[LLVMbugs] [Bug 16109] New: Optimization error - Generates an "orr" command instead of "eor"

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=16109

            Bug ID: 16109
           Summary: Optimization error - Generates an "orr" command
                    instead of "eor"
           Product: libraries
           Version: 3.1
          Hardware: All
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Backend: ARM
          Assignee: unassignedbugs at nondot.org
          Reporter: ashaltiel at nds.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Created attachment 10565
  --> http://llvm.org/bugs/attachment.cgi?id=10565&action=edit
LLVM-IR code example

When a variable is being XOR'ed with immediate values consecutively, the
ARM-backend calculates their commulative xor (correctly), but (in certain
cases) instead of XORing it with the original value, it uses an OR (orr)
instruction.

In the attached .ll file - notice the lines
  %C033 = ptrtoint i8* %LGE1 to i64
  %A34 = xor i64 %C033, 1889002644
  %A35 = xor i64 %A34, 148

Which translate (using llc -march=arm) to:
        ldr     r0, .LCPI0_2
        orr     r4, r4, r0

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20130522/c4f0f057/attachment.html>