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.

Compiler/AM5726: link error due to optimization level option

Part Number: AM5726

Tool/software: TI C/C++ Compiler

Hi,

My customers are developing software for C66x DSP in the AM572x, but a link error that seems to be a compiler problem has occurred.
The link error is that some symbol information can not be referenced.

error #10234-D: unresolved symbols remain

The version of the compiler they are using is 8.2.2.
This problem depends on the optimization level of the compiler and this error occurs when -o3 is used.
However, this problem did not occur with -o3 when compiler version 8.1.6 was used.

Is this a compiler bug? or can we avoid this problem with compiler options?

Best reagrds,
H.U

  • This is very odd.  I have never seen a problem with unresolved symbols be caused by a change in optimization level.  This is likely to take a few posts back and forth before we narrow this down.  

    Please show exactly how the linker is invoked, and all of the resulting diagnostics.  I don't think this will allow me to solve the problem.  But I should be able to ask better questions.

    Thanks and regards,

    -George

  • Hi George,

    Thank you for your quick reply.
    It is difficult for me to post all of the error messages here because it contains confidential customer information.
    Is it possible to contact you via TI local FAE? I can provide all error messages and their customer project.

    Best regards,
    H.U

  • H.U said:
    It is difficult for me to post all of the error messages here because it contains confidential customer information.

    I understand.  You can send me private messages through the forum.  Hover your mouse over my screen name or avatar.  A dialog box pops up.  Click on Send a private message.  

    H.U said:
    Is it possible to contact you via TI local FAE?

    I prefer forum messaging.  But, if you prefer to work this way, that's fine too.  Ask the FAE to send an e-mail just to me, copying no one else.

    Thanks and regards,

    -George

  • Thank you for sending in a test case through other channels.  I have reproduced the same diagnostics.  I strongly suspect a bug in the compiler.  I am in the middle of determining what happened well enough that I can turn in a bug report.

    Thanks and regards,

    -George

  • Hi George,

    Thank you for your reply.

    Results of verification by customer, It is known that v8.2.x produces code with much better performance than v8.1.x with the -o2 option.
    So, We hope this problem is solved soon and can use the -o3 option.

    Best regards,
    H.U

  • Thank you again for sending in the test case.  I filed CODEGEN-4623 in the SDOWP system to have this investigated.  You are welcome to follow it with the SDOWP link below in my signature.

    Thanks and regards,

    -George