[LLVMbugs] [Bug 7260] New: Bogus(?) floating-point overflow

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Sun May 30 09:40:51 PDT 2010


           Summary: Bogus(?) floating-point overflow
           Product: libraries
           Version: trunk
          Platform: Macintosh
        OS/Version: MacOS X
            Status: NEW
          Severity: normal
          Priority: P
         Component: Backend: X86
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: luc_j_bourhis at mac.com
                CC: llvmbugs at cs.uiuc.edu

Created an attachment (id=4949)
 --> (http://llvm.org/bugs/attachment.cgi?id=4949)
self-contained source code body to reproduce the problem. 

LLVM revision 105128
CLANG revision 105014
MacOS 10.6.3
XCode 3.2.2
MacPro 1st generation

I attached a self-contained source code to reproduce the problem:
(1) make
(2) ./tst_eigen: that should result in a floating-point exception
(3) ./tst_eigen mask_overflow: the program should enter an infinite loop

That shows that a floating-point overflow happens, and it can only be in the
function real_symmetric_eigen.

The program runs normally if CXXFLAGS does not feature -O3.

It also runs fine if g++ is used, even with -O3.

The body of that function real_symmetric_eigen is truly a black-box which I
have not studied. My motivation to report it is that this is a blocker to
enable clang as a compiler for the project this code comes from, the
Computational Crystallographic Toolbox (CCTBX).

Configure bugmail: http://llvm.org/bugs/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

More information about the llvm-bugs mailing list