https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67755
--- Comment #4 from Jeffrey A. Law <law at redhat dot com> --- It looks like a logic error in the case where we need to make sure there's enough "count" going to the final block of jump thread path. I've got a patch, but I'm still working through the code and the final counts/probabilities to make sure they all make sense.
