http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48603
--- Comment #9 from rguenther at suse dot de <rguenther at suse dot de> 2011-04-19 08:52:11 UTC --- On Tue, 19 Apr 2011, jan.kratochvil at redhat dot com wrote: > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48603 > > --- Comment #8 from Jan Kratochvil <jan.kratochvil at redhat dot com> > 2011-04-19 08:30:52 UTC --- > It is a fully redundant information. It just makes the processing in > consumers > easier. There was a first GDB patch which did not need > DW_AT{,_MIPS}_linkage_name but it had some issues: > http://sourceware.org/ml/gdb-patches/2011-02/msg00029.html > > So either there should be some DW_AT_GNU_assumed_name having no namespaces or > GDB should be reworked some way to deal without it. It the current way is a > real LTO blocker for GCC. I'm considering two ways of fixing the LTO issue, either make dwarf2out.c not ICE (thus, simply avoid this extension - possibly with, but hopefully without debugging impact). Or compute a DECL_ASSEMBLER_NAME for such typedefs during compile-time, which has some memory usage impact. What I'm trying to figure out is if there is an even better solution completely avoiding the need for this kind of late language-specific type mangling. Richard.
