http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53602
--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> 2012-06-08 07:57:23 UTC --- Or perhaps just handle this particular case for now, when there has been jump threading right after the sp adjustment sequence that cancelled each other. Thus, if we reach end of bb after the insn without seeing a CALL or or throwing insn and the bb has a single fallthru successor only, keep looking in the successor bb.
