Log in or register to post comments

error occurs at about the same time each day

December 21, 2010 - 8:03am #1

The "CPU may be pegged" seems to only occur at about the same time each day. I've observed this for a few days now and it's a bit strange.

I work with the QCAR SDK every day from 9.30 till 5.30 and this error starts occurring about every second build &;;;;; run at about 3.30GMT each day.
Usually happens after cleaning from coding in JNI.

Just thought I'd bring it to attention in case it's not just me.

Also logcat if it's of interest:

12-21 15:53:44.339: INFO/QCAR(661): Creating OpenGL ES 2.0 context
12-21 15:53:44.349: DEBUG/QCAR(661): GLRenderer::onSurfaceCreated
12-21 15:53:44.349: DEBUG/QCAR(661): Java_com_qualcomm_QCARSamples_ImageTargets_ImageTargetsRenderer_initRendering
12-21 15:53:44.399: DEBUG/QCAR(661): GLRenderer::onSurfaceChanged
12-21 15:53:44.399: DEBUG/QCAR(661): Java_com_qualcomm_QCARSamples_ImageTargets_ImageTargetsRenderer_updateRendering
12-21 15:53:49.489: DEBUG/dalvikvm(409): GC_EXPLICIT freed 5616 objects / 341344 bytes in 66ms
12-21 15:53:54.580: ERROR/Adreno200-ES20(661): gsl_device_waittimestamp failed in rb_timestamp_wait_on_timestamp
12-21 15:53:55.613: WARN/SharedBufferStack(661): waitForCondition(LockCondition) timed out (identity=18, status=0). CPU may be pegged. trying again.
12-21 15:53:56.610: WARN/SharedBufferStack(661): waitForCondition(LockCondition) timed out (identity=18, status=0). CPU may be pegged. trying again.
12-21 15:53:57.610: WARN/SharedBufferStack(661): waitForCondition(LockCondition) timed out (identity=18, status=0). CPU may be pegged. trying again.
12-21 15:53:58.609: WARN/SharedBufferStack(661): waitForCondition(LockCondition) timed out (identity=18, status=0). CPU may be pegged. trying again.
12-21 15:53:59.609: WARN/SharedBufferStack(661): waitForCondition(LockCondition) timed out (identity=18, status=0). CPU may be pegged. trying again.

Rob.

robertross
Joined:
November 2, 2010
Posts: 200
Top

Re: error occurs at about the same time each day

December 22, 2010 - 1:59am #3

Sorry its 3.30pm Greenwich Mean Time (GMT).

Yeah that makes more sense. I didn't think it could been from a long session because I reboot the device so often but I do use "adb reboot". I'll trying just rebooting from the device instead from now on and see if that helps any.

- What device / Android OS are you using?
- Nexus One Android 2.2
- Have you upgraded to the latest QCAR SDK, v0.10.0?
- Yes.
- What are you rendering, about how many polygons?
- Just one plane.
- Are you using OpenGL ES 1.1 or 2.0?
- OpenGL ES 2.0

Like I say it only really shows the problem after working with JNI for many hours. It's not a very good test case.

Rob.

robertross
Joined:
November 2, 2010
Posts: 200
Top

Re: error occurs at about the same time each day

December 21, 2010 - 9:11am #2

Thanks for the report. Any clues to help reproduce this bug are welcome, it's something that we also see on occasion but without any repeatability.

Just to double check, when you say 3:30 GMT do you mean PM? A time-of-day causality doesn't make much sense, but we'll consider it. It's more likely that you're experiencing this crash after hours of device use, I tend to see it more often at the end of a long session. I find that popping the battery out and waiting several seconds fixes the problem better than rebooting via "adb reboot".

Also, a few other pieces of information will help us here:

- What device / Android OS are you using?
- Have you upgraded to the latest QCAR SDK, v0.10.0?
- What are you rendering, about how many polygons?
- Are you using OpenGL ES 1.1 or 2.0?

If your project really shows this problem repeatably, could you possibly zip it up and send it to us at

? A reproducible test case would go a long way towards solving this problem :)

Thanks!

- Kim

ksiva
Joined:
October 4, 2010
Posts: 2340
Top
Log in or register to post comments