[Q] Nexus 10 hangs every night - something in libEGL/OpenGL? - Nexus 10 Q&A, Help & Troubleshooting

Hi!
I have a Nexus 10 and it's been working flawlessly since I bought it, running CyanogenMod 10.1 most of the time. After I upgraded to 4.4 it started to hang, become unresponsive when being left on with screen turned off for a couple hours. Wouldn't last a night without hanging, regardless if it was being charged or not. Every time after that I had to hard reboot (press power button for longer than 8-10 seconds). This would happen on stock 4.4, stock 4.3, CM 10.2, stock kernel, francoKernel. Even on fresh install with no apps installed.
When I left it connected to USB with ADB logging, I could see that the last thing in the log before it hangs was usually something related to libEGL, GL, GL Operation, glDrawArrays.
For example:
Code:
11-23 13:08:34.250 W/System.err( 1012): java.lang.RuntimeException: GL Operation 'glDrawArrays' caused error 505!
(I found somewhere that this is GL Out of Memory error)
Code:
11-23 13:08:38.955 W/ ( 1012): GLES-MALI OOM error: execution failed (gles_fbp_read_pixels at vendor/arm/mali6xx/gles/src/fb/mali_gles_fb_read_pixels.c:693)
Code:
12-15 21:06:40.813 D/libEGL (16985): glGetIntegerv(GL_FRAMEBUFFER_BINDING_OES, (GLint *) 0x426e3be0);
12-15 21:06:40.813 D/libEGL (16985): glBindFramebuffer(GL_FRAMEBUFFER_OES, 2);
12-15 21:06:40.813 D/libEGL (16985): glGetError();
This is the latest log on CM 10.2 with OpenGL tracing in logcat (just the last part):
http://pastebin.com/6rpmfYca
The full log:
http://www61.zippyshare.com/v/5060489/file.html
This is some previous log:
http://pastebin.com/J9dNBBZS
What is causing this? What app/service is it dying in (there is MovieMaker in the log before the issue happens, but WTH?)? How can I debug this further?

_anarion_ said:
Hi!
I have a Nexus 10 and it's been working flawlessly since I bought it, running CyanogenMod 10.1 most of the time. After I upgraded to 4.4 it started to hang, become unresponsive when being left on with screen turned off for a couple hours. Wouldn't last a night without hanging, regardless if it was being charged or not. Every time after that I had to hard reboot (press power button for longer than 8-10 seconds). This would happen on stock 4.4, stock 4.3, CM 10.2, stock kernel, francoKernel. Even on fresh install with no apps installed.
When I left it connected to USB with ADB logging, I could see that the last thing in the log before it hangs was usually something related to libEGL, GL, GL Operation, glDrawArrays.
For example:
Code:
11-23 13:08:34.250 W/System.err( 1012): java.lang.RuntimeException: GL Operation 'glDrawArrays' caused error 505!
(I found somewhere that this is GL Out of Memory error)
Code:
11-23 13:08:38.955 W/ ( 1012): GLES-MALI OOM error: execution failed (gles_fbp_read_pixels at vendor/arm/mali6xx/gles/src/fb/mali_gles_fb_read_pixels.c:693)
Code:
12-15 21:06:40.813 D/libEGL (16985): glGetIntegerv(GL_FRAMEBUFFER_BINDING_OES, (GLint *) 0x426e3be0);
12-15 21:06:40.813 D/libEGL (16985): glBindFramebuffer(GL_FRAMEBUFFER_OES, 2);
12-15 21:06:40.813 D/libEGL (16985): glGetError();
This is the latest log on CM 10.2 with OpenGL tracing in logcat (just the last part):
http://pastebin.com/6rpmfYca
The full log:
http://www61.zippyshare.com/v/5060489/file.html
This is some previous log:
http://pastebin.com/J9dNBBZS
What is causing this? What app/service is it dying in (there is MovieMaker in the log before the issue happens, but WTH?)? How can I debug this further?
Click to expand...
Click to collapse
I can't offer a solution, but I have the same problem. Mine started when I moved to KitKat. I thought the problem was occurring when on a charger and the charge level reached FULL. But I also have the OLD MovieMaker app loaded. I am going to uninstall it and see if that has any effect.
Have you had any luck tracking down the problem?

