• State
  • Locked Locked
  • Replies 1 reply
  • Subscribers 98 subscribers
  • Views 576 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DM8168 Memory corruption issue

Eugenia Guzman - RidgeRun
Intellectual 300 points
Other Parts Discussed in Thread: CCSTUDIO

Hi all,

We have a custom board based on DM8168 with a Micron MT41K128M16JT DDR3 running at 796MHz. We have a gstreamer application that makes dual capture (1080p@30) and dual display, also it is scaling and encoding each one. We are facing random failures, sometimes the board runs fine and sometimes we got a kernel crash but this kernel crash is not consistent, we don't get the same kernel error in each crash. We think that it is something related with memory corruption, similar to the problem found in the following post:

http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/716/p/252042/906685#pi317334=1&pi317334filter=all&pi317334scroll=false

We don't think it is a software related issue because we ran the same test in a DM8168 EVM board and it didn't crash, so we discarded a software issue. How can we determine if it is a memory corruption issue? Is there a way to do it using CCS?

We are using the EZSDK 5.05.02.

Thanks,

Eugenia

  • 0
    TI__Guru**** 170625 points

    Eugenia,

    Eugenia Guzman - RidgeRun said:
    We have a custom board based on DM8168 with a Micron MT41K128M16JT DDR3 running at 796MHz.

    Check if this 796MHz is still supported in the new DM816x documentation (TRM, datasheet and Silicon Errata) from March 2015. Besides DDR3 max allowed frequencies, check also HDVPSS, ARM A8, DSP, HDVICP2, etc.

    Eugenia Guzman - RidgeRun said:
    We don't think it is a software related issue because we ran the same test in a DM8168 EVM board and it didn't crash, so we discarded a software issue.

    I agree.

    Eugenia Guzman - RidgeRun said:
    We think that it is something related with memory corruption

    Eugenia Guzman - RidgeRun said:
    ow can we determine if it is a memory corruption issue? Is there a way to do it using CCS?

    You can use the u-boot mtest:

    You can use also the DDR3 HW diagnostic test, CCStudio project, available at the Spectrum Digital web site.

    See also the below wiki pages:

    You can check also the silicon revision and speed grade of the DM816x device used in TI EVM and your custom board. See datasheet and silicon errata for more info on differences.

    Regards,
    Pavel