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.

CCS/TMS320C6678: Crash when trying to watch variable

Part Number: TMS320C6678

Tool/software: Code Composer Studio

Hi,

I recently upgraded to CCS9.2.0.00013, installed on windows 7 x64. I can’t use the debugger, since the application always closes unexpectedly when I try to watch variables in the “Expressions” window, or when I switch to “Variables” window.

I’m using XDS200 connected to a TMS320C6678 and I would like to debug my code ported to CCS9, but it is actually impossible with this issue.

I already reported the error through the automatic procedure which activates when restarting CCS after unexpected crashes. I would like to know if there might be any parameter setting that I could try to mitigate the problem, while waiting for a new CCS update that would solve the problem.

 

Also, it looks like incremental building is not preserved through CCS sessions. My project is always recompiled completely whenever CCS is restarted, even if no file was changed. After first compilation, within a CCS session, incremental building works fine instead. I haven’t experimented it with previous CCS versions. Should it be the correct behavior?

 

Kind Regards,

Alessandro

  • Alessandro Dallai said:
    Also, it looks like incremental building is not preserved through CCS sessions. My project is always recompiled completely whenever CCS is restarted, even if no file was changed. After first compilation, within a CCS session, incremental building works fine instead.

    Perhaps a good thing it does as CCS9.1 tends to drop editor code lines from CTRL+Z back that have changed after last compile. Please tell me how or where to turn on the No incremental build switch. Seem to recall that build option might be a project dependent setting?

    Anyway perhaps Also install CCS9.1 and see if the Monitor debug issue occurs there too. I have 3 versions of CCS installed just to confirm odd issues as you bring up in this post.

  • Alessandro Dallai said:
    I already reported the error through the automatic procedure which activates when restarting CCS after unexpected crashes.

    Could you also attach the debug server dump file to this thread?

    https://software-dl.ti.com/ccs/esd/documents/ccs_diagnostic-logs.html

    Alessandro Dallai said:
    Also, it looks like incremental building is not preserved through CCS sessions. My project is always recompiled completely whenever CCS is restarted, even if no file was changed. After first compilation, within a CCS session, incremental building works fine instead. I haven’t experimented it with previous CCS versions. Should it be the correct behavior?

    Assuming the project is not cleaned in any way, it should always default to an incremental build - regardless of if CCS was restarted or not.

    Can you try using a clean new workspace folder and import your project and see if the issue persists there?

    Thanks

    ki

  • Hi BP,

    BP101 said:

    Perhaps a good thing it does as CCS9.1 tends to drop editor code lines from CTRL+Z back that have changed after last compile. Please tell me how or where to turn on the No incremental build switch. Seem to recall that build option might be a project dependent setting?

    If you want to turn off incremental build, and therefore rebuild the whole project every time, please try this setting:

    Go to project settings, and in the “Build” page click the “Builder” tab. On the bottom of the tab, edit the “Build (incremental build)” box and replace “all” with “clean all”. I tried and it worked for me.

    Still I prefer incremental build :)

     

    Kind Regards,

    Alessandro

  • Hi Ki,

    I imported my project in a clean new workspace, as you suggested, and this solved the issue with incremental building. Thanks.

     

    The problem with the debugger is still there though.

    I tried to attach the dump file of the debug server to my replay on this forum, but I was not able to do it as an error is issued every time a click on replay. Please find the file attached through WeTransfer: https://we.tl/t-Ktwgcu2vzk

     

    Thank you for your much appreciated help.

    Kind Regards,

    Alessandro

  • Alessandro Dallai said:
    Please find the file attached through WeTransfer: https://we.tl/t-Ktwgcu2vzk

    Unfortunately that link is block by IT. I cannot access it. Can you try zipping up the dmp file and attaching the zip?

  • Ki said:

    Unfortunately that link is block by IT. I cannot access it. Can you try zipping up the dmp file and attaching the zip?

    Ok, please find attached the dump file zipped. Hope this time it works!

    Regards,

    Alessandro

    8540.dump.zip

  • Got it, thanks! I will send it to the engineering team for analysis and keep you updated.

    ki

  • Alessandro,

    Any chance you could provide the executable (or some sort of reproducible test case)? 