rlrbjr said:
I can't offer a solution, but I have the same problem. Mine started when I moved to KitKat. I thought the problem was occurring when on a charger and the charge level reached FULL. But I also have the OLD MovieMaker app loaded. I am going to uninstall it and see if that has any effect.
Have you had any luck tracking down the problem?
Click to expand...
Click to collapse
Hi!
I was able to find some old threads about this aptly named Sleep of Death issue, here for example.
Unfortunately no conclusion, most people seem to return their devices :/
Now I think this is indeed some massive memory leak, becasue I was able to see ADB logs with all processes stopping, and the screen blinking with lock screen once when this happened.
As for MovieMaker, this might be misleading - I froze it in TitaniumBackup but the issue still exists.
My plan now is to write a script that would run
Code:
adb shell dumpsys meminfo
every couple seconds/minutes and plot the memory usage until the issue strikes.
I will report here if I find anything.
Best regards,
anarion

I uninstalled Movie Editor (rather than freezing it) and my system is much better (so far). I have not had a single lockup. It still too early to know if this is a god long term solution but I will keep you informed.
Sent from my Nexus 10 using Tapatalk 4

Related

Possible reason for HD2 CoreDroid HD GB 2.3.3 V2 slowness

Seems like having SD card inserted causes high IOWait.
User 1%, System 3%, IOW 95%, IRQ 0%
User 5 + Nice 0 + Sys 10 + Idle 0 + IOW 291 + IRQ 0 + SIRQ 0 = 306
Currently running smsBackup+ that restores sms messages from Gmail (1 message in 4 sec)
This also seems to cause reported installation issues:
(thread http://forum.xda-developers.com/showthread.php?t=926507)
Installed several times myself... waited up to 0.5 hour.. still splash screen.
Several processes (com.android.phone, etc) got "Force" or "Wait" messages.
Only after connecting ADB console was i able to see that android was still initializing.. waited for longer and in the end ~1 - 1.5 hours phone booted up.
Are you affected? .. Try removing SD card.. does HD2 boot up within 10 min.? .. then probably you have the same issue.
(not really usable with no SD card.. settings are lost on each boot)
This is discussed very often in the developing thread that you already mentioned: Dev Thread
The first time it needs a lot of time - in my case it was more then one hour before the splash screen was away - for the FCs just press the "Wait" button and wait - after the first time it will become less and less and the device will work quite fast (just the known bugs are still there).
Some people told that they had to wait for 2 or one even for 4 hours before the splash screen was away. I had very strong problems with installing and it did not disappear after 6 hours.
If you have that try the following:
1.) data wipe
2.) d-cache wipe
3.) starting in MAGLDR
4.) USB flasher / DAF.exe
5.) recovery menu / cwm partition sd card (1024/0M)
6.) mounting and transfering .zip
7.) install .zip
8.) reboot
hope i got you right

