<html>
    <head>
      <base href="https://bugs.llvm.org/">
    </head>
    <body><table border="1" cellspacing="0" cellpadding="8">
        <tr>
          <th>Bug ID</th>
          <td><a class="bz_bug_link 
          bz_status_NEW "
   title="NEW - Suboptimal codegen for CAS loop"
   href="https://bugs.llvm.org/show_bug.cgi?id=51104">51104</a>
          </td>
        </tr>

        <tr>
          <th>Summary</th>
          <td>Suboptimal codegen for CAS loop
          </td>
        </tr>

        <tr>
          <th>Product</th>
          <td>clang
          </td>
        </tr>

        <tr>
          <th>Version</th>
          <td>trunk
          </td>
        </tr>

        <tr>
          <th>Hardware</th>
          <td>PC
          </td>
        </tr>

        <tr>
          <th>OS</th>
          <td>Linux
          </td>
        </tr>

        <tr>
          <th>Status</th>
          <td>NEW
          </td>
        </tr>

        <tr>
          <th>Severity</th>
          <td>enhancement
          </td>
        </tr>

        <tr>
          <th>Priority</th>
          <td>P
          </td>
        </tr>

        <tr>
          <th>Component</th>
          <td>LLVM Codegen
          </td>
        </tr>

        <tr>
          <th>Assignee</th>
          <td>unassignedclangbugs@nondot.org
          </td>
        </tr>

        <tr>
          <th>Reporter</th>
          <td>gonzalo.gadeschi@gmail.com
          </td>
        </tr>

        <tr>
          <th>CC</th>
          <td>llvm-bugs@lists.llvm.org, neeilans@live.com, richard-llvm@metafoo.co.uk
          </td>
        </tr></table>
      <p>
        <div>
        <pre>Given this CAS loop:

#include <atomic>
int fetch_max(std::atomic<int>& mem, int val) {
    int read = mem.load(std::memory_order_relaxed);
    int write;
    do {
        write = std::max(read, val);
    } while( !mem.compare_exchange_weak(read, write) );
    return read;
}

Using -O3 -fno-exceptions -g0 -march=skylake, gcc generates:

fetch_max(std::atomic<int>&, int):
        mov     eax, DWORD PTR [rdi]
.L2:
        cmp     eax, esi
        mov     edx, esi
        cmovge  edx, eax
        lock cmpxchg    DWORD PTR [rdi], edx
        jne     .L2
        ret

but clang generates:

fetch_max(std::atomic<int>&, int):             # @fetch_max(std::atomic<int>&,
int)
        movl    (%rdi), %ecx
        cmpl    %esi, %ecx
        movl    %ecx, %edx
        cmovll  %esi, %edx
        movl    %ecx, %eax
        lock            cmpxchgl        %edx, (%rdi)
        je      .LBB0_3
.LBB0_1:                                # =>This Inner Loop Header: Depth=1
        movl    %eax, %ecx
        cmpl    %esi, %eax
        movl    %eax, %edx
        cmovll  %esi, %edx
        lock            cmpxchgl        %edx, (%rdi)
        jne     .LBB0_1
.LBB0_3:
        movl    %ecx, %eax
        retq

If instead of return read, I change the example to return write, then the
codegen improves significantly.</pre>
        </div>
      </p>


      <hr>
      <span>You are receiving this mail because:</span>

      <ul>
          <li>You are on the CC list for the bug.</li>
      </ul>
    </body>
</html>