I've read about restarting bme service when battery charge drops suddenly to 4%. And it works. However, I found that it also affects when battery is charged: if I restart BME when battery reaches 100%, it drops to a lower level (e.g. 93%) and charing continues. If I keep restarting BME using a while loop with a small delay between restarts (70 seconds), it almost never reaches 100%.

I wonder what is really happening, and also if it is dangerous?! May it cause overcharging? Why it doesn't reach 100%?

(It seems that this way battery lasts longer, and sudden drops in charge level are smaller).

I'm not sure about the dangers of such actions, but this commit has good description of how charging behaves once it's done: https://git.merproject.org/mer-core/...853bd74785bcec
I guess charging stops once bme detects that voltage didn't increase after certain amount of charging time and that is considered as the new "full capacity". Restarting bme causes this to go into charging/detecting cycle at least once more.

I don't know how it actually works, but bmestat output is very unreliable. When it shows 100%, my current capacity is equal to max capacity, while 'last full capacity' is less than that! But when I restart bme, it shows a lower capacity as the 'current' capacity (sometimes it is still higher than 'last full capacity'!!!).

Also, it always reports charging percent using 'max capacity' rather than 'last full capacity'; which is weird: either it should use 'last full capacity' as 100% full (how my laptop behaves), or your battery should never reach 100% when battery capacity decreases.

The voltage was almost fixed even when I restarted bme. But I can see negative current from battery which probably means that it is actually charging (it becomes positive when bme reaches 100%, but restarting it made it negative again).

I should do more tests, but I suspect that the 'current' number might be useful, as it seems that it worked reliably (except when bme thinks that battery is fully charged, where it starts to discharge) and it was becoming less (near zero!) as I kept restarting bme. But I will need to test more to make sure that it is actually the case. Next time I'm going to keep restarting bme to see if the current value ever reaches zero or very small near zero.

BTW, my battery decreased from 99% to 93% in 9:22 hours with 5mA average idle consumption reported by Battery Usage app (almost idle all the time). Which is something I don't remember if I have ever seen on my phone. (but I guess others have seen. I think my battery is faulty anyway).