[Bug tree-optimization/21829] New: [4.1 Regression] missed jump threading after unroller

[pinskia at gcc dot gnu dot org]

Take the following code:
int test(int v)
{
  int x = 0;
  for (int u=0;u<2;u++)
  {
    if (u>v)  // v is input-arg the compiler can't deside at compiletime
    {
      if (u%2==1) // can only happen for u==1 (so loops for 0 and 2 does not do
        x++;      // anything. Hoped gcc would notice when unrolling.
    }
  }  
  return x;
}

We get:
<bb 0>:
  if (v_8 < 0) goto <L1>; else goto <L15>;

<L15>:;
  if (v_8 <= 0) goto <L1>; else goto <L3>;

<L1>:;

  # x_1 = PHI <0(3), 1(1)>;
<L3>:;
  return x_1;

}

Notice that the if in L15 is always true.
The 4.0 branch gets this correct.

-- 
           Summary: [4.1 Regression] missed jump threading after unroller
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: minor
          Priority: P2
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: pinskia at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=21829