https://gcc.gnu.org/bugzilla/show_bug.cgi?id=117919
Filip Kastl <pheeck at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Assignee|unassigned at gcc dot gnu.org |pheeck at gcc dot gnu.org --- Comment #2 from Filip Kastl <pheeck at gcc dot gnu.org> --- Thanks for Cc-ing me. Apparently, SCCOPY copy-propagates on this statement _25 = &MEM[(struct path *)&writer]._M_pathname; and that somehow makes # .MEM_37 = PHI <.MEM_52(11), .MEM_53(12)> into _37 = PHI <> And then SCCOPY crashes because it expects PHIs to have arguments. Not sure how this happens. Maybe the copy propagation shouldn't be done and doing it corrupts GCC's memory? Will look into this some more.