    If not, can you reproduce the crash with debug server logging enabled? Then zip up the log and attach it to this thread:

    https://software-dl.ti.com/ccs/esd/documents/ccs_diagnostic-logs.html

    Thanks

    ki

  • Please refer to this link for the debug server logging:

    https://software-dl.ti.com/ccs/esd/documents/ccs_diagnostic-logs.html

    Thanks

    ki

  • Hi Alessandro,

    Yes I noticed that "all" is default properties settings for incremental build. Also noticed CCS 9.1 one or few changes changes are not always being detected since the checksum in LM Flash is not changing. Wonder if by switching the project to "clean all" then back to "all" it corrects your incremental build issue? 

  • Hi Ki,

    Let’s start with the debug server log. Please find it attached.

    It is a very short log anyway, with no significant information in my opinion.

     

    I may provide you with the executable (the .out file), including skeleton debug information, but I don’t want to public it on the forum. Is there a way to directly send you the file?

     

    Moreover, I found that CCS crashes when I try to watch particular variables in the “Expressions” window, while I experiment no problem with others.

    I will investigate this further as soon as I can spend some more time on this task.

     

    Kind Regards,

    Alessandro

    8863.log.zip

     

  • Hi Ki,

    Let’s start with the debug server log. Please find it attached.

    It is a very short log anyway, with no significant information in my opinion.

     

    I may provide you with the executable (the .out file), including skeleton debug information, but I don’t want to public it on the forum. Is there a way to directly send you the file?

     

    Moreover, I found that CCS crashes when I try to watch particular variables in the “Expressions” window, while I experiment no problem with others.

    I will investigate this further as soon as I can spend some more time on this task.

     

    Kind Regards,

    Alessandro

    17504.log.zip

     

  • Alessandro Dallai said:
    I may provide you with the executable (the .out file), including skeleton debug information, but I don’t want to public it on the forum. Is there a way to directly send you the file?

    Please start a private conversation with me. If you are not sure how to start one, please see the last FAQ at the bottom of: https://e2e.ti.com/support/tools/ccs/f/81/t/3131

  • Hello Alessandro,

    If you are still experiencing this issue, please start a private conversation with me.

    Thanks

    ki

  • Hi Ki,

    I started a private conversation as you suggested.

    Thanks for your support.

    Alessandro

  • Thank you for providing a reproducible test case. I have escalated the issue to engineering and filed a bug for the issue. Tracking ID: CCBT-2559

    Thanks

    ki

  • Hi Ki,

    I’m curious to track the bug while waiting for a future release, but unfortunately I can’t find the ID you provided on the bug tracking portal. Did the ID changed in the meanwhile?

     

    Regards,

    Alessandro

  • Yes the ID did change to CODEGEN-6914 when it was identified that the root cause was not with CCS but the code gen tools.

    However, that ID will also not appear in the portal. We are currently migrating our portal to a new system and working through some issues with it. 

    You can track the bug with this link:

    https://sir.ext.ti.com/jira/browse/EXT_CCBT-1

    Note that it still references the old ID (one of the issues with the new system). 

  • Alessandro,

    We would like to see how ElabType and SElabModuleParams are defined. Would it be possible to provide the related source/header files that contain this information (elab.h, etc)?

    Thanks

    ki

  • Hi Ki,

    I’m very much impressed of the care you and TI are putting on this issue. Good job!

    I sent you a private message with a package containing all the header files from my code. I’m going to show here just the definition of ‘ElabType’ and ‘SElabModuleParams’:

     

    enum ElabType {};

     

    struct SElabModuleParams

    {

    ElabType m_eElabType;

    };

     

    Maybe the problem is connected with the empty enum definition? I will undertake some tryouts tomorrow, and will tell you the results.

     

    Regards,

    Alessandro

  • The test case has an enumeration type with no enumerators. This is not legal in C, but in C++, it is legal to have no enumerators, in which case [C++98 7.2 "Enumeration declarations" p5] "If the enumerator list is empty, the underlying type is as if the enumeration had a single enumerator with value 0." In other words, the enum is a complete type with non-zero size. However, the compiler mistakenly thinks that this enum is incomplete, so it does not emit a DW_AT_byte_size attribute in the DWARF information for the enumeration type. This lack causes CCS to crash when attempting to display a variable declared with that enumeration type.

    The workaround is to add a dummy enumerator to the enum type, like so:

    enum e { fake = 0 };

  • Hi Archaeologist,

     

    Thanks for your detailed and appreciated answer. With the workaround you suggested, I have no more problems with the debugger. I’m now looking forward to experimenting future releases of CCS with the bug in the compiler removed.

     

    Kind Regards,

    Alessandro