https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119896
--- Comment #5 from James K. Lowden <jklowden at gcc dot gnu.org> --- > Good. I just saw that Robert's account is already set up. Yes. I've just completed my world tour with cppcheck, always an adventure. And I applied for my own cfarm account, because your bt was helpful and confounding. The trace shows a lookup of the program name. By time it's processing "based-p", it's on line 26 of the input. The program name has long been established, and probably already used 25 times. For it to fail there suggests something exceedingly odd. So, I'd like to have a look myself. ;-) The symbol table is a memory-mapped file. As it grows, the file is extended and new maps added. (No old map is removed.) It seems unlikely to me that's an issue here, because I doubt we're at a point where the map is being extended already. But something is confusing the lookup.
