[LLVMbugs] [Bug 15968] New: constexpr code unnecessarily generated and called

[bugzilla-daemon at llvm.org]

http://llvm.org/bugs/show_bug.cgi?id=15968

            Bug ID: 15968
           Summary: constexpr code unnecessarily generated and called
           Product: clang
           Version: trunk
          Hardware: PC
                OS: All
            Status: NEW
          Severity: normal
          Priority: P
         Component: C++11
          Assignee: unassignedclangbugs at nondot.org
          Reporter: hhinnant at apple.com
                CC: dgregor at apple.com, llvmbugs at cs.uiuc.edu
    Classification: Unclassified

Created attachment 10497
  --> http://llvm.org/bugs/attachment.cgi?id=10497&action=edit
date.cpp

I've been working on a constexpr date class that validates its input.  If all
of the input is compile-time information, the validation should happen at
compile time.

The enclosed file ends with:

int
main()
{
    constexpr ymd_date ymd(year(2013), month(5), day(11));
    return ymd.year();
}

Using tip-of-trunk clang, and -O3, the generated assembly for main is:


    .section    __TEXT,__text,regular,pure_instructions
    .globl    _main
    .align    4, 0x90
_main:                                  ## @main
    .cfi_startproc
## BB#0:                                ## %entry
    pushq    %rbp
Ltmp2:
    .cfi_def_cfa_offset 16
Ltmp3:
    .cfi_offset %rbp, -16
    movq    %rsp, %rbp
Ltmp4:
    .cfi_def_cfa_register %rbp
    movl    $2013, %edi             ## imm = 0x7DD
    movl    $5, %esi
    movl    $11, %edx
    callq    __ZN8ymd_date16check_invariantsE4year5month3day
    movl    $2013, %eax             ## imm = 0x7DD
    popq    %rbp
    ret
    .cfi_endproc

However the compiler should be able to deduce that the call to
__ZN8ymd_date16check_invariantsE4year5month3day is unnecessary because:

1.  It is constexpr.
2.  It is called in the process of constructing a constexpr.
3.  It compiled.

Thus it can't possibly have any run time impact.  What I'm looking for is:

    .section    __TEXT,__text,regular,pure_instructions
    .globl    _main
    .align    4, 0x90
_main:                                  ## @main
    .cfi_startproc
## BB#0:                                ## %entry
    pushq    %rbp
Ltmp2:
    .cfi_def_cfa_offset 16
Ltmp3:
    .cfi_offset %rbp, -16
    movq    %rsp, %rbp
Ltmp4:
    .cfi_def_cfa_register %rbp
    movl    $2013, %eax             ## imm = 0x7DD
    popq    %rbp
    ret
    .cfi_endproc

Which is *all* that gets generated if I change the enclosure to:

constexpr ymd_date ymd(year(2013), month(5), day(11));

int
main()
{
    return ymd.year();
}

In the example the programmer can quickly loose track of what input needs to be
validated and which doesn't (because it is compile-time information).  And this
seems to me to be one of the things that constexpr could do 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/20130511/a03b2e9b/attachment.html>