[llvm-bugs] [Bug 28829] New: Missed optimization when mixing volatile with normal access

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=28829

            Bug ID: 28829
           Summary: Missed optimization when mixing volatile with normal
                    access
           Product: new-bugs
           Version: 3.8
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: new bugs
          Assignee: unassignedbugs at nondot.org
          Reporter: ahmad at a3f.at
                CC: llvm-bugs at lists.llvm.org
    Classification: Unclassified

Godbolt link: https://godbolt.org/g/9brNCj

    int x;
    static int y;
    int f() {     x = 1; *(volatile int *)(&x) = x; x = 2; return x;}
    int g() {     y = 1; *(volatile int *)(&y) = y; y = 2; return y;}
    int h() { int z = 1; *(volatile int *)(&z) = z; z = 2; return z;}

with clang 3.8 -O2 on AMD64 compiles to:

    f:                                      # @f
        movl    $1, x(%rip)
        movl    $1, x(%rip)
        movl    $2, x(%rip)
        movl    $2, %eax
        retq

    g:                                      # @g
        movl    $1, y(%rip)
        movl    $1, y(%rip)
        movl    $2, y(%rip)
        movl    $2, %eax
        retq

    h:                                      # @h
        movl    $1, -4(%rsp)
        movl    $1, -4(%rsp)
        movl    $2, %eax
        retq

Is this intended behavior? My opinion is that the first $1 written isn't
necessary and that the $2 written to y should have been optimized out as well.

-- 
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/20160803/e001dfbc/attachment-0001.html>