[MX4] Battery drains quickly/device overheats

[Sturm Flut]

Arale has poor battery life compared to krillin, despite the battery having a 50% higher capacity. Not only does the device heat up considerably during normal use, also standby times are quite short for such a device. Here is some data I've collected:

arale fully charged, all radios off: 93% remaining after 9 hours of idling on my desk
arale fully charged, WiFi and 3G on: 72% remaining after 9 hours of idling on my desk

The second data set translates to a standby time of about 30 hours, while krillin can do several days with radios enabled.

I've already tried to do some debugging using /proc/timer_stats, cpustat and friends, but haven't found anything particularly interesting.

[Sturm Flut (sturmflut)]

I had a single process running amok on a single core for 23 minutes today. The whole device, really all of the surface, heated up so much that it was uncomfortable to touch it. It is hard to believe that a single CPU core can cause this amount of power dissipation.

[Harald Walker (4-walker)]

Don't know if relevant but on my MX4 scoperegistry is constantly using about 50% of the CPU.

[Pat McGowan (pat-mcgowan)]

That is bug #1470750 which has a fix landing soon

[Harald Walker (4-walker)]

Thanks Pat. After rebooting the process is gone now and so far my MX4 doesn't get hot when using it.

[You-Sheng Yang (vicamo)]

That 2nd dataset looks reasonable to me because Arale kernel just doesn't have correct WiFi suspend function implemented. From https://bugs.launchpad.net/tangxi/+bug/1471718 we know even with WoWLAN supported in wpasupplicant, suspend with WiFi connected still consumes 41mA all the time, and the best its Android build can do is 33mA or so. Compared to average consumed current ~10mA when suspend without WiFi connected, 41mA means 4 times power consumption, and you see 4 times battery level drop here, so that suspending with WiFi connected part just makes sense to me because we don't really have a way to improve chip vendor's WiFi driver right now.

However, 7% battery level drop within 9hr with radio off means 31mA consumed. That's 3 times higher than expected ~10mA and deserves some more investigation.

[George Vassilakes (george-vassilakes)]

I have experienced similar issues with my battery draining even when the phone is on standby.
I have checked using top on terminal and none of the process uses more than 1% of CPU.
Is there any information logged I can send you or any tests to run to try and help to find the cause?

[You-Sheng Yang (vicamo)]

@george-vassilakes, may I have your syslog? `adb pull /var/log/syslog .`

Colin Ian King has also other tools for diagnosing power problems.
Quote from one of his mail in other threads:

1. Use suspend-blocker [1] to parse the /var/log/syslog to get some
idea of the deep suspend characteristics.

I suggest doing:

rm /var/log/syslog
kill -9 $(pidof rsyslogd)

.. let phone run .. then.. grab a copy of /var/log/syslog and run
suspend-blocker on this:

suspend-blocker -b -d -H -r syslog

2. Capture cpu activity

Use cpustat [2] to capture process activity, e.g.

cpustat 60 > cpustat.log &

.. let phone run .. then stop cpustat (SIGKILL) and look at the log to
see if there any CPU hogging processes

3. Capture fork-exec-exits

Sometimes threads that come alive for a few milliseconds can suck CPU
and aren't easily traceable. So perhaps capturing activity with forkstat
[3] is an idea.

sudo forkstat >forkstat.log &

..again let phone run.. stop forkstat (SIGKILL) and look at the log to
see if we get any busy activity.

4. Look at wakeup-events.

Wakeup-events can sometimes show when a device and a process are context
switching a lot and causing lots of interrupts or context switches. I
suggest running eventstat to capture these

sudo eventstat 60 > eventstat.log &

..again let phone run.. stop eventstat (SIGKILL) and look at the log to
see if we get any busy activity.

5. Check for extraneous file system activity.

Perhaps we have a rogue app that is hammering the filesystem. We can see
what is happening using fnotifystat [5]

sudo fnotifystat -mT > fnotifystat.log

..again let phone run.. stop fnotifystat (SIGKILL) and look at the log
to see if we get any busy activity.

