OK, it fails due to lack of calibration data + the behavior of bq27x00_battery module, that stops giving any data on actual max capacity when device is uncalibrated, instead of kindly giving exactly what the hardware is telling it (default value of 2048 mAh). I tried to convince Pali that arbitrary breaking sysfsnode ''because'' is a bad idea, and that it should report whatever chip is telling it, no matter if we, (again) arbitrary decided it's good or bad... Whatever. Probably, I should stop taking care and just put some "ERROR: Battery doesn't seem to be calibrated and kernel module is broken in this case, send bug reports to (...)" into BNF when sysnode fails, instead of it failing silently.

Anyway, you will get BNF reports back if you (as root):

Then, check what it reports, but I bet it's just uncalibrated. No idea why it havent calibrated itself - I would suggest charging it to max, again, disconnecting charger (checking if bnf says "VDQ: 1"), then using it normally, until it shuts itself down. If it doesn't calibrate *again*, we might want to debug it further - Pali ensured that device doesn't shut down before calibration takes place, and it seemed to work well. Maybe you just discovered a case, where it doesn't

Or you just haven't had VDQ bit set since start, for some reason, or it got invalidated (for example, by shutting device down or rebooting it too many times - play it safe, don't shut down/reboot between charging to full and shutting down itself due to low power. And check BNf output if VDQ is still 1, from time to time... It should turn to 0 in the very last moment before shutdown, jsut after device calibrates itself).

/Estel

// Edit

Kossuth beat me to it. Thanks for the compliments, although I must admit that my guide is a little outdated, in the sense that it describes pre- BME-Replacement scenario, and while using BME replacement it should be much, much more simple - just install everything, charge to full, and discharge 'till it shuts down. unlike standard BMe, BME-Replacement *is* supposed to *not* shut down before calibration takes place.

Of course, things from the guide aren't obsolete in the meaning that it won't work - is just more "complicated" than it may be required for end user using BME Replacement . Of course, when something goes wrong, it's still perfectly usable for debugging, thanks to big sections explaining inner workings and why we're doing this or that.