http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48603
--- Comment #4 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-04-14 10:06:53 UTC --- (In reply to comment #3) > This extension is not about DW_AT_linkage_name vs. DW_AT_MIPS_linkage_name. > It is about DW_AT_{,MIPS_}linkage_name for DW_TAG_structure_type which did not > have any meaning before. > > The extension is used by GDB DWARF reader, more info at: > http://sourceware.org/ml/gdb-patches/2011-03/msg00916.html > = > It is just an earlier indication of how will be the DW_TAG_typedef DIE named > for an anonymous struct/class/union. > > I do not know if it has anything to do with LTO inside GCC. FYI this > DW_AT_{,MIPS_}linkage_name is fully qualified = already contains namespaces. Yes, it's exactly the mangling which we can't perform at LTO time and thus to preserve this extension we'd have to compute DECL_ASSEMBLER_NAME for all type-decls as well, not just for entities that will have symbols.