• State TI Thinks Resolved
  • Locked Locked
  • Replies 10 replies
  • Answers 1 answer
  • Subscribers 102 subscribers
  • Views 1081 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.

AM3517 Boot Fails, ECC failure

Steven Kane
Prodigy 90 points
Other Parts Discussed in Thread: AM3517

Hello,

    While trying to boot my SOM which is on the Logic PD Zoom AM3517 Experimenter kit: http://logicpd.com/products/development-kits/zoom-am3517-experimenter-kit  that has the AM3517 microprocessor, I get the following error, or one similar (this is the entire boot log):

 

 

 

Texas Instruments X-Loader 1.44 (Dec  8 2009 - 22:58:46)

ECC Failed, page 0x00080300

48 0 8d e2 34 50 8d e2 e 10 a0 e1 f 0 85 e8 d 0 a0 e1 4a 8 0 eb 0 0 a0 e1 0 0 a0 e1 e4 d2 1f a5 1 d7 4d e2 88 d0 4d e2 0 e0 8d e5 0 e0 4f e1 4 e0 8d e5 13 d0 a0 e3 d f0 69 e1 f e0 a0 e1 e f0 b0 e1 48 d0 4d e2 ff 1f 8d e8 14 23 1f a5 1 27 42 e2 88 20 42 e2 c 0 92 e8 48 0 8d e2 34 50 8d e2 e 10 a0 e1 f 0 85 e8 d 0 a0 e1 20 8 0 eb 0 0 a0 e1 0 0 a0 e1 44 d3 1f e5 1 d7 4d e2 88 d0 4d e2 0 e0 8d e5 0 e0 4f e1 4 e0 8d e5 13 d0 a0 e3 d f0 69 e1 f e0 a0 e1 e f0 b0 e1 48 d0 4d a2 ff 1f 8d e8 74 23 1f e5 1 27 42 e2 88 20 42 e2 c0 92 e8 48 0 8d e2 34 50 8d e2 e 10 a0 e1 f 0 85 e8 d 0 a0 e1 11 80 eb 0 0 a0 e1 0 0 a0 e1 bf 5e 2d e9 0 70 a0 e1 3 0 50 e3 1 c0 a0 3 70 0 60 1 3 0 57 a3 1f 0 0 a 30 f 30 ee

Is this memory error fixable or should the entire board be replaced?

 

Thanks,

 

   Steven

  • 0
    TI__Mastermind 44306 points

    Steven,

    this indicates there is an issue loading from NAND. As the boot loader does give some output I don't think the board is broken.

    Try to reflash the board. There should be enough info on the wiki (and the tools) how to do that.

    Regards.

  • 0
    TI__Intellectual 1910 points

    Steven,

       You might also want to try running the DDR memory test that is part of LogicPD's Board Support Library, to confirm that your EVM board is OK.

    http://ap-fpdsp-swapps.dal.design.ti.com/index.php/Shiva_EVM#Board_Support_Library_.28BSL.29

    Regards, Clay

     

  • 0 in reply to Clay Montgomery
    Prodigy 90 points

    Clay,

       That link appears to be broken. Also, I did not see any DDR memory test on logic PDs support pages: http://support.logicpd.com or in the BSL that I downloaded from there (1016315A_AM35xx_BSL.zip). Do you have another link I can try?

     

    Thanks,

     

        Steven

  • 0 in reply to Steven Kane
    TI__Intellectual 1910 points

    Steven,

        That was a page internal to TI.  Sorry.  Use the download page at LogicPD intsead which is public but requires that you have created an account on their site.

    http://support.logicpd.com/auth/downloads/Zoom%20AM3517%20EVM%20Development%20Kit/

    It looks like you have already downloaded the BSL.  The memory test is at:

    \tests\experimenter\ram\src\test_ram.c

    And be sure to read the documentation at:

    \Doxygen\html\index.html

    It appears that you must build with Code Composer Studio in order to use the BSL memory tests on DDR, because CodeSourcery will only allow the test to execute out of DDR.

     

    Regards, Clay

  • 0 in reply to Clay Montgomery
    Expert 1390 points
    Hi Everyone !!!!!! Sorry to quote on your query!! can anyone just share the bsl for am3517 EVM board for interrupt configuration[Ethernet,CAN,UART,GPtimers,WDT]. I have bundle of BSL but I m missig the IRQ files also the startup and handler files. Would request if any one among you have got these file please share them.
  • 0
    Expert 1125 points

    Hi,

    Having similar problem with AM3517EVM x-loader image I am building myself. Using the signed one: x-loader.bin.ift

    Downloaded the AM35x-OMAP35x-PSP-SDK-03.00.01.06.tgz and built the image according to instructions.

    BTW, the EVM, works perfectly with TI's prebuilt images.

    No problems with memory, nor NAND nor SDRAM seen before.

    Where is the inconsistency and what's wrong with ECC read ?

    Why does it fail on 0x80300, shouldn't it be constrained to page size which is 0x800?

    Thanks ahead

    Yakir

  • 0 in reply to Gopher79
    Guru 31670 points

    Yakir,

    Could you check the hardware interface in your custom board? You can try the analysis we did in the post given below.

    http://e2e.ti.com/support/dsp/davinci_digital_media_processors/f/716/p/235461/827420.aspx#827420

  • 0 in reply to Renjith Thomas
    Expert 1125 points

    Regarding the ECC failure, there were two issues for me, all solved.

    First, it was misunderstanding of the uboot code I am using, which occurs also to be the source code for the x-loader. So, my problem was that I had used external MLO image from other x-loader project, SDK xxx.01.06. I suppose it didn't match it terms of ECC. Anyway when burned the MLO and u-boot.img from same build, the problem was gone.

    Second, using MLO and u-boot images programmed by flash_tool v.1.6 resulted in issue of x-loader failing to read the u-boot memory from nand, failing on ECC check. Found similar issue at: http://e2e.ti.com/support/dsp/sitara_arm174_microprocessors/f/416/t/167726.aspx

    Seems due to a bug is eccoffset of the flash tool (trying 2 while actually writing 1), the x-loader missed the preprogrammed ecc bytes location. So, I have created a workaround by moving eccpos in the u-boot to 1 and using eccoffset 1. Worksforme!

  • 0 in reply to Gopher79
    Expert 1125 points

    Interesting that after my workaround I am programming MLO and u-boot and they boot fine but ....

    If I apply "nand bad" I get bad blocks in the location of the programmed sectors (0x0, 0x80000, 0xa0000 and 0xb0000).

    These are surely ghosts since it all boots fine. Also scrubbed the nand and these blocks were gone. Reprogramming is a success.

    I would be keen if the flash_tool bug is resolved to go back to correct eccoffset 2 which I suspect is causing this confusion during nand program...

  • 0 in reply to Gopher79
    Guru 31670 points

    Yakir,

    AFAIK eccoffset got changed many times before in different revisions of TI NAND driver. I couldnt understand why these many times changes are done, but I believe in most of the cases if you take the latest source packages, this kind of issues may not happen.