Enhance if-conversion for automatic arrays

Automatic arrays that are not address-taken should not be subject to
store data races.  This applies to OMP SIMD in-branch lowered
functions result array which for the testcase otherwise prevents
vectorization with SSE and for AVX and AVX512 ends up with spurious
.MASK_STORE to the stack surviving.

This inefficiency was noted in PR111793.

I've introduced ref_can_have_store_data_races, commonizing uses
of flag_store_data_races in if-conversion, cselim and store motion.

	PR tree-optimization/111793
	* tree-ssa-alias.h (ref_can_have_store_data_races): Declare.
	* tree-ssa-alias.cc (ref_can_have_store_data_races): New
	function.
	* tree-if-conv.cc (ifcvt_memrefs_wont_trap): Use
	ref_can_have_store_data_races to allow more unconditional
	stores.
	* tree-ssa-loop-im.cc (execute_sm): Likewise.
	* tree-ssa-phiopt.cc (cond_store_replacement): Likewise.

	* gcc.dg/vect/vect-simd-clone-21.c: New testcase.

1 files changed, 5 insertions, 6 deletions

diff --git a/gcc/tree-if-conv.cc b/gcc/tree-if-conv.cc
index c4c3ed4..57992b6 100644
--- a/gcc/tree-if-conv.cc
+++ b/gcc/tree-if-conv.cc
@@ -936,12 +936,11 @@ ifcvt_memrefs_wont_trap (gimple *stmt, vec<data_reference_p> drs)
 
       /* an unconditionaly write won't trap if the base is written
          to unconditionally.  */
-      if (base_master_dr
-	  && DR_BASE_W_UNCONDITIONALLY (*base_master_dr))
-	return flag_store_data_races;
-      /* or the base is known to be not readonly.  */
-      else if (base_object_writable (DR_REF (a)))
-	return flag_store_data_races;
+      if ((base_master_dr
+	   && DR_BASE_W_UNCONDITIONALLY (*base_master_dr))
+	  /* or the base is known to be not readonly.  */
+	  || base_object_writable (DR_REF (a)))
+	return !ref_can_have_store_data_races (base);
     }
 
   return false;