[LLVMbugs] [Bug 5551] New: Bug during llvm optimization with different results depending of architecture

[bugzilla-daemon at cs.uiuc.edu]

http://llvm.org/bugs/show_bug.cgi?id=5551

           Summary: Bug during llvm optimization with different results
                    depending of architecture
           Product: new-bugs
           Version: trunk
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: new bugs
        AssignedTo: unassignedbugs at nondot.org
        ReportedBy: quickslyver at free.fr
                CC: llvmbugs at cs.uiuc.edu


With this test case from gcc torture suite I have various results:

/* Origin: Joseph Myers <jsm28 at cam.ac.uk>.

   This tests for inconsistency in whether wide STRING_CSTs use the host
   or the target endianness.  */

extern void exit (int);
extern void abort (void);

int
main (void)
{
  if (L"a" "b"[1] != L'b')
    abort ();
  exit (0);
}

with clang at -O0 : generated code seems correct

with clang at -O3 :
on 32 bits little endian architectures it is optimized to an exit call (OK)
on 16 bits architectures like msp430 it is wrongly optimized to an abort call
(ERROR)

on 32 bits big-endian architecture it is not optimized (Optimization missing?)


-- 
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.