https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71403
--- Comment #6 from Jeffrey A. Law <law at redhat dot com> --- But what doesn't make sense here Jan is jump threading, has had the ability to peel loop iterations forever, so if we're storing that data prior to threading and expecting it to persist accurately across jump threading, then it's fundamentally broken. However, for this specific case, I think the key here is the early threading allows DOM to essentially combine two loops with a new (higher) iteration count. So any saved data WRT loop iterations is totally bogus. What I don't see is any reasonable API for invaliding the saved iteration data. It appears to be attached to the loop with no way to update it.