NOTE: fnotifystat will record ALL file activity by default, including
activity on /proc, /sys etc. You probably should exclude this using the
-x option

Also, fnotifystat will dump data to a file, so that also generates file
activity which it will log, so that makes is a little problematic to run
in the background.

These tools should capture enough data to allow one to get an idea of
what may be causing a system to be busy when it is in deep sleep.

Colin

== Tools ==

Latest versions can always be found in ppa:colin-king/white

[1] http://kernel.ubuntu.com/~cking/suspend-blocker/
[2] http://kernel.ubuntu.com/~cking/cpustat/
[3] http://kernel.ubuntu.com/~cking/forkstat/
[4] http://kernel.ubuntu.com/~cking/eventstat
[5] http://kernel.ubuntu.com/~cking/fnotifystat/

[Krzysztof Tataradziński (ktatar156)]

After OTA-5 it's even worst. 40 % in 10 hours of little usage of phone, wifi and mobile connection off most of time. In similar period before OTA-5 I think was 20-25%.

[Harald Walker (4-walker)]

Sounds like Ubuntu Touch needs a front end GUI app that shows you exactly which app or process used how many resources.

[Stefan Mikulaj (stefanmikulaj)]

 Even after the OTA5 my MX4 is getting hot, like 50-60 st C.
It got better after OTA5 for about a week or so, but after that it just like a radiator.
It happens when browsing the internet or in Ubuntu Store.
I have tried to swap the channels from stable version 3 to rc version 5, the heating is not so bad at the moment but it was the same with stable version 3 for a week or so.

This need to be fixed ASAP as it affect the battery big time.
With such a huge battery should the phone be able to last for 2 days, not barely 1.

[Dylan Coakley (dylan-coakley)]

I can confirm this happens to me. Either while web browsing or playing games (which creates heat really fast). I'm running OTA 5.

[janj (blackfed)]

Fully updated to 15.04-3 armhf(20150720-002916)

I received this phone three days ago. It discharges very quick when using above 2G connection, when using (trying to) GPS, surfing and video.

Also gets fairly hot though not to a temperature I think above 50 celcius.

The charging proces is out of hand. The animation and charging level is very un-reliable. Freezes when charging with the phone turned on - less with the phone shut down while charging. It seems though that charging finishes when at 100 percent, as the phone is cold after a while (after reaching 100 %).

Top in terminal shows no significant cpu-use from single processes (related to scope-registry fix)

Theres a significant magnetic field at the bottom, back side.

[Marco Graziotti (graziottimarco)]

> Arale r4 - Ubuntu 15.04 - armhf (20150825-170563)

Same problem for me, my Meizu MX4 have overheating when I use it for 10 or 15 minute in YouTube or similar websites that requires more work for the hardware.

Also overheat happen when I'm on a call for 10 minutes or more. The phone burning at the top.

I see that the top of the phone it's more subjected to overheat than bottom, probably because in this part there are CPU and GPU.

There's a scheduling for release a bugfix?

[You-Sheng Yang (vicamo)]

Hi, there is an internal tracking bug https://bugs.launchpad.net/tangxi/+bug/1446196 that has been closed with FIX-RELEASED for OTA-5, which back ports thermal manager from Android. It should help overheating problem as the statistics tell us so. There is also another internal bug https://bugs.launchpad.net/tangxi/+bug/1434433 that landed a work-around to vendor WiFi driver that doesn't really suspend correctly for OTA-4. This helps battery drain with WiFi turned on a lot from our observation. There is still one bug in the hunt related to media-hub high cpu usage, which indirectly brings down thermal performance as a side effect.

[Yuan-Chen Cheng (ycheng-twn)]

unless we have report against new revision (likt OTA5 or OTA6), we won't do further work on it.

[Krzysztof Tataradziński (ktatar156)]

For me, after OTA-6 problem still exist. I don't face overheating for now, but battery still drain really fast - it's hard to survive 24 hours for me.

