How to resolve extensive CPU consumption after UBports update?

marc_aurel

@AlainW94 Thanks, this was indeed a plausible assumption. Unfortunately enabling the flight mode, doesn't change the behaviour:

%CPU   %USR   %SYS   PID S  CPU   Time Task
29.41  21.01   8.40  2262 S    3  4.28m NetworkManager
21.85  18.49   3.36   812 S    2  3.69m dbus-daemon
15.13   6.72   8.40     1 S    2  2.09m /sbin/init
 7.56   0.84   6.72 18258 R    2  3.24s cpustat
 4.20   3.36   0.84 19903 S    0 27.49s unity8
 3.36   2.52   0.84  2471 R    2 45.48s upstart-dbus-bridge
 2.52   1.68   0.84  4142 R    3 18.83s /usr/lib/arm-linux-gnueabihf/indicator-network/indicator-network-service
 2.52   1.68   0.84  3625 S    2 18.67s /usr/lib/NetworkManager/nm-dispatcher
 2.52   1.68   0.84 28866 S    3  8.04s unity8-dash
 2.52   0.84   1.68  5008 R    2 35.56s /opt/click.ubuntu.com/dekko2.dekkoproject/current/lib/arm-linux-gnueabihf/bin/dekkod
 2.52   1.68   0.84  3556 S    2 20.62s /usr/lib/arm-linux-gnueabihf/sync-monitor/sync-monitor
 1.68   1.68   0.00 20085 R    2  0.24s upstart
 1.68   0.84   0.84  1005 S    3 19.59s /lib/systemd/systemd-udevd
 1.68   1.68   0.00 22193 S    2 13.34s /usr/bin/zeitgeist-daemon
 1.68   0.84   0.84   832 S    2 12.93s upstart-file-bridge
 1.68   0.84   0.84  5654 S    2 16.77s /usr/lib/evolution/evolution-calendar-factory-subprocess
 1.68   0.84   0.84  2163 S    0 13.54s /usr/lib/udisks2/udisksd
 1.68   0.00   1.68 27986 S    3  6.70s maliit-server
 1.68   0.84   0.84  3420 S    2 19.56s /usr/lib/arm-linux-gnueabihf/thumbnailer/thumbnailer-service
 1.68   1.68   0.00  1845 R    0 48.79s upstart
 1.68   1.68   0.00  6573 S    2 19.16s /usr/bin/msyncd
 1.68   0.00   1.68   176 S    0 13.06s [mmcqd/0]
 1.68   1.68   0.00  5610 S    2 17.24s /usr/lib/evolution/evolution-calendar-factory-subprocess
 1.68   0.84   0.84  3355 S    2 17.56s /usr/lib/evolution/evolution-addressbook-factory-subprocess
 1.68   0.84   0.84   997 S    2 17.93s upstart-udev-bridge
 1.68   0.84   0.84  2724 R    2 35.31s /usr/lib/telepathy/mission-control-5
 1.68   1.68   0.00  5199 S    2 16.77s /usr/lib/evolution/evolution-calendar-factory-subprocess
 1.68   0.00   1.68    36 S    0  6.66s [kworker/u:1]
 0.84   0.84   0.00   807 S    0 11.67s rsyslogd
 0.84   0.84   0.00  2074 R    0 13.53s upstart-udev-bridge
 0.84   0.00   0.84   297 S    2  4.24s [jbd2/mmcblk0p20]
 0.84   0.00   0.84 21454 S    2  0.17s trust-stored-skeleton
 0.84   0.84   0.00   941 S    3  5.14s /system/bin/logd
 0.84   0.00   0.84     4 S    0  1.19s [kworker/0:0]
 0.84   0.00   0.84     2 S    2  1.80s [kthreadd]
 0.84   0.84   0.00  2708 S    3 17.76s upstart-dbus-bridge
 0.84   0.00   0.84  3021 S    2 12.58s upstart-socket-bridge
 0.84   0.00   0.84  1031 S    3  2.32s /usr/lib/accountsservice/accounts-daemon
 0.84   0.00   0.84 18969 S    2  0.06s [kworker/2:1]
 0.84   0.00   0.84 31587 S    0  2.63s [kworker/u:7]
 0.84   0.00   0.84   840 S    3  4.35s /sbin/ueventd
 0.84   0.84   0.00 27982 S    0  2.78s /usr/lib/ubuntu-push-client/ubuntu-push-client
 0.84   0.00   0.84   681 S    2 12.61s upstart-local-bridge
 0.84   0.00   0.84   305 S    0  3.43s [loop0]

guru

You can strace the NetworkManager proc to try to understand what it keeps so busy, run:

sudo su
strace -o /tmp/nm.tr -p PID
Ctrl-C
vi /tmp/nm.tr

PID is the proc id, 2262 in the above screen.

marc_aurel

@guru Thanks. Here is the document.

I have limited skills to interpret the entries of the log. There is a repeating entry with futex. I think this could be reated to the error on the screenshot, I've posted.

AlainW94

@marc_aurel They are a lot of entry related to anbox0, maybe an issue with Anbox ?

guru

Try to solve this:

$ grep /sys/devices/virtual/net/anbox0/driver nm.tr
readlinkat(AT_FDCWD, "/sys/devices/virtual/net/anbox0/driver", 0xfaa008, 99) = -1 ENOENT (No such file or directory)
readlinkat(AT_FDCWD, "/sys/devices/virtual/net/anbox0/driver", 0xfb1440, 99) = -1 ENOENT (No such file or directory)
readlinkat(AT_FDCWD, "/sys/devices/virtual/net/anbox0/driver", 0xf87600, 99) = -1 ENOENT (No such file or directory)
readlinkat(AT_FDCWD, "/sys/devices/virtual/net/anbox0/driver", 0xfab388, 99) = -1 ENOENT (No such file or directory)

marc_aurel

@guru Thanks for your suggestion. I have reinstalled the boot image as described in the documentation. Now it seems to be solved:

adb shell
sudo reboot -f bootloader # 'adb shell' will exit after this command
wget http://cdimage.ubports.com/anbox-images/anbox-boot-$CODENAME.img
sudo fastboot flash $PARTITIONNAME anbox-boot-$CODENAME.img
sudo fastboot reboot
rm anbox-boot-$CODENAME.img

It seems, that the updates of UBports doesn't consider an Anbox installation and cause a conflict.

guru

@marc_aurel said in How to resolve extensive CPU consumption after UBports update?:

@guru Thanks for your suggestion. I have reinstalled the boot image as described in the documentation. Not it seems to be solved:
....

Is it solved NOW or NOT?

marc_aurel

@guru Sorry. I don't know, how the "Not" came into the sentence. The device works fast now.

guru

You might concider edit your post.

Any way, glad you solved your problem, even if I want to add, that I'd not install and use any Android app without having the source.

marc_aurel

@guru Makes sense.

I also agree according Android apps. I deon't need them either. I test Anbox for our planned device for a conumer market, because most people of our focus group ask clearly for Android support.

The good thing of Anbox is, that this approach is a sandboxed container with no access to system ressources outside of the container. It's not yet possible to shut down and start Anbox to be sure, that the apps are only running, if I need them, a banking app for example. But I tthink, it could be possible. I tink it would be even possible to add more controll to this Android emulation.