This is OK with me. I suspect gcc's potential inlining and/or our target-specific memcpy()'s might offset any performance lost anyway. Reviewed-by: DJ Delorie <dj@redhat.com>