Unable to use Glowlight

Hi guys, I've got a little problem here. I played with my Nook Simple Touch Glowlight and even rooted it. But I couldn't install apps for some reason, so I decided to restore and root it again with other tool (Glownooter) so I could have Google's App Market. Here comes my first fault. I installed [NST]Touch-Formatter v2 .zip - I didn't notice it is only for the normal NSG - using CWM. Everything worked fine, I registered my Nook and then, using CWM, installed glownooter. But instead of showing me Android homescreen, it rebooted itself. And it got stuck booting (five dots loading) until the battery died. So I charged it and then I tried it again (NST Touch Formatter, Glownooter...). And "surprisingly", I ended up with the same result. I tried that about four or five times and then I gave up for two days. Today, I wanted to read something, so I used only the Touch Formatter, registered it, uploaded the book on it... And held the "n" button to turn the glowlight on. But nothing happened. So I went to settings and there wasn't anything about glowlight. Well then, I had a backup, so I restored from backup. When it booted, it kept me showing the screen "your system is damged" (or something like that) "you must perform a system restart". Tried that 2 times. Didn't help. I did factory reset (holding the two bottom buttons during boot up). After that, I had trouble with connecting to my wifi, so I skipped the registration proccess.
Now, I can use it, but without Glowlight and that was the main reason why I bought Nook and not Kindle. I guess, the Touch Formatter somehow installed the normal NSG software, so the device thinks, it's the old model. And that may also be the reason why Glownooter didn't work. Then again, my opinion has 90% to be wrong - given to what I did with my nook, and that this is my first Android device.
I did the backup with Windows, everything else using Terminal in OS X (if that info helps).
Thanks for any idea that comes up onto your mind.
P.S.: Sorry for mistakes in my text, I'm not a native speaker and I've been learning English for 3 years (plus it's 1 a.m. ).
,
The good thing about that screen that says "You must reboot your Nook" is that it means that your Nook is almost working.
At least in some cases it's caused by not being able to mount partitions.
Here's an example:
Code:
I//system/bin/fsck_msdos( 744): ** /dev/block//vold/179:6
I//system/bin/fsck_msdos( 744): Invalid cluster size: 0
I/logwrapper( 744): /system/bin/fsck_msdos terminated by exit(8)
D/vold ( 744): Filesystem check failed (unknown exit code 8)
E/vold ( 744): vfat filesystem check failed on 179:6 (I/O error)
D/SurfaceFlinger( 811): Frame buffer posted; elapsed time = 34 msecs
D/MountListener( 811): handleEvent volume_checking:/media
D/MountListener( 811): handleEvent volume_damaged:/media
I/ActivityManager( 811): Starting activity: Intent { flg=0x10000000 cmp=android/com.android.interna
l.app.RebootRequiredActivity (has extras) }
I/RebootRequiredActivity( 811): getReason(): received reason [REASON_MEDIA_UNMOUNTABLE [/media]]
The easy solution is to check whether all the partitions mount.
You can do that with noogie and a Linux system or Clockwork Recovery.
If you still can't figure it out, get ADB over USB working on your Nook and get a logcat.

[Q] Encrypt phone feature bugged on Defy starting from 4.0

Hi Guys,
Has anyone tried the encryption feature found in Security settings? I'm using a recent nightly CM11 from Quarx/Blechd0se (Thanks, brilliant work!) quite stably for days now, but fail to make the encrypt phone function work...
When fully charged and USB powered, which are requirements, the successive encrypt phone buttons end up on a green android picture. This screen misses text and a progress bar which should normally appear on top of the screen, as seen on some screenshots. Even left for several hours, nothing progresses and a simple press on back button cancels action immediately which proves nothing has been encrypted in the meanwhile.
Took a Catlog and the only interesting message is the following:
Code:
01-02 01:34:23.732 E/Cryptfs ( 1197): Cannot get size of block device
In case it helps, find attached the output of mount command : View attachment mnt.txt
Could you have a look / help me out? (add to bug list if it applies)
In fact, for me, this function needs to be activated for connecting to my professional mails ...
Thanks in advance! :good:
Cheers
Further tested...
Hello all,
Just some more info... I flashed back to Quarx' CM9 and quickly tested once more the encryption and face the same problem but I didn't catlog, neither checked the mounts... Then I flashed Epsylon3's CM9 and quickly checked with the same result too...
But then I checked at catlog and now I get another type of error... Which is interesting and maybe promising as I already read some posts about this message...
Code:
04-14 21:11:13.101 E/Cryptfs ( 2219): Orig filesystem overlaps crypto footer region. Cannot encrypt in place.
Problem seems to be more general than only Quarx/Blechd0se's CM11 but a friend uses a CM9 flavor on a Samsung Galaxy S and has encrypted the phone successfully...
All in all it's linked with the filesystem (block device or not), how it is formated (yaffs/ext3/ext4) and mounted (mount/mount_all)...
Some interesting reading on jira.cyanogenmod.org/browse/CYAN-87 (not linked, I don't have 10 posts yet)...
Hope to find, test and come back with some more info soon...:fingers-crossed:
Cheers!

