On 04/10/11 19:57, Alan Hourihane wrote: > Hi Vincent, > > I'm bumping into this with 4.5.2, any clues ? > > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47908 > I think the gcc_unreachable() might actually be bogus and should be removed given the need for targetm.sched.first_cycle_multipass_dfa_lookahead() calls this as well. Alan.