https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66301
Martin Jambor <jamborm at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |NEW Last reconfirmed| |2015-05-28 Ever confirmed|0 |1 --- Comment #2 from Martin Jambor <jamborm at gcc dot gnu.org> --- Well, the problem is that in eliminate_dom_walker::before_dom_children in tree-ssa-pre.c we have: if (dump_enabled_p ()) dump_possible_polymorphic_call_targets (dump_file, ...) but dump_enabled_p can return true when dump_file is NULL if alt_dump_file is not NULL. I suppose that dump_enabled_p was introduced as a part of the semi-overhaul of dumping interface we got in 2012 but that never quite caught on, and that it simply can't be mixed with old-style dumping which just tests dump_file. The obvious fix would be to change the condition to "if (dump_file)". Or we might convert the whole thing to new style dumping, but I admit I have never quite understood it.