[boehm (8-boehm)]

I have also OTA-6 on MX4 and my phone gets very, very hot during a call so that i feel a little headache after 15min phone call. my battery drains really fast -> it is usually empty in 18hours

[Krzysztof Tataradziński (ktatar156)]

I can report that for now, while phone is idle, battery drain is really low! 2-3% per 8 hours (WiFi and mobile data OFF, GPS ON). But when I use phone, drain is really fast.

[Krzysztof Tataradziński (ktatar156)]

For now, I'm 40 hours on (most time phone idle, with mobile data and WiFi off) and still got 77%. Problem is when we have mobile data turned on all the time. By night (8 hours), while phone is unused and with mobile data turned on - ~15 % usage of battery - I think that is little to much, comparing to 1-2% per 8 hours (yes, for now I'm getting that great score! ;) ) with mobile data off.

[Stefan Mikulaj (stefanmikulaj)]

After OTA-9 MX4 overheats and drain battery fast when browsing net.
I was on RC-proposed before and it overheats and drained battery fast as well, but it was different from day to day due to updating every day.
I have performed a full factory reset, also fastboot -w, flashed recovery and manually flashed 0TA-9,
set up the phone from scratch and now it overheats while browsing the internet.
Is there any progress with this bug?
I also have Nexus 4 and it doesn't have this problem. Is it hardware or software then?
Any reply would be nice.
Thank you.

[Stefan Mikulaj (stefanmikulaj)]

SYSTEM LOG FROM MY MX4 , SEE THE END OF LOG, WHEN I HAVE USED MOBILE 3G CONNECTION TO BROWSE THE NET. WITHIN COUPLE OF MINUTES THE PHONE GETS VERY HOT AND BATTERY GOES DOWN VERY FAST.

Feb 15 06:46:56 ubuntu-phablet kernel: [ 3220.908769] [AUTOSLEEP][try_to_suspend]abort due to autosleep_state: 0
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3220.908814] WQ warning! work (try_to_suspend, c0eb4b2c) execute more than 1 sec, time: 4919306692 ns
Feb 15 06:46:56 ubuntu-phablet powerd[755]: handle_requestWakeup from :1.78 - ubuntu push client, 1455519116
Feb 15 06:46:56 ubuntu-phablet powerd[755]: void {anonymous}::reset_alarm_clock() found 1 remaining wakeup requests
Feb 15 06:46:56 ubuntu-phablet powerd[755]: setting hardware wakeup time to 2016-02-15 06:51:56 for ubuntu push client
Feb 15 06:46:56 ubuntu-phablet powerd[755]: starting hardware alarm worker thread
Feb 15 06:46:56 ubuntu-phablet powerd[755]: calling wait_for_next_alarm
Feb 15 06:46:56 ubuntu-phablet powerd[755]: handle_clearSysState from :1.78, cookie: d7707a6d-15d4-4dfa-abfc-d83eda888d88
Feb 15 06:46:56 ubuntu-phablet powerd[755]: clear_sys_request: ubuntu push client - :1.78
Feb 15 06:46:56 ubuntu-phablet powerd[755]: name_watch_remove: looking for :1.78
Feb 15 06:46:56 ubuntu-phablet powerd[755]: name_watch: ref_count for :1.78 is now 0
Feb 15 06:46:56 ubuntu-phablet powerd[755]: no longer watching :1.78, there are no more requests
Feb 15 06:46:56 ubuntu-phablet powerd[755]: libsuspend: acquire_wake_lock: powerd_power_request
Feb 15 06:46:56 ubuntu-phablet powerd[755]: Enqueue state change to SUSPEND
Feb 15 06:46:56 ubuntu-phablet powerd[755]: libsuspend: prepare_suspend.
Feb 15 06:46:56 ubuntu-phablet powerd[755]: Emitting signal for transition to state SUSPEND (0)
Feb 15 06:46:56 ubuntu-phablet powerd[755]: entering suspend
Feb 15 06:46:56 ubuntu-phablet powerd[755]: libsuspend: calling enter_suspend
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.053679] [Ker_PM][request_suspend_state]sleep (0->3) at 3220964970170 (2016-02-15 06:46:56.147820462 UTC)
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.053715] [Ker_PM][request_suspend_state]sys_sync_work_queue early_sys_sync_work
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.053781] [Ker_PM][request_suspend_state]suspend_work_queue early_suspend_work
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.054684] mtk-tpd->fwu_show_product_id_1987:75TPKW:TW46:S3508:1793047
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.054713] mtk-tpd->synaptics_rmi4_early_suspend_3334:mode[1]
Feb 15 06:46:56 ubuntu-phablet powerd[755]: libsuspend: enter_suspend succeeded
Feb 15 06:46:56 ubuntu-phablet powerd[755]: Transition to SUSPEND complete
Feb 15 06:46:56 ubuntu-phablet powerd[755]: libsuspend: release_wake_lock: powerd_power_request
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.060563] mtk-tpd->synaptics_rmi4_control_gesture_664:write gesture all[0]mask(0)
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.060578] mtk-tpd->synaptics_rmi4_gesture_enable_546:gesture enable
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.060844] [FB Driver] enter early_suspend
Feb 15 06:46:56 ubuntu-phablet kernel: [ 3221.060862] [PWM] disp_pwm_set_backlight_cmdq(id = 0x1, level_1024 = 0...

