Avoid more invalid GIMPLE with register bases

The Ada frontend, for example with gnat.dg/inline2_pkg.adb, tends
to create VIEW_CONVERT expressions with aggregate type even of
non-aggregate entities.  In this case for example

  return <retval> = (BIT_FIELD_REF <VIEW_CONVERT_EXPR<struct inline2_pkg__ieee_short_real>(number), 16, 16> & 32640) != 32640;

currently gimplification and SSA rewrite turn this into

  _1 = BIT_FIELD_REF <VIEW_CONVERT_EXPR<struct inline2_pkg__ieee_short_real>(number_2(D));

which is two operations on a register.  While as seen with PR109652
we might not want to completely rule out register to aggregate type
VIEW_CONVERTs we definitely do not want to stack multiple ops here.

The solution is to make sure the gimplifier puts a non-register as
the base object.  For the above this will add

  number.1 = number;

and use number.1 in the compound reference.  Code generation is
unchanged, FRE optimizes this to BIT_FIELD_REF <number_2(D), ...>.
I think BIT_FIELD_REF <VIEW_CONVERT (x), ...> could be always
rewritten into BIT_FIELD_REF <x, ...>, but that's a separate thing.

	* gimplify.cc (gimplify_compound_lval): When there's a
	non-register type produced by one of the handled component
	operations make sure we get a non-register base.

diff --git a/gcc/gimplify.cc b/gcc/gimplify.cc
index c38a962dd0421c5db637f46a64030e02172c82dd..3740a8979af2a7a40e74011344171e151c982621 100644 (file)
--- a/gcc/gimplify.cc
+++ b/gcc/gimplify.cc
@@ -3264,6 +3264,11 @@ gimplify_compound_lval (tree *expr_p, gimple_seq *pre_p, gimple_seq *post_p,
            }
          need_non_reg = true;
        }
+      else if (!is_gimple_reg_type (TREE_TYPE (t)))
+       /* When the result of an operation, in particular a VIEW_CONVERT_EXPR
+          is a non-register type then require the base object to be a
+          non-register as well.  */
+       need_non_reg = true;
     }
 
   /* Step 2 is to gimplify the base expression.  Make sure lvalue is set