[LLVMbugs] [Bug 17113] New: extra zeroing move operation executed

bugzilla-daemon at llvm.org bugzilla-daemon at llvm.org
Thu Sep 5 14:45:51 PDT 2013


            Bug ID: 17113
           Summary: extra zeroing move operation executed
           Product: libraries
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: Backend: X86
          Assignee: unassignedbugs at nondot.org
          Reporter: kkhoo at perfwizard.com
                CC: llvmbugs at cs.uiuc.edu
    Classification: Unclassified

$ ./clang -v
clang version 3.4 (trunk 189776)
Target: x86_64-apple-darwin11.4.2
Thread model: posix

$ cat power_of_2.c 
unsigned int power_of_two(unsigned int x) {
    return x && !(x & (x - 1));

$ ./clang -S -O3 -fomit-frame-pointer  power_of_2.c -o /dev/stdout 
    .section    __TEXT,__text,regular,pure_instructions
    .globl    _power_of_two
    .align    4, 0x90
_power_of_two:                          ## @power_of_two
## BB#0:                                ## %entry
                                        ## kill: EDI<def> EDI<kill> RDI<def>
    xorl    %eax, %eax
    testl    %edi, %edi
    je    LBB0_2
## BB#1:                                ## %land.rhs
    leal    -1(%rdi), %eax
    testl    %edi, %eax
    sete    %al
LBB0_2:                                 ## %land.end
    movzbl    %al, %eax


I hope I'm not missing something in reading this asm...

In the case where x==0, we execute the 'test' to see if x==0 and then 'je'
(jump if equal to 0) to the 'movzb' instruction. However, we've already ensured
that %eax is zeroed out at this point via the 'xor'. Couldn't we have jumped
straight to the 'ret'?

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/20130905/f4c3c6af/attachment.html>

More information about the llvm-bugs mailing list