http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52401
--- Comment #3 from Bin Tian <tianbin at cernet dot edu.cn> 2012-02-28 02:59:03 UTC --- (In reply to comment #1) > Confirmed. There are other cases with linker-scripts / linker options that > are not handled well. > > This one is more a linker bug though. The input a.o should be translated > to the new replacement input that went though LTO processing, instead it > seems to be appended. > > Please file a binutils bug. http://sourceware.org/bugzilla/show_bug.cgi?id=13779