t = init_composite_type ("__ppc_builtin_type_vec64", TYPE_CODE_UNION);
append_composite_type_field (t, "uint64", builtin_type_int64);
append_composite_type_field (t, "v2_float",
- init_vector_type (builtin_type_float, 2));
+ init_vector_type (builtin_type (gdbarch)
+ ->builtin_float, 2));
append_composite_type_field (t, "v2_int32",
init_vector_type (builtin_type_int32, 2));
append_composite_type_field (t, "v4_int16",
&& regnum >= tdep->ppc_fp0_regnum
&& regnum < tdep->ppc_fp0_regnum + ppc_num_fprs
&& TYPE_CODE (type) == TYPE_CODE_FLT
- && TYPE_LENGTH (type) != TYPE_LENGTH (builtin_type_double));
+ && TYPE_LENGTH (type)
+ != TYPE_LENGTH (builtin_type (gdbarch)->builtin_double));
}
static void
struct type *type,
gdb_byte *to)
{
+ struct gdbarch *gdbarch = get_frame_arch (frame);
gdb_byte from[MAX_REGISTER_SIZE];
gdb_assert (TYPE_CODE (type) == TYPE_CODE_FLT);
get_frame_register (frame, regnum, from);
- convert_typed_floating (from, builtin_type_double, to, type);
+ convert_typed_floating (from, builtin_type (gdbarch)->builtin_double,
+ to, type);
}
static void
struct type *type,
const gdb_byte *from)
{
+ struct gdbarch *gdbarch = get_frame_arch (frame);
gdb_byte to[MAX_REGISTER_SIZE];
gdb_assert (TYPE_CODE (type) == TYPE_CODE_FLT);
- convert_typed_floating (from, type, to, builtin_type_double);
+ convert_typed_floating (from, type,
+ to, builtin_type (gdbarch)->builtin_double);
put_frame_register (frame, regnum, to);
}