https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97090
--- Comment #5 from David Malcolm <dmalcolm at gcc dot gnu.org> --- PR 97621 reports it as starting on powerpc64*-linux-gnu with r11-4434, which was a fix for non-determinism in -fanalyzer, so perhaps this is a flaky test that the non-determinism fixes have made fail more reliably (which comment #1 suggests as well).
