<html>
    <head>
      <base href="http://llvm.org/bugs/" />
    </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 --- - Agressively merge BBs of predict-taken paths"
   href="http://llvm.org/bugs/show_bug.cgi?id=16426">16426</a>
          </td>
        </tr>

        <tr>
          <th>Summary</th>
          <td>Agressively merge BBs of predict-taken paths
          </td>
        </tr>

        <tr>
          <th>Product</th>
          <td>libraries
          </td>
        </tr>

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

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

        <tr>
          <th>OS</th>
          <td>All
          </td>
        </tr>

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

        <tr>
          <th>Severity</th>
          <td>normal
          </td>
        </tr>

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

        <tr>
          <th>Component</th>
          <td>Scalar Optimizations
          </td>
        </tr>

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

        <tr>
          <th>Reporter</th>
          <td>nunoplopes@sapo.pt
          </td>
        </tr>

        <tr>
          <th>CC</th>
          <td>llvmbugs@cs.uiuc.edu
          </td>
        </tr>

        <tr>
          <th>Classification</th>
          <td>Unclassified
          </td>
        </tr></table>
      <p>
        <div>
        <pre>I think LLVM should be more agressive in merging BBs of predict-taken paths.

E.g.:
size_t safe_address(size_t nmemb, size_t size, size_t offset) {
  unsigned long res;
  if (__builtin_expect(
       __builtin_expect(__builtin_umul_overflow(nmemb, size, &res), 0) ||
       __builtin_expect( __builtin_uadd_overflow(res, offset, &res), 0), 0)) {
    return 0;
  }
  return res;
}

(multiple __builtin_expect because of another unrelated bug).


This code compiles to:

define i32 @safe_address(i32 %nmemb, i32 %size, i32 %offset) {
entry:
  %0 = tail call { i32, i1 } @llvm.umul.with.overflow.i32(i32 %nmemb, i32
%size)
  %1 = extractvalue { i32, i1 } %0, 1
  br i1 %1, label %lor.end, label %lor.rhs, !prof !0

lor.rhs:                                          ; preds = %entry
  %2 = extractvalue { i32, i1 } %0, 0
  %3 = tail call { i32, i1 } @llvm.uadd.with.overflow.i32(i32 %2, i32 %offset)
  %4 = extractvalue { i32, i1 } %3, 1
  %5 = extractvalue { i32, i1 } %3, 0
  %phitmp = select i1 %4, i32 0, i32 %5
  br label %lor.end

lor.end:                                          ; preds = %lor.rhs, %entry
  %6 = phi i32 [ 0, %entry ], [ %phitmp, %lor.rhs ]
  ret i32 %6
}

!0 = metadata !{metadata !"branch_weights", i32 4, i32 64}

declare { i32, i1 } @llvm.umul.with.overflow.i32(i32, i32) #1
declare { i32, i1 } @llvm.uadd.with.overflow.i32(i32, i32) #1
attributes #1 = { nounwind readnone }


Since we know that the branch to 'lor.rhs' is assumed to be taken and that
there's no instruction with side-effects in that BB, we should merge the
'entry' and 'lor.rhs' BBs.  Merging BBs helps later passes, like instruction
selection.</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>