https://gcc.gnu.org/bugzilla/show_bug.cgi?id=113675 --- Comment #3 from Andrew Pinski <pinskia at gcc dot gnu.org> --- Actually the issue is not %p but rather the constraint. Anyways it is still a dup.