[REF] Known identified battery drainers

JustinW3 · Mar 12, 2012

Entropy512 said:
You claim now to have used BBS and CPUSpy - but there isn't a single shred of evidence in your existing posts prior to this that indicate you have used it. No wakelock lists, no indication of what your deepsleep percentages were, nothing useful. All you've done is post some stats from the stock Android battery statistics screen which are basically useless.

Fine, I will post info from BBS next time I experience the drain, I disabled Facebook notifications and disabled the automatic refresh of the status updates for the widget to see if that was causing wakelocks. I have previously posted stats from CPUSpy but didn't in this case because they weren't any good. I had forgotten to reset the timers so there were like 3 days of data in there.

So far since I made the mentioned changes, the idle seems to be extremely improved. The AOS says it's been on the battery for 4+ hrs. CPUSpy is accounting for 3h 12 min and showing 96% of that in deep sleep. About phone says battery is still at 99%.

akaleshnikov · Mar 13, 2012

Entropy, attached are my wakelocks.txt and dmesg.txt

I would've analyzed in excel myself but I've never done formulaic work with it, and couldn't figure out how to do it.

Total wake time: 1hr 13m
Battery Drain: 4%

CPU Spy information:
80% deep sleep 59:16
9% 200mhz 06:43
6% 800mhz
2% 1000mhz
1% 500mhz

Firmware: CheckROM RevoHD V6
Kernel: Siyah 2.6.13 (noop I/O scheduler, smartassV2 governor, underclocked 1000mhz and undervolted)
Radio: KH3

I'll try and get a longer reading by doing it overnight, but I use Sleep as An Android to wake me up in the morning and that'd skew the data heavily

Spo0f · Mar 14, 2012

Hi guys,

while i'm reading through the thread, here is some wireshark from me. I see a little of the mentioned bug, few APR request, 2-3 IPv6 requests and huge amount of things that i cannot identify. Expect Yahoo ofc, but it does not seems to be that huge battery drain.

The problem is that at workside i get HUGE battery drain on WiFi - like 50% for 8 hours.

Can someone help me with this? Usually i have huge locks from wlan_wake and deleted_wake_lokcs.

This is with LA4/LA6.

I've identify that the biggest packages are send and received to Google and something like "CDNetworks".

Thank you in advance.

duowing · Mar 15, 2012

I've been following this thread and thanks to it have pretty much solved any issues with my phone constantly waking on 3G. With 3G on I pretty much have great battery life. With Wi-fi on my phone never goes to sleep. I've used wireshark over say a 10 minute period to try and analyze things. I see stuff, but I'm not exactly sure what it is. The biggest thing is that it appears my router/modem is continuously sending out ARP requests. It goes through what appears every single address and just continues to do it over and over and over. I don't know if this is the reason I'm not sleeping on wi-fi? I got up in the morning unplugged my phone rebooted it, reset cpu spy, and after a little over an hour of time on. Didn't touch the phone at all, I had 7 minutes of deep sleep. The rest remained at 200Mhz. I know this is the SII forum and I'm using the Galaxy Note, but hopefully you guys can give me some info. Is this just my phone or is it something going ridiculous with my wireless?

I just did a quick couple minute poll on my Galaxy Nexus on Wi-Fi and it looks like it's seeing these same ARP requests, but it looks like my Nexus at least goes into deep sleep. I added the dump from the Nexus as well. These are at different days/different times so unfortunately network activity may be different between the two. Hopefully someone can compare these and see if they have any recommendations.

Thanks!

alacrify · Mar 15, 2012

duowing said:
I've been following this thread and thanks to it have pretty much solved any issues with my phone constantly waking on 3G. With 3G on I pretty much have great battery life. With Wi-fi on my phone never goes to sleep. I've used wireshark over say a 10 minute period to try and analyze things. I see stuff, but I'm not exactly sure what it is. The biggest thing is that it appears my router/modem is continuously sending out ARP requests. It goes through what appears every single address and just continues to do it over and over and over. I don't know if this is the reason I'm not sleeping on wi-fi? I got up in the morning unplugged my phone rebooted it, reset cpu spy, and after a little over an hour of time on. Didn't touch the phone at all, I had 7 minutes of deep sleep. The rest remained at 200Mhz. I know this is the SII forum and I'm using the Galaxy Note, but hopefully you guys can give me some info. Is this just my phone or is it something going ridiculous with my wireless?

