Partial Fix for 2416967: Fix potential memory leak in PasswordUnlockScreen.
PasswordUnlockScreen was previously registering for multiple callbacks which *sometimes*
caused a memory leak because a reference stuck around in the callback list. I wasn't able to
track down the cause, but it looks like an interaction between switching between various lockscreen
modes and ordering of CONFIGURATION_CHANGED events.
I found one instance where the callback was being registered twice (fixed). However, I'm
still able to reproduce the bug but far less frequently.
As a workaround, I've added code to prevent adding additional callbacks and added logging to
detect when a callback is already registered.
Also fixed an instance where we'd recreate the password unlock screen in LockPatternKeyguardView.updateScreen()
so it should be quite a bit faster switching orientations now.
Renamed UnlockScreen.java to PatternUnlockScreen.java for consistency.
4 files changed
