Fix ICE in vect_slp_analyze_node_operations_1

This patch fixes bug 90681.  It was caused by trying to SLP vectorize a non
groupped load.  We've fixed it by tweaking a bit the implementation: mark
masked loads as not vectorizable, but support them as an special case.  Then
the detect them in the test for normal non-groupped loads that was already
there.

From-SVN: r271856

gcc/ChangeLog

+2019-06-03  Alejandro Martinez  <alejandro.martinezvicente@arm.com>
+
+	PR tree-optimization/90681
+	* internal-fn.c (mask_load_direct): Mark as non-vectorizable again.
+	* tree-vect-slp.c (vect_build_slp_tree_1): Add masked loads as a
+	special case for SLP, but fail on non-groupped loads.
+
 2019-06-03  Martin Liska  <mliska@suse.cz>
 
 	* cfg.c (debug): Use TDF_DETAILS for debug and

gcc/internal-fn.c

@@ -100,7 +100,7 @@ init_internal_fns ()
 /* Create static initializers for the information returned by
    direct_internal_fn.  */
 #define not_direct { -2, -2, false }
-#define mask_load_direct { -1, 2, true }
+#define mask_load_direct { -1, 2, false }
 #define load_lanes_direct { -1, -1, false }
 #define mask_load_lanes_direct { -1, -1, false }
 #define gather_load_direct { -1, -1, false }

gcc/testsuite/ChangeLog

+2019-06-03  Alejandro Martinez  <alejandro.martinezvicente@arm.com>
+
+	PR tree-optimization/90681
+	* gfortran.dg/vect/pr90681.f: New test.
+
 2019-06-03  Richard Biener  <rguenther@suse.de>
 
 	PR testsuite/90713

gcc/testsuite/gfortran.dg/vect/pr90681.f

+C { dg-do compile }
+C { dg-additional-options "-march=armv8.2-a+sve" { target { aarch64*-*-* } } }
+      SUBROUTINE HMU (H1)
+      COMMON DD(107)
+      DIMENSION H1(NORBS,*)
+            DO 70 J1 = IA,I1
+               H1(I1,J1) = 0
+               JO1 = J1
+               IF (JO1.EQ.1) THEN
+                   H1(I1,J1) = DD(NI)
+               END IF
+   70       CONTINUE
+      END

gcc/tree-vect-slp.c

@@ -661,6 +661,7 @@ vect_build_slp_tree_1 (unsigned char *swap,
   machine_mode optab_op2_mode;
   machine_mode vec_mode;
   stmt_vec_info first_load = NULL, prev_first_load = NULL;
+  bool load_p = false;
 
   /* For every stmt in NODE find its def stmt/s.  */
   stmt_vec_info stmt_info;
@@ -714,13 +715,16 @@ vect_build_slp_tree_1 (unsigned char *swap,
       if (gcall *call_stmt = dyn_cast <gcall *> (stmt))
 	{
 	  rhs_code = CALL_EXPR;
-	  if ((gimple_call_internal_p (call_stmt)
-	       && (!vectorizable_internal_fn_p
-		   (gimple_call_internal_fn (call_stmt))))
-	      || gimple_call_tail_p (call_stmt)
-	      || gimple_call_noreturn_p (call_stmt)
-	      || !gimple_call_nothrow_p (call_stmt)
-	      || gimple_call_chain (call_stmt))
+
+	  if (gimple_call_internal_p (stmt, IFN_MASK_LOAD))
+	    load_p = true;
+	  else if ((gimple_call_internal_p (call_stmt)
+		    && (!vectorizable_internal_fn_p
+			(gimple_call_internal_fn (call_stmt))))
+		   || gimple_call_tail_p (call_stmt)
+		   || gimple_call_noreturn_p (call_stmt)
+		   || !gimple_call_nothrow_p (call_stmt)
+		   || gimple_call_chain (call_stmt))
 	    {
 	      if (dump_enabled_p ())
 		dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location,
@@ -732,7 +736,10 @@ vect_build_slp_tree_1 (unsigned char *swap,
 	    }
 	}
       else
-	rhs_code = gimple_assign_rhs_code (stmt);
+	{
+	  rhs_code = gimple_assign_rhs_code (stmt);
+	  load_p = TREE_CODE_CLASS (rhs_code) == tcc_reference;
+	}
 
       /* Check the operation.  */
       if (i == 0)
@@ -899,7 +906,7 @@ vect_build_slp_tree_1 (unsigned char *swap,
         } /* Grouped access.  */
       else
 	{
-	  if (TREE_CODE_CLASS (rhs_code) == tcc_reference)
+	  if (load_p)
 	    {
 	      /* Not grouped load.  */
 	      if (dump_enabled_p ())