2009-10-23 Richard Guenther <rguenther@suse.de>

	PR tree-optimization/41778
	* tree-ssa-pre.c (do_regular_insertion): Only insert if a
	redundancy along a path in the CFG we want to optimize for speed
	is going to be removed.
	(execute_pre): Do partial-PRE only if the function is to be
	optimized for speed.
	(gate_pre): Do not turn off all of PRE when not optimizing a
	function for speed.

	* gcc.dg/tree-ssa/ssa-pre-26.c: New testcase.


git-svn-id: https://gcc.gnu.org/svn/gcc/trunk@153491 138bc75d-0d04-0410-961f-82ee72b054a4

4 files changed, 53 insertions, 5 deletions

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index e03bd07eb34..e0f70c48843 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,14 @@
+2009-10-23  Richard Guenther  <rguenther@suse.de>
+
+	PR tree-optimization/41778
+	* tree-ssa-pre.c (do_regular_insertion): Only insert if a
+	redundancy along a path in the CFG we want to optimize for speed
+	is going to be removed.
+	(execute_pre): Do partial-PRE only if the function is to be
+	optimized for speed.
+	(gate_pre): Do not turn off all of PRE when not optimizing a
+	function for speed.
+
 2009-10-23  Kaveh R. Ghazi  <ghazi@caip.rutgers.edu>
 
 	* builtins.c (fold_builtin_cabs): Use validate_arg().
diff --git a/gcc/testsuite/ChangeLog b/gcc/testsuite/ChangeLog
index ba31a604cea..aa2522805c2 100644
--- a/gcc/testsuite/ChangeLog
+++ b/gcc/testsuite/ChangeLog
@@ -1,3 +1,8 @@
+2009-10-23  Richard Guenther  <rguenther@suse.de>
+
+	PR tree-optimization/41778
+	* gcc.dg/tree-ssa/ssa-pre-26.c: New testcase.
+
 2009-10-22  Richard Guenther  <rguenther@suse.de>
 
 	PR lto/41791
diff --git a/gcc/testsuite/gcc.dg/tree-ssa/ssa-pre-26.c b/gcc/testsuite/gcc.dg/tree-ssa/ssa-pre-26.c
new file mode 100644
index 00000000000..978b7abab3a
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/tree-ssa/ssa-pre-26.c
@@ -0,0 +1,27 @@
+/* { dg-do compile } */
+/* { dg-options "-Os -fdump-tree-pre-details" } */
+
+typedef union
+{
+  int* data;
+} SA;
+
+typedef struct
+{
+  int reserved;
+  char* array;
+}SB;
+
+typedef struct
+{
+  int status;
+}SC;
+
+void foo(SA* pResult, SB* method, SC* self)
+{
+  if (method->array[0] == 'L' && !self->status && pResult->data != 0)
+    pResult->data = pResult->data;
+}
+
+/* { dg-final { scan-tree-dump "Deleted redundant store" "pre" } } */
+/* { dg-final { cleanup-tree-dump "pre" } } */
diff --git a/gcc/tree-ssa-pre.c b/gcc/tree-ssa-pre.c
index 5da6c63b400..6ae05b5a866 100644
--- a/gcc/tree-ssa-pre.c
+++ b/gcc/tree-ssa-pre.c
@@ -3405,6 +3405,7 @@ do_regular_insertion (basic_block block, basic_block dom)
 	  pre_expr eprime = NULL;
 	  edge_iterator ei;
 	  pre_expr edoubleprime = NULL;
+	  bool do_insertion = false;
 
 	  val = get_expr_value_id (expr);
 	  if (bitmap_set_contains_value (PHI_GEN (block), val))
@@ -3456,6 +3457,10 @@ do_regular_insertion (basic_block block, basic_block dom)
 		{
 		  avail[bprime->index] = edoubleprime;
 		  by_some = true;
+		  /* We want to perform insertions to remove a redundancy on
+		     a path in the CFG we want to optimize for speed.  */
+		  if (optimize_edge_for_speed_p (pred))
+		    do_insertion = true;
 		  if (first_s == NULL)
 		    first_s = edoubleprime;
 		  else if (!pre_expr_eq (first_s, edoubleprime))
@@ -3466,7 +3471,8 @@ do_regular_insertion (basic_block block, basic_block dom)
 	     already existing along every predecessor, and
 	     it's defined by some predecessor, it is
 	     partially redundant.  */
-	  if (!cant_insert && !all_same && by_some && dbg_cnt (treepre_insert))
+	  if (!cant_insert && !all_same && by_some && do_insertion
+	      && dbg_cnt (treepre_insert))
 	    {
 	      if (insert_into_preds_of_block (block, get_expression_id (expr),
 					      avail))
@@ -4526,11 +4532,11 @@ fini_pre (bool do_fre)
    only wants to do full redundancy elimination.  */
 
 static unsigned int
-execute_pre (bool do_fre ATTRIBUTE_UNUSED)
+execute_pre (bool do_fre)
 {
   unsigned int todo = 0;
 
-  do_partial_partial = optimize > 2;
+  do_partial_partial = optimize > 2 && optimize_function_for_speed_p (cfun);
 
   /* This has to happen before SCCVN runs because
      loop_optimizer_init may create new phis, etc.  */
@@ -4615,8 +4621,7 @@ do_pre (void)
 static bool
 gate_pre (void)
 {
-  /* PRE tends to generate bigger code.  */
-  return flag_tree_pre != 0 && optimize_function_for_speed_p (cfun);
+  return flag_tree_pre != 0;
 }
 
 struct gimple_opt_pass pass_pre =