https://gcc.gnu.org/bugzilla/show_bug.cgi?id=121933
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |muecker at gwdg dot de
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #3)
> I am suspecting r16-3747-gafa74d37e8170d exposed this.
Yes looking like it.
Before we had:
start:
after we get:
start: [non-local];
And then I think the unnesting pass generates a second non-local lable rather
than keeping and reusing the same one:
<D.4612>: [non-local]
start: [non-local]
That is just a missed optimization I think.
Let me see if I can come up with a testcase for 15 because the real issue is in
speculation pass.
