Improve ivopts handling of forced scales

This patch improves the ivopts address cost calculation for modes
in which an index must be scaled rather than unscaled.  Previously
we would only try the scaled form if the unscaled form was valid.

Many of the SVE tests rely on this when matching scaled indices.

2017-11-09  Richard Sandiford  <richard.sandiford@linaro.org>
	    Alan Hayward  <alan.hayward@arm.com>
	    David Sherwood  <david.sherwood@arm.com>

gcc/
	* tree-ssa-loop-ivopts.c (get_address_cost): Try using a
	scaled index even if the unscaled address was invalid.
	Don't increase the complexity of using a scale in that case.

Co-Authored-By: Alan Hayward <alan.hayward@arm.com>
Co-Authored-By: David Sherwood <david.sherwood@arm.com>
From-SVN: r254585

diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index c4e9444..eb0e6c2 100644 (file)
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -2,6 +2,14 @@
            Alan Hayward  <alan.hayward@arm.com>
            David Sherwood  <david.sherwood@arm.com>
 
+       * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
+       scaled index even if the unscaled address was invalid.
+       Don't increase the complexity of using a scale in that case.
+
+2017-11-09  Richard Sandiford  <richard.sandiford@linaro.org>
+           Alan Hayward  <alan.hayward@arm.com>
+           David Sherwood  <david.sherwood@arm.com>
+
        * doc/rtl.texi: Rewrite the subreg rules so that they partition
        the inner register into REGMODE_NATURAL_SIZE bytes rather than
        UNITS_PER_WORD bytes.

diff --git a/gcc/tree-ssa-loop-ivopts.c b/gcc/tree-ssa-loop-ivopts.c
index e16a482..65794b2 100644 (file)
--- a/gcc/tree-ssa-loop-ivopts.c
+++ b/gcc/tree-ssa-loop-ivopts.c
@@ -4331,18 +4331,25 @@ get_address_cost (struct ivopts_data *data, struct iv_use *use,
   machine_mode addr_mode = TYPE_MODE (type);
   machine_mode mem_mode = TYPE_MODE (TREE_TYPE (*use->op_p));
   addr_space_t as = TYPE_ADDR_SPACE (TREE_TYPE (use->iv->base));
+  /* Only true if ratio != 1.  */
+  bool ok_with_ratio_p = false;
+  bool ok_without_ratio_p = false;
 
   if (!aff_combination_const_p (aff_inv))
     {
       parts.index = integer_one_node;
       /* Addressing mode "base + index".  */
-      if (valid_mem_ref_p (mem_mode, as, &parts))
+      ok_without_ratio_p = valid_mem_ref_p (mem_mode, as, &parts);
+      if (ratio != 1)
        {
          parts.step = wide_int_to_tree (type, ratio);
          /* Addressing mode "base + index << scale".  */
-         if (ratio != 1 && !valid_mem_ref_p (mem_mode, as, &parts))
+         ok_with_ratio_p = valid_mem_ref_p (mem_mode, as, &parts);
+         if (!ok_with_ratio_p)
            parts.step = NULL_TREE;
-
+       }
+      if (ok_with_ratio_p || ok_without_ratio_p)
+       {
          if (aff_inv->offset != 0)
            {
              parts.offset = wide_int_to_tree (sizetype, aff_inv->offset);
@@ -4440,7 +4447,9 @@ get_address_cost (struct ivopts_data *data, struct iv_use *use,
 
   if (parts.symbol != NULL_TREE)
     cost.complexity += 1;
-  if (parts.step != NULL_TREE && !integer_onep (parts.step))
+  /* Don't increase the complexity of adding a scaled index if it's
+     the only kind of index that the target allows.  */
+  if (parts.step != NULL_TREE && ok_without_ratio_p)
     cost.complexity += 1;
   if (parts.base != NULL_TREE && parts.index != NULL_TREE)
     cost.complexity += 1;