I'm using a Fairphone 2 with Fairphone open OS, under Android 7, (build-number:
fp2_sibon-userdebug 7.1.2 19.11.2 sibon-24df0be9), up to date. My battery loses systematically 50% to 90% overnight, and after having tried all sorts of things to stop it, my brother checked in a terminal (using Termux) with the
top command which processes were using the CPU. A process named
irq/288-wcd9xxx uses 15-20% of the CPU permanently, which could explain why my battery simply dies in 12 hours. Here's a screenshot of what I get.
kill -9 has absolutely no effect on it (it seems to work when running the command, but the process is still there afterwards), does anyone have any idea about what it is, and how to kill it, or stop it?
Thank you Irfan Latif for your detailed answer, my brother and me started having a look into this, and for the moment, not much improvement:
grep -rI $(readlink /proc/$(pgrep -f irq/288-wcd9xxx)/exe) /*/etc/init/ /init.*.rc indeed doesn't yield any output, and ppid is 2, so as you thought, it appears to be a kernel thread.
(We still had a look at the
getprop | grep init.svc. list, and tried stopping
audiod just to see, but it didn't change anything.)
As this phone is a Fairphone (modular phone), we can remove the suspect hardware and reboot to see if it changes anything (suggestion by this fairphone thread I posted too, actually already saying it's a "kernel process", hadn't understood the implications: It's hardware interrupts coming from the audio codec).
But removing hardware didn't appear to change anything either, maybe swapping will work (my brother has the same phone ‒with a different android version‒, we'll try next time we meet), but at this stage we doubt it.
Last, we will try soon your suggestions to control the CPU usage of the process, maybe it will be a temporary fix, but this now seems to my brother to need a full system re-installation, as from the moment we did the installation (the standard fairphone open OS install), the battery always appeared to be excessively draining.
P.S. Another process,
irq/195-ngd_sli, seems to be eating up as well 5 to 7% permanently. We hadn't paid much attention to it, but it seems now just as abnormal; we'll dig this too.