https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106057
--- Comment #2 from Jan Hubicka <hubicka at gcc dot gnu.org> --- The second testcase (with longjmp) invalid since longjmp can clobber automatic variable and making the variable static breaks the testcase since we believe htat longjmp reads global memory state (it doesn't). The first testcase is also kind-of bordeline since it is question whether const/pure can throw. Performance wise we probably do not need to care about longjmp but with cxa_throw we may want to get side effects modeled right...