Fix for constant rebooting after upgrade to 2.1.0

**TLDR** - If your shield keeps rebooting after the 2.1.0 update, unplug your ethernet cable and/or disable your wifi as a temporary fix.
I recently got a 16GB model. After the latest firmware update (to 2.1.0) I started to get constant rebooting. My first thought was to use fastboot to get back to 2.0.0 - I believe you can do this without wiping your user data, as the official images are signed with the nvidia key and I believe their bootloader would allow you to downgrade - can anyone confirm this? See https: // developer .nvidia. com/ gameworksdownload (sorry, can't link with <10 posts...) for nvidia 2.0.0 image.
However I wanted to see what was happening first. Running
Code:
$ adb logcat
...lots of output
...the system just rebooted, so let's watch the logs!
D/AndroidRuntime( 4318): Shutting down VM
E/AndroidRuntime( 4318): *** FATAL EXCEPTION IN SYSTEM PROCESS: main
E/AndroidRuntime( 4318): java.lang.IndexOutOfBoundsException: Invalid index 0, size is 0
E/AndroidRuntime( 4318): at java.util.ArrayList.throwIndexOutOfBoundsException(ArrayList.java:255)
E/AndroidRuntime( 4318): at java.util.ArrayList.get(ArrayList.java:308)
E/AndroidRuntime( 4318): at com.android.server.pm.PackageManagerService$PMLocationListener.getCountryFromLocation(PackageManagerService.java:14375)
E/AndroidRuntime( 4318): at com.android.server.pm.PackageManagerService$PMLocationListener.onLocationChanged(PackageManagerService.java:14282)
E/AndroidRuntime( 4318): at android.location.LocationManager$ListenerTransport._handleMessage(LocationManager.java:281)
E/AndroidRuntime( 4318): at android.location.LocationManager$ListenerTransport.access$000(LocationManager.java:210)
E/AndroidRuntime( 4318): at android.location.LocationManager$ListenerTransport$1.handleMessage(LocationManager.java:226)
E/AndroidRuntime( 4318): at android.os.Handler.dispatchMessage(Handler.java:104)
E/AndroidRuntime( 4318): at android.os.Looper.loop(Looper.java:135)
E/AndroidRuntime( 4318): at com.android.server.SystemServer.run(SystemServer.java:276)
E/AndroidRuntime( 4318): at com.android.server.SystemServer.main(SystemServer.java:177)
E/AndroidRuntime( 4318): at java.lang.reflect.Method.invoke(Native Method)
E/AndroidRuntime( 4318): at java.lang.reflect.Method.invoke(Method.java:372)
E/AndroidRuntime( 4318): at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:903)
E/AndroidRuntime( 4318): at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:698)
I/Process ( 4318): Sending signal. PID: 4318 SIG: 9
E/installd( 226): eof
E/installd( 226): failed to read size
I/installd( 226): closing connection
For those of you that don't speak stacktrace or android internals, what happened here is that something (presumably, my network state tracker noticed either an IP change or something else from my ethernet) sent a message to the internal LocationManager saying "something happened, update your location estimate". In the process of doing this LocationManager crashes, and then the entire android framework crashes in response to this critical component (system_server) crash. Then the init binary helpfully restarts everything, and in a few minutes it crashes again.
Unplugging the ethernet cable has resulted in a stable system. I haven't experienced the crash while on WiFi, but I wouldn't rule it out yet.
While I have not looked in extreme detail, my suspicion is that something about the firmware update broke google mobile services - I see an error trying to mount the data for GMS in the logs, and noticed the stacktrace above seems to be failing when the LocationManager uses the PackageManager which uses installd which then fails with some kind of file read.
Code:
E/Vold ( 206): Failed to find mounted volume for /storage/sdcard1/Android/data/com.google.android.gms/files/
W/Vold ( 206): Returning OperationFailed - no handler for errno 0
W/ContextImpl( 6163): Failed to ensure directory: /storage/sdcard1/Android/data/com.google.android.gms/files
Turning off location services seems to stop this too
I'm 100% that no ethernet stops the rebooting (got a few hours of testing and no crashes). Tried turning off location services after I saw the logs of the error, and I've had about 30 min with no crashes, so that's obviously a much more preferred way to address this issue until nvidia fixes the bug
well interesting to know about this issue.
slammertime said:
