[tetris] Lockscreen freeze root cause identified - UdfpsController race condition

/e/OS Devices Nothing
1

Following up on this locked thread and GitLab issue #8521.

After digging into the android_frameworks_base source I believe I’ve identified the specific race condition causing the lockscreen
freeze on CMF Phone 1.

Root cause:

dispatchOnUiReady() schedules a 50ms delayed callback via mHandler.postDelayed() to enable HBM (high brightness mode) on the panel
sysfs node. If the finger lifts before that 50ms elapses, onFingerUp() runs — writes HBM “off” and tears down the overlay. Then
50ms later the delayed callback fires anyway and writes HBM “on”, leaving the display driver in an inconsistent state → full
freeze.

The December 2024 fix (d28bdd8c) addressed the Goodix HAL double-acquire but didn’t cancel this pending callback. That’s why
freezes reduced but didn’t stop entirely on later builds.

Fix:

One line in UdfpsController.java, in onFingerUp() before the sysfs write:

mHandler.removeCallbacksAndMessages(null);

Status: Currently testing via LSPosed hook on e/OS 3.5-a14-20260211580870 on my CMF Phone 1. Hook is confirmed active. Will report
back after a few days.

If it holds I’ll submit a proper MR to android_frameworks_base. Would appreciate if someone with GitLab access could flag this on
issue #8521.

1 Like
2

Finally! This is the killer bug that always teases me on the least favorable moment….

Hope to see the effect of your finding soon.

3

Update: Been running the hook for a few days now with no freezes. Stress tested by lifting my finger quickly and repeatedly to force the race condition, the hook is firing with no errors
in the logs. I am optimistic this was the issue but will keep monitoring over the next few days before calling it fixed and submitting the MR.

4

Sadly it locked up on me once, it was first time in almost a week. So not sure if fix helped or not will have to dig into it more. Just updated to 3.6 so will work from that.