re PR lto/66752 (spec2000 255.vortex performance compiled with GCC is ~20% lower than with CLANG)

	PR lto/66752
	* tree-ssa-threadedge.c (simplify_conrol_stmt_condition): If we are
	unable to find X NE 0 in the tables, return X as the simplified
	condition.
	(fsm_find_control_statement_thread_paths): If nodes in NEXT_PATH are
	in VISISTED_BBS, then return failure.  Else add nodes from NEXT_PATH
	to VISISTED_BBS.  */
	* tree-ssa-threadupdate.c (duplicate_thread_path): Fix up edge flags
	after removing the control flow statement and unnecessary edges.

testsuite/
	PR lto/66752
	* gcc.dg/tree-ssa/pr66752-2.c: New test.
	* gcc.dg/torture/pr66752-1.c: New test
	* g++.dg/torture/pr66752-2.C: New test.

From-SVN: r226125

1 files changed, 25 insertions, 0 deletions

diff --git a/gcc/tree-ssa-threadedge.c b/gcc/tree-ssa-threadedge.c
index 7164122..5228951 100644
--- a/gcc/tree-ssa-threadedge.c
+++ b/gcc/tree-ssa-threadedge.c
@@ -553,6 +553,16 @@ simplify_control_stmt_condition (edge e,
           || !is_gimple_min_invariant (cached_lhs))
         cached_lhs = (*simplify) (dummy_cond, stmt);
 
+      /* If we were just testing that an integral type was != 0, and that
+	 failed, just return the first operand.  This gives the FSM code a
+	 chance to optimize the path.  */
+      if (cached_lhs == NULL
+	  && cond_code == NE_EXPR
+	  && INTEGRAL_TYPE_P (TREE_TYPE (op0))
+	  && TREE_CODE (op0) == SSA_NAME
+	  && integer_zerop (op1))
+	return op0;
+
       return cached_lhs;
     }
 
@@ -974,6 +984,21 @@ fsm_find_control_statement_thread_paths (tree expr,
 	  return;
 	}
 
+      /* Make sure we haven't already visited any of the nodes in
+ 	 NEXT_PATH.  Don't add them here to avoid pollution.  */
+      for (unsigned int i = 0; i < next_path->length () - 1; i++)
+	{
+	  if (visited_bbs->contains ((*next_path)[i]))
+	    {
+	      vec_free (next_path);
+	      return;
+	    }
+	}
+
+      /* Now add the nodes to VISISTED_BBS.  */
+      for (unsigned int i = 0; i < next_path->length () - 1; i++)
+	visited_bbs->add ((*next_path)[i]);
+
       /* Append all the nodes from NEXT_PATH to PATH.  */
       vec_safe_splice (path, next_path);
       next_path_length = next_path->length ();