• State Resolved
  • Locked Locked
  • Replies 4 replies
  • Subscribers 52 subscribers
  • Views 2295 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.

compiling error message "invalid instruction schedule generated"

Takahiro Ogo
Genius 5940 points



Hi
I've got the following error messages while compiling a certain c code.
error: file xxxxxxx.c : invalid instruction schedule generated; try reducing speed optimization levels

However, this OBJ file is generated after compiling and it seems to work correctly.

What kind of problem is this? I have to reduce the optimize level?
If possible, I want to use it as it is.

#CGT version 6.1.13
Build Option:
 -mv6740
 -O3
 --symdebug:none

o2 and o1 are the same situation.
When I stop optimization, this error message disappears .

-Takao

 

  • 0
    TI__Guru* 84285 points

    This is known bug SDSCM00023977.  See SDOWP for updates on this issue.

  • 0 in reply to Archaeologist
    Genius 5940 points

    Thank you for the reply.
    I saw the bug report of SDSCM00023977. Please let me confirm this.

    1)
    Even if this error message is appeared, OBJ is generated. And it seems to work correctly on board.
    I have to reduce the optimize level?

    2)
    This error message does not disappear even if I add -ms0/-ms1 to compiling option.
    Why does not workaround work?

    #CGT version 6.1.13
    Build Option:
     -mv6740
     -O3
     --symdebug:none
     -ms0 (or -ms1)

    Regards.

  • 0 in reply to Takahiro Ogo
    TI__Guru* 84285 points

    Takao said:

    Even if this error message is appeared, OBJ is generated. And it seems to work correctly on board.

    Although the compiler does still create an object file, it is corrupted.  If the program ever executes the instruction that causes the problem, it will probably fail, but it is not guaranteed to fail.  The instruction in question is probably a double-precision floating-point instruction.

    Takao said:

    Why does not workaround work?

    I'm sorry, the suggested workaround is not guaranteed to work.  The suggested workaround hopes that slightly different optimization option combinations might cause the compiler-generated code to be different enough that the bug doesn't occur.  The only guaranteed way to avoid this error is to compile that file with a CPU version flag that does not have native floating-point support, such as -mv64plus.

  • 0 in reply to Archaeologist
    Genius 5940 points

    Thank you so much!

    -Takao