https://sourceware.org/bugzilla/show_bug.cgi?id=32860
--- Comment #2 from Michael Matz <matz at suse dot de> --- I really think this ultimately boils down to the compiler not providing a correct symbol table in pr31482a.c, i.e. one not refering to 'abort' in any way. The only way the linker could do something about this incomplete-symtab situation is basically to always feed all IR objects from static archives to the LTO phase, even if they are seemingly unused, and hope for the compiler to throw them away if they turn out to be really unused. But that's a major deviation from the normal semantics of static archives :-/ -- You are receiving this mail because: You are on the CC list for the bug.