[Stefan Mikulaj (stefanmikulaj-k)]

OTA-9, stable channel, MX4.
After charging to above 90 % in the evening, I have found my MX4 with depleted battery to 69 % after about 6.5 hours during night.
Please see the syslog attached.
All applications were closed, only one active was the Alarm/Clock.

[Stefan Mikulaj (stefanmikulaj-k)]

MX4 battery drain continue.
Wifi , mobile data off, gps, location... all off. Battery charged to 100% in the evening and after 6.5 hours reads 71 %.
No unlclosed application in background. Only set up alarm to 02:25 in the morning.
It's impossible to have such a drain with no application run in the background, this is either a serious software problem or my battery is defective or the indicator is completely wrong or what?
 The phone last all day though, with a light usage. And always getting hot when browsing web on mobile data connection.
Is there something else that I can do to provide some other logs or ???

[Stefan Mikulaj (stefanmikulaj-k)]

Yesterday I said enough, I have dirty flashed my MX4 with the OTA-9.
Fully charged battery in evening dropped to 96% only within approximately 10 hours.
Here is the syslog for the period of time when the device was 100% inactive, screen off, mobile data enabled, alarm off, gps off, wifi off.
How come that after I flashed it, the device battery dropped only 4 %??? Even with mobile data enabled???

[Harald Walker (4-walker)]

Stefan, after upgrading to OTA-9 I had a quick battery drain overnight once. Something like that happened many OTA releases ago as well and was related to some kind of stuck background process (there is a bug report for it and it should be fixed). A clean reboot solved it back then and also now. With wifi, gps and bluetooth off, I charge my OTA-9 MX4 maybe once per week, which I think is ok.

[Harald Walker (4-walker)]

With OTA-9.1 I can reproduce battery drain by using the camera app on the MX4
1. Charge phone to 100%
2. Start camera app.
3. Turn off phone (not shutdown, just let it be inactive)
4. Wait several hours. Battery drains 50% in 8 hours.
5. Restart phone.
6. Wait several hours.

Leaving the camera app running or quitting it does not make a difference. Must be some kind of background process.
After restarting the phone, battery usage is normal again.

[Julia Palandri (julia-palandri)]

I've seen this as well on Turbo, both with camera app and with telegram didn't getting connection (also, the phone got incredibly hot)

[Stefan Mikulaj (stefanmikulaj-k)]

I noticed that the phone overheats only on mobile data connection, probably due more strain putting on chipset when in low signal area (3g) never overheats on wifi connection.