https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124107
--- Comment #3 from Eric Botcazou <ebotcazou at gcc dot gnu.org> --- > The bug still exists without Unchecked_Access, the RM is clear that > finalization of the subpool must remove it from the pool. Unchecked_Access > is just when this bug can cause the major issues. There's unfortunately no > way to do what I'm doing in the linked PR without Unchecked_Access, but it > should all be safe according to the RM. You cannot invoke the RM once you circumvent it by means of 'Unchecked_Access though; in particular, your testcase explicitly creates a dangling pointer and therefore all bets are off.