I'm 100% that no ethernet stops the rebooting (got a few hours of testing and no crashes). Tried turning off location services after I saw the logs of the error, and I've had about 30 min with no crashes, so that's obviously a much more preferred way to address this issue until nvidia fixes the bug
Click to expand...
Click to collapse
Has having locations services turned off solved the random restarts for you inevitably?
Yes, it seems so
Sorry for resurrecting an old thread, but I'm still having this issue.
Tried disabling location, changing languages, every other suggestion here and elsewhere online.. still getting reboots ~3-5 mins after the system starts, and random lockups. Only started happening post 2.1. The device is basically unusable at this point.
I'm just about to factory wipe/reinstall. Is there any other solution?
Same
grivad said:
Sorry for resurrecting an old thread, but I'm still having this issue.
Tried disabling location, changing languages, every other suggestion here and elsewhere online.. still getting reboots ~3-5 mins after the system starts, and random lockups. Only started happening post 2.1. The device is basically unusable at this point.
I'm just about to factory wipe/reinstall. Is there any other solution?
Click to expand...
Click to collapse
I am having the same issue. Talked to support, and they told me to unpair, and reset my devices that were attached to it, and unplug the hdmi and power, and still no dice.
peppy6582 said:
I am having the same issue. Talked to support, and they told me to unpair, and reset my devices that were attached to it, and unplug the hdmi and power, and still no dice.
Click to expand...
Click to collapse
I got sick of it, wiped the device, and did a fresh install of 2.1 (IIRC, whatever factory image was available a few weeks back) and applied the OTA once prompted. Zero issues since. It took some time installing and configuring everything again, but having a completely stable device made it worth it. It was unusable before, as it'd reboot every few minutes no matter what it was doing.
It looks like they have the 3.0 image up now, if you're thinking of reinstalling as well.
https://developer.nvidia.com/shield-open-source
grivad said:
I got sick of it, wiped the device, and did a fresh install of 2.1 (IIRC, whatever factory image was available a few weeks back) and applied the OTA once prompted. Zero issues since. It took some time installing and configuring everything again, but having a completely stable device made it worth it. It was unusable before, as it'd reboot every few minutes no matter what it was doing.
It looks like they have the 3.0 image up now, if you're thinking of reinstalling as well.
https://developer.nvidia.com/shield-open-source
Click to expand...
Click to collapse
Yeah, I actually did that, with a fastboot -w after the system.img part, just to make sure.
hey guys, do you still have the problem with the random reboot?
I have the same problem with the 500gb version but I didn't root this device or whatever, it isn't even my intention to root it.
Any idea what it is?? Nvidia want to send me a new one (or refurbished item) but I have like the idea that it will not be solved if I get a new one.
I read to much people with the same problems, even on 2 links I read from someone who got a new one and the problem was still there :s
I'm a little scared to send it back , because I fear it will not be solved.
Any idea guys???

Changing dex2oat watchdog timeout

Hi,
Using Redmi Note 3 MTK on Epic Rom, Android 5.0.1.
Lately when updating Chrome and Android System Webview (base APK of 50+MB), the update always fail. After some research, it was actually the dex2oat getting timed out after 6 minutes.
1. I used these codes in build.prop
dalvik.vm.image-dex2oat-filter=speed
dalvik.vm.dex2oat-filter=interpret-only
It worked, but I lost ART on the newly updated apps. The smaller size .dex files in the dalvik cache confirms this.
2. I then reverted to
dalvik.vm.image-dex2oat-filter=speed
dalvik.vm.dex2oat-filter=space
Uninstalled and reupdated ASW, which failed, still exceeds the 6 min.
3. I then reasoned that turning off the wacthdog would work
dalvik.vm.image-dex2oat-filter=speed
dalvik.vm.dex2oat-filter=speed
dalvik.vm.dex2oat-flags=--no-watch-dog
Uninstalled and reupdated ASW, went on for a good 10 min+ (I didn't time it), and the phone hangs.
Anyone knows how to change the dex2oat timeout, then I can increase it progressively from 6 to whatever value that works.
Thank you.

Categories

Resources