[llvm-bugs] [Bug 31703] New: DCE optimization lost after manual inlining

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=31703

            Bug ID: 31703
           Summary: DCE optimization lost after manual inlining
           Product: clang
           Version: 3.9
          Hardware: PC
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: C++
          Assignee: unassignedclangbugs at nondot.org
          Reporter: skvadrik at gmail.com
                CC: dgregor at apple.com, llvm-bugs at lists.llvm.org
    Classification: Unclassified

Created attachment 17870
  --> https://llvm.org/bugs/attachment.cgi?id=17870&action=edit
a.c, b.c

Consider the following two snippets, a.c and b.c (in attach):

// a.c:
static void f(const char *s)
{
    for (; *s++ == '0'; );
}
int main(int argc, char **argv)
{
    const char *s0 = argv[1];
    for (int x = 0; x < 1000000000; ++x) f(s0);
    return 0;
}

// b.c is the same as a.c with 'f' inlined:
int main(int argc, char **argv)
{
    const char *s0 = argv[1];
    for (int x = 0; x < 1000000000; ++x) {
        for (const char *s = s0; *s++ == '0'; );
    }
    return 0;
}

Clang manages to optimize a.c to just 'return 0':

$ clang++-3.9 -c -O2 a.c -oa.o && objdump -d a.o
clang-3.9: warning: treating 'c' input as 'c++' when in C++ mode, this behavior
is deprecated

a.o:     file format elf64-x86-64


Disassembly of section .text:

0000000000000000 <main>:
   0:   31 c0                   xor    %eax,%eax
   2:   c3                      retq 

However, after manual inlining of 'f' this optimization fails:

$ clang++-3.9 -c -O2 b.c -ob.o && objdump -d b.o
clang-3.9: warning: treating 'c' input as 'c++' when in C++ mode, this behavior
is deprecated

b.o:     file format elf64-x86-64


Disassembly of section .text:

0000000000000000 <main>:
   0:   48 8b 46 08             mov    0x8(%rsi),%rax
   4:   31 c9                   xor    %ecx,%ecx
   6:   66 2e 0f 1f 84 00 00    nopw   %cs:0x0(%rax,%rax,1)
   d:   00 00 00 
  10:   48 89 c2                mov    %rax,%rdx
  13:   66 66 66 66 2e 0f 1f    data16 data16 data16 nopw %cs:0x0(%rax,%rax,1)
  1a:   84 00 00 00 00 00 
  20:   80 3a 30                cmpb   $0x30,(%rdx)
  23:   48 8d 52 01             lea    0x1(%rdx),%rdx
  27:   74 f7                   je     20 <main+0x20>
  29:   48 89 c2                mov    %rax,%rdx
  2c:   0f 1f 40 00             nopl   0x0(%rax)
  30:   80 3a 30                cmpb   $0x30,(%rdx)
  33:   48 8d 52 01             lea    0x1(%rdx),%rdx
  37:   74 f7                   je     30 <main+0x30>
  39:   48 89 c2                mov    %rax,%rdx
  3c:   0f 1f 40 00             nopl   0x0(%rax)
  40:   80 3a 30                cmpb   $0x30,(%rdx)
  43:   48 8d 52 01             lea    0x1(%rdx),%rdx
  47:   74 f7                   je     40 <main+0x40>
  49:   48 89 c2                mov    %rax,%rdx
  4c:   0f 1f 40 00             nopl   0x0(%rax)
  50:   80 3a 30                cmpb   $0x30,(%rdx)
  53:   48 8d 52 01             lea    0x1(%rdx),%rdx
  57:   74 f7                   je     50 <main+0x50>
  59:   48 89 c2                mov    %rax,%rdx
  5c:   0f 1f 40 00             nopl   0x0(%rax)
  60:   80 3a 30                cmpb   $0x30,(%rdx)
  63:   48 8d 52 01             lea    0x1(%rdx),%rdx
  67:   74 f7                   je     60 <main+0x60>
  69:   83 c1 05                add    $0x5,%ecx
  6c:   81 f9 00 ca 9a 3b       cmp    $0x3b9aca00,%ecx
  72:   75 9c                   jne    10 <main+0x10>
  74:   31 c0                   xor    %eax,%eax
  76:   c3                      retq

-- 
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/20170120/cf0a8a0d/attachment.html>