment.
if fan_boost is the same as a 100% duty cycle pwm... then yes, exposing fan_boost is not necessary.
I am now wondering if using the TRIGGER_1 register might be better?
On our tests, switch through TRIGGER_1 triggers a WMI event. Tested with super key, fan boost and lightbar. On trigger 2 If I am not wrong, we could increase/decrease keyboard backlight. I can double check if you find interesting to have it documented.
That is the only benefit we found of trigger register. Unfortunately, TRIGGER_1_SILENT_MODE doesn't seem to switch what we call "Silent Mode" so we do it by writing FAN_CTRL instead.
Also, I've noticed fan.c functions are atomic, protected by a mutex and pdev.c aren't. As this is triggered by an human action (pressing Fn+F5) race conditions seem unlikely... however, we can move "silent mode" code to fan.c and surround it with a mutex
* Added Slimbook silent mode sysfs access
|
Sorry for the mess, I haven't much experience with this on github. I expected PR to be smarter. |
* Reduced verbosity * Using hex literals for events
… a proper led interface will be used but we need this for testing
* Added missing attribute condition check
…odels * hero/titan may be managed in manual mode as it looks like automatic does not emit a performance event
No description provided.