stm32: Support compiling with object representation D.

With this and previous patches the stm32 port can now be compiled using
object representation D (nan boxing).  Note that native code and frozen mpy
files with float constants are currently not supported with this object
representation.

1 files changed, 2 insertions, 2 deletions

diff --git a/ports/stm32/accel.c b/ports/stm32/accel.c
index 49674eb2d..0d7708c09 100644
--- a/ports/stm32/accel.c
+++ b/ports/stm32/accel.c
@@ -122,7 +122,7 @@ STATIC mp_obj_t pyb_accel_make_new(const mp_obj_type_t *type, size_t n_args, siz
     pyb_accel_obj.base.type = &pyb_accel_type;
     accel_start();
 
-    return &pyb_accel_obj;
+    return MP_OBJ_FROM_PTR(&pyb_accel_obj);
 }
 
 STATIC mp_obj_t read_axis(int axis) {
@@ -166,7 +166,7 @@ STATIC MP_DEFINE_CONST_FUN_OBJ_1(pyb_accel_tilt_obj, pyb_accel_tilt);
 /// \method filtered_xyz()
 /// Get a 3-tuple of filtered x, y and z values.
 STATIC mp_obj_t pyb_accel_filtered_xyz(mp_obj_t self_in) {
-    pyb_accel_obj_t *self = self_in;
+    pyb_accel_obj_t *self = MP_OBJ_TO_PTR(self_in);
 
     memmove(self->buf, self->buf + NUM_AXIS, NUM_AXIS * (FILT_DEPTH - 1) * sizeof(int16_t));