[Bug c++/51962] New: Compiling with -O3 and using the same input produces a different result

[Bug c++/51962] New: Compiling with -O3 and using the same input produces a different result

[mario.achkar at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51962

             Bug #: 51962
           Summary: Compiling with -O3 and using the same input produces a
                    different result
    Classification: Unclassified
           Product: gcc
           Version: 4.6.2
            Status: UNCONFIRMED
          Severity: critical
          Priority: P3
         Component: c++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: mario.achkar@gmail.com


Created attachment 26423
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=26423
The simple C++ code that causes the "bug".

Compile the following simple code without -O3, and run.

Now compile it with -O3 option (for optimization), run again.

Surprisingly 2 different outputs appear.

I have made the program as minimal as possible and traced back the error to the
following line:

if (((k-x[j])*(k-x[j])+(y[j]-ya)*(y[j]-ya))<=r[j]*r[j]) { found1 = true; } 
(line 17 in the code)

Comment this and the problem is solved.

This caused a huge problem that took me around an hour to figure out that it
was related to compiler optimization. I have provided the simplest case I could
find causing the error.

Thank you for your time and hope you fix it soon.
Please keep me updated.

[Bug c++/51962] Compiling with -O3 and using the same input produces a different result

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51962

Jonathan Wakely <redi at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|critical                    |normal

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-01-23 11:19:39 UTC ---
You do not initialise found1.  Set that to false and your problem probably goes
away.

[Bug c++/51962] Compiling with -O3 and using the same input produces a different result

[mario.achkar at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51962

Mario Achkar <mario.achkar at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID

--- Comment #2 from Mario Achkar <mario.achkar at gmail dot com> 2012-01-23 11:24:54 UTC ---
yes you are right. That was hard to find. It's weird though how both work
differently when it comes to this.

[Bug c++/51962] Compiling with -O3 and using the same input produces a different result

[mario.achkar at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51962

--- Comment #3 from Mario Achkar <mario.achkar at gmail dot com> 2012-01-23 11:27:48 UTC ---
(In reply to comment #1)
> You do not initialise found1.  Set that to false and your problem probably goes
> away.
Thanks for the fast reply.

[Bug c++/51962] Compiling with -O3 and using the same input produces a different result

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51962

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-01-23 11:33:39 UTC ---
It was very easy to find using valgrind, or a static analysis tool.