[Bug middle-end/63184] [5/6/7 Regression] Fails to simplify comparison

[law at redhat dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63184

Jeffrey A. Law <law at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |law at redhat dot com

--- Comment #8 from Jeffrey A. Law <law at redhat dot com> ---
So while we don't optimize this on the tree level, the addressing lowering that
was introduced for gcc-6 sets things up so the RTL optimizers can detect the
equivalences and remove the tests.

So ISTM we can do two things here.  Keep this as a regression since it'd be
nice for the SSA/gimple optimizers to catch this.  Or mark it was fixed for
gcc-6 and gcc-7 since the RTL optimizers are able to clean this stuff up.