I just did a quick couple minute poll on my Galaxy Nexus on Wi-Fi and it looks like it's seeing these same ARP requests, but it looks like my Nexus at least goes into deep sleep. I added the dump from the Nexus as well. These are at different days/different times so unfortunately network activity may be different between the two. Hopefully someone can compare these and see if they have any recommendations.

Thanks!

I didn't look at the shark dump (don't have the resources at work), but sounds like the Note has the same sucky problem the GSII does with UCKK6 - it doesn't filter out the ARP requests like it should. I fixed this by killing the services on my Win 7 network that were doing the continual broadcasts. You can find that here: http://xdaforums.com/showpost.php?p=23337687&postcount=514

If that doesn't do it, maybe someone else here will chime in, but you're probably better off in the Note forums building off each others knowledge.

Entropy512 · Mar 15, 2012

duowing said:
I've been following this thread and thanks to it have pretty much solved any issues with my phone constantly waking on 3G. With 3G on I pretty much have great battery life. With Wi-fi on my phone never goes to sleep. I've used wireshark over say a 10 minute period to try and analyze things. I see stuff, but I'm not exactly sure what it is. The biggest thing is that it appears my router/modem is continuously sending out ARP requests. It goes through what appears every single address and just continues to do it over and over and over. I don't know if this is the reason I'm not sleeping on wi-fi? I got up in the morning unplugged my phone rebooted it, reset cpu spy, and after a little over an hour of time on. Didn't touch the phone at all, I had 7 minutes of deep sleep. The rest remained at 200Mhz. I know this is the SII forum and I'm using the Galaxy Note, but hopefully you guys can give me some info. Is this just my phone or is it something going ridiculous with my wireless?

I just did a quick couple minute poll on my Galaxy Nexus on Wi-Fi and it looks like it's seeing these same ARP requests, but it looks like my Nexus at least goes into deep sleep. I added the dump from the Nexus as well. These are at different days/different times so unfortunately network activity may be different between the two. Hopefully someone can compare these and see if they have any recommendations.

Thanks!

I'll need to look at the dumps when I get time (I've been busy lately) - but which firmware are you on?

Some bases (XXKI3, UCKK6, probably XWKK5) block the ARP filters from working right, on other bases (XWKL1, XWLA4) the ARP filters seem to work well.

duowing · Mar 15, 2012

I'm on the LB1 Rom for the note, I'm now running a custom Rom and I just flashed the LA4 Baseband as someone said it seems to work better with data connections. So I'm giving that a try. I saw there was a new LC1 Rom/Kernel with LB2 baseband just released for the Note. Supposedly this will be the last GB rom before ICS so I'm going to give that a try. As for 3G I checked after about an hour and a half in CPU spy and my deep sleep was I believe 96%. So I'm good as far as programs running in the background goes. At least over mobile data. Thanks for the help guys. I was looking around in the Note section and couldn't find any real topics on this.

jsolares · Mar 15, 2012

Latitude was eating through my battery, now i wonder if the drain i had on the ICS roms wasn't also due to Latitude, i signed out of latitude and disabled wifi for location.

simplicityy · Mar 16, 2012

I keep wanting to install Skype to make use of my phone's front camera, but I know it's a wreck of an app. Better app? Or do you think since Microsoft bought out Skype recently that they will fix it?

JustinW3 · Mar 18, 2012

simplicityy said:
I keep wanting to install Skype to make use of my phone's front camera, but I know it's a wreck of an app. Better app? Or do you think since Microsoft bought out Skype recently that they will fix it?

I wouldn't hold my breath for MS fixing Skype. Why not keep Skype and just freeze it on TiBu when you're not using it. Then if you want to use it, thaw it.

