On Tue, 10 Dec 2024, Gerecke, Jason wrote:
> From: Jason Gerecke <jason.gerecke@xxxxxxxxx>
>
> Assigning a non-standard brightness value to an LED can cause the value
> to slowly drift downward over time as the effects of integer division
> accumulate. Each time that an LED is triggered, a series of set and get
> calls occur. For example, if we assume a tablet with max_hlv = 100, then
> when brightness is set to "200" through sysfs, the hlv value written to
> hardware will be `200*100/255 = 78`. If the LED trigger is later activated,
> the hlv value will be used to determine the brightness: `78*255/100 = 198`.
> This lower brightness then used to set the brightness of the next LED.
> However, `198*100/255 = 77`, so the next LED ends up slightly dimmer.
> Each subsequent trigger activation will cause the brightness to continue
> drifting down until we reach a point where the result of integer divsion
> does not introduce any new error.
>
> This commit corrects the issue by being more careful about how we handle
> scaling between the two ranges (0..max_{h,l}lv) and (0..LED_FULL).
>
> Signed-off-by: Jason Gerecke <jason.gerecke@xxxxxxxxx>
Both applied, thanks.
--
Jiri Kosina
SUSE Labs
