maemo.org - Talk - Methods for determining battery drainage / cause

maemo.org - Talk (https://talk.maemo.org/index.php)

- Nokia N900 (https://talk.maemo.org/forumdisplay.php?f=44)

- - Methods for determining battery drainage / cause (https://talk.maemo.org/showthread.php?t=59611)

Re: Methods for determining battery drainage / cause

Is it working fine in Desktop Command Widgets? I had to abandon it, it kept going bonkers. What command?

Re: Methods for determining battery drainage / cause

Here is what I did when there was something causing unusually high battery usage.

1. Having had BatteryGraph installed, I was able to go back and compare the daily graphs to find the approximate beginning of the issue.
2. Then I went into FapMan, and had it list all installed/updated items by date.
3. Systematically removed two or three items every day, checking the results in between.
4. Repeated that until the battery drain was back to the previous averages.
5. Having it narrowed down to two items, I was able to find the culprit (in my case MaeNotify).

Re: Methods for determining battery drainage / cause

Quote:

Originally Posted by ndi (Post 775776)

Is it working fine in Desktop Command Widgets? I had to abandon it, it kept going bonkers. What command?

believe it is

Code:

sudo kernel-config limits 125 600

Just change the 125 and 600 to whatever values you want to use. Also make sure you disable the updating for everything except being clicked. On mine as soon as a saved the new command, it would go back to the desktop, so just click the wrench again to get into settings and make sure you take care of the updating options as mentioned.

I have 3 on mine. One for underclocking while I'm sleeping, one for normal use, and one for some more guts.

I don't like leaving it overclocked as it's just not necessary. To be honest I'm not even really seeing any real speed ups when overclocked, I just like this kernel because it allows me to underclock way lower than the stock one.

2d

Re: Methods for determining battery drainage / cause

Quote:

Originally Posted by tho (Post 775804)

Thanks man, good advice and nice method. I'll use that myself from now on.

2d

Re: Methods for determining battery drainage / cause

AlMehdi

I managed to reproduce my power drain problem with the device stuck in C0. I noticed this only happens after booting from a shutdown and will stay until I do another (or a number of) shutdowns. Interestingly the nokia boot up animation is cut short when this happens.

Attached is the syslog and also my powertop reading (monitored for 10min)

I went through the log and can't find anything useful so I'm wondering if you can take a look?

I shutdown the n900 and powered back on. syslogd starts logging on line 678 - 20:17:37 Nokia-N900 syslogd 1.5.0#5maemo7+0m5: restart.
After the system booted I then went into xterm and ran powertop for 10min. After that I left the device idle till line 1689: Aug 5 20:57:37 Nokia-N900 -- MARK --

I started leafpad to have a look at my powertop output and then started wifi to copy the output of syslog and powertop onto my pc.

The only thing I noticed is that I have some bad eraseblock errors on line 841-853, which is worrying. What can i do about that???

thanks!!

EDIT:

I did a line by line comparison of a 'good' boot vs a 'bad' boot syslog. What i've found is that the dsp crashed on boot and had to recover. line 1227:

Code:

Nokia-N900 dsp-manager: DSP crash detected: 0

Aug  5 20:17:54 Nokia-N900 kernel: [   26.195526] ***** DSPMMU FAULT ***** IRQStatus 0x2

Aug  5 20:17:54 Nokia-N900 kernel: [   26.195556] ***** DSPMMU FAULT ***** faultAddr 0x1ecc6228

Aug  5 20:17:54 Nokia-N900 kernel: [   26.195587] WMD_DEH_Notify: ********** DEVICE EXCEPTION **********

Aug  5 20:17:54 Nokia-N900 kernel: [   26.195587] WMD_DEH_Notify: DSP_MMUFAULT,errInfo = 0x0

Aug  5 20:17:54 Nokia-N900 kernel: [   26.195617] WMD_DEH_Notify: DSP_MMUFAULT, High Address = 0x1ecc

Aug  5 20:17:54 Nokia-N900 kernel: [   26.195617] WMD_DEH_Notify: DSP_MMUFAULT, Low Address = 0x6228

Aug  5 20:17:54 Nokia-N900 kernel: [   26.195648] WMD_DEH_Notify: DSP_MMUFAULT, fault address = 0x1ecc6228

Aug  5 20:17:54 Nokia-N900 hildon-welcome[1211]: omxh264dec0: unrecoverable error: The component is in invalid state (0x8000100a)

This would actually explain why the startup animation stops 1/2way during boot. Thats the only meanful thing I could find. I've also attached the boot sequence of a good boot.

Re: Methods for determining battery drainage / cause

Quote:

Originally Posted by 2disbetter (Post 775309)

Got it, just working on the command line for enabling the 125-250mhz setting on the kernel.

Edit: Got it setup in Alarmed and in the Desktop Command Widgets.

Pretty nice, lets see if it helps my battery. ;)

2d

Well I woke up this morning to my phone being rebooted. It was stuck asking if I wanted to exit offline mode. Still I only lost 5% battery that night (8 hour period). I have all the things my alarmed does set up in a desktop command widget, this way I don't have to have a time on it occuring, and can just trigger it before going to bed.

It seems my battery problems have dissippated though as I only have 3 hours left of work (been here since 0800), and I'm still at 68% and have been using the phone a lot.

I think the real issue was MaeNotify. I also think installing the power kernel helped me out a lot (as I under clock a lot although the verdict is still out on whether or not this helped. This is becuase a process might run for 10 seconds at 600mhz but take 60 seconds at 250mhz. So I'm still trying to see if proportionally I'm getting a gain out of underclocking).

2d

Re: Methods for determining battery drainage / cause

It depends on your usage profile. Me, e.g., I do little work time-wise, maybe an hour sum total, but I really lean on it. Upgrades, graphics apps, VNC, stuff like that. Overall, I get a "feel like" 30-40% improvement since OC to 900 MHz. I handled a stock-clocked today and it feels sluggish, I kept waiting on the device to finish. It might be that it spends more time waiting for me than the other way around.

Re: Methods for determining battery drainage / cause

Quote:

Originally Posted by rexii23 (Post 776478)

Hey, you run on very high frequency 1.15GHz ! And it could be an exact reason why DSP crashed - it may be just not able to run this.

Re: Methods for determining battery drainage / cause

OneNAND are specified for minimum amount of usable sectors. For example atleast 256 megs as in n900. There's almost always a list of bad blocks already when they leave the factory. It's nothing to worry about, the OS doesn't use the bad blocks.

Re: Methods for determining battery drainage / cause

Quote:

Originally Posted by egoshin (Post 776799)

Hey, you run on very high frequency 1.15GHz ! And it could be an exact reason why DSP crashed - it may be just not able to run this.

no its not running at 1.15ghz! its due to the titan's kernel. Powertop doesnt pick up the proper frequencies. I'm running at 250-750mhz right now. However loading the default frequencies and voltages could not fix the problem.

@shadowjk. thanks for that info. That puts my mind at ease!