Will this work? I'm not 100% on it, can somebody else comment on this idea?

sirhc81 · Mar 18, 2012

my results

hello entropy, just wanted to post my results and say thank you for all of your fine work, i'm currently running unnamed 1.3.2 w/standard kernel. if i can improve the battery life any further just let me know

sirhc81 · Mar 18, 2012

sirhc81 said:
hello entropy, just wanted to post my results and say thank you for all of your fine work, i'm currently running unnamed 1.3.2 w/standard kernel. if i can improve the battery life any further just let me know

and here's my battery drain....

chamonix · Mar 18, 2012

sirhc81 said:
hello entropy, just wanted to post my results and say thank you for all of your fine work, i'm currently running unnamed 1.3.2 w/standard kernel. if i can improve the battery life any further just let me know

You main problem is network traffic. Looking at the alarms and the fact that wifi is on I believe that google maps is besides facebook the reason for the overhead. You could try to disable wifi when not using it to see if part of the awake time goes down (as google maps caching options distinguish between Wifi and Cell)

dandrumheller · Mar 28, 2012

multipdp kernel wakelock

wondering if anyone can help me track down a kernel wakelock that I am seeing on ICS that I wasn't on GB?

I'm currently on ShoStock2 RC3, and I'm seeing this "multipdp" kernel wakelock in BBS. It is by far the highest time wakelock on the device. As I write this it is at 42m of kernel wakelock in 8h6m of uptime, with 435 counts, so it's firing pretty close to once a minute. I don't ever recall seeing it on GB. I'm currently monitoring it while on the mobile data network, and will check to see if it behaves similarly on WiFi.

Google results showed a couple mentions of it in the BBS thread and the i9100 forum here on XDA, and it was also mentioned in a poor battery life thread for i9100 on AndroidForums. But in all these cases, the post was never followed up by an answer that addressed it (or at least I didn't see it - read about 100 posts past it's mention in i9100 forum here...)

Could this be caused by an app misbehaving because it's not built properly for ICS? I'm running the same apps I had on GB ROMs, and not noticing any "out of the ordinary" data use by installed apps.

In addition to this (perhaps related) is an apparent discrepancy between reported awake time in BBS and the details of the android OS item in the built in battery monitoring in ICS. Android OS seems to report significantly higher awake time than BBS. Right now, Android OS reports 3h18m of awake time, while BBS reports 2h37m.

Hope someone can point me in a direction to try to troubleshoot this...

Thanks,

dd

Entropy512 · Mar 30, 2012

dandrumheller said:
wondering if anyone can help me track down a kernel wakelock that I am seeing on ICS that I wasn't on GB?

I'm currently on ShoStock2 RC3, and I'm seeing this "multipdp" kernel wakelock in BBS. It is by far the highest time wakelock on the device. As I write this it is at 42m of kernel wakelock in 8h6m of uptime, with 435 counts, so it's firing pretty close to once a minute. I don't ever recall seeing it on GB. I'm currently monitoring it while on the mobile data network, and will check to see if it behaves similarly on WiFi.

Google results showed a couple mentions of it in the BBS thread and the i9100 forum here on XDA, and it was also mentioned in a poor battery life thread for i9100 on AndroidForums. But in all these cases, the post was never followed up by an answer that addressed it (or at least I didn't see it - read about 100 posts past it's mention in i9100 forum here...)

Could this be caused by an app misbehaving because it's not built properly for ICS? I'm running the same apps I had on GB ROMs, and not noticing any "out of the ordinary" data use by installed apps.

In addition to this (perhaps related) is an apparent discrepancy between reported awake time in BBS and the details of the android OS item in the built in battery monitoring in ICS. Android OS seems to report significantly higher awake time than BBS. Right now, Android OS reports 3h18m of awake time, while BBS reports 2h37m.

Hope someone can point me in a direction to try to troubleshoot this...

Thanks,

dd

I'll need some time to do research, but based on the name, multipdp may be the new name for svnet-dormancy.

As to Android OS reporting high awake time - that's because Android 4.0.3 and below have a display bug that adds screen-on time to that statistic instead of subtracting it.

dandrumheller · Mar 30, 2012

Entropy512 said:
I'll need some time to do research, but based on the name, multipdp may be the new name for svnet-dormancy.

As to Android OS reporting high awake time - that's because Android 4.0.3 and below have a display bug that adds screen-on time to that statistic instead of subtracting it.

Thank you for getting back to me on this Entropy. Sorry I missed the simple math on the conflicting awake times. Since that's the case, then AOS should be reporting (BBS awake + (2*screen on))? Or (1*screen on) if BBS is also reporting screen on as awake?

