Reply
Thread Tools
Posts: 3 | Thanked: 0 times | Joined on Dec 2013
#1
Hello Maemo-Folks!

I hope you are able to help me with this one.
I bought me a used N9 / black / 16 GB.

Like always, when you get a device, I flushed the device.
Not only to swipe away the data of the previous owner, but also to get rid of all those chinese stuff (weibo, etc.).

Steps
flasher -f -F xxxmain.bin --erase-user-data=secure
flasher -f -F xxxmain.bin -F xxxemmc.bin --flash-only=mmc

Binaries used 2BCA0F8F_DFL61_HARMATTAN_40.2012.21-3.ALPS.14_EMMC_ALPS.bin

68BA2DA0_DFL61_HARMATTAN_40.2012.21-3.232.05.2_PR_LEGACY_232_ARM_RM-696_PRD_signed.bin

All worked fine, Harmattan is chill.

Okay, enough of the girlie-talk, let's jump to the problem.
After a couple of minutes, the device was restarting itself. Given not much thought into it, but it happend again and again.

So what seems to be the problem here. According to "top" the /usr/sbin/csd process starts with around 400 MB RAM and grows within time.

At the end, around 900 MB RAM the systems calls

[Dec 15 12:40:08 (2013) kernel: [ 3051.546936] Out of memory: kill process 2923 (sh) score 519 or a child
Dec 15 12:40:08 (2013) kernel: [ 3051.546936] Killed process 2923 (sh)
Dec 15 12:40:08 (2013) kernel: [ 3051.630554] Out of memory: kill process 2933 (sh) score 519 or a child
Dec 15 12:40:08 (2013) kernel: [ 3051.630615] Killed process 2933 (sh)
Dec 15 12:40:08 (2013) kernel: [ 3051.925079] Out of memory: kill process 2879 (sh) score 430 or a child
Dec 15 12:40:08 (2013) kernel: [ 3051.925079] Killed process 2942 (initctl)
Dec 15 12:40:09 (2013) kernel: [ 3052.120513] Out of memory: kill process 976 (immvibed) score 378 or a child
Dec 15 12:40:09 (2013) kernel: [ 3052.120544] Killed process 976 (immvibed)
Dec 15 12:40:09 (2013) kernel: [ 3052.130401] Out of memory: kill process 2883 (aegis-exec) score 377 or a child
Dec 15 12:40:09 (2013) kernel: [ 3052.130432] Killed process 2883 (aegis-exec)
Dec 15 12:40:10 (2013) kernel: [ 3053.174713] Out of memory: kill process 2902 (sensord) score 340 or a child
Dec 15 12:40:10 (2013) kernel: [ 3053.174713] Killed process 2902 (sensord)


as long as nothing is there to kill anymore and the restart happens.

Unfortunately I am not able - even as devel-su - to kill the pid.

So my question is, how do I stop this memory leak?

Thank you guys in advance
Cheers
-ZF
 
Community Council | Posts: 4,920 | Thanked: 12,867 times | Joined on May 2012 @ Southerrn Finland
#2
Are you able to use it at all before it restarts?

csd is the cellular functions daemon (at least I guess so, since it launches plugin stuff like call, csnet, gprs, sim, sms...)

It could be that you have some mismatch between your firmware and your user SW...?
Maybe if you flashed another variant's code to your device?
 
Posts: 3 | Thanked: 0 times | Joined on Dec 2013
#3
Originally Posted by juiceme View Post
Are you able to use it at all before it restarts?

csd is the cellular functions daemon (at least I guess so, since it launches plugin stuff like call, csnet, gprs, sim, sms...)

It could be that you have some mismatch between your firmware and your user SW...?
Maybe if you flashed another variant's code to your device?

Hello juiceme!

Thank you for your reply.
Yes indeed I am able to use the phone without any difficulties as long as enough memory is there.

With "free -m" it pretty easy shows me how my free ram decreases.

About the firmware: Well I guess it should be the latest HARMATTAN image, no?


-ZF
 
Community Council | Posts: 4,920 | Thanked: 12,867 times | Joined on May 2012 @ Southerrn Finland
#4
Originally Posted by zeigefinger View Post
Hello juiceme!

Thank you for your reply.
Yes indeed I am able to use the phone without any difficulties as long as enough memory is there.

With "free -m" it pretty easy shows me how my free ram decreases.

About the firmware: Well I guess it should be the latest HARMATTAN image, no?
No, what I meant is that you said you originally had chinese software on the device, and you flashed with some other variant, right?

Now I am not sure about this but I vaguely remember people are claiming that different variants (especially the chinese) have also slightly different firmware. AFAIK they should all be compatible but some people were claiming that chinese CMD would be better in some cases.

Now all this talk of one FW performing better than some other could of course be just placebo effect, but If you only flashed the device main SW without firmware, and if there are some differences, then it could cause something like this, no?
 
Posts: 3 | Thanked: 0 times | Joined on Dec 2013
#5
Originally Posted by juiceme View Post
No, what I meant is that you said you originally had chinese software on the device, and you flashed with some other variant, right?

Now I am not sure about this but I vaguely remember people are claiming that different variants (especially the chinese) have also slightly different firmware. AFAIK they should all be compatible but some people were claiming that chinese CMD would be better in some cases.

Now all this talk of one FW performing better than some other could of course be just placebo effect, but If you only flashed the device main SW without firmware, and if there are some differences, then it could cause something like this, no?


Hard to tell right now.
But I am going to flush it once more with different firmwares anyway.
Will take some time, but maybe there is one combination which has not mmory leak.

Will let you know afterwards.
Cheers,
-ZF
 
Reply


 
Forum Jump


All times are GMT. The time now is 04:38.