I looked through the code in the mx4 kernel now a bit more and the reason for the hold wakelock is that there is simply data flowing. What triggers the data flow or what data is actually flowing is not clear yet but I don't see any HCI frames receiving on the bluetooth stack side so far when this happens.
Next steps are to find a proper fix:
1. Find out what is actually sending data and why
2. Look into the difference between mx4 and krillin on the kernel side and see if that helps us
3. Trace the trigger through the whole MTK stack and see what is actually causing the activity
I looked through the code in the mx4 kernel now a bit more and the reason for the hold wakelock is that there is simply data flowing. What triggers the data flow or what data is actually flowing is not clear yet but I don't see any HCI frames receiving on the bluetooth stack side so far when this happens.
Next steps are to find a proper fix:
1. Find out what is actually sending data and why
2. Look into the difference between mx4 and krillin on the kernel side and see if that helps us
3. Trace the trigger through the whole MTK stack and see what is actually causing the activity