https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71316
Jeffrey A. Law <law at redhat dot com> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|ASSIGNED |RESOLVED
Resolution|--- |FIXED
--- Comment #3 from Jeffrey A. Law <law at redhat dot com> ---
And the loose end turns out to be a non-issue as well.
Given that the first threading pass was not finding any backward/FSM threads,
but VRP1 before my change did find backward/FSM threads, I wanted to make sure
nothing was amiss.
The backward/FSM stuff that was previously being found by VRP1 were a direct
result of the context sensitive constant propagation. The interaction between
context sensitive cprop and the backwards threader is a known issue and is
being looked at independently of this BZ.
--
So the net result here is that with the recent threading changes we actually
get the code we want, just a pass earlier in the pipeline -- which is a
positive in my book. What I think what we want to do is to verify that we pick
up the two old style jump threads in VRP1.
