• State
  • Locked Locked
  • Replies 8 replies
  • Subscribers 54 subscribers
  • Views 853 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.

CCSv5 and CCSv6 occasionally hang on compiler error

Eddie3909
Guru 10385 points


CCS5.5, CGT 7.4.4

CCS6.0 CGT 7.4.8

Window 7

I originally thought this was a CCSv6 issue, but now that we're using CCSv5, we've observed it there too.

This has happened several times to myself and colleagues.

Steps

1) Build project

2) compiler finds error in a file, console displays error, but no more files are compiled.

3) CCS has to be killed with task manager because its hung at this point

The file error in once case was a simple variable declaration missing.

After restarting CCS, the file and project compiled OK.

Its random when this happens and is unfortunately infrequent. 

I tried to find a similar post but there are a lot of hits for CCSv5 + hang.

Cheers2u

Eddie

  • 0
    TI__Guru**** 468021 points

    Hi Eddie,

    Sorry to hear of your continual struggles with our tools.

    Without a reproducible test case, it is really difficult to determine the root cause of the hangs.

    Eddie3909 said:
    3) CCS has to be killed with task manager because its hung at this point

    Do you get the "Not Responding" message for CCS when trying to access it? Or is it an issue of CCS being blocked because the build is not complete? If you let it build in the background, does it still hang?

    ki

  • 0 in reply to Ki
    Intellectual 645 points

    Hi Ki-Soo,

       I am a co-worker of Eddie. 

    I believe it is a case of "Not Responding".

    If I leave it in that state for hours, it still is stuck there. I don't see any progress on the Build console.

    If I make any change to my source file (based on the compiler error message for that file). I can't save the file - it won't save.

    I can't even close the CCS.

    At that point the only way out is to kill  the CCS from Windows Task Manager.

    -- Somnath

     

  • 0 in reply to SNBANIK
    Guru 10385 points

    Hi Ki-Soo

    I'll get you info soon how my system behaves. Its been awhile.

    Hopefully Somnath's info will help.

    Cheers

    Eddie

  • 0 in reply to Eddie3909
    Guru 10385 points

    Hi Ki-Soo

    We have a theory whats causing the hang. The -j N compiler option is set to -j 8. It seems some PCs are OK with this while others need a N thats lower. It seems eclipse is getting confused when there are so many threads executing.

    Although this is slightly different, we have a build server that hangs at the linker phase. Changing to -j 8 fixed the server, but perhaps we need it lower for other PCs.

    Cheers
    Eddie

  • 0 in reply to Eddie3909
    TI__Guru**** 468021 points

    Eddie - that is interesting. Thanks for the troubleshooting and updating us. From what I read of Eclipse documentation, it is a bit of trial and error when trying to determine how many parallel jobs to specify. The standard default for many is simply the number of cores you have but nothing is given. Sometimes you can even increase the number of jobs than cores and you actually see more gains while in other cases you need to specify less jobs than cores. A lot of it depends on the actual application being compiled and which is why the default is currently to have it disabled.

  • 0 in reply to Ki
    Guru 10385 points

    Hi Ki-Soo

    In our build server case, not specifying the -j and letting it figure out how many was bad news. I'm going to ask Somnath to try his PC with something like -j 4 instead of -j 8. Doing so seems to have helped another engineer in our group.

    Wouldn't that be nice if it also helps the errors that never go away in the "Problems" tab? (wishful thinking perhaps)

    Cheers
    Eddie

  • 0 in reply to Eddie3909
    Guru 10385 points

    Rats. I thought we had this fixed by using "-j 4". It just hung on me. Arg.

    I changed it to -j 1 and then it compiled to the error without hanging (see below). Then I changed it back to -j 4 and it didn't hang. 

    >> Compilation failure
    "../Doppler.cluster/Integ.module/src/Doppler_gapFill.c", line 328: error: struct "<unnamed>" has no field "gapCount"
    1 error detected in the compilation of "../Doppler.cluster/Integ.module/src/Doppler_gapFill.c".
    gmake: *** [Doppler.cluster/Integ.module/src/Doppler_gapFill.obj] Error 1
    gmake: Target `all' not remade because of errors.

    Is there any tool I can use to figure out what CCS process is hanging?

    Please let me know if this is fixed in CCSv6. It also had the same problem when I tried it.

    Cheers

    Eddie

  • 0 in reply to Eddie3909
    Guru 10385 points

    This may be related.

    Bug 367429 - Stopping a parallel build (make -j) hangs infinitely
    https://bugs.eclipse.org/bugs/show_bug.cgi?id=367429

    Is there a way TI could provide us with a CCSv5.5.0.00077, Windows 7. C6472 starter.exe with the CREATE_BREAKAWAY_FROM_JOB flag set as per the linked post to the above bug report?
    https://bugs.eclipse.org/bugs/show_bug.cgi?id=333578

    Compiles with -j 1 are taking way too long. ;(

    Also, I re-posted this request on the link below just in case this thread has been closed.

    http://e2e.ti.com/support/development_tools/code_composer_studio/f/81/t/395228

    Cheers

    Eddie