http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58106
--- Comment #3 from Martin Jambor <jamborm at gcc dot gnu.org> --- Created attachment 30708 --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=30708&action=edit Patch The problem is that the rdesc chain creation mechanism cannot handle the case where indirect inlining creates its master_clone and we are thus "normally" cloning an inline tree. Fixed with this (fully tested) patch. I forgot to add the testcase, will do that before submitting. I would also like to measure the chain lengths on at least a few benchmarks first too.