Further observation of the multipdp wakelock indicates that it is only present when on mobile data - does not appear to increase significantly when on WiFi. If I understand correctly what svnet-dormancy is, that would be consistent with multipdp being a renamed svnet-dormancy, would it not?

Entropy512 · Mar 30, 2012

dandrumheller said:
Thank you for getting back to me on this Entropy. Sorry I missed the simple math on the conflicting awake times. Since that's the case, then AOS should be reporting (BBS awake + (2*screen on))? Or (1*screen on) if BBS is also reporting screen on as awake?

Further observation of the multipdp wakelock indicates that it is only present when on mobile data - does not appear to increase significantly when on WiFi. If I understand correctly what svnet-dormancy is, that would be consistent with multipdp being a renamed svnet-dormancy, would it not?

Yes, that is correct.

One of the things new in Android 4 is lumping all wakelocks not accounted for by apps under Android OS. (This should include kernel wakelocks.) Screen time was supposed to be removed from this number, but a math error caused it to be added instead.

I think the math might still have some issues that could lead to some things being double-counted but I'm not sure.

dandrumheller · Mar 30, 2012

Entropy512 said:
Yes, that is correct.

One of the things new in Android 4 is lumping all wakelocks not accounted for by apps under Android OS. (This should include kernel wakelocks.) Screen time was supposed to be removed from this number, but a math error caused it to be added instead.

I think the math might still have some issues that could lead to some things being double-counted but I'm not sure.

Understood. I think you are right about other wakelocks getting double counted - when I have been paying closer attention to wakelock times, quick adding up of wakelocks do seem to indicate a higher total than the total awake time being reported by BBS - but I haven't been on ICS ROMs long enough to have spent the time to do careful math on them. I'll keep an eye on this. Are there any particular areas for me to look at where info would be useful to you?

test9876543 · Apr 1, 2012

Entropy512 said:
I'll need some time to do research, but based on the name, multipdp may be the new name for svnet-dormancy.

As to Android OS reporting high awake time - that's because Android 4.0.3 and below have a display bug that adds screen-on time to that statistic instead of subtracting it.

I've got wakelock issue with multipdp as well and with secril_fd-interface.

I've been on CM9 nightlies for a few months and am currently on the CODEWORKX 2012-03-31 release with his 3.0.26 Kernel, but the same kernel wakelocks have been there on the Samsung kernels before as well. I've got two google accounts setup and one twitter account. I tried googling for any details but only found the sources for multipdp and don't understand enough to figure out what that driver actually does and why it would keep my phone awake all the time without doing anything.
The stats below are taken after charging all night in airplane mode, then unplugging and going back to regular mobile data. Have not received any mails/texts/tweets in the past 1.5 hours, still these two keep my phone awake for some reason. Any help would be much appreciated. If you need more details, let me know. Thanks.

Statistic Type: (0) Since Charged
Since 1 h 28 m 8 s

Awake (): 53 m 13 s (3193 s) Ratio: 60.4%
Screen On (): 11 m 53 s (713 s) Ratio: 13.5%

================
Kernel Wakelocks
================
"multipdp" (): 37 m 13 s (2233 s) Cnt

c/wc/ec)62/0/62 42.1%
"secril_fd-interface" (): 34 m 53 s (2093 s) Cnt

c/wc/ec)157/0/0 39.5%

mzaur · Apr 1, 2012

I'm running UnNamed Rom 2.2.1 which is based on UCKK6 and battery life is sucking. With 6 hours of uptime, I have 33minutes of kernel wakelocks from [P]_NetworkLocationLocator, 15min from svnet, 13min from wlan_rx_wake.

Weird thing is that I set wifi sleep policy to turn off when screen turns off, but the picture below shows that it seems to stay on.. and my phone keeps intermittently coming on for some reason.

Is the only fix to flash to an other ROM?

[REF] Known identified battery drainers

Senior Member

Senior Member

Senior Member

Attachments

Senior Member

Attachments

Senior Member

Senior Recognized Developer

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Senior Member

Recognized Contributor

Senior Member

Senior Recognized Developer

Senior Member

Senior Recognized Developer

Senior Member

Senior Member

Attachments

Senior Member

Attachments

Similar threads

Top Liked Posts